i
  
 
SELECTION FOR REMEDIAL INTERVENTION:   
THE VALIDATION OF THE ACADEMIC PROFICIENCY 
ASSESSMENT BATTERY AS A PREDICTOR OF 
ACADEMIC ACHIEVEMENT 
 
Shafeeka Yusuf Dockrat 
  
 
 
 
 
 
A research report submitted to the Faculty of 
Humanities, University of the Witwatersrand, 
Johannesburg, in fulfilment of the requirements for 
the degree of Master of Arts (Psychology) 
 
Johannesburg, 2005 
 
ii
 ABSTRACT 
 
The Academic Proficiency Assessment battery evaluates language and 
study skills. This study focused on the internal consistency reliability and 
predictive validity of this battery for Information Technology students.   
 
In terms of reliability, the Time management, Note-taking skills, and 
Debilitating stress scale on the achievement anxiety questionnaire were  
found to be internally consistent.   However, the items in the English 
proficiency, Reading comprehension, Memorisation skills, concentration 
skills and motivating stress scale on the achievement anxiety 
questionnaire require modification or replacement. Intercorrelations across 
questionnaires necessitate further streamlining.  
 
In terms of predictive validity, a significant negative relationship was found 
between Note-taking Skills and academic performance (R2adj = 8,3%). 
Matric results remain the best predictor accounting for 11% of the variance 
in CGPA.  Cumulatively, Matric results and Note-taking skills accounted for 
13,34% of the variance. None of the biographical variables significantly 
predicted CGPA.   
 
Despite the apparent lack of relationship between individual predictors and 
the criterion, a discriminant function analysis indicated that all the 
academic proficiencies, with the exception of English proficiency, correctly 
predicted pass or failure in 72% of the cases. The lack of relationship 
between English proficiency and pass/failure can be accounted for in 
terms of the type of courses studied as part of an Information Technology 
programme.  
 
Overall, the results would suggest that Matric results remain the best 
predictor of academic performance in Information Technology courses, but 
at a very low level.  Given the lack of reliability in the majority of the 
 
iii
 subtests of the Academic Proficiency Battery, the use of the APA battery 
for selection for remedial intervention for Information Technology students 
is not yet justified.  
 
 
 
iv 
DECLARATION 
 
I declare that this dissertation is my own unaided work.  It is submitted for 
the degree of Master of Arts in the University of the Witwatersrand, 
Johannesburg.  It has not been submitted before for any other degree or 
examination in any other university. 
 
 
 
____________________________________________ 
Shafeeka Yusuf Dockrat 
 
 
11 February 2005 
 
v 
To Alisha 
 
With apologies for my time taken away from her  
to complete this degree 
 
 
 
vi 
ACKNOWLEDGEMENTS 
 
? My supervisor, Charles Potter 
? Peter Fridjhon and Mike Greyling for statistical support 
? The Directorate of Student Development and Support at the Tshwane 
University of Technology for making the data available 
? My family (Ibraheem, Rashida and Fahmeeda) for their support and 
assistance 
 
vii
 TABLE OF CONTENTS 
SELECTION FOR REMEDIAL INTERVENTION: ...................................... i 
THE VALIDATION OF THE ACADEMIC PROFICIENCY ASSESSMENT 
BATTERY AS A PREDICTOR OF ACADEMIC ACHIEVEMENT .............. i 
Shafeeka Yusuf Dockrat ........................................................................... i 
A research report submitted to the Faculty of Humanities, University 
of the Witwatersrand, Johannesburg, in fulfilment of the 
requirements for the degree of Master of Arts (Psychology)................ i 
Johannesburg, 2005 ................................................................................. i 
ABSTRACT ............................................................................................... ii 
Chapter 1 .................................................................................................. 1 
INTRODUCTION ....................................................................................... 1 
1.1. Need for this research project 1 
1.2 Problem statement 2 
1.3 Research Questions 2 
1.4 Objective 3 
1.5 Brief overview of dissertation 3 
Chapter 2 .................................................................................................. 4 
THE PREDICTION OF ACADEMIC PERFORMANCE.............................. 4 
2.1 Introduction 4 
2.2 Choice of performance criteria 4 
2.3 Factors influencing academic performance 6 
2.4 Intellective factors as predictors of academic performance 8 
2.4.2 History of intelligence testing 10 
 
viii
 2.4.3 Intelligence as a predictor of academic performance 12 
2.4.3.1 Classification of studies 12 
2.4.3.2 Findings at the tertiary level on intelligence as a predictor 
of academic performance 13 
? ? Global measures of ability and performance 13 
? ? Specific ability dimensions used to predict GPA 14 
? ? Several dimensions of intelligence used to predict 
grades in specific courses or course areas 15 
? ? Global ability measure is used to predict grades in 
specific courses 17 
2.5 Personality factors as predictors of academic performance 17 
2.5.1 Classification of personality studies 17 
2.4.5.1 Motivational states as a predictor of academic 
performance 18 
? ? Interest 18 
? ? Achievement motivation 19 
? ? Aptitude 21 
2.5.1.1 Personality style 23 
? ? Personality traits of underachievers and overachievers23 
? ? Big Five factors of personality 24 
2.5.1.2 Self-image 25 
2.5.1.3 Adjustment 26 
2.6 Sociological determinants of academic performance 26 
2.6.1 Demographic and ecological variables 27 
2.6.1.1 Socio-economic status (SES) 27 
2.6.1.2 Gender 28 
2.6.1.3 Miscellaneous ecological and demographic characteristics
  30 
? ? Age 30 
? ? Regional and rural-urban variation 31 
2.6.2 Specific role relationships 32 
2.6.2.1 Educators 33 
 
ix 
2.6.2.2 Parents 35 
2.6.2.3 Peers 36 
2.7 Conclusion 37 
Chapter 3 ................................................................................................ 38 
THE ASSESSMENT OF ACADEMIC PROFICIENCIES ......................... 38 
3.1  Introduction 38 
3.2 Academic proficiencies 42 
3.2.1 Language 42 
3.2.1.1 English Proficiency 43 
3.2.1.2 Reading Skills 45 
3.2.2 Study Skills 46 
3.2.2.1  What do Study Skills entail? 46 
3.2.2.2 Study Skills as a predictor of academic performance 47 
3.2.2.3  Specific study skills as predictors of achievement 47 
? ? Time Management 47 
? Note-Taking 49 
? Memorisation skills 50 
? ? Concentration Skills 51 
? ? Examination stress 51 
3.3 The assessment of academic proficiencies 52 
3.3.1 Learning and Study Strategies Inventory (LASSI) 52 
3.3.1.1 Description of the Inventory 52 
3.3.1.2 Learning and Study Strategies Inventory as a predictor of 
academic performance 53 
3.3.2 Survey of Study Habits and Attitudes (SSHA) 53 
3.3.3.1 Description of the Questionnaire 54 
3.3.3.2 Biggs?s Study Process Questionnaire as a predictor of 
academic performance 54 
3.3.4 Approaches and Study Skills Inventory for Students 55 
3.3.4.1 Description of the Inventory 55 
 
x 
3.3.4.2 Predicting academic performance 55 
3.3.5 van Schoor?s Study Process Questionnaire (SPQ) 55 
3.3.6 Study Skills Questionnaire (SSQUES) 56 
3.3.6.1 Description of the Questionnaire 56 
3.3.6.2 SSQUES as a predictor of academic performance 56 
3.4 Assessment in the South African context 56 
3.4.1 The utility of assessment  measures in contemporary South 
Africa 56 
3.4.2 Escalating diversity and accelerated change 57 
3.5 Conclusion 57 
Chapter 4 ................................................................................................ 59 
METHODOLOGY .................................................................................... 59 
4.1 Introduction 59 
4.2 Research questions 59 
4.2.1  Are all the questionnaire of the Academic Proficiency 59 
Assessment (APA) battery reliable? 59 
4.2.2 If any of the questionnaires in the APA are not reliable, what 
changes can be implemented to enhance its psychometric properties?
  60 
4.2.4 Are there additional variables that predict the academic 
performance of Information Technology students? 61 
4.2.5  An integrated regression model 61 
4.3 Sample 62 
4.3.1  Sampling method 62 
4.3.2  Description of the sample 62 
4.4 Instruments 67 
4.4.1 Academic Proficiency Assessment battery 67 
4.4.1.1 English proficiency 67 
4.4.1.2 Reading skills 67 
4.4.1.3 Time Management 67 
4.4.1.4 Note-taking skills 68 
 
xi 
4.4.1.5 Memorisation skills 68 
4.4.1.6 Concentration 68 
4.4.1.7 Examination stress 68 
4.4.2 Biographical questionnaire 69 
4.4.3 Integrated Tertiary System (ITS) 69 
4.4.3.1 Matric results 69 
4.4.3.1 Course results 70 
4.5 Design and Procedure 71 
4.5.1 Design 71 
4.5.1.1 Non-experimental 71 
4.5.1.2 Correlational 72 
4.5.1.3 Descriptive 73 
4.5.2 Procedure 73 
4.6 Statistical analyses 73 
4.6.1 Statistical methods for research question 1 73 
4.6.1.1 Cronbach?s coefficient alpha (r?) 74 
4.6.1.2 Kuder-Richardson formula 20 (KR-20) 75 
4.6.1.3 Interpretation of reliability coefficients 75 
4.6.2 Statistical methods for research question 2 76 
4.6.2.1 Cronbach?s coefficient alpha 76 
4.6.2.2 Item difficulty 76 
4.6.3  Statistical methods for research question 3 77 
4.6.3.1 Multiple regression 77 
? ? Correlation 78 
? ? Multiple regression equation 79 
? ? Assumptions underlying multiple regression 79 
? ? Model and variable selection 80 
? ? Goodness of fit 81 
4.6.3.2 Discriminant function analysis 81 
4.6.4  Statistical methods for research question 4 82 
4.6.5  An integrated regression model 82 
4.7 Conclusion 83 
 
xii 
Research question 84 
Are all the questionnaire of the Academic Proficiency Assessment 
(APA) battery reliable? 84 
Research question 85 
Research question 86 
An additional regression model 86 
Chapter 5 ................................................................................................ 87 
RESULTS ................................................................................................ 87 
5.1 Introduction 87 
5.2  Results for research question 1 87 
5.3  Results for research question 2 88 
5.3.1  English proficiency 89 
5.3.2 Reading comprehension 92 
5.3.3 Memorisation skills 92 
5.3.4 Concentration 93 
5.3.5 Achievement anxiety  (motivating stress) 94 
5.4 Results for research question 3 95 
5.4.1  Multiple regression analysis 96 
? Multicollinearity 97 
? Multiple regression analysis 100 
5.4.2 Discriminant function analysis 102 
5.5 Research question 4 107 
? Multicollinearity 108 
? Multiple regression 108 
5.6 An integrated regression model 110 
? Multicollinearity 110 
? Multiple regression analysis 110 
5.7 Conclusion 112 
Chapter 6 .............................................................................................. 114 
 
xiii 
CONCLUSION....................................................................................... 114 
6.1       Introduction 114 
6.2 Synopsis of the study 114 
6.2.1  English proficiency 115 
6.2.2  Reading skills 115 
6.2.3  Time Management 116 
6.2.4  Note-taking skills 116 
6.2.5  Memorisation skills 116 
6.2.6  Concentration skills 117 
6.2.7  Examination stress 117 
6.2.8 Additional predictors 118 
6.3 Summary of findings 118 
6.3.1 Are all the questionnaires of the Academic Proficiency 118 
Assessment (APA) battery reliable? 118 
6.3.2 If the APA is not reliable, what changes can be 118 
implemented to enhance its psychometric properties? 118 
6.3.3 What is the predictive validity of the questionnaires in the 119 
APA battery? 119 
6.3.4 Are there any additional variables that could enhance the 120 
predictive validity of the APA battery for Information Technology 
students? 120 
6.4 Strengths of the study 120 
6.4.1 Multiple variables under investigation 120 
6.4.2 Ecological validity 122 
6.5 Limitations of the study 122 
6.5.1 Threats to internal validity 122 
6.5.1.1 Maturation 122 
6.5.1.2 Instrument reactivity 123 
6.5.1.3 History 123 
6.4.2Threats to external validity 123 
6.4.2.1 Population validity 124 
6.4.3 Sample size 124 
 
xiv 
6.5 Recommendations 124 
6.5.1 ?Back to the drawing board? 124 
6.5.2 Intervention programmes 125 
6.5.3   Selection for remedial intervention 125 
6.6 Conclusion 126 
Chapter 7 .............................................................................................. 128 
REFERENCE LIST ................................................................................ 128 
Appendix A........................................................................................... 140 
FIGURES ............................................................................................... 140 
Appendix B........................................................................................... 141 
TABLES................................................................................................. 141 
 
 
xv 
LIST OF FIGURES 
 
 
Figure 1:  Factors related to academic performance 7 
Figure 2:  Factors related to academic performance highlighting 
                factors assessed by the APA battery 
41 
Figure 3:  Study Process Questionnaire (SPQ) 140 
Figure 4:  Gender 64 
Figure 5:  Predominant language of communication 65 
Figure 6:  Age 66 
Figure 7:  Factors investigated in the present study highlighted 121 
 
xvi 
LIST OF TABLES  
 
Table 1: Types of studies relating ability and academic 
               performance   
13 
Table 2:  Results of multiple regression analysis in Kriel?s (2000) 
               study 
16 
Table 3:  McCrae and Costa?s Five-Factor Model of Personality 141 
Table 4:  LASSI scales 142 
Table 5:  Quantification of matriculation symbols using the 
               Swedish formula 
70 
Table 6: Research design table 84 
Table 7: Internal consistency of questionnaires in the APA 
               battery 
88 
Table 8:  Item analysis of English Proficiency questionnaire 90 
Table 9:  Item analysis of Reading Comprehension  
               questionnaire 
92 
Table 10: Internal consistency of Memorisation Skills  
                 questionnaire 
93 
Table 11:  Internal consistency of Concentration Skills  
                 questionnaire 
94 
Table 12:  Internal consistency of Motivating Stress 
                  questionnaire 
95 
Table 13:  Correlation matrix of explanatory variables  99 
Table 14:  R2 for research question 3 100 
Table 15:  Parameter estimates for research question 3 100 
Table 16:  Summary of forward selection for research question 3 101 
Table 17:  Analysis of variance for research question 3 101 
Table 18:  Linear discriminant function for research question 3 104 
Table 19: Classification table for research question 3 104 
Table 20:  Average R2 for discriminant function analysis 105 
Table 21:  Univariate test statistics for research question 3 105 
 
xvii 
Table 22:  Multivariate and exact F statistics for research 
                 question 3 
106 
Table 23:  R2 for research question 4 108 
Table 24:  Parameter estimates for research question 4 109 
Table 25:  Summary of forward selection for research question 4 109 
Table 26: Analysis of variance for research question 4 109 
Table 27: R2 for the integrated regression model 111 
Table 28:  Parameter estimates for the integrated regression  
                 model 
111 
Table 29: Summary of forward selection for the integrated  
                 regression model 
111 
Table 30:  Analysis of variance for the integrated regression  
         model 
112 
 
 
 
1 
Chapter 1 
INTRODUCTION 
 
 
1.1. Need for this research project 
 
In accordance with the White Paper on Higher Education (1997), tertiary 
institutions have a responsibility towards registered students to provide 
them with academic support programmes.  These programmes have to be 
directed towards improving academic performance. 
 
The need for academic support programmes is primarily attributed to the 
lack of academic, language and life skills in most learners from historically 
disadvantaged background; the high failure rate amongst first year 
students; and language policy.  Consequently, the aim of academic 
support is to facilitate holistic development of learners in order to succeed 
academically, and thereby increase throughput figures (Celliers, 2000). 
 
The Directorate for Student Development and Support at the Tshwane 
University of Technology (TUT) attempts to improve student?s academic 
performance by enhancing language and study skills.  Language 
development is accomplished through an English Proficiency course, and 
a Reading course that aims to increase reading speed and 
comprehension. Study skills are improved through a comprehensive study 
methods course that integrates the following modules: time management, 
note-taking skills, memory skills, examination stress, and concentration. 
 
Eno, McLaughlin, Sheldon & Brozovsky (1999) assert that the success of 
such academic intervention programmes depends largely on accurately  
 
2 
identifying students requiring such services.  An assessment that identifies 
students who lack the necessary academic competencies and language 
proficiencies may serve as an ?early warning system? of students 
potentially at risk for academic underachievement. In order to identify such 
students, a battery of tests, collectively referred to as the Academic 
Proficiency Assessment battery (APA battery) is administered.  
 
1.2  Problem statement 
 
The Academic Proficiency Assessment used at Tshwane University of 
Technology is not a psychometric instrument. It is a set of questionnaires 
that have not been tested for reliability and validity.   These questionnaires 
were initially selected simply on the basis of their face validity.  What 
commenced as a seldom administered set of questionnaires to identify 
possible problem areas in terms of academic support proficiencies evolved 
into a formal assessment administered on a large scale.  Considering the 
resources deployed in administering this battery of tests, validity and 
reliability studies become inevitable. 
 
Furthermore, the Academic Proficiency Assessment is being utilised with 
the aim of identifying a student?s proficiency level in terms of language and 
study skills.  The present study also aims to determine if these skills are 
related to academic achievement, as the aim of administering the 
instrument is to identify developmental areas with regard to these skills on 
the assumption that if these skills are enhanced, academic performance 
will also improve too. 
 
1.3  Research Questions 
 
1. Are all the questionnaires of the Academic Proficiency Assessment 
(APA) battery reliable?  
 
3 
2. If any of the questionnaires in the APA battery are not reliable, what 
changes can be implemented to enhance their psychometric 
properties?  
3. What is the predictive validity of the questionnaires in the APA battery? 
4. Are there any additional variables that enhance the predictive validity 
for Information Technology students? 
 
1.4  Objective 
 
The objectives of this study are twofold.  Firstly, to establish the 
psychometric properties of a battery of tests collectively referred to as the 
Academic Proficiency Assessment (APA) battery, and secondly to 
determine what variables predict academic achievement for Information 
Technology students. 
 
1.5  Brief overview of dissertation 
 
The dissertation proceeds in Chapter Two with a comprehensive review of 
existing literature on predictors of academic performance.  Chapter 3 is 
devoted to the Academic Proficiency Assessment battery, particularly a 
discussion of a proficiencies evaluated by the instrument, and current 
issues surrounding the assessment thereof. Chapter 4 provides an 
overview of the methodology of the present research project, specifically 
the sampling, research design, and statistical analyses.  The details and 
results of the data analyses are presented in Chapter 5.  Chapter 6 
concludes the research report, providing a summary of the findings as well 
as the recommendations. 
 
 
 
 
 
 
4 
 
Chapter 2 
THE PREDICTION OF ACADEMIC PERFORMANCE 
 
 
2.1 Introduction 
 
This chapter examines the literature related to the prediction of academic 
performance.  The chapter commences with a discussion of the value 
judgements in the choice of performance criteria, followed by the problems 
involved in the prediction of academic performance.  The remainder of the 
chapter is devoted to an in-depth review of the factors related to academic 
performance. 
 
2.2 Choice of performance criteria 
 
Academic achievement is a complex and multi-faceted construct, 
encompassing numerous aspects.  Academic performance at higher 
education compared to school achievement is more complex.  Monteith 
(1988) defines school achievement as the results of a final examination.  
At school level, a student either passes or fails a grade year.  At tertiary 
level, a learner accumulates credits for individual subjects, even though 
the learner may not necessarily be promoted to the following academic 
year.  Due to this more complex promotion system, academic achievement 
is not easily measurable at higher education.  Further complications 
include diverse study fields, differences in requirements and standards in 
varying fields of study, differences prevalent in different faculties 
(Swanepoel, 2002). 
 
According to Fourie (1992), academic performance is not easily defined, 
and cannot be simply expressed by a single mark or symbol.  There are 
 
5 
no valid norms to evaluate academic achievement.  Academic 
performance has been defined in a number of ways, including success or 
failure in a subject, over- or under-achievement, poor, or no progress at 
academic level, graduating, completion of the qualification within the 
minimum period prescribed by the institution, intellectual curiosity, long-
 term career success of good and poor students, and the quality of verbal 
expression of graduate students (Rademeyer & Scheepers, 1998; Fourie, 
1992; Lavin, 1965).   
 
In the majority of studies of academic achievement, examination results 
are the only index of performance.  Although examination results do not 
completely represent academic progress and adaptation, it is regarded as 
the most significant aspect of these processes.  Examination papers in 
different subjects cover the content of a semester or a year.  These are 
meticulously compiled and marked.  Swanepoel (2002) contends that it is 
therefore fair to accept that examination results provide a valid and reliable 
indication of academic achievement. 
 
Kruger (1972) suggested the use of examination results as performance 
criteria according to the following methods: Firstly, the information of pass 
and fail may be utilised as dichotomous criterion for academic 
achievement.  The limitation of using it in this way is that this criterion does 
not include all the information about academic achievement.  Secondly, 
the information about the number of subjects passed may also be used as 
a guideline, but when the requirements of different fields of study differ, 
the data is not always comparable.  Thirdly, the marks obtained in different 
subjects may used as an average or as a total ? again, the different 
subjects may jeopardise comparability.  The practical problem when 
utilising these methods is to manipulate subject results in such a manner 
that they are comparable. 
 
 
6 
As a result of the variety of study fields, differences in evaluation 
measures and other factors, it is not always possible to reach consensus 
on the choice of performance criteria.  Hence, it is necessary that specific 
points of departure or assumptions be formulated as a basis for the 
formulation of a criterion for tertiary achievement.  Swanepoel (2002) sets 
forth the following assumptions, precedented by Smith (1979) as a point of 
departure when formulating criteria for tertiary selection:  Firstly, it is 
assumed that a student?s final mark in the examination, is an objective 
measure of academic performance.  The second assumption is that the 
mark obtained in a specific subject is equivalent to the same mark 
obtained in another subject.  Factors such as evaluation standards or 
differences in complexity between different subjects are not taken into 
account but condoned, as the pass mark is 50% for all subjects.  Lastly, it 
is assumed that the degree of difficulty of different subjects on the same 
level is the same.   
 
2.3  Factors influencing academic performance 
 
Research on the factors influencing academic performance can be broadly 
group into three categories, namely, intellective factors, non-intellective 
personality characteristics, and sociological factors.  Figure 1 illustrates 
the specific factors that will be highlighted in each category. 
 
7
    _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
  
Fi
 gu
 re
  
1.
  
 
Fa
 ct
 o
 rs
  
re
 la
 te
 d 
to
  
a
 ca
 de
 m
 ic 
pe
 rfo
 rm
 a
 n
 ce
 .
  
8  
 
2.4 Intellective factors as predictors of academic performance 
 
This section focuses on measures of intellectual ability as predictors of 
academic performance.  A summary of prevailing theories of intelligence, 
and the history of intelligence testing, contextualise the research on 
intelligence as a predictor of academic performance. 
 
2.4.1 Theories of intelligence 
 
Before even considering the relationship between intellectual ability and 
academic performance, the meaning of intelligence needs to first be 
understood.  This section takes a brief look at the dominant theories of 
intelligence, specifically those of Galton and Cattell, Jenson and Vernon, 
Spearman, Thurstone, R. Cattell, Piaget, Guilford, Luria, Gardner and 
Sternberg. 
 
Galton and Cattell saw sensory abilities as the basis of intelligence.  They 
proposed that all information regarding the outside world passes through 
one?s senses, leading them to believe that the more perceptive the 
senses, the larger the field upon which a person?s judgement can act 
(Gregory, 1996). 
 
Much later, researchers like Jensen and Vernon reported substantial 
correlations between reaction-time type measures and traditional 
intelligence tests (Gregory, 1996). 
 
Spearman postulated that intelligence consisted two main types of factors, 
namely a general factor (g) and several specific factors (s1, s2, s3, etc).  
Spearman utilised three principles of cognition to explain individual 
differences in the g-factor.  These three factors were the apprehension of 
9  
experience, the eduction of relations, and the eduction of correlations 
(Gregory, 1996). 
 
According to Thurstone, intelligence could be explained more clearly by 
several group factors instead of one general factor.  He derived the 
primary mental abilities:  verbal comprehension, word fluency, number, 
space, associative memory, perceptual speed and inductive reasoning 
(Gregory, 1996). 
 
Cattell proposed intelligence as consisting of two major components, fluid 
and crystallised intelligence.  Fluid intelligence is a largely non-verbal and 
culture-reduced form of intelligence which is related to a person?s inherent 
capacity to learn and solve problems, while crystallised intelligence is 
highly cultural, and used for tasks requiring learned or habitual responses.  
Crystallised intelligence is therefore prevalent when fluid intelligence is 
applied in cultural settings where learnt behaviour is required (Gregory, 
1996). 
 
Piaget suggested that organised patterns of behaviour or mental 
structures called schemas evolve to greater maturity through a process of 
equilibration, where schemas are applied to objects, persons or events 
through the process of assimilation.  If a schema is assimilated, a state of 
equilibrium arises.  If not, the inequilibrium or tension requires the 
adjusting of the schema, until a state of equilibrium can be reached.  The 
adjustment process is referred to as accommodation (Gregory, 1996). 
 
Guilford developed the Structure of Intellect Model.  In terms of this model, 
intelligence is three dimensional, the three dimensions being operations, 
contents and products that produce 5X5X6 or 150 different kinds of 
intellect (Gregory, 1996). 
 
10  
Luria?s Information Processing Theory focuses on two different kinds of 
processing.  Simultaneous processing is typified by the execution of 
several different mental operations at the same time.  Forms of thinking 
and perception requiring spatial analysis require this type of processing.  
Successive processing organises information in a linear series, and is 
required where a proper sequence of activities must be followed.  Most 
forms of information processing are dependent on interplay of both types 
of processing (Gregory, 1996). 
 
Gardner proposed a theory of multiple intelligences.  He suggested that 
intelligence constitutes multiple intelligences that are relatively 
independent, even though the exact nature, extent and number of the 
intelligences has not yet been established.  He tentatively confirmed six 
natural kinds of intelligence, namely linguistic, musical, logical-
 mathematical, spatial, bodily-kinesthetic and personal (Gregory, 1996). 
 
Sternberg?s theory stipulates three aspects of intelligence.  Componential 
intelligence is responsible for intelligent behaviour, experiential intelligence 
deals effectively with novel tasks, and contextual intelligence deals with 
the adaptation to, shaping and selection of real-world environments 
(Gregory, 1996). 
 
2.4.2 History of intelligence testing 
 
In order to understand intelligence testing it is important to consider the 
evolution of intelligence testing, the defining of intelligence, and for the 
purpose of this study, studies on the correlation of academic performance 
and intelligence. 
 
Intelligence measurement is by no means a new concept.  As far back as 
the nineteenth century Galton and Wundt reasoned that all knowledge is 
derived from our senses.  Intelligence testing for them was therefore 
11  
measuring individual differences in sensorial discrimination and 
psychomotor co-ordination.  Cattell followed the same line of reasoning 
and measured intelligence by measuring properties like reaction time, 
sensory perception, mechanical memory and motor skills.  However none 
of these measures were found to be effective predictors of academic 
performance or other facets of intellectual functioning.  Binet invented the 
first modern intelligence scale in 1905.  The main purpose of his test was 
to identify children who would not benefit from the conventional methods 
of formal education.  The test was revised in 1908 by Binet and Simon, 
and again in 1911.  In 1912, the concept of the intelligence quotient (IQ) 
was invented by Stern (Gregory, 1996). 
 
For most of the twentieth century, attention was refocused on reaction 
time with the growth of experimental cognitive psychology, particularly the 
time taken to process cognitive tasks. Jensen?s research during the 
eighties revealed a high correlation between reaction time on elementary 
cognitive tasks and the g factor.  Terman also experimented with tests of 
complex abilities like creative imagination, logical reasoning, 
mathematical, language and learning ability, and insight.  These tests 
showed positive correlations with academic performance. suggesting that 
the measurement of a single trait, viz. Intelligence was acceptable.  
Terman adapted the Binet scale, but later began using the intelligence 
quotient (IQ) as a measure of intelligence using the Stanford Binet scale.  
From a theoretical standpoint, the problem with this scale was that the 
content of items was different for different age groups.  David Wechsler's 
scale grouped all items of a given type into subtests ant arranged the 
items in increasing order of difficulty.  He accepted the notion of the 
deviation IQ scale, which ensured that IQ had a consistent standard 
deviation at each level (Owen & Taljaard, 1996). 
 
 
 
12  
2.4.3 Intelligence as a predictor of academic performance 
 
An extensive body of research on intelligence as a predictor of academic 
performance exists.  This section addresses how this body of research is 
subdivided, and focuses on relevant research. 
 
2.4.3.1 Classification of studies 
 
Studies usually differ on two dimensions.  Firstly, in their concern with 
global as against multidimensional prediction in terms of both the predictor 
variable, as well as academic performance, and secondly on the 
educational level. 
 
In terms of the first classification, global refers to the use of a single, 
overall measure of intelligence and/or academic performance, whereas 
multidimensional refers to the use of a number of specific dimensions of 
intelligence and/or performance.  These distinctions are illustrated in Table 
1 (Lavin, 1965).  Although Lavin?s work may be considered outdated, it is 
the author?s opinion that it still provides an exemplary structure when 
attempting to classify studies on the prediction of academic performance. 
13  
Table 1 
 Types of studies relating ability and academic performance 
 Predictor Criterion 
 Global Multidimensional 
Global Global ability measure 
used to predict grade 
point average 
Global ability measure 
used to predict grades 
in specific courses 
   
Multidimensional Several specific ability 
dimensions used to 
predict grade-point 
average 
Several specific ability 
dimensions used to 
predict grades in 
specific courses 
Note: From The prediction of academic performance  (p.49 ), by D.E. Lavin, New York:  Russel Sage 
Foundation.  Copyright 1965 by Russel Sage Foundation.
  
 
In terms of educational level, study samples usually constitute 
elementary/primary school, high school, tertiary level and post-graduate 
students.  
 
2.4.3.2 Findings at the tertiary level on intelligence as a predictor 
of academic performance 
  
Due to vast amount of research available on intelligence as a predictor of 
academic performance, only findings at the college level will be 
highlighted, as these are most relevant to the present study of first year 
students.  Findings will be organised according to the categories set forth 
in the previous section. 
 
? Global measures of ability and performance 
 
Research findings consistently reveal ability measures having low 
predictive validity for academic performance.  Altink (1988) cited in 
Zaaiman (1998) ascribes the low predictive validity of ability tests to such 
14  
tests measuring constructs that are only hypothetically related to academic 
achievement and are therefore further removed from the performance to 
be predicted than either achievement or scholastic aptitude tests. 
 
? Specific ability dimensions used to predict GPA 
 
Of all the specific ability measures utilised in batteries of predictors, the 
one that constantly emerged as the best single predictor is high school 
marks.  Using academic performance at school is the most conventional 
method of predicting academic performance at tertiary level.  Smit (1992) 
operationally defines school performance as a pupil?s performance in the 
different exams during his/her high school career, with specific reference 
to the performance in the Matric exams.  Standard nine and Matric marks 
have been widely used as admission criteria at South African technikons 
and universities in the past, and continues to play a significant role. This is 
also the criterion primarily used for entrance to Tshwane University of 
Technology. Kotz? (1994) cites research by Behr (1985), Bloom (1976), 
Jencks, et al (1983), Monteith (1988) and Zietsman & Geringer (1985) as 
evidence of school performance being the only best predictor of future 
academic performance, and that it is an even better predictor than 
cognitive potential.   
 
However, research by Potter and Jamotte (1985), and Kotz?, van der 
Merwe and Nel (1996), reveals that Matric results from the former 
Department of Education and Training and equivalent government school 
departments specifically at the lower ranges, are inaccurate reflectors of 
students' academic ability, and that there is a definite inconsistency 
between the results of attendees of "black schools" and university 
academic performance.  Zietsman and Geringer (1985) note the influence 
of milieu disability, which leads to Matric results being a poor predictor of 
academic performance.  Miller (1992), Gourley (1992) and Huysamen and 
Raubenheimer (1999) question the use of Matric performance as a 
15  
predictor of academic success as a result of the different instruction 
approaches at school and tertiary institutions.  
 
Case studies by Pavlich and Orkin (1993) indicate the importance of 
developing alternate admission programmes that identify competent 
students without depending on Matric results, as school performance may 
have been negatively affected by disadvantage.  Further necessitating the 
identification of alternative selection criteria is the introduction of the 
Further Education and Training Certificate (FETC) that will be replacing 
the Matric certificate when the National Qualifications Framework (NQF) 
will be implemented in the South African Education system.  The FETC will 
then become the statutory minimum requirement for entry into all higher 
education programmes, and will be obtained through formal schooling or 
through adult basic education and training and recognition of prior 
learning, or a combination of these options.  The predictive validity of the 
new FETC qualification will have to be investigated before it can be used 
for selection purposes (Zaaiman, 1998). 
 
? Several dimensions of intelligence used to predict grades in 
specific courses or course areas 
 
When the content of predictor batteries varies depending on the particular 
course for which grades are being predicted, this is regarded as the 
differential prediction technique.  Horst has made a meaningful 
contribution to the development of this technique.  His work is based on 
the rationale that while global predictors and criteria emphasise overall 
academic performance, it is likely that the performance of most students 
differ qualitatively from one subject area to another (Lavin, 1965). 
 
This technique is also utilised at Tshwane University of Technology for 
selection.  Selection batteries vary from course to course. In May 2001, 
the first phase of a project begun to identify alternative selection criteria for 
16  
students to admit students to the Information Technology (Information 
Technology) course at Tshwane University of Technology.  450 
prospective students were assessed from May to June 2001 for admission 
into the Information Technology course. 140 students were selected.  Two 
instruments were utilized, namely the Situation Specific Evaluation Expert 
(SpEEx) and the ITTS. The indices used from the SpEEx were 
Conceptualisation, Observance, Insight, Calculations, Environmental 
Exposure, Comparison, Perception, Object Assembly, and Reading 
Comprehension.  The ITTS indices used were Verbal Reasoning, 
Advanced Calculations, Syntax Checking, and Diagramming.  Neither the 
SpEEx battery, the ITTS battery, nor the two combined were able to make 
a significant contribution to the prediction of academic performance of 
Information Technology students at Tshwane University of Technology 
(Kriel, 2002). 
  
Table 2 
Results of multiple regression analysis in Kriel?s (2000) study 
Subject code R2  (SpEEx) R2  (ITTS) 
R2   
(Best 
predictors) 
ITS IIAA 0.13 0.09 0.24 
ITS IIBA 0.13 0.03 0.11 
ISY13AA 0.05 0.05 0.70 
ISY13AB 0.15 - 0.15 
SSF IIAA 0.08 0.10 0.13 
 DSO 15AA 0.11 0.07 0.11 
      From  ??n Uitgebreide keuringsbattery vir studente in Inligtingstegnologie? by H. Kriel, 2002, p.3.   
 
The differential prediction technique is also utilised in another important 
way in South African assessment.  It is used to ascertain test fairness.  In 
psychometric and legal circles, the Cleary approach is currently regarded 
as the most accepted statistical approach to test fairness.  The Cleary 
17  
regression model uses different subgroup criterion-predictor regression 
equations for prediction purposes when the equations differ significantly 
for different subgroups, assuming an unbiased criterion (Zaaiman, 1998). 
 
? Global ability measure is used to predict grades in specific 
courses  
 
Compared to the volume of research in the other categories, relatively little 
research is available in this group of studies.  According to Thorndike 
(1986), global ability measures only provide a measure of general 
cognitive functioning, with Zaaiman (1998) adding that ability tests are 
non-subject specific.   
 
2.5 Personality factors as predictors of academic performance 
 
Although ability measures are the best single type of predictor of academic 
performance, they account only for a portion of the variance.  Hence, it is 
essential to consider nonintellective factors too.  This has led many 
researchers to study academic performance by concentrating on 
personality characteristics.  This section first commences with an 
explanation of the classification of these studies, and then focuses on 
some of the personality variables that have been related to academic 
performance. 
 
2.5.1 Classification of personality studies 
 
These studies are organised into five areas.  Firstly, achievement 
motivation, level of interest, etc. are grouped under motivational states.  A 
second area is personality style, which includes factors like degree of 
interdependence, impulse control, and introversion.  The third area is the 
cognitive level, which is the self-concept.  A fourth group of studies focus 
on manifestations of pathology to account for achievement.  Finally, 
18  
measures of study habits are more directed at the behavioural level.  This 
final category of factors will be discusses at length in Chapter 3, as it is the 
focus of the present study (Lavin, 1965) 
 
Two basic methods of analysis are utilised in research of personality 
variables.  Firstly, the correlational method is used to evaluate the degree 
of relationship between the personality factor and academic performance.  
In these studies, ability is controlled either through a partial correlation 
analysis, or by a multiple correlation, where the contribution of a 
personality variable to a battery of intellective factors is assessed.  With 
the second method of analysis, academic performance is researched by 
composing groups of high and low achievers or under- and over 
achievers, and assessing possible personality differences between such 
groups (Lavin, 1965). 
 
2.4.5.1 Motivational states as a predictor of academic performance 
 
Three motivational states are addressed in terms of their relationship with 
academic performance, namely interest, achievement motivation and 
aptitude. 
 
? Interest 
 
?Interest is an aspect of personality and can be defined as a spontaneous 
attraction to, or preference for, certain activities, as well as a spontaneous 
aversion to other activities?  (Owen & Taljaard, 1996, p.396).  Interest is an 
important component in the prediction of academic performance according 
to De Boer (cited in Engelbrecht, 1999), as a lack of interest can 
negatively impact on performance. 
 
Measures of interest have been utilised in three ways to predict academic 
performance.  In the first manner, scores on particular interest categories 
19  
have been related to performance in corresponding course work.   Another 
approach has been to assess if any type of interests predict academic 
performance.  The third group of studies does not deal with substantive 
interests per se, but with particular general dimensions that transverse 
content, like clarity and intensity of interests (Lavin, 1965). 
 
Singh?s (1986) study (cited in Padma, 2003) typifies the first group of 
studies.  Using Samal?s Vocational Interest Inventory, Singh failed to find a 
relationship between achievement in mathematics, and scientific interest, 
mechanical interest, interest in agriculture, business, social service, art, 
nor official activities. 
 
? Achievement motivation 
 
Achievement motivation may be defined as an individual?s predisposition 
to approach or avoid a competitive situation.  More widely defined, it 
embraces the concept of a desire to excel.  From the mid-1950?s through 
the mid-1970?s, the theory of achievement motivation that attracted the 
most attention in psychological literature was the McClelland-Atkinson 
theory.   This model proposes that achievement motivation depends two 
constructs: the motive to achieve success and fear of failure.  The motive 
to achieve success is believed to represent a student?s intrinsic motivation 
to engage in an activity, and fear of failure is a psychological construct 
related to cognitive state anxiety. The desire to enter such an activity is a 
function of the relative strength of the two constructs.  If a student?s self-
 confidence on a specific task is greater than the fear of failure at the task, 
that person will approach and perform the task. Another leading model 
was the test-anxiety approach, which proposed that fear of test taking or 
fear of failure (test anxiety) was the critical factor in determining whether or 
not an individual would approach or avoid an achievement situation.  
Students suffering from high levels of anxiety would therefore be expected 
to avoid achievement situations.  The third model, namely the Crandall 
20  
approach emphasised the individual?s expectation for reinforcement and 
on an individually set level of excellence.  In practice, a student would 
compare social expectations with minimal standards of success.  If 
expectations are too high or reinforcements too few, the student will drop 
out of the achievement situation (Cox, 1990). 
 
Generally, studies examining achievement motivation as a correlate of 
academic performance, are sceptical about making inferences about this 
variable. Shanmugasundaram (1983), Deshpande (1984) and Sween 
(1984) found a positive relationship between the two variables.  Rajput 
(1984) and Sontakey (1986), on the other hand, failed to find such a 
relationship.  Mitra?s (1985) study found that achievement motivation is 
positively and significantly related to academic performance, but this result 
is not applicable if intelligence is partialled out.  Sharma (1981) studied the 
factors related to academic underachievement of secondary schools 
located in rural areas in a village in India.  She also attempted to find out 
the specific contribution of variables towards high achievement and 
underachievement.  Her results indicated that poor academic motivation 
was one of the factors that contributed to underachievement (Padma, 
2003). 
 
Goldberg (cited in Anderson & Keith, 1997) contends that student 
motivation historically has been regarded as the most significant area of 
difference between lower class and middle-class students.  A greater 
emphasis on motivation is essential in comprehending and enhancing 
minority achievement.  Motivation is a multidimensional construct that 
includes cognitive, environmental and behavioural components, is usually 
defined only by its cognitive component.  Anderson and Keith (1997) cites 
Carroll?s (1989) work which is the noted exception ? rather than focusing 
on beliefs, expectancies, self-efficacy, and other attitudinal expressions of 
motivation, Carroll describes the actual amount of time spent on learning 
as the operational definition of motivation for learning.  Estimates of 
21  
motivation that have included behavioural expressions have been more 
strongly related to achievement than those that have relied on cognitive, 
affective, or attitudinal expressions only. 
 
? Aptitude 
 
One of the earliest definitions of aptitude is provided by Bingham (1937, 
p.17) who defines aptitude as a ?condition or set of characteristics 
regarded as symptomatic of an individual?s ability to acquire with training 
some (usually specified) knowledge, skill or set of responses such as the 
ability to speak a language, to produce music, etc.?.  Bingham contends 
that aptitude is a present condition that refers to the future.  Aptitude tests 
measure current achievement that provides an indication of future 
potentialities.  Aptitude implies a fitness, or suitability for the activities in 
question.  Gekokski accepts Bingham?s definition, and additionally points 
out five characteristics of aptitude, namely, intelligence, interest, 
personality, special abilities and attitude.  He believes that the five 
components are independent form one another and each is to a greater or 
lesser extent involved in a person?s aptitude for a particular task.  Guilford 
(1959) views aptitude differently.  Guilford (1959, p.342) refers to aptitude 
as ?underlying dimensions of ability that prepare a person for a certain 
level of performance in given activities and for learning new things.?  
These underlying dimensions are perceptual, psychometric and 
intellectual abilities.  Snow (1992) defines aptitude as apt, appropriate and 
suitable, implying that the definition of aptitude is situation-specific.  The 
concept subsists in the person-situation interface, rather than solely in the 
person.  Hence aptitudes imply initial states that subsequently affect later 
developments, given specified conditions, meaning a responsiveness to 
learn from a given situation. 
 
Aptitude has been identified as a notable correlate of academic attainment 
in certain fields.    Agarwal?s (cited in Padma, 2003) investigation indicated 
22  
medical aptitude as one of the contributing factors associated with 
proficiency in medical examinations in five different medical colleges.   
 
Owen and Taljaard (1996) assert that the scores from aptitude tests 
should be regarded as useful information that ought to be used with other 
information about a person in order to take certain decisions.  Other 
information includes school performance, interests, attitudes, study habits, 
human relations, and specific likes and dislikes.  Aptitude tests on their 
own should not be used as ?decision makers?. 
 
Zaaiman (1998) concludes from Altink?s (1988), Heyneman?s (1988) and 
Jacobs?s (1995) studies that prevailing scholastic aptitude tests are 
designed to assess subject-related skills, comprehension and insight, 
utilise non-routine problems that require minimal subject-specific 
knowledge. 
 
The Senior Aptitude Test (SAT) is an example of an aptitude test which 
research has failed to uphold as a predictor of academic performance.  
Van Zyl, Terblanche and Jacobs (1992) found no significant difference 
between the SAT performance of successful and unsuccessful social work 
students.  Fourie?s (1990) investigation did also not find the SAT to be an 
effective predictor of academic performance. 
 
However, a number of studies have provided some support for the 
predictive validity of aptitude tests.  Skuy, Zolezzi, Mentis, Fridjhorn and 
Cockcroft (1996) found that Barker?s Pattern Relations Test, which is a test 
of inductive reasoning and reasoning by analogy, predicted performance 
in Accounting for a group of students at the University of Witwatersrand.   
Pienaar (1991) concluded that specific subtests of the HSRC?s Academic 
Aptitude Test (AAT) differentiated between successful and unsuccessful 
learners at higher education level, finding that the academically successful 
students performed significantly better on the Non-verbal Reasoning, 
23  
Verbal Reasoning and Reading Comprehension subtests.  The predictive 
validity of the AAT was also researched by Samkin (1996), whose results 
indicated that the Reading Comprehension explained 14.4% of the 
variance of the final Accounting mark and Number Comprehension 
subtests.  
 
2.5.1.1 Personality style 
 
Studies investigating the relationship between personality style and 
academic performance tend to fall into two groups.  The first group of 
studies aims to identify the difference in personality traits between 
overachievers and underachievers in specific areas.  The second group of 
studies attempt to determine if specific personality traits are related to 
academic performance.   
 
? Personality traits of underachievers and overachievers 
 
Jahan?s research (cited in Padma, 2003) compared the personality profiles 
of over and underachieving students studying in science, arts and 
commerce streams in preuniversity classes.  The sample comprised 758 
male and female students.  Cattell?s High School Personality 
Questionnaire (HSPQ) was used to measure personality.  Thorndike?s 
method of identifying over and under-achievers on the basis of 
discrepancies between actual achievement and that predicted on the basis 
of intelligence was employed.  The significance of difference between the 
means of scores on the fourteen dimensions of HSPQ secured by the over 
and underachievers of the three streams were ascertained by a t-test.  
Overachievers of the science stream were more reserved, intelligent, 
emotionally stable, excitable, obedient, sober, conscientious, shy, self-
 assured, self-sufficient, controlled and relaxed as compared to the 
underachievers.  Overachievers of the arts stream were more warm-
 hearted, intelligent, affected by feelings, undemonstrative, assertive, 
24  
enthusiastic, conscientious, zestful, apprehensive and tense.  In the 
commerce stream, overachievers were reserved, intelligent, affected by 
feelings, sober, conscientious and self-assured (Padma, 2003). 
 
Padma (2003) also discusses Puri?s (1987) study that used the same 
instrument to make a detailed study of the personality traits of 
underachievers of 425 students at 12 intermediate colleges of Lucknow 
city.  Underachievers were found to be more warm-hearted and 
easygoing, inactive, assertive, aggressive, stubborn and dominant, 
impulsive, lively, happy-go-lucky, and socially bold.  They were generally 
over-protected, sensitive, individualist and reflective, apprehensive, 
worrying and troubled. 
 
Snake?s (1986) study cited by Padma (2003) found high achievers in the 
natural sciences to be less excitable, undisciplined, having self-conflict, 
relaxed, tranquil, and more unfrustrated than low achievers.  In the 
biological sciences, high achievers were less excitable, tough minded, 
self-reliant and realistic. 
 
? Big Five factors of personality 
 
Zaaiman (1998) mentions studies by Rothstein et al. (1994), De Fruyt and 
Mervielde (1996) and de Raad and Scholuwenburg (1996) on the 
predictive validity of personality factors for academic achievement, 
referring to the current interest in the Big Five factors of personality.  
McCrae and Costa (1987) proposed a five-factor model of personality.  
McCrae and Costa assert that most personality traits are derived from just 
five critical traits, namely Neuroticism, Extraversion, Openness to 
Experience, Agreeableness and Conscientiousness.  A description of each 
personality factor is described in Table 3 (see Appendix B).   
 
25  
Research thus far has indicated that the factor Conscientiousness shows 
the most promise as a predictor of academic achievement, with mixed 
results found for the other factors with the predictive validity and direction 
of prediction dependent on the performance context and type of 
achievement measure.  To what extent and which personality traits predict 
academic performance was investigated by Chamorro-Premuzic and 
Furnham (2003) in a longitudinal studies of a British university sample.  
Academic performance was measured throughout a three-year period and 
used multiple criteria.  In addition, several indicators of academic 
behaviour were also examined with regard to both academic performance 
and personality traits.  In the first sample, the Big Five personality factors, 
specifically Neuroticism and Conscientiousness, were found to predict 
overall exam marks over and above several academic predictors, 
accounting for more than 10% of unique variance in overall exam marks.  
The results suggest that Neuroticism may impair academic performance, 
while Conscientiousness may lead to higher academic performance 
(Chamorro-Premuzic & Furnham, 2003).  
 
Patel?s study (cited in Padma, 2003) on the relationship between 
neuroticism and high achievement provided contradictory results. The 
sample consisted of medical students drawn from  two medical colleges of 
Ahmedabad city.  Eyesenck?s Neuroticism Inventory was utilised for data 
collection.  No relationship was found between the two variables. 
 
One of the aims of Mitra?s research (cited in Padma, 2003) was to 
ascertain the relationship between extraversion and academic 
performance.    Eyesenck?s Personality Inventory for Juniors adopted in 
Bengalis by Arti Sen was used to measure extraversion.  Extraversion 
correlated positively and significantly with academic performance, but lost 
its significant effect when intelligence was partialled out (Padma, 2003). 
 
2.5.1.2 Self-image 
26  
 
Padma (2003) cites studies by Shanmugasundaram (1983), Singha 
(1983), Sween (1984), and Pathani (1985) which all considered self-
 concept as a factor correlating with academic performance.  Sween used 
the Deo Personality Word List, and found that students with high self-
 concept achieved significantly higher than those with a low self-concept.  
Similarly, Pathni concluded that self-concept is a significant predictor of 
both actual and perceived academic performance, using the Self-concept 
Scale. Shanmugasundaram?s results differed from Sween?s and Pathni?s 
studies. Shanmugasundaram, using Mohsin?s Self-concept Scale failed to 
differentiate between high and low achievers. 
 
2.5.1.3 Adjustment 
 
Numerous studies have been carried out on certain personality factors of 
students as correlates of academic performance.  Personality traits are 
assessed using personality questionnaires or inventories (Padma, 2003). 
 
Studies utilising the MMPI has indicated a positive relationship between 
personality and academic performance.  A multiple correlation of 0.69 is 
found between adaptability and academic  performance.  Underachievers 
were also found to display the following characteristics: more emotionally 
labile, defensive and lacking a warm disposition to others.  Students who 
performed well academically are more self confident and well adapted 
socially (Kotz?, 1994). 
 
2.6 Sociological determinants of academic performance 
 
Studies on sociological correlates of academic performance tend to fall 
into one of two groups.  The first group examines the effects of role 
relationships on academic performance, and the second looks at the 
27  
effects of various ecological and demographic characteristics upon 
academic performance (Lavin, 1965). 
 
2.6.1 Demographic and ecological variables 
 
Demographic and ecological variables are related to academic 
performance as they symbolise specific uniformities of personality.  
Positions in the social structure like socio-economic status and gender 
tend to create particular similarities in personality among the occupants of 
these positions.  Some of these personality characteristics are in turn 
related to academic performance (Lavin, 1965). 
 
 
2.6.1.1 Socio-economic status (SES) 
 
The term socio-economic status (SES) usually refers to the overall social 
position of a family as measured by the parents? economic and social 
achievements.  Indices of SES typically utilise indicators of the 
educational, occupational and economic achievements of the parents.  
Relevant research indicates that low SES as defined by these factors 
contributes to educational disadvantage (Ainley, Graetz, Long & Batten, 
1995). 
 
Educational disadvantage may be characterised in terms of the 
educational opportunities available to a student (Ross, Flarish & Plunkett, 
1988).  It is the net effect of the features of a learner?s environment that 
provide less than normal exposure to factors that motivate and facilitate 
educational growth.  Within the South African context, a learner is 
therefore regarded as disadvantaged if s/he has had insufficient access to 
quality educational services, resulting in a lack of opportunity to fully 
develop her/his academic potential (Zaaiman, 1998). 
 
28  
A student?s home background generally affects the educational 
opportunities available.  The quality of education available to a student 
from a low SES background can be expected to be influenced by the 
student?s peer group, the educational facilities available at the school, and 
the quality of teachers employed by the school (Zaaiman, 1998). 
 
Padma (2003) cites studies by Das (1975), Dwiveldi (1983), Sarah (1983), 
Sharma (1984), Shukla (1984), Rajput (1984), Mehrotra (1986), Misra 
(1986), Singh (1986), Srivastava and Ramaswamy (1986), Sontakey 
(1986), and Trivedi (1987), which indicated a significant correlation 
between socio-economic status and academic performance of students. 
The environment provided to a student by his home has been researched 
by Grover (1979), Gaur (1982), Sarkar (1983), Lail (1984), Jagannadhan 
(1985), Maitra (1985), Paul (1986) and Trivedi (1987) (cited in Padma, 
2003).  The factors revealed in the abovementioned studies as being 
correlates of academic performance include educational environment, 
income, spatial environment, social background, provision of facilities, 
parent-child relationship, home environment, recognition of child?s 
achievement, parental aspiration, forbearance for the child?s wishes, 
parental affection and encouragement (Padma, 2003). 
 
Anderson and Keith (1997) analysed longitudinal data from 8100 high 
school students in a national study to test a model of at-risk students? 
school learning, using structural equations analysis to fit their model to the 
data.  Family socio-economic status was found to be one of the variables 
in their model that exerted a meaningful total effect. 
 
2.6.1.2 Gender 
 
Gender inequality is of international concern in terms of access to higher 
education, more especially in science- and technology-based study.  The 
29  
Department of Education (1997) regards equal access for women as an 
important aim for the transformation of higher education in South Africa. 
 
Anderson and Keith?s (1997) study discussed above also included gender 
as a variable in the model.  Gender was supported by the large, 
longitudinal dataset as a significant and important predictor of academic 
performance for at-risk high school students. 
 
Padma (2003) cites Dhar?s (1986) and Jasuja?s (1983) research.  Dhar?s 
study which found no difference in the academic performance of 
intellectually superior and intellectually very superior Grade 11 boys and 
girls, while Jasuja?s research found that adolescent girls achieved higher 
in the academic field.   
  
Tinklin (2003) used the Scottish School Leavers Survey to examine factors 
related to high attainment and questioned whether they were different for 
males and females.  A strong correlation was found between social 
advantage and high attainment within each gender, but no evidence of 
differential progression rates for males and females was found between 
the ages of 16 and 18.  The only factor that displayed evidence of gender 
differences in achievement was that girls took school more seriously than 
boys, supporting theories that girls and boys experience different peer 
pressures that affect their academic performance. 
 
Potential causes of the observed gender differences in academic 
performance are neurophysiological differences (Kahle, Parker, Rennie & 
Riley, 1993), different learning strategies (Meece & Ecclees, 1993), or 
attitudinal factors (Brusselmans-Dehairs & Henry, 1994).  Humphreys 
(1986) and Young (1991) refer to continuous findings that tests of general 
intelligence, scholastic aptitude and high school grade point average 
underpredict female performance in higher education across course or 
subject choices.  However, Young (1991) contends that such findings may 
30  
result from incomparability of criteria (course differences) rather than 
gender bias in the standardised tests (Zaaiman, 1998). 
 
Concluding from the international literature, gender differences can be 
expected to occur in academic achievement.  However, selection tests 
and other criteria need to be checked for bias on gender in order to ensure 
gender fairness (Zaaiman, 1998). 
 
2.6.1.3 Miscellaneous ecological and demographic characteristics 
 
Besides SES and gender, other ecological and demographic 
characteristics also impact academic performance.  Variables that will be 
discussed are age and regional variation. 
 
? Age 
 
A factor possibly influencing student performance in South Africa is 
student age.  School disruptions lead to many pupils taking longer to 
complete Matric and are substantially older than the norm of 18 or 19 
years when applying for admission to a tertiary institution (Zaaiman, 1998). 
 
Bot (1994) reported that Matric pass rates show a notable decrease with 
age for black students.  In 1993, 50% of candidates under the age of 19 
passed Matric.  In the same year, only 33.3% over the age of 21 passed 
Matric.  A large proportion of black matriculants are 22 years or older ? this 
age group constituted 37% of the total number of black candidates in 
1993.  This is therefore not an age group that can be ignored during 
selection, and may need special assistance to be able to succeed in 
higher education. 
 
Girija?s (1980) comprehensive study discussed by Padma (2003) also 
found age to be one of the thirty-two potential contributors to cumulative 
31  
grade point average (CGPA), and Muffo, Dickey and Bodo (1999) also 
found age to predict first year retention and academic success.  Wince 
and Borden (1995) indicated that age was the most significant 
demographic variable in terms of  GPA prediction, and Farabaug-Dorkins 
(1991) reported that attrition rates of older students are higher than those 
reported for traditional-aged college student freshman. 
 
? Regional and rural-urban variation 
 
The concept of ?rurality? is often associated with the concept of educational 
disadvantage and lower educational participation.  Klitgaard (1986) reports 
that students form urban areas are more likely to be admitted to 
universities than those from rural areas.  Rosseaux (1993) studied rurality 
and educational opportunity in Australia.  Wide variations exist among 
nations about the differentiation between rural and urban areas.  Rural 
locations and populations are commonly defined in terms of population 
density and size, location and economic activity.  In large countries, most 
regions contain a mix of both urban centres and rural areas.  It is often 
assumed that when defining an area as rural or urban, there is a direct 
relationship between the quality and number of educational services 
available and the population size of the area.  Relevant to rurality is the 
concept of remoteness implying isolation from educational services.  In 
remote areas, generally access to services depends on the economic 
power of the community to meet the costs of overcoming the distance 
between the user and the available service (Rousseaux, 1993). 
 
 A classification of a school as being ?rural? rather than ?urban? on the basis 
of expected lower access conforms to the description of disadvantage 
outlined in section 4.7.1.1.  However, in the South African context, it 
cannot naturally be assumed that population size necessarily corresponds 
with the services available.  Many of the black ex-homeland areas have 
large populations, but are under-serviced.  The general SES of the 
32  
population in these areas does not allow for access to services that may 
be available elsewhere.  Students from the previously black-only township 
areas surrounding the metropolitan Gauteng area, as well as learners from 
the current principal provincial cities, had greater access to educational 
and other services that student from the non-urban areas (Zaaiman, 
1998). 
 
Padma (2003) cites a study by Doraiswamy (1985), which aimed to 
determine school differences in performance and compare the standards 
of attainment reached by children of different social composition studying 
in different school locations. Findings indicated that performance of 
students from hilly areas was slightly less.  The performance of students in 
urban areas was better than that of students in rural areas. 
 
What is different at South African universities and technikons is that most 
of these institutions are situated in urban areas.  However, the institution 
serves as a melting pot for students from urban and rural areas.  Girija 
(cited by Padma, 2003) conducted a study of intellectual and non-
 intellectual factors in academic performance of students from professional 
colleges.  Place of permanent residence was one of the predictors of 
CGPA.  Rather than utilising location of institution, location of permanent 
residence (urban/rural) could therefore be a more appropriate predictor in 
South African studies on academic performance at tertiary level. 
 
2.6.2 Specific role relationships 
 
This section focuses on the correlation between the role of significant 
other people in the student?s life and academic performance.  Three 
specific categories are investigated:  educators, parents and peers. 
 
 
 
33  
2.6.2.1 Educators 
 
Niebuhr and Niebuhr (1999) conducted an empirical study of student 
relationships and academic performance.  Student-teacher relationships 
were assessed by a questionnaire, and academic performance was 
measured by GPA. Student-teacher relationships were positively 
correlated with performance, stressing that the importance of warm 
teacher involvement.  Part of the necessary quality condition in the 
classroom is that as teachers allow their students to know them, and like 
them, the students will work harder, thereby increasing their opportunities 
for success. 
 
Anderson and Keith (1997) cite Cool and Keith (1991) who considered a 
composite variable of quality of instruction defined by student ratings of 
teacher interest, quality of instruction and reputation of school within the 
community.  The multidimensional quality of instruction variable directly 
influenced student?s motivation, enrolment in academic coursework, and 
time doing homework, and indirectly through these variables, the teacher?s 
interest and quality of instruction influenced academic performance 
(Anderson and Keith, 1997). 
 
Watson and Munroe (1990) surveyed existing literature on the impact of 
teacher perception on academic performance. Studies report that teachers 
interact differently with students who are perceived to be at different 
achievement levels.  Perceived high achievers received more questions 
and praise from their teachers, teachers consistently interacted more with 
the top third of the class, waited significantly longer for higher-achieving 
students to respond to a question, and teachers with artificially high 
expectations were related to enhanced student performance.  Research 
attempting to induce teacher expectations by providing teachers with 
unreal information has generally not yielded significant results.  However, 
naturalistic studies show the effects of expectancy.  Continued research in 
34  
the area leads to a consensus that teachers? expectations affect teacher-
 student interaction and student outcome, along with the recognition that 
the processes involved are much more complex than originally believed.  
Most teacher expectations of student achievement are negative ones, 
where low expectations lead to lower achievement than might have been 
attained otherwise (Watson & Munroe, 1990). 
 
Some researchers have focused their attention on the effect of the 
processes that occur in teaching-learning activity on the achievement of 
students.  Some of the areas investigated include teaching methods and 
teacher behaviour (Padma, 2003).   
 
Padma (2003) cites Nagpal (1983), Bhalwankar (1985), and Kumari 
(1985) who considered the impact of the method of teaching on academic 
success.  Nagpal compared the effects of the Piagetian method of 
teaching, objective-centred teaching based on the Advanced Curriculum 
Model of Cognitive Learning and the traditional method on the 
achievement in Hindi of primary school students.  Both methods of 
teaching were found to be superior to the traditional method, with the 
objective-centred method showing superiority of the Piagetian method. 
Bhalwankar?s study examined the effects of expository and guided 
discovery method of teaching mathematics on achievement.  The 
effectiveness of the two methods differed according to the different 
instructional objectives as well as the level of intelligence of the students.  
The research conducted by Kumari found no difference in students? 
achievement when taught by either the inductive or deductive strategy of 
instruction, but the combined strategy was more effective than either 
strategy taught separately (Padma, 2003). 
 
Teacher behaviour was investigated by Yadav (cited in Padma, 2003) who 
attempted to find the relationship between indirectness in teacher 
behaviour and pupil achievement.  The Flanders Interaction Analysis 
35  
Category System was used to measure indirectness.   No relationship was 
found between indirectness and student achievement in high school 
biology. 
 
With the increase in student numbers at South African tertiary institutions 
(massification) and limited subsidies, a lecturer is compelled to get through 
the contact faster in class.  Personal contact with students is minimal.  
Schaap en Buys (cited in Engelbrecht, 1999) contend that this directly 
influences a student?s attitude.  A lecturer does not have sufficient class 
time available to devote the necessary interest and attention to each 
student.  Consequently, the student does not believe that the lecturer can 
empathise with his/her problems.  This, in turn negatively affects the trust 
and communication between the lecturer and student, which leads to poor 
academic performance (Engelbrecht, 1999). 
 
2.6.2.2 Parents 
 
Parents? education levels and their beliefs concerning their children as a 
correlate of academic performance was investigated by Hortacsu (1995) 
with a Turkish sample characterised by high levels of interdependence 
among family members, and high levels of parental control over children.  
Parents? level of education as an index of social context has both physical 
and psychological aspects and may therefore exert both direct and indirect 
influences on child outcomes.  Direct effects may originate from 
differences in quality and quantity of educational opportunities available to 
children, and indirect effects may be attributed to values and beliefs 
concerning children.  A high level of parents? education was association 
with endorsement of perspectivitistic beliefs and rejection of categorical 
beliefs for both parents, but the relationship between parent?s beliefs and 
academic performance was different for both parents.  Only the level of 
mother?s education was directly related to the child?s academic 
performance.  The emergence of a significant relationship between 
36  
mother?s level of education and GPA is consistent with earlier research 
involving a similar sample of Turkish children (Hortacsu, Ertem, Kurtoglu & 
Uzer (1990).   Hortacsu (1995) attributes this finding to the fact that 
women are primarily responsible for childrearing in Turkey.  Mothers with 
higher levels of education are capable of facilitating their children?s 
education, whereas women with lower levels of education are not able to 
do so. 
 
The relationship between parenting practices and academic performance 
was investigated by Hae-Song and Bauer (2003).  The sample constituted 
Asian Americans, Hispanics, African Americans and European Americans.  
The results of the study show revealed that European Americans were 
more authoritative that other ethnic groups, and there was also only a 
significant relationship between having an authoritative parenting style and 
academic performance for this group (Hae-Song and Bauer, 2003). 
 
2.6.2.3 Peers 
 
Niebuhr and Niebuhr (1999) also investigated student-peer relationships.  
Student-peer relationships positively correlated with performance as well.  
A possible explanation was that a cohesive student group with high 
academic norms enhances student motivation, which is in turn related to 
academic performance.  The results of this study suggest a need for future 
research to examine the impact that motivation has on student 
relationships. 
 
Engelbrecht (1999) discusses the importance of interpersonal 
relationships.  The vast majority of first-year students exchange familiar 
surroundings with an established circle of friend for a new environment 
with no familiar peers.  The student?s social skills will then determine how 
long it will take to establish a new social network.  This implies that social 
isolation can become a reality, which may negatively impact academic 
37  
performance.  The demands of the new peer group plays an important role 
in the extent to which the student will be able to adjust socially.  An 
academically successful student has better interpersonal relationships and 
social skills than students who do not perform well academically. 
 
Astin?s (1983) Theory of Involvement model emphasises the importance of 
interaction with peers from the perspective of the active involvement in 
institutional activities.  Astin (1983) contends that the more involved 
students are in the total college experience, the more satisfied they will be 
with the institution, they will tend to be well-integrated with their peers and 
perform better academically.  
 
2.7 Conclusion 
 
According to Daniels & Schouten cited in Louw (1990), cognitive factors 
constitute 50% of the variance in academic performance.  According to the 
literature surveyed, intellective factors still emerge as the strongest 
predictor of academic performance.  Information Technology students, the 
subjects in the present study, are also still selected on the basis of an 
intellective factor, namely Matric results which has consistently been 
upheld as the single best predictor of academic performance at tertiary 
level.  However, this criterion is insufficient on its own, or else there would 
be no need for research on non-intellective factors.  While the research 
surveyed in the literature review does not provide any unequivocal 
irrefutable results, it provides a basis for the wider exploration of certain 
factors as possible correlates of academic performance. 
 
Chapter 3 also discusses correlates of academic performance, pertaining 
specifically to those measured by the APA battery, that is language 
abilities and study skills. 
38  
 
 
Chapter 3 
THE ASSESSMENT OF ACADEMIC PROFICIENCIES  
 
 
3.1  Introduction 
 
This chapter focuses on the variables evaluated by the Academic 
Proficiency Assessment battery (APA battery).  The APA was compiled by 
the Directorate of Student Development and Support at the Tshwane 
University of Technology (formerly Technikon Pretoria) in 2001 in order to 
assess academic proficiencies in two areas, namely language and study 
skills.  It is not an instrument used for selection to higher education. It is an 
assessment that aims to identify students who lack the essential academic 
competencies and language proficiencies required at tertiary level thereby 
serving as an ?early warning system? of students possibly at risk for 
academic underachievement. In other words, the APA tries to detect 
students who have already been selected by, and registered at the 
university, but still lack the necessary language and study skills for 
satisfactory academic performance. 
 
Foxcroft and Roodt (2001) also note the utility of assessment for such 
purposes in higher education in South Africa.  Other than assisting in 
admission decisions, another advantage of assessing learners at entry to 
higher education is that the resultant information can be used for the 
development of student support initiatives.  Rather than only identifying 
learners who have the potential to succeed, assessment can serve a more 
developmental purpose that ultimately benefits the learner. 
 
39  
This chapter examines research on the academic proficiencies evaluated 
by the APA battery, and relevant issues in the assessment of academic 
proficiencies.  It should be noted that no validation studies have been 
done to confirm that the APA does indeed measure these proficiencies. 
41
  
 
_
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
  
Fi
 gu
 re
  
2.
  
 
Fa
 ct
 o
 rs
  
re
 la
 te
 d 
to
  
a
 ca
 de
 m
 ic 
pe
 rfo
 rm
 a
 n
 ce
  
hi
 gh
 lig
 ht
 in
 g 
fa
 ct
 o
 rs
  
as
 se
 ss
 e
 d 
by
  
th
 e
  
AP
 A 
ba
 tte
 ry
  
42  
 
3.2   Academic proficiencies 
 
In this section, each of the seven academic proficiencies (English proficiency, 
reading skills, time management, note-taking skills, memory skills, 
examination stress, and concentration skills) will be investigated with regard 
to its operational definition within Tshwane University of Technology?s 
academic skills development programme, and relevant studies on each 
variable as a predictor of academic performance 
 
3.2.1 Language 
 
The significance of language was highlighted in Celliers? (2000) research, 
which indicated that academic support practitioners regarded language as the 
most important skill required to enhance learners? academic success. Since 
the official academic languages at Tshwane University of Technology are 
English and Afrikaans, with lectures being predominantly in English, most 
learners at Tshwane University of Technology are disadvantaged, as English 
is the second or even third/fourth language.  When entering the Technikon, 
these learners experience difficulties with language: reading, writing and 
communicating clearly.  This is one of the most serious problems confronting 
higher education in South Africa, as language is a vehicle that enables 
learners to successfully complete their studies.  Without adequate language 
competence, and communication skills, learners have a significant problem.  
Furthermore, learners cannot be given quick fixes in English.  Language 
acquisition and proficiency requires years of constant practice and usage 
(Celliers, 2000). 
 
Language proficiency is assessed and remediated in two basic ways at 
Tshwane University of Technology, through English proficiency and reading 
skills. 
 
 
 
43  
3.2.1.1 English Proficiency 
 
Language proficiency will be explicitly defined within in its context, how 
English is enhanced, and studies on English proficiency as a correlate of 
academic performance will be discussed in this section. 
 
Tshwane University of Technology?s academic support programme attempts 
to enhance English proficiency using the English Word Power (EWP) 
program.  This is not a basic English literacy course.  It is aimed at learners 
who have basic English skills, but still need to increase their level of English 
communication considerably.  In order to so, EWP attempts to enhance 
English spelling and grammar.  Hence, within the context of this language 
program, English literacy refers to one?s level of English spelling and 
grammar. 
 
The primary outcome of the English Word Power (EWP) program is to enable 
the learner to write and speak English with grammatical integrity.  It also aims 
to enhance one?s vocabulary and comprehension (EWP, 2000).   
 
The program consists of two modules.  The first module aims to equip the 
learner with sound knowledge of English rules for word-usage, spelling and 
general vocabulary.  The second module teaches how to structure sentences 
in each of  the twelve different tenses.  The learner also learns how to 
structure sentences in both the active and passive voice (EWP, 2000). 
 
EWP has been structured to deliver effective results, whilst building self-
 confidence.  The course commences with relatively easy exercises, and gets 
progressively more difficult.  Each section starts with the learning of a rule, 
followed by an exercise to check progress, and ends with an assessment 
(EWP, 2000). 
 
Existing research on the relationship between English proficiency and 
academic performance may be classified into two groups.  The first group 
analyses the general relationship between the two variables, whereas the 
44  
second group focus on the relationship specifically amongst subjects who are 
the recipients of a bilingual education. According to Srivastava and 
Ramaswamy (cited in Padma, 2003) a bilingual education implies having a 
language other than one?s mother tongue as a medium of instruction.  
 
With regard to the first group, Padma (2003) cites Mukhopadhyay?s and 
Deka?s studies. Mukhopadhyay?s indicated that comprehensibility of language 
used in science textbooks is significantly related to different levels of science 
achievement.  Deka found that low proficiency in certain subjects including 
vocabulary and spelling, was significantly and positively related to school 
failure. Eggly, Musial and Smulowitz (1999) study of international graduated 
medical students using two tests of English proficiency, did not find scores on 
an objective test of medical knowledge related to English language skills. 
 
In terms of the second group of studies investigating the relationship between 
English proficiency and academic performance, samples discussed below 
include limited English proficient (LEP) students, refugees and immigrants, 
and people for whom English is not the only language of communication.  
Graham (1987) asserts that there are difficulties in determining the 
relationship between English language proficiency and academic success 
amongst students for whom English is a second language, and who are 
attending an institution where English is the medium of instruction. 
 
Students? inability to express themselves adequately in the medium of 
instruction has a negative effect on academic performance according to 
Hunter, and Stark and Lowther (cited by Engelbrecht, 1999).  The lack of 
understanding and comprehension of concepts and principles implies that a 
student will be unable to conceptualise content, and cannot follow the logical 
sequence of a cardinal principle or idea. In practical terms, the lack of 
language proficiency in the medium of instruction, means that the student will 
not be able to meaningfully be able to talk and write, listen and understand 
what has been said, substantiate one?s viewpoint, nor confidently read, write 
and communicate (Engelbrecht, 1999). 
 
45  
Hawson (1996) reviewed different perspectives on the relationship between 
academic performance and second-language learning, failing to find 
consensus on which factors affect academic outcomes for second language 
learners. 
 
Saville-Troike (1991) too, concluded that tests of English language proficiency 
alone are not related to academic performance for limited English proficient 
(LEP) students.  Academic competence is more related to knowing how to 
use language as a tool in acquiring knowledge and performing analytic 
processes, but those skills are more related to language competence in 
general. 
 
A study by Taylor (1978) investigated the relationship between oral English 
language proficiency and scholastic achievement among Mexican-Americans 
and Anglo-Americans. A significant relationship was found between English 
proficiency and scholastic achievement of these fifth grade students. 
 
Friedenberg and Curry (1981) did not find scores on the Michigan Test of 
English Language Proficiency a strong predictor of academic success.  
However, they found a significant relationship between English proficiency 
and grade point average amongst Cuban American students attending a 
bilingual teacher education programme in Florida.   
 
In Bosher and Rowekamp?s (1992) investigation of numerous factors that may 
be related to the academic success of refugees and immigrants in higher 
education, English language proficiency emerged as one of the most 
important predictors.  
 
3.2.1.2 Reading Skills 
 
In Tshwane University of Technology?s academic support programme, reading 
skills are enhanced using the Readers are Leaders program.  This program is 
not an initial reading program, but aims to develop essential reading skills, the 
two cardinal reading skills being reading speed and comprehension.  
46  
Developing reading speed or comprehension individually without the other 
promotes a mechanistic reading style.  Reading speed without 
comprehension is useless, and maximum comprehension inhibits optimal 
scholastic achievement (Fourie and Venter, 2000) 
 
Out of the four studies reviewed in this section, three studies found a 
correlation between reading skills and academic performance. 
 
At-risk college students with reading strategy training earned significantly 
higher grade point averages than those who received basic reading skills 
instruction in reading-intensive courses (Stallworth-Clark, 1996).  A study by 
Fields and Cosgrove (2001) examining the relationship between reading level 
and performance in a general psychology class found a significant 
relationship between reading level and grade-point average.  Celliers? (2000) 
research undertaken in her own practice revealed that reading enhanced 
general academic performance.  Reading programmes developed fluency, 
speed, vocabulary and comprehension, therefore overall language 
proficiency. 
 
Srivastava (1984) cited in Padma (2003) investigated the relationship 
between reading ability and school achievement using 480 boys and girls 
from public and private schools.  Srivastava?s results contradicted the other 
studies reviewed by failing to indicate any significant relationship between 
various measures of reading ability and achievement in different school 
subjects. 
 
3.2.2 Study Skills 
 
3.2.2.1  What do Study Skills entail? 
 
Study methods refer to the general planning, organising and work procedure 
adhered to attain an academic goal.  It encompasses learning skills and 
conditions for mastering the content.  Learning styles and learning strategies 
also form part of study methods (Kotz?, 1994). 
47  
 
Study methods are different from study habits.  Study habits are established 
patterns of acting that are learnt by repetition.  Positive study habits involve 
responsible learning with the aim of academic success.  Motivation and study 
attitude impact strongly on study habits but not on study methods (Kotz?, 
1994). 
 
Kotz? (1994) cites studies by Bell (1985), du Toit (1970), Margrain (1978), 
Monteith (1984, 1987), Watkins (1982) and Weigel and Weigel (1967) all 
found a positive relationship between study habits and academic 
performance.  Watkins (1982) found that study methods account for a 10% 
variance in academic performance at tertiary institutions. 
 
The Tshwane University of Technology Study Methods Course covers the 
specific modules discussed below, specifically time management, notetaking 
skills, memory skills, examination stress, and concentration.  These modules 
are also covered within the first year ?Life Skills? curriculum as part of 
academic skills development. 
 
3.2.2.2  Study Skills as a predictor of academic performance 
 
Kriel (2000) found that first year students who attended the ?Life Skills? course 
at Tshwane University of Technology, on average, performed better than 
those who had not attended the course in all eight subjects for which they 
were registered. 
 
3.2.2.3  Specific study skills as predictors of achievement 
 
? Time Management 
 
Time Management involves more than just being punctual.  It entails a sense 
of organisation and planning that predisposes a student to submitting tasks on 
time (Brown & Holtzman, 1983). 
 
48  
Studies investigating the relationship between time management and 
academic performance have provided some evidence for the existence of a 
significant relationship between the two variables.  These studies include 
research by Smith and Smith (1988), Zhang and Richarde (1999), Lahmers 
and Zulauf (2000), and Girija?s (1980) study (cited in Padma, 2003). 
 
Smith and Smith (1988) assert that successful learners are people who have 
succeeded in organising themselves in such a manner that they have realised 
their full potential as students.  Full time study varies from any other form of 
work in one important way:  in any other occupation a person has fixed times 
to work that keep the person busy most of the day.  However, at a tertiary 
institution, aside from the actual time of lectures and laboratories/practicals, 
there are no fixed times for working at an academic course.  It is entirely the 
learner?s responsibility to decide how to utilise the time optimally.  Failure to 
do so, results in not getting the most out of one?s studying.   
 
The causal structure between academic performance and cognitive, 
psychological and facilitative variables during the freshman year in college 
was studied by Zhang and Richarde (1999).  Data from 455 freshmen was 
used in the analysis.  Their structural equation model on Freshman Academic 
performance indicated that the ability to prioritise study was essential to 
freshmen academic performance. 
 
A recursive approach was utilised by Lahmers and Zulauf (2000) to identify 
factors associated with academic time use and academic performance for 79 
college students. A regression analysis revealed that time management skills 
and study time were associated with quarter grade point average. 
 
Girija?s (1980) study (cited in Padma, 2003) study of intellectual and non-
 intellectual factors in academic performance of students from professional 
colleges also included ?planning work efficiently? as a non-intellectual factor 
that served as a potential contributor to CGPA. 
 
 
49  
? Note-Taking 
 
Note-taking refers to the process of making notes from written communication 
or written messages. Within the context of this study, note-taking refers to 
making notes specifically from the textbook.  The student has to get his 
information from a list of books given to him by a lecturer, and has to isolate 
essential information from what is less important in the textbook (Mutua, 
Omulando & Otiende, 1992). 
 
Several researchers have focused on the impact of note-taking on learning.  
Mutua, et al, (1992), Brown (1991), Driskell (2003) and Hult (1984) reported 
positive effects resulting from the note-taking process. 
 
Mutua, et al (1992, p.35), asserts that good note-taking habits help to train a 
critical mind, assisting the learner to pass his examinations, and ?develop a 
discerning, astute and appreciative mind which is quick at grasping relevant 
information?.   Good note-taking also enhances memory, as the extraction and 
writing of the notes sets the important facts in the learner?s mind. 
 
Brown (1991) conducted a study to evaluate the impact of teaching a 
structured notetaking procedure to students.  Students who learned the 
technique reported higher success rates than students who had taken the 
same course the previous term.   
 
Driskell?s (2003) study also researched the effect of a system of guided 
notetaking on academic performance of entering University of Idaho freshmen 
with low predicted GPA?s.  Students were given verbal instruction, skills 
demonstration, class practice, and individual assignment in the notetaking.  
An analysis of grade point averages revealed that the instructional 
programme in notetaking had a significant positive impact on academic 
performance of the freshmen as a general group.  The program had an 
especially marked effect on male subjects, and on those with quantitative 
majors rather than verbal majors. 
 
50  
Hult (1984) attempted to assess if notetaking effectiveness was positively 
related to learning.  After a comparison of the notes of high and low effective 
notetakers, Hult concluded that note taking as an activity, and independent of 
note review, was positively correlated to learning. 
 
Contradicting the previous three studies, Howe?s (1976) review of studies on 
notetaking revealed no evidence that notetaking activities have a positive 
impact on learning.   Howe (1976) suggested that future research needs to 
emphasise types of notetaking strategies and the use students make of notes. 
 
? Memorisation skills 
 
Memorisation skills refer to the different methods, aids and strategies used to 
enhance recall of significant information (Mutua, et al, 1992).  The purpose of 
any assessment procedure is to test understanding.  This is demonstrated by 
a student?s ability to select only the facts and arguments that the question 
asks for.  For this a firm grasp of the essential body of knowledge of the 
subject is required, and this essential body of knowledge has to be 
remembered.  Memory skills improve one?s ability to remember information 
(Williams, 1989). 
 
Baehre & Gentile?s (1991) classroom evaluation of 58 eighth grade students 
also found that students using the keyword mnemonic technique resulted in 
improved performance when learning social studies facts.   
 
Memory was one of the seven modules in the University of Auckland?s 
Intensive Learning Skills Course for university students on academic 
probation. Students who attended the course improved significantly in their 
academic performance than students on probation who did not take the 
course (Manalo, 1996). 
 
 
 
 
51  
? Concentration Skills 
 
Concentration refers to mental tenacity.  It is the ability to give one?s mind an 
order and have the mind obey.  Furthermore, it is the discipline of focussing 
on a chosen activity and ignoring all irrelevant matters, as well as having the 
strength to impose on time, people, and events one?s priorities (Horn, 1991). 
 
The benefits of optimal concentration are an increased output, optimal 
performance, decrease in time spent studying, enhanced self-confidence, and 
a student can accomplish desired goals and transcend one?s abilities and 
circumstances (Horn, 1991). 
 
Concentration was also one of the modules in the University of Auckland?s 
Intensive Learning Skills Course where academic performance improved 
significantly after attending the course (Manalo, 1996). 
 
? Examination stress 
 
Examination stress can be positive or negative.  Positive stress is known as 
motivating stress.  It is characterised by greater mental alertness and 
enthusiasm, and results in enhanced academic performance.  Negative stress 
is paralysing stress.  This can result in a decline in health and affects different 
people in various ways.  This negative stress condition also leads to a decline 
in academic performance (Beaty & Barling, 1982). 
 
Stress is a two-stage process.  Firstly, a person becomes aware of the 
different physical sensations.  Secondly, the person interprets and labels each 
of the physical sensations.  Typical sensations are sweating, increased 
heart/pulse rate, panic, ?butterflies in the stomach?.  Whether one experiences 
these sensations as a help or barrier, determines whether the person labels 
the stress as motivating or paralysing (Beaty & Barling, 1982). 
 
Paralysing stress levels are regulated in two predominant ways in study 
methods programmes.  Firstly, through instruction in the use of relaxation 
52  
techniques, and secondly by equipping students with appropriate test-taking 
skills. 
 
Test-taking techniques deal with the mental and physical components of 
stress (Ellis, 1997).  Norton and Crowley (1995) questioned if students can be 
helped to learn how to learn by evaluating a British learning programme for 
first year degree students.  The workshops focused on essay writing and test-
 taking skills.  Regular attendees of the workshops had higher examination 
scores, but the results did not reveal as clear-cut results for differences in 
overall academic performance. 
 
Carrier?s (1984) study of 87 high school students indicated that anxiety was 
not necessarily detrimental to performance.  Griffore?s (1975) results 
supported the predicted relationship between achievement anxiety reaction 
type and academic performance, with the inverted-U hypothesis partially 
accounting for the relationship.   
 
Padma (2003) cites Girija?s (1980) comprehensive study, which also included 
performance anxiety as a possible factor affecting academic performance. 
 
3.3 The assessment of academic proficiencies 
 
This section reviews existing instruments assessing academic proficiencies.  
The Wrenn Study Habits Inventory, the Approaches to Studying Inventory, the 
Study Process Questionnaire and the Learning and Study Strategies 
Inventory (LASSI) and the Survey of Study Habits and Attitudes (SSHA) has 
been used in the assessment of study skills.   
 
3.3.1 Learning and Study Strategies Inventory (LASSI) 
 
3.3.1.1 Description of the Inventory 
 
The LASSI is a 10-scale, 80 item assessment of students? awareness of the 
use of learning and study strategies, related to skill, will and self-regulation 
53  
components of strategic learning.  The assessment covers both covert and 
overt thoughts, behaviours, attitudes and beliefs relevant to successful 
learning and that can be modified through appropriate educational 
interventions, like learning and study skills courses (Learning and Study 
Strategies Inventory, 2003). 
 
The LASSI provides standardised scores and national (American) norms for 
ten scales, namely attitude towards college, level of motivation, time 
management skills, and anxiety about academic performance, concentration 
level, information processing skills, and ability to select main ideas, use of 
study aids, test preparation abilities, and utilisation of test strategies.  Table 4 
(Appendix B) provides some indication of the reliability of the instrument 
(Electronic Learning and Study Strategies Inventory, 2003). 
 
The LASSI is both diagnostic and prescriptive.  It provides students with a 
diagnosis of their strengths and weaknesses, and is prescriptive as it provides 
feedback on areas where students may require intervention (Learning and 
Study Strategies Inventory, 2003). 
 
3.3.1.2 Learning and Study Strategies Inventory as a predictor of 
academic performance 
 
Rugsaken, Robertson and Jones (1998) investigated the efficacy of the LASSI 
as an additional predictor of college students? academic performance.  This 
study was very similar to the present research project.  Common variables 
included traditional predictors (high school scores), academic performance, 
and a study skills questionnaire.  The results of Rugsaken et al?s study 
indicated that the LASSI scores provided a slight, but not significant increase 
in predictability of academic performance. 
 
3.3.2 Survey of Study Habits and Attitudes (SSHA) 
 
The SSHA is basically the only assessment categorised as a psychometric 
test in South Africa in the field of evaluation of study methods.  However, the 
54  
reason it is no longer used as part of the academic development programme 
is because the score on the Study Habits scale only gives an overall score of 
the efficacy of study methods.  The scoring system does not specify which 
specific aspect of the study method is a problem.  Although valuable 
information may be gleaned from the assessment if individual responses to 
items are analysed qualitatively, this is logistically impractical and extremely 
time-consuming if done on a large-scale. 
 
3.3.3 Biggs?s Study Process Questionnaire (SPQ) 
 
3.3.3.1 Description of the Questionnaire 
 
Biggs? Study Process Questionnaire is a diagnostic tool that attempts to 
identify the learning approaches of students.  The SPQ consists of 42 items.  
It provides feedback on the learning approaches in three domains:  surface, 
deep and achieving domains (see Appendix A).  Each domain is further 
subdivided into motives and strategies used (Hua, Williams & Hoi, 2003). 
 
In order to ease facilitation and administration of the questionnaire, a web-
 based version of the questionnaire has been developed, that automatically 
generates the feedback to students and class profile reports to tutors (Hua, et 
al, 2003). 
 
3.3.3.2  Biggs?s Study Process Questionnaire as a predictor of 
academic performance 
 
The correlation of the SPQ with academic performance indicates that students 
classified as surface learners tend to have lower academic performance.  
However, the low-achieving students were not necessarily at risk of failing the 
course (mathematics).  Hence, the SPQ as a diagnostic tool to identify at-risk 
students is questionable.  The researchers contended that there were ?noise? 
factors prevalent, which were not accounted for in the study (Hua, et al, 
2003). 
 
55  
3.3.4 Approaches and Study Skills Inventory for Students 
 
3.3.4.1 Description of the Inventory 
 
The Approaches and Study Skills Inventory for Students appears to be very 
similar to the Biggs Study Process Questionnaire.  It also proposes three 
approaches to learning, surface, deep and strategic.  The Biggs instrument 
has chivvying as the third approach. 
 
3.3.4.2 Predicting academic performance 
 
Diseth and Martinsen (2003) analysed the relationship between the three 
approaches to learning and academic performance.  A sample of 192 
undergraduate psychology students with a mean age of 21.7 years 
participated.  Surprisingly, the deep approach did not predict achievement, 
while the other two approaches significantly predicted academic performance. 
 
An earlier study by Diseth (2002) found a curvilinear relationship between the 
surface approach and academic performance.  Multiple regression analysis 
showed interaction effects between deep-strategic and surface-strategic 
approaches to learning as predictors of academic performance, supporting 
the construct validity of approaches to learning. 
 
3.3.5 van Schoor?s Study Process Questionnaire (SPQ) 
 
Bearing the same name as the questionnaire by Biggs, it is however a 
different questionnaire.  The questionnaire is utilised by the University of 
South Africa?s Bureau for Student Counselling and Career Development.  
Unisa is a distance learning institution, and this questionnaire is appropriately 
web-based.  The SPQ assists a student to identify possible problem areas in 
the way he or she studies.  It is utilised to assess study habits, attitudes and 
environment, and can be used to prepare a study programme.  The 70 items 
evaluate organisation of the study process, time management, external 
interruptions in the study space, concentration, internal/external motivation, 
56  
examination skills, classroom behaviour, library skills, attitude towards 
learning, psychosomatic symptoms, personal problems, uncritical learning 
approach, study venue, and academic confidence (Study Process 
Questionnaire, 2002). 
 
3.3.6 Study Skills Questionnaire (SSQUES) 
 
3.3.6.1 Description of the Questionnaire 
 
The SSQUES is a 30-item questionnaire that assesses a student?s study skills 
in four areas:  reading comprehension, memorisation, concentration 
management and test taking (McCombs and Dobrovolny, 1980). 
 
3.3.6.2 SSQUES as a predictor of academic performance 
 
McCombs and Dobrovolny (1980) evaluated the potential reliability and 
predictive validity of the SSQUES in predicting student performance, and 
identifying students requiring some type of study skills remediation. The 
SSQUES demonstrated significant reliability, construct validity and predictive 
validity in predicting student achievement and to discriminate between 
students who would perform well as opposed to those who would perform 
poorly. 
 
3.4 Assessment in the South African context 
 
3.4.1 The utility of assessment  measures in contemporary South Africa 
 
Academics still maintain that testing has a significant role.  The need for tests 
in a multicultural South Africa is greater than elsewhere as valid assessment 
is a necessary condition for equity and the effective management of personal 
development.  Psychological assessment continues to  assist in decision-
 making processes provided it is used fairly, ethically and responsibly.  Testing 
is not regarded as perfect, but comparable to alternative assessment 
techniques utilised in selection, placement and guidance and contributes to a 
57  
more reliable, fair and cost-effective results (Plug, 1996). 
 
3.4.2 Escalating diversity and accelerated change 
 
Erasmus (1997) contends that the most critical issues pertaining to reliability 
and validity in testing in South Africa are escalating diversity and accelerated 
change. 
 
Erasmus (1997) stresses the role of cultures in the performance on 
psychometric tests.  He takes into account the references that suggest that 
Black Americans perform lower on intelligence tests than Whites and 
rationalises this as the tendency of American test constructors to select test 
items to favour the white culture since that culture represents the majority.  
This is very applicable to the South African situation too.  Instruments 
historically used in South Africa were usually imported from other Western 
countries, and thereafter adapted to predominantly accommodate South 
African Whites. 
 
The demographic character of contemporary tertiary institutions has altered 
drastically from that of the past.  There is a rapid decline in the test takers 
conforming to the norms of the 'adapted adopted' instruments.  A possible 
alternative to overcoming this test bias may be differential norms, where 
different cultural groups have different norms.  However, this may be viewed 
as discriminatory (Erasmus, 1997). 
 
Erasmus (1997) also discusses his research that indicates the speed at which 
test norms become outdated.  Performance of a given group of testees 
changes from one year to the next.  In the past environments were relatively 
stable, static and to a large degree homogenous. 
 
3.5 Conclusion 
 
Based on the studies reviewed in this chapter, a considerable body of 
research exists indicating the likelihood of language abilities and study skills 
58  
contributing to academic performance.  Numerous batteries of tests have also 
been developed to assess such proficiencies. Given the existence of other 
batteries of tests, the purpose of this study is to identify if, the Academic 
Proficiency Assessment?s (APA?s) psychometric properties validate its 
addition it to the portfolio of such assessments.  It does so by examining the 
internal consistency of each of the questionnaires in the APA.  
Recommendations are then made on how each questionnaire may be 
enhanced. 
 
The study then attempts to validate the APA as a predictor of academic 
achievement.  This is done by evaluating the extent to which each of the tests 
is a predictor of academic achievement by using a multiple regression 
analysis, and a discriminant function analysis.   
 
59  
 
 
Chapter 4 
METHODOLOGY 
 
 
4.1 Introduction 
 
This chapter addresses the methodology of the present research project.  The 
research questions are stated and variables specified, the sample and 
sampling method is elaborated upon, the instruments utilised are discussed, 
and the research design and procedure is explained.  Finally, the methods of 
statistical analyses are introduced. 
 
4.2 Research questions 
 
4.2.1  Are all the questionnaire of the Academic Proficiency  
Assessment (APA) battery reliable?  
 
This research question addressed the psychometric properties of the APA, 
with specific reference to the internal consistency reliability of the 
questionnaires forming part of the battery.  
 
The questionnaires that constituted the APA battery were: 
? English proficiency 
? Reading comprehension 
? Time management 
? Note-taking skills 
? Memorisation skills 
? Concentration skills 
? Achievement anxiety  (debilitating stress scale) 
? Achievement anxiety (motivating stress scale) 
60  
 
4.2.2 If any of the questionnaires in the APA are not reliable, what 
changes can be implemented to enhance its psychometric 
properties? 
 
After addressing the first research question, for those of the APA 
questionnaires listed below that were found to be unreliable, the second 
research question investigated the reasons for the inadequate reliability, and 
options to enhance the reliabilities of the relevant questionnaires: 
? English proficiency 
? Reading comprehension 
? Time management 
? Note-taking skills 
? Memorisation skills 
? Concentration skills 
? Achievement anxiety  (debilitating stress scale) 
? Achievement anxiety (motivating stress scale) 
 
4.2.3 What is the predictive validity of the tests in the APA battery for 
Information Technology students? 
 
This research question investigated the validity of the questionnaires in the 
APA battery as a predictor of academic performance for Information 
Technology students. 
 
The response variable was academic performance and the predictor variables 
were: 
? English proficiency 
? Reading comprehension 
? Time management 
? Note-taking skills 
? Memorisation skills 
? Concentration skills 
61  
? Debilitating stress  
? Motivating stress  
 
4.2.4 Are there additional variables that predict the academic 
performance of Information Technology students? 
 
Although the literature survey indicated that were many additional variables 
that could have been included in this research, the scope of the present study 
limited the investigation of additional predictors to biographical variables and 
Matric results.  The three biographical variables (age, gender and 
predominant language of communication) solicited in the biographical 
questionnaire were investigated as predictors of academic performance.  
Matric results were also included as a variable, as Matric has consistently 
emerged as the best predictor in a number of previous studies, and is also the 
primary selection tool for admission to the Information Technology course. 
 
This research question was added as an additional question, and not included 
with research question three as the sample size prevented statistical analysis 
with the chosen statistical techniques using all the variables (academic 
proficiencies, biographical variables and Matric results) in a single analysis.  
Most authors suggest that one should have at least ten to twenty times as 
many respondents as one has variables, otherwise the estimates of the 
regression line are probably very unstable and unlikely to replicate if one were 
to do the study over (StatSoft, Inc., 2001).  Although the size of the sample 
was 133 respondents, depending on the variables in question, due to the 
missing cases, the sample size ranged from 97 to 107.  This implied that a 
maximum of ten predictor variables could be included ina regression, whereas 
the study consists of thirteen predictor variables.  
 
4.2.5  An integrated regression model 
 
In order to consolidate the results of research questions three and four, an 
additional multiple regression analysis was executed, including as predictor 
62  
variables, the variables that were found significant in research questions three 
and four. 
 
4.3 Sample 
 
The type of sampling as well as the subjects who constitute the sample is 
discussed in this section. 
 
4.3.1  Sampling method 
 
A nonprobability convenience sampling method was used, as the sample had 
not been randomly selected. More specifically, purposive sampling had been 
used.  There had been an intentional attempt to obtain a representative 
sample by utilising a presumably typical group of students studying a 
particular course as the sample (Kerlinger, 1986). The present research 
project was the second phase investigating correlates of academic 
performance for Information Technology students at Tshwane University of 
Technology.  As SpEEx, the supplementary selection tool for admission to 
Tshwane University of Technology was found to have limited utility in the 
prediction of academic performance (Kriel, 2002), alternate predictors were 
sought, thereby having presented Information Technology students as a 
convenient sample. 
 
4.3.2  Description of the sample 
 
The sample constituted 133 first year Information Technology students at 
Tshwane University of Technology.  The sample was made up of 53 females 
and 79 males (Figure 4).  Afrikaans was the most spoken language amongst 
the Information Technology students.   Afrikaans was the predominant 
language of communication for 28 students, 18 were English, and the rest of 
the students? predominant language was one of the official African languages 
(Figure 5).  Home language was classified in this way, as the University's 
official languages at the time were English and Afrikaans.  Examinations could 
be written in one of the two languages. Most of the subjects were between 19 
63  
and 24 years (Figure 6).  However, only 23% of the sample was 18 or 19 
years old ? the age usually expected for first year students.  25% of the 
sample was older than 21, indicating that adult learners also made up a 
substantial proportion of the group. 
64  
 
 
 
 
 
 
 
 
 
  Figure 4.  Gender. 
Females Males
65  
   
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Figure 5.  Predominant language of communication. 
English Afrikaans African language
66  
 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Figure 4.  Age. 
 Figure 6.  Age. 
67  
 
4.4 Instruments 
 
Data for the research was obtained from three sources:  the APA battery 
database,  questionnaires eliciting biographical details and Matric results, and 
the Integrated Tertiary System database. 
 
4.4.1 Academic Proficiency Assessment battery 
 
The battery of questionnaires that constituted the Academic Proficiency 
Assessment (APA) battery were the primary focus of this study. 
 
4.4.1.1 English proficiency 
 
The English proficiency questionnaire is directly related to the ?English Word 
Power? English enhancement programme.  It consisted of 100 multiple-choice 
items testing spelling and grammar, and was a criterion-referenced 
achievement test. 
 
4.4.1.2 Reading skills 
 
The reading skills test had two components.  Reading speed was indicated in 
words per minute by timing how long a student took to read a passage on the 
appropriate grade level. The result of the reading speed test was the only 
norm-referenced item amongst all the questionnaires in the battery. 
Thereafter, a comprehension test on the passage was answered to assess 
understanding of the passage read.  The comprehension test was also a 
criterion-referenced achievement test. 
 
4.4.1.3 Time Management 
 
Time management was measured using a questionnaire adapted from the 
Delay Avoidance scale on the Survey of Study Habits and Attitudes (du Toit, 
1970).  
68  
 
4.4.1.4 Note-taking skills 
 
The use of appropriate note-taking and summarising skills were evaluated 
from an adaptation of Brown?s (1994) checklist. 
 
4.4.1.5 Memorisation skills 
 
The utilisation of memory techniques was assessed using a nine-item 
questionnaire (Horn, 1991). 
 
4.4.1.6 Concentration 
 
Concentration skills while studying were tested using Horn?s (1991) modified 
nine-item questionnaire. 
 
4.4.1.7 Examination stress 
 
Examination stress levels were identified using an adaptation of the 
Achievement Anxiety Test.  This questionnaire had two subscales namely 
paralysing exam stress and motivating exam stress (Alpert & Haber cited in 
Barling & Beaty, 1982). 
 
The dimensionality and utility of the AAT was investigated by Watson (1988).  
Her analyses suggested the viability of a unidimensional construct, and had 
potential for separating debilitating and facilitating test anxiety aspects. 
 
Plake?s (1981) investigation of the concurrent and predictive validity of the 
AAT raised concern about the quality of the difference score as a result of the 
nonindependence of the two scores.  Furthermore, the underlying structure of 
the instrument seems to be more complex than initially hypothesised.  Plake 
(1981, April) also examined the factor structure of the AAT.  A principal 
component factor analysis specifying a two-factor solution was selected to 
provide a check on the soundness of considering AAT as comprising two 
69  
independent factors.  Results of the two-factor varimax rotation showed that 
16 of the 19 items had sufficiently high loadings on either of the factors, using 
0.40 as the cut-off value. 
 
4.4.2 Biographical questionnaire 
 
Biographical details (age, gender and predominant language of 
communication) were obtained from a biographical questionnaire that was 
completed by the subjects during the same session that the APA battery was 
completed.  The biographical questionnaire also included a consent section, 
which asked the subject if the researcher could utilise his/her Matric, APA and 
course results for research purposes. 
 
4.4.3 Integrated Tertiary System (ITS) 
 
The ITS is the central database utilised by Tshwane University of Technology.   
Course results were obtained from this database. 
 
4.4.3.1 Matric results 
 
Three details of the Matric results were captured:  subjects, symbols and 
grades (higher/standard/lower).   
 
The conventional methods of quantifying matriculation symbols are by 
conversion to an M-score or by using the Swedish Formula. For the present 
study, the matriculation symbols were converted using the Swedish formulae, 
and aggregated.  The results were obtained for six subjects were utilised.  If a 
student had results available for more than six subjects, the lowest scores 
were omitted. 
 
  
 
 
70  
 
Table 5 
 Quantification of matriculation symbols using the Swedish formula 
Matric symbols HG SG 
A 8 6 
B 7 5 
C 6 4 
D 5 3 
E 4 2 
F 3 1 
 
 
4.4.3.1 Course results 
 
In the first year of studying Information Technology at Tshwane University of 
Technology, eight modules were usually studied.  These modules spanned 
four subjects, the first module of the subject being studied in the first 
semester, and the second module in the subsequent semester.  The four 
subjects were Information Systems, Development Software, Systems 
Software and Information Technology Skills. A continuous measure of 
academic performance was the cumulative grade point average (CGPA) 
which was obtained from the mean of the results of the eight modules.  
 
However, the CGPA may not necessarily be the best measure of academic 
performance, particularly when selecting for remedial intervention.  When 
selecting for remedial intervention, it may be necessary to differentiate 
between successful and unsuccessful students, as unsuccessful students are 
likely to require remediation.  A dichotomous measure of academic 
performance was categorising students as successful or unsuccessful. A 
successful student would be a student likely to obtain the Information 
Technology qualification in three years, which is the minimum duration of the 
course.  A student was coded as successful if he/she passed six, seven or 
eight of the eight subjects, as such a student could possibly qualify in three 
71  
years without remedial intervention.  A student was coded as unsuccessful if 
zero to five of the eight subjects were passed.  Students were classified in this 
manner, as this would 
 
Swanepoel (2002) and Smith?s (1979) assumptions are adopted for the 
present study: 
1. It is assumed that a student?s final mark for each subject, is an objective 
measure of academic performance.   
2. The mark obtained in a specific subject is equivalent to the same mark 
obtained in another subject.   
3. The degree of difficulty of the four different subjects is the same.   
 
4.5      Design and Procedure 
 
The research design and procedure for conducting the study are explained in 
this section 
 
4.5.1 Design  
 
This was a non-experimental, correlational and descriptive study.   These 
concepts will be elaborated upon in this section. 
 
4.5.1.1 Non-experimental 
 
The study was non-experimental, as the independent variable (academic 
performance) was not manipulated, there was no control over extraneous 
variables, and there was no random assignment of subjects to different 
conditions. The role of the researcher was essentially that of an observer who 
was attempting to quantify the relationship between academic performance 
and the various independent variables (academic proficiencies, biographical 
variables and Matric results) to determine if there was a correlation.  No 
experimental treatment was executed (Potter, 1998).  
 
72  
This type of research is especially useful for descriptive and construct-
 seeking, rather than for inferential purposes.  Its explanatory power can be 
enhanced through cross validation/replication.  However, the disadvantage of 
this design is that there are many potential threats to the internal validity of the 
study, like the third variable problem, and problems due to causal arrow 
ambiguity (Potter, 1998). 
 
4.5.1.2 Correlational 
 
Correlation refers to a co- or joint relationship between or amongst two 
variables.  Correlational research entails obtaining pairs of observations on a 
group of subjects, comparing the scores, and calculating an index 
representing the relationship between the two sets of scores.  In the present 
study, a relationship between academic performance and the various 
independent variables (academic proficiencies, biographical variables and 
Matric results) was sought.  The relationships found indicated a relationship of 
association rather than of causation (Potter, 1998).    
 
The present study was one approach to a correlational design ? a longitudinal 
design.  A longitudinal design is usually used for predictive validation 
purposes.  The predictor variables, the APA battery was administered at the 
beginning of the academic year, and the response variables were obtained at 
a different point in time, that is the subject results were obtained at the end of 
the year.  This form of prediction should be utilised cautiously.  This design is 
criticised due to lack of randomisation, control over variables by the 
researcher, and the absence of a control group, and history was a genuine 
threat to internal validity.  As the literature survey indicated, a myriad of other 
factors have been linked to academic performance. Only a few of those 
factors have been investigated in this study.  It is likely, and even probable, 
that some of those additional factors may have affected student academic 
performance during the duration of the study (Potter, 1998). 
 
 
 
73  
4.5.1.3 Descriptive 
 
Descriptive research has an investigatory focus that tends to attempt to 
carefully map out a situation or set of events in order to describe what is 
occurring behaviourally.  This type of focus is usually considered a necessary 
first step in the development of a programme of research, as it establishes the 
foundation of any future ventures.  However, it is seldom regarded as 
sufficient, as a point will be reached where the reasons and processes 
underlying the behaviours will be questioned (Rosenthal & Rosnow, 1991).   
 
This study is descriptive, as it aimed to depict the psychometric properties of 
the APA.  It is the starting point for further research into enhancing and 
validating the APA. 
 
4.5.2 Procedure 
 
The Academic Proficiency Assessment battery was administered to the 
sample group when they commenced with their course.  A biographical 
questionnaire was also administered during the testing session.  At the end of 
the academic year, academic performance was correlated with the 
biographical variables and the APA battery. Students who had received any 
remedial intervention during the year were excluded from the study, as their 
APA scores were likely to have changed after the intervention. 
 
4.6 Statistical analyses 
 
A variety of statistical methods were utilised to analyse the data.  The data 
were processed using Enterprise Guide. 
 
4.6.1 Statistical methods for research question 1 
 
In order to assess the internal consistency reliability of the questionnaires, 
Cronbach?s coefficient alpha (r?) and the Kuder-Richardson formula 20 (KR-
 20) were utilised. 
74  
 
4.6.1.1 Cronbach?s coefficient alpha (r?) 
 
There are two basic approaches of estimating the reliability of an individual 
test without developing alternate forms and without administering the test 
twice to the same subjects.  The first approach correlates the results from one 
half of the test with other half of the test, and is referred to as split-half 
reliability. The second approach, which was the one utilised in the present 
study, examines the internal consistency of individual test items, determining 
whether the items tend to measure the same factors (Gregory, 1996). 
 
Cronbach (1951) criticised the reliance on a single split, advocating taking a 
more typical value such as the mean of the split-half coefficients resulting 
from all possible splittings of a test to estimate the reliability of a psychological 
test.  Coefficient alpha may be regarded as the mean of all possible split-half 
coefficients, corrected by the Spearman-Brown formula.  Coefficient alpha 
ranges from 0 to 1.00 (Gregory, 1996). 
 
A high coefficient alpha implies that not matter how the test is split, scores on 
the two halves almost always correlate to a very high degree.  This would 
indicate that the individual test items must be homogenous, that is, the items 
tend to measure a single trait.  If the individual items were multifactorial, then 
many of the possible split-half combinations would include large numbers off 
one trait at the expense of the other traits measured by the test.  In those 
cases, the two split halves would measure different characteristics, correlating 
minimally or not at all, lowering the average split-half correlation among all 
possible half-tests.  Hence, coefficient alpha may be regarded as an index of 
the degree to which a test measures a single factor.  Homogenous tests that 
are unifactorial yield high values of coefficient alpha.  Therefore, coefficient 
alpha is referred to as an index of internal consistency (Gregory, 1996). 
 
To assess internal consistency reliability of the English Proficiency of the Time 
Management, Note-Taking Skills, and Memorisation Skills, Concentration 
Skills, and Achievement anxiety questionnaires, Cronbach?s coefficient alpha 
75  
(r?) was used as an index of internal consistency. For the Achievement 
Anxiety questionnaire, the debilitating stress and motivating stress scales 
were analysed separately.  Other factors held constant, coefficient alpha will 
be high to the extent that many items are included in the scale, and the items 
are highly correlated with one another (Hatcher, 1994).   
 
4.6.1.2 Kuder-Richardson formula 20 (KR-20) 
 
The scores on the Reading Comprehension questionnaire were on a nominal 
scale, where responses were scored either incorrect (zero) or correct (one).  
Hence, Kuder-Richardson formula 20 (KR-20) was a more appropriate 
measure of internal consistency. The KR-20 formula is relevant to the special 
case where each test item is scored 0 or 1 (e.g., wrong or right). Coefficient 
alpha is basically the general application of the KR-20 formula, which is a 
more specific formula (Gregory, 1996). 
 
4.6.1.3 Interpretation of reliability coefficients 
 
What is an acceptable level of reliability?  There is no clear-cut answer to this 
question.  Many authors suggest that reliability should be at least .90 for 
decisions about individuals, with .95 being the desired standared.  Hatcher 
(1994) mentions Nunnally?s (1978) widely used rule of thumb of .70, and 
many standard tests with reliabilities as low as .70 have proven to be very 
useful.  Tests with even lower reliabilities have been useful in research 
(Gregory, 1996).  Owen and Taljaard (1996) state a reliability coefficient of .60 
can provide useful information as long as the test results are interpreted 
appropriately.  On the contrary, Galpin (2001) contends that a value under 0.6 
is regarded as indicating poor reliability.  Thorndike (1951) asserts that 
reliability coefficients as low as .40 or .50 are even acceptable if urgent 
decisions need to be made, and the only measurements available are 
unreliable.  For the purpose of this study, Nunally?s general rule of thumb of 
.70 will be applied as the acceptable standard of reliability. 
 
 
76  
4.6.2 Statistical methods for research question 2 
 
An analysis of the Cronbach?s coefficient alpha and the item-difficulty values 
for the two achievement tests (English proficiency and Reading 
comprehension) provided the guidelines for recommendations for enhancing 
the internal consistency of the questionnaires. 
 
4.6.2.1 Cronbach?s coefficient alpha 
 
An inadequate coefficient alpha, may be improved by dropping from the scale 
those items that demonstrate poor item-total correlations.  An item-total 
correlation is the correlation between the individual item, and the sum of the 
remaining items in the scale.  A low item-total correlation is regarded as 
evidence that the item is not measuring the same construct that is measured 
by the items in the scale (Hatcher, 1994). 
 
4.6.2.2 Item difficulty 
 
The difficulty value of an item (pi) is the proportion or percentage of individuals 
who answer the items correctly. 
 
Number of people who answered the item correctly 
       pI      =       
Number of people who took the measure 
        X 100 
 
The higher the percentage of correct responses, the easier the item.  The 
optimal level of item difficulty is 50%, ranging between 30% and 70%, in order 
to for the test to maximise the information provided about the differences 
between testees.  The exception to this rule is true/false or multiple-choice 
items.  The optimal level of item difficulty for these two formats is computed 
from the formula: 
 
 
 
77  
1.0  + g 
Optimal level of   pI      =       
2 
        X 100 
   
1 
                  where        g   = Chance success level  = 
Number of options 
 
 
4.6.3  Statistical methods for research question 3 
 
This research question addressed the predictive validity of the APA battery as 
a predictor of academic performance. In a predictive validation study, scores 
are used to estimate measures obtained at a later time.  The criterion 
measures are obtained in the future, usually months or years after the test 
scores are obtained.  Predictive validity is one approach to criterion-related 
validity. Criterion-related validity is exhibited when a test is shown to be 
effective in estimating a testee?s performance on some outcome measure.  
The variable of primary interest is the outcome measure, regarded as the 
criterion.  The test score is useful to the extent that it accurately predicts the 
criterion.  In this case the criterion is academic performance which is obtained 
almost eleven months after the APA questionnaires are administered  
(Gregory, 1996). 
 
4.6.3.1 Multiple regression 
 
When tests are utilised for prediction purposes, it is necessary to develop a 
regression equation that describes the best-fitting straight line for estimating 
academic performance from the APA?s scores.  The relationship between the 
APA?s scores and academic performance may be expressed in various ways.  
The most common approach is to compute the correlation between the two 
(rxy).  This is known as the validity coefficient.  The higher the validity 
coefficient rxy the more accurate the APA battery is in predicting academic 
performance (Gregory, 1996). In order to compute this validity coefficient, 
multiple regression analyses was used to analyse the data captured for 
determining the predictive validity of the APA battery. The multiple regression 
78  
procedure comprised two main parts.  In the first part of the analysis, a 
correlation matrix was extracted.  The second part entailed performing the 
actual multiple regression analyses (StatSoft, Inc., 2001). 
 
? Correlation 
 
Correlation is an index of the relationship between two variables.  Pearson?s r, 
is the most widely used correlation coefficient .  r ranges from ?1.00 to +1.00.  
(StatSoft, Inc., 2001).  Hatcher (1994) provides an approximate guide for 
interpreting the strength of the relationship between two variables: 
  ?1.00 =  perfect correlation 
  ?.80   =  strong correlation 
  ?.50   =  moderate correlation 
  ?.20   =  weak correlation 
  ?.00   =  no correlation 
 
The assumptions for Pearson?s r are that both the predictor and criterion 
variables should be at least an interval level of measurement, there should be 
random sampling, a linear relationship between the criterion variable and 
predictor variable, and the pairs of scores should follow a bivariate normal 
distribution (Hatcher, 1994). 
 
Two or more explanatory variables in a regression model being highly 
correlated is referred to as multicollinearity.  Multicollinearity makes it difficult 
to separate the individual effects of the correlated variables on the dependent 
variable (Vogiatzi, 2002).  One way to assess the extent of the 
multicollinearity between two variables is to compute the simple correlation 
coefficient between two explanatory variables.  When two or more 
independent variables in a regression model are correlated, they contribute 
redundant information.  Hence, even though each independent variable 
contributes information for describing and predicting the dependent variable, 
some of the information overlaps (Bowerman & O? Connell, 1990). 
 
79  
Multicollinearity is not a problem if the aim of the analysis is to simply predict 
Y from a set of X variables.  The predictions will still be accurate, and the R2adj 
will still quantify how well the model predicts the Y values.  However, if it is 
imperative to understand the individual contribution of each X variable to the 
prediction of Y, then multicollinearity is a serious problem.  One problem is 
that the individual p value can be high, even though the variable is important.  
So, a variable may be excluded from the model although it may possibly be 
significant.  A second problem is that the confidence intervals on the 
regression coefficients will be very wide, and may even include zero.  Thus, 
one cannot even be confident whether an increase in the X value is related to 
a change in Y (Motulsky, 2002). 
 
? Multiple regression equation 
 
The actual multiple regression analysis involves finding a regression equation 
to predict academic performance on the basis of a number of predictors (the 
questionnaires in the APA battery). The equation is represented as: 
 
y = a + b1x1 + b2x2 + b3x3 + ... + bnxn 
 
 
where  y  is the dependent variable, a is the intercept, b is the coefficient of 
the respective x, and x is the independent variable/predictor/regressor.  The 
coefficient indicates what relationship the specific variable has with the 
dependent variable, when all the other independent variables are held 
constant (Howell, 1992; Lea, 1997). 
 
? Assumptions underlying multiple regression 
 
The assumptions for multiple regression are that all the variables are in the 
equation, errors are uncorrelated, errors have the same variance throughout, 
there is no change of regime, there are no outliers, there are no influential 
points, and residuals are normally distributed (Galpin, 2001). 
 
 
80  
? Model and variable selection 
 
When there is a large set of independent variables, multiple regression 
attempts to determine which subset of them are most useful for predicting the 
dependent variable.  A screening procedure is required to identify the most 
important subset of independent variables. The stepwise selection procedure 
is a useful technique of variable selection in the exploratory phases of 
research, or for the purpose of prediction.  Stepwise methods include forward 
selection, backward elimination and the combined method (Lea, 1997;  
Menard, 1995). 
 
For the present analysis a forward selection model was utilised.  In a forward 
selection model, all simple linear regression models involving only one 
independent variable at a time are considered, in order to determine which 
gives the best regression.  That variable is then included in the regression 
equation.  Once that variable is entered into the regression model, it is never 
removed.  The remaining variables are then considered in turn to identify 
which variable gives the greatest improvement in the fit of the regression 
equation, given the variables already in the model.  At the second step, the 
variable is included which in conjunction with the variable already in the 
model, provides the greatest improvement over the best equation with only 
the variable selected in the first step.  This process continues step by step, 
with the variable about to be entered into the model being tested at each step, 
to determine if it will have a significant regression coefficient on being 
included in the equation.  If not, the process terminates (Galpin, 2001). 
 
Hosmer and Lemeshow (1999) contend that the criterion for including a 
variable in a model varies amongst scientific disciplines.  The aim is to 
minimise the number of variables in a model so that the resultant model may 
be numerically stable.  Rothman and Greenland (1998), on the other hand, 
advocate including all or many intuitively relevant variables, irrespective of 
their statistical significance.  This may lead to the model being ?overfit? 
resulting in a numerically unstable estimate. 
 
81  
The forward selection model was preferred for the present study, as it is 
important to identify the contribution of each predictor variable to the 
regression model.  By being able to determine each individual predictor?s 
contribution to the regression model, it enables selection of the variables with 
the highest validity to be retained in the final APA battery (Kriel, 2001). 
 
? Goodness of fit 
 
Having found the best straight line, the next consideration was how well it 
described the data.  Goodness of fit may be assessed from the R2- adjusted 
(R2adj) or the F-value. 
 
The R2adj ranges from zero to one, but is usually represented as a percentage 
in practice. It indicates how well the regressors taken together explain the 
variation in the dependent variable.  Lea (1997) provides the following rough 
guidelines for interpretation of the R2adj value: 
  > 90%         = rare in psychological data - questionable 
  75% ? 90% = very good 
  50% ? 74% = good 
  25% ? 49% = fair 
  < 25%         = poor, perhaps unacceptable 
 
The F statistic in the ANOVA portion of the regression output indicates if  the 
regressors taken together are significantly associated with the dependent 
variable. The higher the F value, the more significant it will be for given 
degrees of freedom. 
 
4.6.3.2 Discriminant function analysis 
 
This analysis was included in addition to the multiple regression analysis 
because CGPA may not necessarily be the optimal indicator of academic 
achievement, particularly when selecting for remedial intervention.  When 
selecting students requiring remedial aid, there may be a need to discriminate 
between successful and unsuccessful students, which renders the dependent 
82  
variable dichotomous.  In such a case multiple regression is not a suitable 
technique as the dependent variable has to be continuous in order to use 
multiple regression, and discriminant function analysis is appropriate. 
 
Discriminant function analysis is a special case of multiple regression in which 
there is a dichotomous variable.  The discriminant function is the set of 
optimal weights that best discriminates whether subjects are members of one 
or another group, given the predictor variable (Rosenthal & Rosnow, 1991). 
 
A backward elimination method was utilised.  With the backward elimination 
method, all variables are brought into the equation.  The coefficients of each 
variable are examined individually, and the variable with the least significant 
coefficient is eliminated. The equation is recalculated, and the process 
repeated.  The process stops when all coefficients are significant at the 
desired level (Galpin, 2001).  This method is considered a reasonable 
procedure, specifically for researchers who prefer to commence with all the 
possible independent variables in the model, and not overlook any important 
variables (Bowerman & O? Connell, 1990). 
 
4.6.4  Statistical methods for research question 4 
 
Multiple regression analysis was also used for the fourth research question, 
as it entailed finding a regression equation to predict academic performance 
on the basis of a number of predictors.  In this research question the criterion 
variable was CGPA, and the predictor variables were age, gender, and 
predominant language of communication and Matric results.  As gender and 
predominant language of communication were on a nominal scale, and one of 
the assumptions of multiple regression is that the variables should be at least 
on an interval level of measurement, dummy variables will be used. 
 
4.6.5  An integrated regression model 
 
As the sample size was too small to include the predictor variables from 
research questions three and four in a single multiple regression analysis, 
83  
another set of statistical analyses is executed in addition to those performed 
for research questions three and four.   This multiple regression analysis 
contained as predictor variables, the variables that emerged as significant in 
the previous two research questions.  The response variable was CGPA. 
 
4.7 Conclusion 
 
In answering the research questions, the assumption was that unreliable 
questionnaires would have limited value for predictive purposes. The first level 
of analysis thus focused on the reliability of the tests used. Unreliable 
questionnaires were analysed to indicate areas in which each instrument 
could be improved. 
 
Once the reliability of each of the tests had been established, the second level 
of analysis involved examination of the questionnaires as individual 
predictors, to establish the amount of variance in academic performance 
which could  be predicted through their individual use. The predictive value of 
the questionnaires was then considered in combination. As not all the 
questionnaires may have been reliable, the predictive value of additional 
variables (biographical variables and Matric results) was also established.  
 
The final level of analysis involved drawing conclusions relating to future use 
of the instruments in the battery. This was based on examination of the 
reliability and validity analyses of each test, and an assessment of their  
predictive value.    
 
Table 6 provides a summary of the research design.
84
  
 
Ta
 bl
 e
  
6 
R
 e
 se
 ar
 ch
  
de
 si
 gn
  
ta
 bl
 e
  
R
 es
 ea
 rc
 h 
qu
 es
 tio
 n
  
D
 a
 ta
  
In
 ve
 st
 ig
 a
 tin
 g 
St
 a
 tis
 tic
 a
 l 
a
 n
 a
 lys
 is
  
In
 te
 rp
 re
 ta
 tio
 n
  
1 
A
 re
  
al
 l t
 he
  
qu
 es
 tio
 n
 n
 ai
 re
  
o
 f t
 he
  
Ac
 ad
 em
 ic
  
Pr
 o
 fic
 ie
 n
 cy
  
As
 se
 ss
 m
 en
 t 
(A
 PA
 ) b
 at
 te
 ry
  
re
 lia
 bl
 e?
  
 
Ea
 ch
  
ite
 m
 ?s
  
sc
 o
 re
  
in
  
e
 a
 ch
  
o
 f t
 he
  
fo
 llo
 w
 in
 g 
te
 st
 s:
  
?
  
En
 gl
 is
 h 
pr
 o
 fic
 ie
 nc
 y 
?
  
R
 e
 a
 di
 n
 g 
co
 m
 pr
 e
 he
 ns
 io
 n
  
?
  
Ti
 m
 e
  
m
 a
 n
 a
 ge
 m
 e
 n
 t 
?
  
N
 o
 te
 -
 ta
 kin
 g 
sk
 ills
  
?
  
M
 e
 m
 o
 ris
 a
 tio
 n
  
sk
 ills
  
?
  
Co
 n
 ce
 n
 tra
 tio
 n
  
sk
 ills
  
?
  
Ac
 hi
 e
 ve
 m
 e
 n
 t a
 n
 xi
 e
 ty
  
 
(de
 bi
 lit
 a
 tin
 g 
st
 re
 ss
 ) 
?
  
Ac
 hi
 e
 ve
 m
 e
 n
 t a
 n
 xi
 e
 ty
  
(m
 o
 tiv
 a
 tin
 g 
st
 re
 ss
 ) 
 
In
 te
 rn
 a
 l 
co
 ns
 is
 te
 nc
 y 
re
 lia
 bi
 lit
 y 
Cr
 o
 n
 ba
 ch
 ?s
  
co
 e
 ffi
 ci
 e
 n
 t 
a
 lp
 ha
  
 KR
 -
 20
  
fo
 r 
Re
 a
 di
 n
 g 
Co
 m
 p.
  
o
 n
 ly 
.
 70
  
=
  
ac
 ce
 pt
 ab
 le
  
 
2 
If 
an
 y 
o
 f t
 he
  
qu
 es
 tio
 n
 n
 a
 ire
 s 
in
  
th
 e
  
AP
 A 
ar
 e 
n
 o
 t r
 el
 ia
 bl
 e,
  
w
 ha
 t 
ch
 a
 n
 ge
 s
  
ca
 n
  
be
  
im
 pl
 em
 en
 te
 d 
to
  
en
 ha
 n
 ce
  
its
  
ps
 yc
 ho
 m
 et
 ric
  
pr
 o
 pe
 rt
 ie
 s?
  
Ea
 ch
  
ite
 m
 ?s
  
sc
 o
 re
  
in
  
e
 a
 ch
  
o
 f t
 he
  
fo
 llo
 w
 in
 g 
u
 n
 re
 lia
 bl
 e
  
te
 st
 s:
  
?
  
En
 gl
 is
 h 
pr
 o
 fic
 ie
 nc
 y 
?
  
R
 e
 a
 di
 n
 g 
co
 m
 pr
 e
 he
 ns
 io
 n
  
?
  
M
 e
 m
 o
 ris
 a
 tio
 n
  
sk
 ills
  
?
  
Co
 n
 ce
 n
 tra
 tio
 n
  
sk
 ills
  
?
  
Ac
 hi
 e
 ve
 m
 e
 n
 t a
 n
 xi
 e
 ty
  
(m
 o
 tiv
 a
 tin
 g 
st
 re
 ss
 ) 
 
Ite
 m
 -
  
to
 ta
 l 
co
 rr
 e
 la
 tio
 n
  
      Ite
 m
  
di
 ffi
 cu
 lty
  
Cr
 o
 n
 ba
 ch
 ?s
  
co
 e
 ffi
 ci
 e
 n
 t 
a
 lp
 ha
  
 KR
 -
 20
  
fo
 r 
Re
 a
 di
 n
 g 
Co
 m
 p.
  
 Ite
 m
  
di
 ffi
 cu
 lty
  
(p I
  
) v
 a
 lu
 e
  
.
 70
  
=
  
ac
 ce
 pt
 ab
 le
  
       O
 pt
 im
 a
 l 
le
 v
 e
 l 
o
 f 
p I 
fo
 rm
 u
 la
  
85
  
 
R
 es
 ea
 rc
 h 
qu
 es
 tio
 n
  
D
 a
 ta
  
In
 ve
 st
 ig
 a
 tin
 g 
St
 a
 tis
 tic
 a
 l 
a
 n
 a
 lys
 is
  
In
 te
 rp
 re
 ta
 tio
 n
  
3 
W
 ha
 t i
 s 
its
  
pr
 ed
 ic
 tiv
 e
  
va
 lid
 ity
  
o
 f t
 he
  
qu
 es
 tio
 n
 n
 a
 ire
 s 
in
  
th
 e
  
AP
 A 
ba
 tte
 ry
  
fo
 r 
In
 fo
 rm
 at
 io
 n
  
Te
 c
 hn
 o
 lo
 gy
  
st
 u
 de
 n
 ts
 ? 
 
Re
 sp
 on
 se
  
v
 ar
 ia
 bl
 e
 : 
 
CG
 PA
  
 Pr
 e
 di
 ct
 o
 r 
va
 ria
 bl
 es
  
(al
 l t
 he
  
to
 ta
 l 
sc
 or
 es
  
o
 n
  
th
 e
  
te
 st
 s 
in
  
th
 e
  
AP
 A)
 : 
?
  
En
 gl
 is
 h 
pr
 o
 fic
 ie
 nc
 y 
?
  
R
 e
 a
 di
 n
 g 
co
 m
 pr
 e
 he
 ns
 io
 n
  
?
  
Ti
 m
 e
  
m
 a
 n
 a
 ge
 m
 e
 n
 t 
?
  
N
 o
 te
 -
 ta
 kin
 g 
sk
 ills
  
?
  
M
 e
 m
 o
 ris
 a
 tio
 n
  
sk
 ills
  
?
  
Co
 n
 ce
 n
 tra
 tio
 n
  
sk
 ills
  
?
  
D
 e
 bi
 lit
 a
 tin
 g 
st
 re
 ss
  
?
  
M
 o
 tiv
 a
 tin
 g 
st
 re
 ss
  
Pr
 e
 di
 ct
 iv
 e
  
va
 lid
 ity
  
In
 te
 r-
 co
 rr
 e
 la
 tio
 n
 s 
    M
 u
 lti
 pl
 e
  
re
 gr
 e
 ss
 io
 n
  
a
 n
 a
 lys
 is
  
 Fo
 rw
 a
 rd
  
se
 le
 ct
 io
 n
  
 
?
 1.
 00
  
=
  
pe
 rfe
 ct
  
 
?
 .
 80
  
 
 
=
  
 
st
 ro
 n
 g 
 
?
 .
 50
  
 
 
=
  
 
m
 o
 de
 ra
 te
  
 
?
 .
 20
  
 
 
=
  
 
w
 e
 a
 k 
 
?
 .
 00
  
 
 
=
  
 
n
 o
  
co
 rr
 e
 la
 tio
 n
  
 
 >
  
90
 %
  
=
  
qu
 es
 tio
 n
 a
 bl
 e
  
75
 %
  
?
  
90
 %
  
=
  
ve
 ry
  
go
 o
 d 
50
 %
  
?
  
74
 %
  
=
  
go
 o
 d 
25
 %
  
?
  
49
 %
  
=
  
fa
 ir 
<
  
25
 %
  
=
  
po
 or
  
 
 
Re
 sp
 on
 se
  
v
 ar
 ia
 bl
 e
 : 
 
Su
 cc
 es
 sf
 u
 l 
st
 u
 de
 n
 t (p
 a
 ss
 e
 d 
6/
 7/
 8 
su
 bje
 ct
 s) 
 Pr
 e
 di
 ct
 o
 r 
va
 ria
 bl
 es
 : 
Al
 l t
 he
  
to
 ta
 l 
sc
 or
 es
  
o
 n
  
th
 e
  
te
 st
 s 
in
  
th
 e
  
AP
 A 
Pr
 e
 di
 ct
 iv
 e
  
va
 lid
 ity
  
Di
 sc
 rim
 in
 a
 n
 t 
fu
 nc
 tio
 n
  
a
 n
 a
 lys
 is
  
 Ba
 ck
 w
 a
 rd
  
st
 e
 pw
 is
 e
  
>
  
90
 %
  
=
  
qu
 es
 tio
 n
 a
 bl
 e
  
75
 %
  
?
  
90
 %
  
=
  
ve
 ry
  
go
 o
 d 
50
 %
  
?
  
74
 %
  
=
  
go
 o
 d 
25
 %
  
?
  
49
 %
  
=
  
fa
 ir 
<
  
25
 %
  
=
  
po
 or
  
 
4 
A
 re
  
th
 er
 e
  
an
 y 
ad
 di
 tio
 n
 al
  
va
 ria
 bl
 es
  
th
 at
  
en
 ha
 n
 ce
  
th
 e 
pr
 ed
 ic
 tiv
 e
  
va
 lid
 ity
  
fo
 r 
In
 fo
 rm
 a
 tio
 n
  
Te
 ch
 n
 o
 lo
 gy
  
st
 u
 de
 n
 ts
 ? 
Re
 sp
 on
 se
  
v
 ar
 ia
 bl
 e
 : 
 
CG
 PA
  
 Pr
 e
 di
 ct
 o
 r 
va
 ria
 bl
 es
 : 
?
  
M
 a
 tri
 c 
re
 su
 lts
  
?
  
Ag
 e
  
?
  
G
 e
 n
 de
 r 
?
  
La
 n
 gu
 a
 ge
  
Pr
 e
 di
 ct
 iv
 e
  
va
 lid
 ity
  
M
 u
 lti
 pl
 e
  
re
 gr
 e
 ss
 io
 n
  
a
 n
 a
 lys
 is
  
 Fo
 rw
 a
 rd
  
se
 le
 ct
 io
 n
  
>
  
90
 %
  
=
  
qu
 es
 tio
 n
 a
 bl
 e
  
75
 %
  
?
  
90
 %
  
=
  
ve
 ry
  
go
 o
 d 
50
 %
  
?
  
74
 %
  
=
  
go
 o
 d 
25
 %
  
?
  
49
 %
  
=
  
fa
 ir 
<
  
25
 %
  
=
  
po
 or
  
86
  
 
 
R
 e
 se
 ar
 ch
  
qu
 es
 tio
 n
  
D
 a
 ta
  
In
 ve
 st
 ig
 a
 tin
 g 
St
 a
 tis
 tic
 a
 l 
a
 n
 a
 lys
 is
  
In
 te
 rp
 re
 ta
 tio
 n
  
 
An
  
ad
 di
 tio
 n
 a
 l r
 eg
 re
 s
 si
 o
 n
  
m
 o
 de
 l 
Re
 sp
 on
 se
  
v
 ar
 ia
 bl
 e
 : 
 
CG
 PA
  
 Si
 gn
 ific
 a
 n
 t p
 re
 di
 ct
 o
 r 
v
 ar
 ia
 bl
 es
  
fro
 m
  
re
 se
 a
 rc
 h 
qu
 es
 tio
 ns
  
3 
an
 d 
4:
  
?
  
M
 a
 tri
 c 
re
 su
 lts
  
?
  
N
 o
 te
 -
 ta
 kin
 g 
sk
 ills
  
 
Pr
 e
 di
 ct
 iv
 e
  
va
 lid
 ity
  
In
 te
 r-
 co
 rr
 e
 la
 tio
 n
 s 
    M
 u
 lti
 pl
 e
  
re
 gr
 e
 ss
 io
 n
  
a
 n
 a
 lys
 is
  
 Fo
 rw
 a
 rd
  
se
 le
 ct
 io
 n
  
?
 1.
 00
  
=
  
pe
 rfe
 ct
  
 
?
 .
 80
  
 
 
=
  
 
st
 ro
 n
 g 
 
?
 .
 50
  
 
 
=
  
 
m
 o
 de
 ra
 te
  
 
?
 .
 20
  
 
 
=
  
 
w
 e
 a
 k 
 
?
 .
 00
  
 
 
=
  
 
n
 o
  
co
 rr
 e
 la
 tio
 n
  
 >
  
90
 %
  
=
  
qu
 es
 tio
 n
 a
 bl
 e
  
75
 %
  
?
  
90
 %
  
=
  
ve
 ry
  
go
 o
 d 
50
 %
  
?
  
74
 %
  
=
  
go
 o
 d 
25
 %
  
?
  
49
 %
  
=
  
fa
 ir 
<
  
25
 %
  
=
  
po
 or
  
 
87  
 
 
Chapter 5 
RESULTS  
 
 
5.1 Introduction 
 
Chapter 5 presents the results of the data analysed.  For each research 
question, the question will be stated, followed by where the relevant data 
to test the hypothesis was located.  The statistical method utilised to 
analyse the data will be indicated, and the results and interpretation of the 
results will also be discussed. 
 
5.2  Results for research question 1 
 
Are all the questionnaires of the Academic Proficiency Assessment 
(APA) battery reliable?  
 
In order to evaluate if the questionnaires in the Academic Proficiency 
Assessment battery were internally consistent, the internal consistency 
reliability within the questionnaires were calculated. The internal 
consistency within questionnaires were each examined. 
 
The data used to examine the internal consistency were located in the 
items on the English Proficiency, Reading Comprehension, Time 
Management, Notetaking Skills, Memory Skills, Concentration Skills, and 
Achievement Anxiety questionnaires. The Kuder-Richardson formula 20 
was utilised to calculate the internal consistency of the English proficiency 
and Reading comprehension questionnaires as these tests were scored 
on a two-point scale (right/wrong).  Cronbach?s coefficient alpha was used 
88  
to assess the internal consistency reliability for the remaining 
questionnaires. 
 
Table 7 
 Internal consistency of questionnaires in the APA battery 
  Internal consistency 
English proficiency 
Reading comprehension 
Time management 
Note-taking skills 
Memorisation skills 
Concentration skills 
Achievement anxiety  (debilitating stress) 
Achievement anxiety (motivating stress) 
0.53 
0.40 
0.84 
0.77 
0.52 
0.66 
0.73 
0.38 
 
 
According to the general rule of thumb a reliability value of 0.7 is adequate  
in the early stages of validation research (Nunnally and Bernstein, 1994). 
Questionnaires evaluating time management, note-taking, and 
achievement anxiety (debilitating stress) have adequate internal 
consistency. The questionnaires assessing English Proficiency, Reading 
Comprehension, Memorisation Skills, Concentration Skills and 
Achievement Anxiety (motivating stress) exhibited inadequate internal 
consistency values. 
 
5.3  Results for research question 2 
 
If any of the questionnaires in the APA are not reliable, what changes 
can be implemented to enhance the psychometric properties of those 
questionnaires?  
 
89  
The questionnaires assessing English Proficiency, Reading 
Comprehension, Concentration Skills, Memorisation Skills and 
Achievement Anxiety (motivating stress) exhibited inadequate internal 
consistency values.  The internal consistency of these five questionnaires 
will be discussed in more detail, in order to determine how to enhance 
their internal consistency reliability.  Two factors affect the internal 
consistency coefficient, namely the correlations between the items, and 
the number of items.  Hence, in order to increase the internal consistency, 
the correlations between items must be increased, or there needs to be a 
greater number of items (Murphy & Davidshofer, 1996). 
 
5.3.1  English proficiency 
 
The internal consistency of the English proficiency questionnaire was not 
high enough.  None of the items correlated very highly with the total.  
Increasing the number of items on this questionnaire is not an option, as 
this is already a lengthy scale with one hundred items.  The correlations 
with the total ranged from 0 to .38.  Some items do not have an item-total 
correlation as all the respondents responded exactly the same.  These 
eight items (E9, E86, E88, E90, E91, E94, E99 and E100) need to be 
dropped completely from the scale and replaced with new different items. 
There are also four items (E68, E72, E78 and E79) with an item-total 
correlation of 0.00. These items also need to be dropped.  Twenty-seven 
items correlate negatively with the total, and also need to be discarded. 
Eliminating the thirty-nine items instantly raises the KR20 value from 0.53 
to 0.79. 
90  
Table 8 
Item analysis of English Proficiency questionnaire 
Item Correlation 
with Total KR 20 
Item  
difficulty 
Optimal 
level of 
difficulty 
E1 -.18 0.55 74.58 75 
E2 -.10 0.53 88.14* 75 
E3 0.18 0.51 38.98** 75 
E4 0.19 0.52 96.55* 75 
E5 0.08 0.52 91.53* 75 
E6 -.00 0.53 45.76** 67 
E7 -.16 0.53 6.78** 67 
E8 -.10 0.54 28.81** 67 
E9 . 0.53 0** 67 
E10 0.15 0.53 96.55* 67 
E11 -.03 0.53 62.71* 75 
E12 0.30 0.51 94.92* 75 
E13 0.06 0.53 50.85** 75 
E14 0.15 0.52 94.92* 75 
E15 0.22 0.51 83.05 75 
E16 0.34 0.50 35.39** 75 
E17 0.11 0.52 89.83* 75 
E18 0.10 0.52 77.97 75 
E19 0.03 0.52 86.44* 75 
E20 0.15 0.52 94.92* 67 
E21 0.08 0.52 96.61* 67 
E22 0.27 0.51 81.36* 67 
E23 -.05 0.53 94.92* 67 
E24 0.34 0.50 54.24** 75 
E25 -.21 0.54 81.36* 67 
E26 0.12 0.52 52.54* 67 
E27 0.10 0.52 71.19 67 
E28 0.17 0.52 79.66 75 
E29 0.22 0.51 43.10** 75 
E30 0.08 0.52 79.66 75 
E31 0.18 0.52 84.75* 67 
E32 0.28 0.52 91.38* 67 
E33 -.24 0.55 64.41 75 
E34 0.08 0.52 96.61* 75 
E35 0.02 0.53 96.61* 67 
E36 0.47 0.49 43.10** 75 
E37 -.03 0.53 89.83* 75 
E38 -.04 0.53 81.36 75 
E39 0.14 0.52 66.10 75 
E40 0.33 0.50 72.88 75 
E41 -.02 0.53 27.12** 75 
E42 0.37 0.51 8.47** 67 
E43 -.10 0.54 55.93** 75 
E44 -.11 0.54 38.98** 67 
E45 0.10 0.52 50.85** 75 
E46 0.03 0.53 93.22* 75 
E47 -.11 0.54 66.10 75 
E48 0.38 0.50 18.64** 75 
E49 -.01 0.53 22.03** 75 
91  
Item Correlation 
with Total KR 20 
Item 
difficulty 
Optimal 
level of 
difficulty 
E50 -.08 0.54 54.24 67 
E51 0.28 0.51 22.03** 75 
E52 -.31 0.56 42.37** 75 
E53 0.23 0.51 66.10 75 
E54 -.01 0.53 88.14* 75 
E55 -.02 0.53 44.07** 75 
E56 -.04 0.53 98.25* 75 
E57 -.15 0.54 94.92* 67 
E58 0.21 0.52 86.44 75 
E59 0.17 0.52 37.29** 75 
E60 0.19 0.51 47.46** 75 
E61 0.06 0.52 94.92* 75 
E62 0.04 0.53 67.80 75 
E63 0.17 0.52 38.98** 75 
E64 0.08 0.52 98.31* 75 
E65 0.10 0.52 93.22* 75 
E66 0.19 0.51 62.71** 75 
E67 -.03 0.53 86.44 75 
E68 0.00 0.53 83.05 75 
E69 -.17 0.54 22.03** 75 
E70 0.28 0.52 96.61* 75 
E71 0.21 0.51 77.97 75 
E72 0.00 0.53 88.14* 75 
E73 -.07 0.53 83.05 75 
E74 0.22 0.51 52.54** 75 
E75 0.17 0.52 91.53* 75 
E76 0.03 0.53 94.92* 75 
E77 0.17 0.52 94.92* 75 
E78 0.00 0.53 37.29** 67 
E79 0.00 0.53 94.92* 75 
E80 0.23 0.52 91.53* 75 
E81 -.15 0.54 83.05 75 
E82 0.15 0.52 45.76** 75 
E83 0.15 0.52 88.14* 75 
E84 0.27 0.52 98.21* 75 
E85 0.12 0.52 86.44 75 
E86 . 0.53 100* 75 
E87 0.30 0.51 96.61* 75 
E88 . 0.53 98.31* 75 
E89 0.30 0.50 69.49 75 
E90 . 0.53 100* 75 
E91 . 0.53 100* 75 
E92 -.22 0.55 30.51** 75 
E93 0.30 0.51 96.61* 75 
E94 . 0.53 100* 75 
E95 0.23 0.52 94.82* 75 
E96 0.32 0.51 96.61* 75 
E97 0.22 0.51 66.10 75 
E98 0.08 0.52 94.92* 75 
E99 . 0.53 100* 75 
E100 . 0.53 98.31* 67 
Note. * indicates too easy.  ** indicates too difficult. 
92  
5.3.2 Reading comprehension 
 
The individual items correlation with the total reading comprehension 
score range from 0.08 to 0.32.  None of the items have a sufficiently high 
correlation with the total.  All the items need to be revised in order to raise 
the internal consistency of the Reading Comprehension questionnaire to 
an acceptable level.  As this test has already very few items, discarding 
items is not desirable, as too many items would have to be removed. 
 
Table 9 
Item analysis of Reading Comprehension questionnaire 
Item 
Correlation 
with total 
KR 20 
Item 
difficulty 
(pi) 
RC1 0.13 0.39 63.49 
RC2 -0.09 0.47 62.70 
RC3 0.17 0.37 29.37 
RC4 0.10 0.40 73.02 
RC5 0.17 0.37 73.02 
RC6 0.21 0.36 58.73 
RC7 0.19 0.37 65.87 
RC8 0.08 0.41 76.19 
RC9 0.30 0.32 63.49 
RC10 0.32 0.31 75.40 
Note.  Optimal level of difficulty is 66.67%. 
 
5.3.3 Memorisation skills 
 
None of the items correlated highly with the total.  This questionnaire 
requires complete revision in order to enhance the internal consistency of 
the scale.  The questionnaire either needs to be replaced with another 
questionnaire assessing memorisation skills or all the items need to be 
93  
rewritten to ensure a more homogenous questionnaire.  This test has even 
fewer items than the Reading Comprehension test.  Eliminating items from 
this test would make the test too short. 
 
 Table 10 
 Internal consistency of Memorisation Skills questionnaire 
Item 
Correlation with 
Total 
Alpha 
MS1 0.18 0.50 
MS2 0.18 0.50 
MS3 0.29 0.47 
MS4 0.11 0.53 
MS5 0.29 0.47 
MS6 0.21 0.49 
MS7 0.38 0.44 
MS8 0.52 0.39 
MS9 -0.04 0.57 
 
 
5.3.4 Concentration 
 
This test also has only ten items, but the internal consistency is higher 
than the other ?inconsistent? tests.  Eliminating the two items with the 
lowest values (CS2 and CS10) raises the value to 0.69. 
 
94  
Table 11 
Internal consistency of Concentration Skills questionnaire 
Item 
Correlation with 
Total 
Alpha 
CS1 0.40 0.61 
CS2 0.15 0.67 
CS3 0.41 0.61 
CS4 0.34 0.63 
CS5 0.46 0.60 
CS6 0.24 0.65 
CS7 0.36 0.62 
CS8 0.31 0.63 
CS9 0.43 0.61 
CS10 0.09 0.68 
 
 
5.3.5 Achievement anxiety  (motivating stress) 
 
This questionnaire presents the same problem as the preceding one.  The 
questionnaire either needs to be replaced with another questionnaire 
assessing memorisation skills or all the items need to be rewritten to 
ensure a more homogenous questionnaire. 
 
  
95  
Table 12 
 Internal consistency of Motivating Stress questionnaire 
 
Correlation with 
Total 
Alpha 
AAT3 0.53 0.17 
AAT1 -0.30 0.53 
AAT2 -0.30 0.51 
AAT5 0.05 0.38 
AAT11 0.25 0.29 
AAT12 0.31 0.26 
AAT13 0.40 0.19 
AAT14 0.24 0.29 
AAT19 0.35 0.24 
 
 
 
5.4 Results for research question 3 
 
What is the predictive validity of the questionnaires in the APA 
battery? 
 
This research question was addressed using two different types of 
analyses:  multiple regression and discriminant function analysis.   The 
rationale for including two different statistical techniques was explained in 
section 4.6.3.2. 
 
Given the low reliabilities of a number of subtests, increases the likelihood 
of a multiple regression analysis capitalising on random features in the 
data.  The analysis was nonetheless executed as the present study is 
exploratory, and if evidence supporting the predictive validity of the APA 
battery was found, it would have motivated the procedure of further 
refining and enhancing the psychometric properties of the battery. 
96  
  
5.4.1  Multiple regression analysis 
 
The dependent variable was CGPA. The CGPA was obtained from the 
average of eight modules: two modules each of Information Systems, 
Technical Programming, Systems Software and Development Software. 
 
The total scores obtained for each of the questionnaires in the Academic 
Proficiency Assessment battery (APA battery) were used as the 
explanatory variables.   These variables were: 
ring2 English proficiency 
ring2 Reading speed 
ring2 Reading comprehension 
ring2 Time management 
ring2 Note-taking skills  
ring2 Memorisation 
ring2 Concentration 
ring2 Debilitating stress  
ring2 Motivating stress 
 
Although the APA battery is to a large extent not sufficiently reliable, it was 
nonetheless utilised to calculate a regression equation as the nature of the 
study is conceptual and exploratory.  Data was also modified in an attempt 
to enhance the reliability of the battery. 
 
As a result, certain scores required adjustment for the analysis.  In the 
English proficiency questionnaire, the scores for the items to be dropped 
were eliminated from the total score.  This enabled analysis using a more 
reliable battery.  The distribution of the English proficiency scores also 
showed a sharp upward and downward curve at both extremes, indicative 
of the tails of the distribution being too heavy to be regarded as normal. 
These scores therefore had to undergo a square-root transformation to 
97  
meet the assumption of normality required for multiple regression.  The 
scores from two items that needed to be dropped from the Concentration 
Skills questionnaire were also excluded from the total Concentration score.  
The Reading Speed, and Debilitating Stress distributions were positively 
skewed, and the scores required log transformations in order to achieve 
normality. The Note-taking Skills distribution displayed a flattening at the 
extremes, indicating that the tails were lighter than normal (Galpin, 2001).  
Hence, the Note-taking skills scores were transformed using a cube-root 
transformation (P. Fridjhon, personal communication, December, 2004). 
 
? Multicollinearity 
 
Before commencing with the multiple regression analysis, the correlation 
and/or multicollinearity between the independent variables were 
investigated, ustails of thing Pearson?s Correlation Coefficient (Table 13).   
 
Correlation coefficients ranged from ?0.00822 to ?0.67888.  A large 
number of the correlations was significant enough to indicate the presence 
of multicollinearity. 
 
The correlation between English proficiency and reading speed was 
understandable. The more proficient one?s English is, the faster English 
word recognition would be, thereby contributing to the increase in reading 
speed.  The reason for having these two tests separately, is that the 
development of English proficiency, and the enhancement of reading 
speed, are facilitated as two independent courses.  Hence, diagnosis for 
selection to an English proficiency programme, or a reading enhancement 
programme is executed independently. 
 
There was also a significant correlation between English proficiency and 
memory skills, and a significant correlation between Reading 
98  
comprehension and motivating stress. These correlations are likely to be 
incidental. 
 
Fourteen of the fifteen correlations of the study skills questionnaires were 
significantly correlated at the 0.05 level of significance, and the remaining 
one correlation was significant at the 0.1 significance level.  The 
correlation matrix indicates a significant correlation between CGPA and 
Time management, Note-taking skills, and Memorisation skills.  But, Time 
management, Note-taking skills, and Memorisation skills are significantly 
correlated too.  This has notable implications, both in terms of the present 
statistical analyses, as well as the far wider implication of the construction 
of the assessment itself.   In terms of the statistical analysis, the R2adj will 
still quantify how well the model predicts academic performance.  But, 
because of the multicollinearity, as well as the poor reliability of the 
questionnaires, gauging the individual contribution of each independent 
variable to the prediction of academic performance is problematic, as 
some variables may be excluded from the model although they may be 
significant (Motulsky, 2002). Others may also be included even though 
they are unreliable.  In terms of the logic of the assessment battery as a 
whole, the high intercorrelation across the questionnaires, indicates that 
there is considerable overlap in the respective questionnaires.  This 
implies that the assessment battery could be shortened considerably, in 
order to avoid reassessing the same constructs. 
 
 
 
 
99
  
 
Ta
 bl
 e
  
13
  
Co
 rr
 e
 la
 tio
 n
  
m
 a
 tri
 x 
o
 f e
 xp
 la
 n
 a
 to
 ry
  
va
 ria
 bl
 e
 s 
 
Pe
 ar
 so
 n 
Co
 rr
 el
 a
 tio
 n 
Co
 ef
 fic
 ie
 nt
 s 
 
Pr
 o
 b 
>
  
|r| 
un
 de
 r 
H
 0:
  
R
 ho
 =
 0 
 
CG
 PA
  
En
 gli
 sh
  
pr
 o
 fic
 ie
 nc
 y 
R
 e
 a
 din
 g 
sp
 e
 e
 d 
R
 e
 a
 din
 g 
co
 m
 p.
  
Ti
 m
 e
  
m
 a
 n.
  
N
 ot
 e-
 ta
 kin
 g 
sk
 ills
  
M
 em
 or
 y 
sk
 ills
  
Co
 nc
 e
 n-
 tra
 tio
 n 
sk
 ills
  
D
 e
 bil
 ita
 tin
 g 
st
 re
 ss
  
M
 o
 tiv
 a
 tin
 g 
st
 re
 ss
  
Ag
 e
  
En
 gli
 sh
  
pr
 of
 ici
 e
 nc
 y 
0.
 03
  
0.
 77
 82
  
 
 
 
 
 
 
 
 
 
 
R
 ea
 din
 g 
sp
 e
 ed
  
-
 0.
 02
  
0.
 83
  
0.
 25
  
0.
 01
 10
  
 
 
 
 
 
 
 
 
 
 
R
 ea
 din
 g 
co
 m
 p.
  
0.
 09
  
0.
 30
 52
  
-
 0.
 01
  
0.
 93
 30
  
 
0.
 02
  
0.
 81
 51
  
 
 
 
 
 
 
 
 
 
Ti
 m
 e 
m
 a
 n.
  
-
 0.
 19
  
0.
 04
 55
  
-
 0.
 04
  
0.
 68
 82
  
 
0.
 05
  
0.
 58
 64
  
 
0.
 03
  
0.
 74
 28
  
 
 
 
 
 
 
 
 
N
 ot
 e
 -
 ta
 kin
 g 
sk
 ills
  
-
 0.
 29
  
0.
 02
 8 
0.
 09
  
0.
 37
 17
  
 
0.
 09
  
0.
 35
 97
  
 
-
 0.
 15
  
0.
 13
 22
  
 
0.
 52
  
<
 .
 00
 01
  
 
 
 
 
 
 
 
M
 em
 o
 ry
  
Sk
 ills
  
-
 0.
 21
  
0.
 03
 50
  
0.
 35
  
0.
 00
 03
  
 
0.
 16
  
0.
 10
 87
  
 
-
 0.
 17
  
0.
 07
 99
  
 
0.
 31
  
0.
 00
 13
  
 
0.
 50
  
<
 .
 00
 01
  
 
 
 
 
 
 
Co
 nc
 e
 n-
 tra
 tio
 n 
sk
 ill 
-
 0.
 11
  
0.
 28
 24
  
0.
 07
  
0.
 49
 20
  
 
-
 0.
 03
  
0.
 77
 50
  
 
-
 0.
 09
  
0.
 35
 46
  
 
0.
 68
  
<
 .
 00
 01
  
 
0.
 37
  
0.
 00
 01
  
 
0.
 30
  
0.
 00
 18
  
 
 
 
 
 
D
 e
 bil
 ita
 tin
 g 
st
 re
 ss
  
0.
 13
  
0.
 17
 98
  
0.
 19
  
0.
 05
 26
  
 
0.
 08
  
0.
 43
 87
  
 
-
 0.
 13
  
0.
 18
 82
  
 
-
 0.
 49
  
<
 .
 00
 01
  
 
-
 0.
 34
  
0.
 00
 04
  
 
-
 0.
 20
  
0.
 03
 90
  
 
-
 0.
 50
  
<
 .
 00
 01
  
 
 
 
 
M
 o
 tiv
 a
 tin
 g 
st
 re
 ss
  
-
 0.
 13
  
0.
 16
 89
  
0.
 20
 29
 3 
0.
 03
 61
  
 
0.
 04
 48
 2 
0.
 64
 67
  
 
-
 0.
 22
 53
 3 
0.
 01
 96
  
 
0.
 28
 66
 0 
0.
 00
 28
  
 
0.
 34
 57
 8 
0.
 00
 03
  
 
0.
 39
 26
 2 
<
 .
 00
 01
  
 
0.
 35
 58
 8 
0.
 00
 02
  
 
-
 0.
 16
 31
 2 
0.
 09
 32
  
 
 
 
Ag
 e
  
-
 0.
 01
  
0.
 88
 76
  
-
 0.
 11
  
0.
 25
 56
  
-
 0.
 13
  
0.
 18
 39
  
-
 0.
 02
  
0.
 84
 56
  
0.
 08
  
0.
 40
 91
  
-
 0.
 07
  
0.
 49
 99
  
-
 0.
 11
  
0.
 24
 50
  
0.
 00
 7 
0.
 94
 36
  
0.
 00
 6 
0.
 95
 44
  
-
 0.
 15
  
0.
 13
 67
  
 
M
 at
 ric
  
re
 su
 lts
  
0.
 32
  
0.
 00
 11
  
0.
 11
  
0.
 29
 75
  
0.
 05
  
0.
 60
 11
  
0.
 00
 1 
0.
 99
 10
  
-
 0.
 18
  
0.
 07
 33
  
-
 0.
 20
  
0.
 04
 19
  
-
 0.
 09
  
-
 0.
 35
 20
  
-
 0.
 07
  
0.
 50
 85
  
 
0.
 07
  
0.
 47
 05
  
-
 0.
 15
  
0.
 14
 32
  
-
 0.
 12
  
0.
 22
 44
  
N
 ot
 e.
  
Si
 gn
 ific
 an
 t c
 or
 re
 la
 tio
 n
 s 
in
  
ita
 lic
 s.
100  
? Multiple regression analysis 
 
A forward selection model was preferred for the analysis, as it enabled the 
researcher to examine the contribution of each predictor variable to the 
regression model independently.  The significance level to enter the model 
was set at 0.05.  The significance level is usually between 0.05 and 0.1, 
with 0.05 being the common practice (Bowerman & O?Connell, 1990). 
 
The results of the multiple regression analysis are presented in Tables 14 
to 17. 
 
Table 14 
R2 for research question 3 
R-Square 0.0826 
Adj R-Sq 0.0737 
 
 
Table 15 
Parameter estimates for research question 3 
Variable Parameter Estimate 
Standard 
Error 
Type II 
SS 
F 
Value Pr > F 
Intercept 84.95 10.35 7103.94 67.35 <.0001 
Note-
 taking 
skills 
-8.24 2.69 987.30 9.36 0.0028 
 
 
 
 
 
 
 
 
 
101  
Table 16 
Summary of forward selection for research question 3 
Step Variable Entered 
Number 
Vars In 
Partial 
R-
 Square 
Model 
R-
 Square 
C(p) F Value Pr > F 
1 Note-taking 
skills 1 0.08 0.08 -3.37 9.36 0.0028 
 
 
Table 17 
Analysis of variance for research question 3 
Source DF Sum of Squares 
Mean 
Square 
F 
Value Pr > F 
Model 1 987.30 987.30 9.36 0.0028 
Error 104 10970 105.48   
Corrected 
Total 105 11958    
 
 
The regression was a rather poor fit with a R2adj value of 8,3%, but the 
overall relationship was significant (F1,104 = 9,36, p < 0.05).  The only 
variable that met the 0.05 significance level for entry into the model was 
Note-Taking Skills.  With other variables held constant, CGPA was 
negatively related to Note-taking Skills.  
 
The regression equation is: 
 
CGPA = 85 ? 8,24 Note-taking Skills 
 
In terms of answering the research question, the statistical analysis 
suggested that the Note-Taking skills questionnaire was the only variable 
that contributed as a predictor of academic performance for Information 
Technology students.  Unfortunately, the relationship was an inverse one, 
implying that if a student performed poorly on this questionnaire, there was 
102  
a likelihood that he/she would have a good CGPA!  This finding certainly 
cannot be implemented when selecting for remedial intervention.  
Although statistically justifiable, the APA battery is likely to lose face 
validity with the students upon communication of the results.   
 
At this stage of the analyses, based on the results of research question 
one and the multiple regression equation in the present research question, 
it may be concluded that the APA is unreliable and invalid. 
 
5.4.2 Discriminant function analysis 
 
This analysis was included in addition to the multiple regression analysis, 
in an attempt to establish whether the tests were predicting performance 
criteria on the Information Technology course, and in particular whether 
the students were academically successful or not. 
 
Although a number of the subtests yielded unsatisfactory internal 
consistency values, the discriminant function analysis was undertaken in 
order to determine if the battery had potential validity, thereby providing a 
motive for the exercise of substantially modifying the test battery and 
rewriting the items. 
 
The dependent variable was a dichotomous variable: successful or 
unsuccessful students. A student was coded as successful if he/she 
passed six, seven or eight of the eight subjects, and unsuccessful if zero 
to five of the eight subjects were passed.  This classisfication was used as 
students passing all eight subjects, or failing one or two subjects were still 
able to complete their national diploma in the minimum duration of three 
years. 
 
The same explanatory variables and transformed scores were used as for 
the previous analysis.   These variables were: 
103  
ring2 English proficiency 
ring2 Reading speed 
ring2 Reading comprehension 
ring2 Time management 
ring2 Note-taking skills  
ring2 Memorisation 
ring2 Concentration 
ring2 Debilitating stress  
ring2 Motivating stress 
 
A backward stepwise discriminant function analysis was also executed. 
This model was preferred, as the variables that were not contributing to 
the prediction of a student as successful or unsuccessful, could be 
removed.  Only one variable, namely English Proficiency was eliminated 
from the model. 
 
The results of the discriminant function analysis are tabulated in Tables 18 
to 22. 
 
104  
Table 18 
Linear discriminant function for research question 3 
Linear Discriminant Function for Successful students 
Variable 0 1 
Constant -604.39 -600.94 
Reading speed 113.78 113.24 
Time management -1.15 -1.21 
Debilitating stress 223.05 220.28 
Note-taking skills 42.46 42.08 
Reading comprehension 6.28 6.28 
Memorisation skills -0.68 -0.73 
Motivating stress -0.04 -0.06 
Concentration skills 387.64 393.49 
 
 
Table 19 
Classification table for research question 3 
Number of Observations and Percent 
Classified into Successful students 
From Successful student 0 1 Total 
0 
22 
73.33 
8 
26.67 
30 
100 
1 
22 
28.57 
55 
71.43 
77 
100 
Total 
44 
41.12 
63 
58.88 
107 
100 
Priors 
0.5 
 
0.5 
 
 
 
 
 
 
 
 
105  
Table 20 
Average R2 for discriminant function analysis 
Average R-Square 
Unweighted 0.01 
Weighted by Variance 0.02 
 
 
Table 21 
Univariate test statistics for research question 3 
F Statistics, Num DF=1, Den DF=105 
Variable 
Total 
Std. 
Dev. 
Pooled 
Std. 
Dev. 
Bet. 
Std. 
Dev. 
R2      
R2/  
(1-R2) 
F 
Value 
Pr > F 
Reading 
speed 
0.13 0.13 0.02 0.01 0.01 0.77 0.38 
Time man. 10.74 10.68 2.13 0.02 0.02 2.13 0.15 
Debilitating 
stress 
0.12 0.12 0.01 0.004 0.004 0.43 0.52 
Reading 
comp. 
1.85 1.85 0.07 0.001 0.001 0.08 0.77 
Memory 
skills 
5.73 5.69 1.26 0.02 0.03 2.64 0.11 
Motivating 
stress 
4.65 4.65 0.57 0.01 0.01 0.79 0.38 
Conc.  skills 0.09 0.09 0.003 0.001 0.001 0.09 0.77 
Note-taking 
skills 
0.37 0.37 0.08 0.02 0.02 2.43 0.12 
 
 
 
 
106  
Table 22 
Multivariate and exact F statistics for research question 3 
S=1 M=3 N=48 
Statistic Value 
F 
Value 
Num D
 F 
Den 
DF 
Pr > F 
Wilks' Lambda 0.91 1.17 8 98 0.33 
Pillai's Trace 0.09 1.17 8 98 0.33 
Hotelling-Lawley Trace 0.10 1.17 8 98 0.33 
Roy's Greatest Root 0.10 1.17 8 98 0.33 
 
72% of the cases were correctly predicted.  This is a good level of 
accuracy within social science research, indicating that the proficiencies 
(except English proficiency, which was the only variable eliminated) 
measured by the APA battery, predicted the likelihood of a student being 
successful or unsuccessful to a fairly good extent. 
 
The discriminant regression equation is: 
 
Success =  -600.94 - 0.06 Motivating stress - 0.73 Memorisation skills - 
1.21 Time management + 6.28 Reading comprehension + 42.08 Note-
 taking skills + 113.24 Reading speed + 220.28 Debilitating stress + 393.49 
Concentration skills 
 
Except for English proficiency, all the independent variables contributed to 
the prediction of academic success.  Two things need to be noted 
regarding this result.  Firstly, the English proficiency questionnaire scores 
used eliminated the scores of the items that needed to be dropped.  
Hence a reliable questionnaire was used, but yet it was the only 
questionnaire that failed to contribute to the prediction of academic 
success.  A possible reason for this could be, that the programming 
languages studied in Information Technology, is a completely different 
type of ?language? with different grammatical rules compared to English.  
107  
Secondly, although eight of the tests contributed significantly to the 
prediction of academic success, three of these are unreliable. Revision of 
these unreliable questionnaires is firstly required, but these results indicate 
that the battery may have potential utility.  This issue could be revisited 
after revision of the APA battery. 
 
5.5 Research question 4 
 
Are there any additional variables that could enhance the predictive 
validity of the APA battery for Information Technology students? 
 
Multiple regression analysis was used for this part of the analysis.  As with 
the previous multiple regression for Research Question Three, a forward 
selection model was again used and the significance level to enter the 
model was set at 0.05.   
 
Availble biographical information solicited in the biographical questionnaire 
were included as explanatory variables.  Matric results were also included, 
based on consistent research on Matric results as a good predictor of 
academic performance, and Matric results are also the primary selection 
tool for admission to the Information Technology course. 
 
CGPA was the dependent variable. The four explanatory variables were 
investigated for this research question was: 
ring2 Age 
ring2 Gender 
ring2 Predominant language of communication 
ring2 Matric results (symbols of each subject converted using the 
Swedish formula, and aggregated). 
 
The data for the first three variables was obtained from the biographical 
questionnaire.  The Matric results were obtained from the ITS system.  
108  
Matric results were transformed to normality using a log transformation, in 
order to meet the assumptions for multiple regression.  Gender and 
predominant language of communication were coded using dummy 
variables. 
  
? Multicollinearity 
 
The correlations between age and Matric results were examined in Table 
11 in order to determine the presence of multicollinearity.  Gender and 
predominant language of communication were excluded from this 
analysis, as they are nominal variables, and one of the assumptions for 
performing correlations is that the data must at least be on an interval level 
of measurement (Hatcher, 1994).  The correlation between age and Matric 
results (-0.12) may be regarded as a weak and insignificant correlation.  
Hence, multicollinearity is not a problem. 
 
? Multiple regression 
 
A forward stepwise selection model was used.  The results of the analysis 
are tabulated in Tables 23 to 26.   
 
Table 23 
R2 for research question 4 
R-Square 0.1097 
Adj R-Sq 0.1004 
 
 
   
 
 
 
109  
  Table 24 
Parameter estimates for research question 4 
Variable Parameter Estimate 
Standard 
Error 
Type II 
SS 
F 
Value Pr > F 
Intercept 7.67 13.33 32.89 0.33 0.5662 
Matric 
results 32.84 9.55 1174.38 11.83 0.0009 
 
 
Table 25 
Summary of forward selection for research question 4 
Step Variable Entered 
Number 
Vars In 
Partial 
R-
 Square 
Model 
R-
 Square 
C(p) F Value Pr > F 
1 Matric 
results 1 0.11 0.11 0.40 11.83 0.0009 
 
 
Table 26 
Analysis of variance for research question 4 
Source DF Sum of Squares 
Mean 
Square 
F 
Value Pr > F 
Model 1 1174.38 1174.38 11.83 0.0009 
Error 96 9530.23 99.27   
Corrected Total 97 10705    
 
 
This regression was also a poor fit with a R2adj value of 11%, but the 
overall relationship was significant (F1,96 = 11.83, p < 0.05).  The only 
variable that met the 0.05 significance level for entry into the model was 
Matric results.  None of the biographical variables were included in the 
model.  With other variables held constant, CGPA was positively related to 
Matric results.  
 
 
110  
The regression equation is: 
 
CGPA = 7,67 + 32,84 Matric results 
 
This result supports one of the present criteria for selection for admission 
into the Information Technology course at Tshwane University of 
Technology. If the present admission criteria are maintained, the 
regression model can be used to identify students at-risk according to this 
equation, and recommend these students for remedial intervention. 
 
5.6 An integrated regression model 
 
As the sample size prevented including the predictor variables from 
research questions three and four in a single multiple regression analysis, 
another set of statistical analyses were executed in addition to those 
performed for research questions three and four.   This multiple regression 
analysis contains as predictor variables, the variables that emerged as 
significant in the last two research questions.  The response variable is 
CGPA. 
 
? Multicollinearity 
 
The correlation between Matric results and Note-taking skills presented in 
Table 11 was examined in order to detect the presence of multicollinearity. 
Although the correlation between Matric results and Note-taking skills is 
significant (P<0.05), it is a weak negative relationship (r  = -0.2039). 
 
? Multiple regression analysis 
 
The results of the multiple regression analysis are presented in Tables 27 
to 30. 
 
111  
Table 27 
R2 for the integrated regression model 
R-Square 0.1502 
Adj R-Sq 0.1334 
 
  
  Table 28 
Parameter estimates the integrated regression model 
 
 
 
Table 29 
Summary of forward selection for the integrated regression model 
Step Variable Entered 
No. 
Vars 
In 
Partial 
R-
 Square 
Model 
R-
 Square 
R-
 Square 
change 
C(p) F Value Pr > F 
1 Matric 
results 1 0.10 0.10 0.10 5.41 11.37 0.0011 
2 
Note-
 taking 
skills 
2 0.04 0.14 0.04 3.00 4.41 0.0383 
 
 
 
 
 
 
Variable Parameter Estimate 
Standard 
Error 
Type II 
SS 
F 
Value Pr > F 
Intercept 35.84 18.89 356.02 3.60 0.0609 
Matric 
results 28.35 9.75 836.14 8.45 0.0045 
Note-
 taking 
skills 
-5.77 2.74 436.77 4.41 0.0383 
112  
 
Table 30 
Analysis of variance for the integrated regression model 
Source DF Sum of Squares 
Mean 
Square 
F 
Value Pr > F 
Model 2 1601.15 800.58 8.09 0.0006 
Error 96 9499.35 98.95   
Corrected Total 98 11100    
 
 
Both Matric results and Note-taking skills met the 0.05 significance level 
for entry into the model was Matric results.  Cumulatively, Matric results 
and Note-taking skills contributed 13,34% towards accounting for the 
variance in CGPA, which is a poor fit.  The overall relationship was 
significant (F2,96 = 8.09, p < 0.05). 
 
The regression equation is: 
 
CGPA = 35.84 + 28.35 Matric results ? 5.77 Note-taking skills 
 
 
5.7 Conclusion 
 
Three questionnaires (time management, the note-taking skills, and the 
debilitating stress scale on the achievement anxiety test) yielded adequate 
Cronbach alpha values, while five tests (English proficiency, reading 
comprehension, memory skills, concentration skills and motivating stress 
scale on the achievement anxiety questionnaire) produced Cronbach 
alpha coefficients less than 0.70, indicating an inadequate level of internal 
consistency reliability. Thirty-nine items in the English proficiency 
questionnaire need to be eliminated completely.  The remaining items will 
benefit from modification.  All the items in the Reading Comprehension 
questionnaire need to be revised.  Two items in the Concentration Skills 
113  
require removal. The Memory Skills questionnaire can be dropped and 
replaced with a new scale. The motivating stress subscale on the 
Achievement Anxiety questionnaire also needs to be eliminated. 
 
If the results of the multiple regression analysis are to be accepted using 
CGPA as a performance indicator, very little can be extracted from the 
scores of the questionnaires in the APA when selecting for remedial 
intervention.  The only significant relationship was with Note-taking Skills, 
and that too was a negative relationship.  It should be noted that due to 
the multicollinearity, some variables may have been excluded from the 
model although they may be significant. 
 
However, the results of the discriminant function analysis, where the 
number of subjects passed was used as a performance criterion, indicated  
that a good level of accuracy could be achieved in predicting successful 
and unsuccessful students were the APA battery revised.  All the 
academic proficiencies except English proficiency contributed to the 
prediction. 
 
Overall, the results would suggest that matric results remain the best 
predictor of academic performance, but at a very low level.  Given the lack 
of reliability in the majority of the tests, the use of the APA battery for 
selection for admission to TUT, or for selection for remedial intervention, is 
not yet justified. 
 
 
114  
 
 
Chapter 6 
CONCLUSION 
 
 
6.1       Introduction 
 
This final chapter presents a synopsis of the study, a summary of the 
major findings, the strengths and limitations of the study, some 
recommendations for future research, as well as the conclusions that may 
be drawn from the present study. 
 
6.2 Synopsis of the study 
 
According to the literature surveyed, intellective factors still emerge as the 
strongest predictor of academic performance.  Information Technology 
students, the subjects in the present study, are also still selected for 
admission to Tshwane University of Technology on the basis of an 
intellective factor, namely Matric results which has consistently been 
upheld as the single best predictor of academic performance at tertiary 
level.  However, this criterion is insufficient on its own, or else there would 
be no need for research on non-intellective factors.  While the research 
surveyed in the literature review does not provide any unequivocal 
irrefutable results, it provides a basis for the wider exploration of certain 
factors as possible correlates of academic performance. 
 
A substantial body of research exists indicating the likelihood of the 
variables that the APA aims to assess, namely language abilities and 
study skills contributing to academic performance.  Numerous 
assessments have also been developed to assess such proficiencies, 
115  
APA attempting to add to the collection of such assessments.  The present 
study attempts to firstly gauge the internal consistency reliability of each of 
the questionnaires in the APA battery, and secondly to validate the APA 
battery as a predictor of academic achievement by evaluating the extent to 
which each of the proficiencies in the test is a predictor of academic 
achievement.  
 
6.2.1  English proficiency 
 
The internal consistency of the English proficiency questionnaire was 
inadequate.  None of the items correlated very highly with the total. Eight 
items did not have an item-total correlation as all the respondents 
responded exactly the same, and needed to be dropped completely from 
the scale and replaced with new different items. There were four items 
with an item-total correlation of 0.00 that need to be dropped, and twenty-
 seven items correlate negatively with the total and also need to be 
discarded. Eliminating the thirty-nine items instantly raises the KR20 value 
from 0.53 to 0.79. 
 
English proficiency did not correlate with academic performance for 
Information Technology students at TUT. 
 
6.2.2  Reading skills 
 
The reading skills test has two components: reading speed 
comprehension. Both reading speed and reading comprehension 
contributed to the prediction of the likelihood of  a student being successful  
following the results of a linear discriminant function analysis. However, 
none of the items in the Reading Comprehension test have a sufficiently 
high correlation with the total.  All the items need to be revised in order to 
raise the internal consistency. 
 
116  
6.2.3  Time Management 
 
The Time management questionnaire exhibited a high degree of internal 
consistency with a reliability of 0.84. 
 
According to the linear discriminant function analysis Time management 
scores were able to contribute to prediction of the likelihood of a student 
being successful, but had a low negative regression coefficient. 
 
6.2.4  Note-taking skills 
 
The note-taking skills questionnaire was adequately internally consistent 
yielding a reliability value of 0.77. 
 
A multiple regression analysis with all the APA questionnaires found Note-
 taking skills to be the only significant predictor of academic performance, 
accounting for 8,3% of the variance.  However, there was a negative 
relationship between the variables. The linear discriminant function 
analysis indicated that Note-taking skills contribute to prediction of the 
likelihood of a student being successful. 
 
Considering that the two sets of analyses show the direction of the 
relationship differently, it would be unwise to utilise the scores of the Note-
 taking skills subtest for selection purposes. 
 
6.2.5  Memorisation skills 
 
A linear discriminant function analysis found Memorisation skills to be able 
to successfully predict the likelihood of a student being successful.  
However, bearing in mind that the memorisation skills questionnaire 
exhibited an unsatisfactory reliability of 0.52, one needs to sceptical about 
making inferences regarding this relationship. 
117  
6.2.6  Concentration skills 
 
Eliminating two items from the concentration skills questionnaire raised the 
reliability value from 0.66 to 0.69. 
 
The linear discriminant function analysis also indicated that Concentration 
skills are able to contribute to predicting the likelihood of a student being 
successful. According to the results of this analysis, Concentration skills 
had by far the highest regression coefficient out of the eight subtests that 
were included in the model.  Based on these results, it would be to the 
advantage of Information Technology students who obtain low scores on 
this subtest, to receive remedial assistance to enhance their concentration 
skills. 
 
6.2.7  Examination stress 
 
The Achievement Anxiety Test had two scales, debilitating and motivating 
stress.  Although the debilitating stress scale was sufficiently reliable 
(0.73), the motivating stress scale exhibited unsatisfactory internal 
consistency (0.38).  The questionnaire either needs to be replaced with 
another questionnaire assessing memorisation skills or all the items need 
to be rewritten to ensure a more internally consistent questionnaire.   
 
According to the linear discriminant function analysis both scales are able 
to contribute to predicting the likelihood of a student being successful.  
However, the motivating stress scale has very low reliability, hence very 
little confidence may be attached to its predictive validity.  The debilitating 
stress scale had the second highest regression coefficient out of the eight 
subtests that were included in the model.  As it is a reliable scale, students 
obtaining a high score on this scale of the Achievement Anxiety subtest, 
would benefit from stress management therapy. 
 
118  
6.2.8 Additional predictors 
 
None of the biographical variables, namely age, gender and predominant 
language of communication emerged as significant predictors of academic 
performance.  Consistent with the literature, Matric results emerged as the 
best predictor variable in this study too, validating Matric results as one of 
the present  admission criteria for the Information Technology course at 
TUT.   
 
6.3 Summary of findings 
 
An overview of the findings will be presented addressing the research 
questions. 
 
6.3.1 Are all the questionnaires of the Academic Proficiency  
  Assessment (APA) battery reliable? 
 
The time management questionnaire, the note-taking skills questionnaire, 
and the debilitating stress scale on the achievement anxiety test yielded 
adequate Cronbach alpha values.   The English proficiency, reading 
comprehension, memorisation skills, concentration skills and motivating 
stress scale on the achievement anxiety questionnaire produced 
Cronbach alpha coefficients less than 0.70, indicating an inadequate level 
of internal consistency reliability. 
 
6.3.2 If the APA is not reliable, what changes can be  
implemented to enhance its psychometric properties? 
 
Several questionnaires require modification. Several items in the English 
proficiency questionnaire need to be eliminated completely.  The 
remaining items will benefit from modification.  All the items in the Reading 
Comprehension questionnaire need to be revised.  Two items in the 
Concentration Skills require removal. The Memory Skills questionnaire can 
119  
be dropped and replaced with a new scale.  The Study Skills 
Questionnaire (SSQUES) has a subscale on memorisation that could be 
investigated.  The motivating stress subscale on the Achievement Anxiety 
questionnaire also needs to be eliminated. 
 
6.3.3 What is the predictive validity of the questionnaires in the  
APA battery? 
 
The results of the multiple regression analysis using CGPA as a 
performance indicator, reveal that very little can be extracted from the 
scores of the questionnaires in the APA when selecting for remedial 
intervention.  The only significant relationship was with Note-taking Skills 
(R2adj = 8.3%), and that too was a negative relationship.  Though, as a 
result of the multicollinearity, some variables may have been excluded 
from the model although they may have been significant.  The 
intercorrelations amongst the variables also indicate that the 
questionnaires tend to measure the same factors, and can therefore be 
streamlined even further. 
 
However, the results of the discriminant function analysis, where the 
number of subjects passed was used as a performance criterion, indicated 
that a good level of accuracy (72%) in predicting successful and 
unsuccessful students could be established by weighting the scores from 
the variables in the analysis.  This would suggest that prediction of 
academic performance is possible, provided that the existing battery is 
substantially modified.   
 
Given the lack of reliability in the majority of the tests, the use of the APA 
battery for selection for remedial intervention, is not yet justified. The high 
proportion of pass rates of students predicted by the tests in combination 
would suggest that the battery has potential, and that work on improving 
its psychometric qualities should be undertaken.   
120  
6.3.4 Are there any additional variables that could enhance the  
predictive validity of the APA battery for Information 
Technology students? 
 
None of the biographical variables significantly predicted CGPA. Matric 
results was positively related to CGPA (R2adj = 11%), validating Matric 
results as one of the present criteria for selection for admission into the 
Information Technology course at TUT.  If the present admission criteria 
are maintained, the regression model can be used to identify students at-
 risk according to this equation, and recommend these students for 
remedial intervention. 
 
6.4 Strengths of the study 
 
The two main strengths of the study are that it investigates multiple 
variables and that it likely to be ecologically valid in South Africa. 
 
6.4.1 Multiple variables under investigation 
 
From the literature surveyed, it is apparent that there is not a single factor 
that determines the academic performance of students.  The academic 
success or failure of a student cannot be attributed to a unitary factor, but 
to a multitude of variables that work either independently or in conjunction 
with other factors (Engelbrecht, 1999).  The present study investigated 
several variables as predictors of academic performance. These factors 
are highlighted in Figure 6. 
 
 
12
 1 
 
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
 _
  
Fi
 gu
 re
  
6.
  
 
Fa
 ct
 o
 rs
  
in
 ve
 st
 ig
 a
 te
 d 
in
  
th
 e
  
pr
 es
 e
 n
 t s
 tu
 dy
  
hi
 gh
 lig
 ht
 e
 d.
  
122  
 
6.4.2 Ecological validity 
 
Ecological validity implies the extent of generalisability from one context to 
another, specifically across geographic areas (Potter, 1998).  It is highly 
likely that the findings could be generalised to Information Technology 
students at other Technikons/Universities of Technology within South 
Africa.  Tshwane University of Technology has the largest student 
population amongst all South Africans, and students originate from all 
regions within the country. 
 
6.5 Limitations of the study 
 
The limitations of the study were the factors threatening the internal and 
external validity of the study, and the sample size. 
 
6.5.1 Threats to internal validity 
 
Internal validity refers to the degree of validity of the statements made 
regarding the relationship between the variables (Rosenthal & Rosnow, 
1991).  Maturation, instrument reactivity and history could have possibly 
reduced the internal validity of the present study, as follows:   
 
6.5.1.1 Maturation 
 
The period between the administering of the APA battery and the final 
examination was almost a year, which presented a substantial duration for 
the occurrence of processes that may have differentially changed in the 
subjects.  Those processes may have introduced other influences into the 
study that may have confounded the phenomenon observed, and the 
results obtained (Potter, 1998).  This could have been overcome by 
administering the APA battery shortly before the final examinations.  
123  
However, that would have raised an ethical issue, as there would have 
been insufficient time to provide effective intervention to students amongst 
whom developmental areas were identified. 
 
6.5.1.2 Instrument reactivity 
 
Instrument reactivity refers to the APA battery reacting with the variables it 
measures, thereby distorting the true measurements.  Within the social 
sciences, it is often difficult to know when reactivity is severe enough t be 
problematic.   There is a vast body of literature on the effects an 
experimenter can have on the behaviour of subjects, a number of which 
could be regarded as instrument reactivity (Potter, 1998). 
 
6.5.1.3 History 
 
Many events could have potentially affected the subjects in addition to the 
independent variables.  History refers to such environmental events 
besides the independent variables that may have occurred over the period 
of the investigation, which may possibly have impacted the results of all 
subjects in the study (Potter, 1998).  For example, the literature review 
cites research on specific role relationships as a sociological factor related 
to academic performance. 
 
6.4.2Threats to external validity 
 
External validity is indicative of the generalisability of the study.  Factors 
threatening external validity impair the ability to generalise the results of 
the study to other situations and contexts (Kerlinger, 1986).  Population 
validity was the most serious threat to external validity in the present 
study. 
 
 
124  
 
6.4.2.1 Population validity 
 
The problem of generalisability prevails, as the sample selected is so 
specialised that conclusions may only be generalised to a limited 
population (Potter, 1998). In terms of the predictors considered in this 
study, it may be difficult to generalise the conclusions to all courses 
presented at Tshwane University of Technology, especially with theory-
 oriented courses like Tourism Management, Nature Conservation or 
Horticulture. 
 
6.4.3 Sample size 
 
The sample size affected the manner in which the data were analysed. 
The sample size prevented the inclusion of all the variables in a single 
multiple regression analysis. There should have at least ten to twenty 
times as many respondents as variables (StatSoft, Inc., 2001).  Although 
the size of the sample was 133 respondents, depending on the variables 
included, due to the missing cases, the sample size ranged from 97 to 
107.  This implied that a maximum of ten predictor variables could be 
included in the regression, whereas the study consists of thirteen predictor 
variables. Hence, instead of one multiple regression analysis, there were 
three analyses. 
 
6.5 Recommendations  
 
6.5.1 ?Back to the drawing board? 
 
The changes suggested in Research question 2 should be implemented, 
in order to enhance the internal consistency of the test.  When modifying 
the questionnaires, there are some other findings that need to be 
considered too.  The intercorrelations across several of the questionnaires 
125  
suggest that the content of the questionnaires overlap to a large extent.  
Further analyses need to be executed, in order to streamline the 
questionnaires even further, thereby eliminating redundant items and 
questionnaires.  This would reduce the length of the APA battery, reduce 
testing time and enhance the psychometric properties of the battery.  
Although Note-taking Skills was one of the internally consistent tests, this 
test has to be revisited in its entirety, in an attempt to determine the 
inverse correlation with academic performance. 
 
6.5.2 Intervention programmes 
 
A significant correlation between English proficiency and reading speed 
was found, having implications for the manner in which English, and 
reading development courses are presented.  Presently, they are 
facilitated individually as independent programmes.  Programme 
managers should investigate the impact of integrating the two courses for 
maximum impact. 
 
6.5.3   Selection for remedial intervention 
 
Although the results of the discriminant function analysis indicate that all 
the academic proficiencies, except English proficiency, contribute to 
predicting whether a student will be successful or unsuccessful, it is 
questionable if the  results of the APA battery can be used as a basis for 
selection for remedial intervention.  Out of the eight predictors, four 
questionnaires are not internally consistent.  From the remaining four 
tests, Note-taking had a negative relationship with academic performance.  
Only three are left:  Reading speed, Time management and the 
Debilitating Stress scale on the Achievement Anxiety test.  With only three 
out of the nine scales having utility in the prediction of academic 
performance, the APA battery as a whole does not appear suitable for 
selection for remedial intervention. 
126  
 
6.6 Conclusion 
 
The present research project addressed two issues.  Firstly, the 
psychometric properties of the Academic Proficiency Assessment battery, 
and secondly the predictors of academic performance. 
 
Regarding the APA battery, the Time management, Note-taking skills, and 
Debilitating stress scale on the achievement anxiety questionnaire were 
found to be internally consistent.   The internal consistency reliability within 
several questionnaires was inadequate.  The English proficiency, Reading 
comprehension, Memorisation skills, Concentration skills and Motivating 
stress scale on the achievement anxiety questionnaire questionnaires 
require modification or replacement.  The intercorrelations across the 
questionnaires also suggest that the questionnaires may be evaluating the 
same factor, and would therefore benefit from further revision. 
 
In terms of the relationship between the APA battery and academic 
performance, a significant negative relationship was found between Note-
 taking Skills and academic performance, yielding a R2adj value of 8,3%. 
However, as this is an inverse relationship it would suggest minimal 
practical utility.  While these results indicated that the test battery 
accounted for low levels of variance in CGPA, a discriminant function 
analysis correctly predicted pass or failure in 72% of the students 
registered for the Information Technology course. This would indicate that 
all the proficiencies measured by the APA, with the exception of English 
proficiency, predicted the likelihood of a student being successful or 
unsuccessful to a fairly good extent.  
 
The present study is also consistent with the volume of existing research 
supporting Matric results as an effective predictor of academic 
performance.  Matric results, however, were positively related to CGPA 
127  
accounting for 11% of the variance in CGPA.  Although a poor fit, these 
results would suggest the value of using Matric results as one of the 
criteria for selection for admission into the Information Technology course 
at Tshwane University of Technology.  
 
Cumulatively, Matric results and Note-taking skills contributed 13,34% 
towards accounting for the variance in CGPA, which was also a poor fit.  It 
should be noted that due to the presence of multicollinearity, some 
variables may have been excluded from the model although they may be 
significant.  The intercorrelations across the questionnaires also suggest 
that the questionnaires may be evaluating the same factor, necessitating 
further revision and streamlining of the APA battery. 
 
The biographical variables age, gender and predominant language of 
communication failed to have a significant relationship with academic 
performance in this study. 
 
Overall the results of the study would thus suggest that matric results 
remain the best predictor of academic performance, but predict a low 
amount of the variance in academic performance.  
 
Given the lack of reliability in the majority of the tests, the use of the APA 
battery for selection for remedial intervention, is not yet justified. The high 
proportion of pass rates of students predicted by the tests in combination 
would suggest that work on improving the psychometric qualities of the 
APA should be undertaken, and would also indicate that additional 
individual predictors other than those examined in this study, should be 
examined in future studies with Information Technology students.  
 
128  
 
 
Chapter 7 
REFERENCE LIST 
 
 
Ainley, J., Graetz, B., Long, M. & Batten, M. (1995).  Socio-economic  
  status and school education.  Canberra:  Australian Government  
 Publishing Service. 
 
Anderson, E.S. & Keith, T.Z. (1997).  A longitudinal test of a model of  
academic success for at-risk high school students.  Journal of  
Educational Research, 90(5), 259-269. 
 
Baehre, M.R. & Gentile, J.R. (1991).  Cumulative effects of the keyword  
 mnemonic and distributed practice in learning social studies facts: A  
 classroom evaluation.   Journal of Research in Education, 1(1),  
 p.70- 78. 
 
Beaty, D. & Barling, J. (1982).  Positive exam results ? Without stress.   
 Isanda:  McGraw- Hill Book Company. 
 
Bingham, W. van D. (1937). Aptitudes and aptitude testing.  New York:   
Harper Bros. 
 
Bosher, S. & Rowenkamp, J. (1992).  Language proficiency and academic  
 success: The refugee/immigrant in higher education  (Report No.  
 HE026173).  Minnesota:   University of Minnesota.  (ERIC Document  
 Reproduction Service No. ED353914). 
 
Bot, M.  (1994).  Age and Standard 10 examination results.  Edusource  
Data News, No.7/December 1994.  Johannesburg:  The Education  
Foundation. 
 
Bowerman, B.L. & O? Connell, R.T.  (1990).  Linear statistical models ? An  
applied approach.  Pacific Grove, CA:  Duxbury. 
 
Brown, F.E. (1991).  The effect of structured notetaking on student  
success in calculus 1.  Community/Junior College Quarterly of  
Research and Practice, 15(3), 261-270. 
 
Brown, W.F. (1994).  Taking class notes.  United States of America:  The  
American College Testing Program. 
 
129  
Brusselmans-Dehairs, C. & Henry, G.F. (1994).  Introduction to special  
issue on gender and mathematics.  International Journal of  
Educational Research, 21(4), 353-371. 
 
Carrier, C. (1984).  The effects of facilitative and debilitative achievement  
anxiety on notetaking.  Journal of Educational Research, 77(3), 133- 
138. 
 
Celliers, M. (2000). The academic development model for university and  
technikon students ? Meeting the demands of the 21st century.  
Unpublished Magister Educationis thesis, University of South Africa,  
Pretoria. 
 
Chamorro-Premuzic, Y. & Furnham, A. (2003).  Personality predicts  
academic performance:  Evidence from two longitudinal university  
samples.  Journal of Research in Personality, 37(4), 319-339. 
 
Cox, R.H. (1990).  Sport Psychology ? Concepts and Applications (3rd  
ed.).  Wisconsin:  Brown and Benchmark. 
 
Cronbach, L.J. (1951).  Coefficient alpha and the internal structure of  
tests.  Psychometrica, 16, 297-334. 
 
Department of Education.  (1997).  Education White Paper 3:  A  
programme for the  Transformation of higher education.  General  
Notice 1196 in the Government Gazette 18207 of 1997-08-15.   
Pretoria. 
 
Diseth, A. (2002).  The relationship between intelligence, approaches to  
learning and academic performance.    Scandinavian Journal of  
Educational Research, 46(2), 219-232. 
 
Diseth, A. & Martinsen, O. (2003).  Approaches to learning, cognitive style,  
and motives as predictors of academic performance.  Educational  
Psychology, 23(20), 195-208. 
 
Dockrat, S.Y. & Grobler, H. (1999, June).  Reading skills enhancement:   
An impact study.  Paper presented at the Annual Tshwane University 
of Technology Academic Support Seminar.  Access and Retention:   
Implications for Tertiary Institutions, South Africa. 
 
Driskell, J.L.  (2003).  A study of the effectiveness of a guided note-taking  
and study skills system upon the level of academic success among  
entering university of Idaho Freshman (Report no. CS202889).   
University of Idaho (Eric Document Reproduction  
Service No. ED127622).  
 
 
130  
du Plessis, P. (1996). Learning and study strategies inventory as a tool in  
academic support.  Retrieved April 14, 2003, from  
http://herdsa.org.au/confs/1996/duplessis.html. 
 
du Toit, L.B.H. (1970).  Survey of study habits and attitudes.  Form C 
 (Students).  Pretoria:  Human Sciences Research Council. 
 
du Toit, L.B.H. (1970).  Survey of study habits and attitudes.  Manual.   
Pretoria:  Human Sciences Research Council. 
 
Eggly, S. Musial, J. & Smulowitz, J. (1999).  The relationship between  
 English language  proficiency and success as a medical resident.   
 English for Specific Purposes.  18(2),  201 ?208. 
 
Electronic Learning and Study Strategies Inventory.  Retrieved April 14,  
2003, from http://www.gwu.edu/~counsel/elec_learn.html. 
 
Ellis, D.B. (1997).  Becoming a master student: tools, techniques, hints,  
ideas, illustrations, instructions, examples, methods, procedures,  
processes, skills, resources, and suggestions for success.  Boston:  
Houghton Mifflin. 
 
Engelbrecht, G.S.  (1999).  Riglyne vir ?n akademiese  
ontwikkelingsprogram vir die Fakulteit Ekonomiesewetenskappe aan  
die Technikon Pretoria. Unpublished Philosophiae Doctor thesis,  
University of Pretoria. 
 
English Word Power  [Computer software].  Johannesburg:  EWP  
Solutions CC. 
 
Eno, D., McLaughlin, G.W., Sheldon, P. & Brozovsky, P. (1999).   
Predicting freshman success based on high school record and other  
measures.  Virginia Tech IRPA Report, 96(51). 
 
Enterprise Guide [Computer software].  Cary:  SAS Institute Inc. 
 
Erasmus, P.F. (1997).  Questions of standardisation, validity and reliability  
of assessment tools.  Erasmus, P.F. (Ed.).  Unfair labour practice,  
the worker and assessment:  What every human resource  
practitioner should know.  Johannesburg:  RNATA. 
 
Farbaugh-Dorkins C. (1991).  Beginning to understand why older students  
drop out of college.  Paper presented at the Annual Forum of the  
Association for Institutional Research. 
 
 
 
 
131  
Fields, H. & Cosgrove, J. (2001).  Performance in General Psychology and  
reading level (Report No. SLCC-TM-01-6).  Missouri:  Saint Louis  
Community College. (ERIC Document Reproduction Service No.  
ED454920). 
 
Fourie, C.M. (1992).  ?n Omskrywing van die meting van akademiese  
prestasie vanuit die RAU perspektief.  Journal of Higher Education,  
6(1), 40-45. 
 
Fourie, C.M. (1990).  Die invloed van die voorgestelde  
matrikulasievrystellingsvereistes van die GMR op eerstejaarinname.   
RAU Bulletin vir dosente, 18(2), 45-54. 
 
Fourie, & Venter (2000).  Lector 2000 [Computer software].  Pretoria:   
Ecoltech. 
 
Foxcroft, C. & Roodt, G. (2001).  An introduction to psychological  
assessment in the South African context.  Cape Town:  Oxford  
University Press. 
 
Friedenberg, J.E. & Curry, W.C. (1981, May).  English proficiency and the  
 bilingual university student [Abstract].  Paper presented at the Annual  
 International Bilingual Education Conference, Boston, MA. 
 
Galpin, J. (2001).  Statistical research design and analysis (Study guide  
for STATS511).  Johannesburg:  University of Witwatersrand. 
 
Graham, J.G. (1987).  English language proficiency and the prediction of  
academic Success.  TESOL Quarterly, 21(3), 505 ?521. 
 
Gregory, R.J. (1996).  Psychological testing:  History, principles and  
applications (2nd ed.)  Boston:  Allyn and Bacon. 
 
Guilford, J.P. (1959).  Personality.  New York:  Mc-Graw Hill. 
 
Gourley, B.M.  (1992).  Financial management in universities.  South  
African Journal of ducation, 6(3), 70-72. 
 
Hatcher, L. (1994).  A step-by-step approach to using the SAS system for  
factor analysis and structural equation modelling.  North Carolina:   
SAS Institute. 
 
Hae-Song, P. & Bauer, S. (2003).  Parenting practices, ethnicity, socio- 
economic status and academic achievement in adolescents.   
Educational Administration Abstracts, 38(3), 309-311. 
 
Hardy, L., Jones, G. & Gould, D. (1996).  Understanding psychological  
preparation for sport.  Theory and practice of elite performers.  
 Chichester:  John Wiley & Sons. 
132  
 
Hawson, A. (1996).  A neuroscientific perspective on second-language  
learning and academic performance. 
 
Horn, S. (1991).  Concentration!  How to focus for success.  California:   
Crisp Publications, Inc. 
 
 
Hortacsu, N. (1995). Parents? education levels, parents? beliefs, and child  
outcomes.  Journal of Genetic Psychology, 156(3), 373-384. 
 
Hortacsu, N., Ertem, L., Kurtoglu, H. & Uzer, B. (1990).  Family  
background and individual measures as predictors of Turkish primary  
school children?s academic performance.   Journal of Psychology,  
124, 329-337. 
 
Hosmer, D.W. & Lemeshow, S. (1999). Applied survival analysis:  
Regression modeling of Time to event data. New York:  Wiley- 
Interscience. 
 
Hosmer, D.W. & Lemeshow, S. (2000).  Applied logistic regression (2nd  
ed.) New York:  Wiley-Interscience. 
 
Howell, D.C. (1992).  Statistical methods for psychology (3rd ed.)  
California:  Duxbury Press. 
 
Hua, T.M., Williams, S. & Hoi, P.S. (2003). Using the Biggs? Study Process  
Questionnaire as a diagnostic tool to identify ?at-risk? students ? a  
Preliminary Study.  Retrieved on April 14, 2003, from  
http://ipdweb.np.edu.sg/It/sept01/spq.htm. 
 
Humphreys, L.G. (1986).  An analysis and evaluation of test and item bias  
in the prediction Context.  Journal of Applied Psychology, 71(2), 327- 
333. 
 
Huysaman, G.K. (1999).  Psychometric explanations for the poor  
predictability of the tertiary-academic performance of educationally  
disadvantaged students.    South African Journal of Education, 13(1),  
132-138. 
 
Hult, R.E. (1984, February).  Differential effects of note taking ability and  
  lecture encoding structure on student learning [Abstract].  Paper  
  presented at the Annual Meeting of  the Eastern Educational  
  Research Association.  West Palm Beach, FL. 
 
Kahle, J.B., Parker, L.H., Rennie, L.J. & Riley, D. (1993).  Gender  
differences in science education:  Building a model.  Educational  
Psychologist, 28(4), 379-404. 
 
133  
Klitgaard, R.  (1986).  Elitism and meritocracy in developing countries ?  
Selection policies for higher education. Maryland:   The Johns  
Hopkins University Press. 
 
Kotz?,H.N. (1994). Keurringsmodelle vir universiteitstudierrigtings: 'n  
  Psigometriese  ondersoek. Unpublished Philosophiae Doctor thesis,  
  Potchefstroom University for Christian Higher Education. 
 
Kriel, H. (2000).  Addressing student development:  The impact of a life  
skills Programme on the academic performance of first-year  
students. Technikon Pretoria, Bureau for Academic Support. 
 
Kriel, H.  (2001).  The validation of a potential assessment battery for  
Engineering Technology students.  Unpublished Magister Artium  
thesis, University of Pretoria. 
 
Kriel, H. (2002).  ?n Uitgebreide keuringsbattery vir studente in  
Inligtingstegnologie. Technikon Pretoria, Bureau for Academic  
Support. 
 
Kruger, H.B.  (1972).  Die voorspelbaarheiud van akademiese prestasie  
deur die verskille tussen nie-verbale en verbale toetsintelligensie- 
prestasies. Unpublished Philosophiae Doctor thesis, Potchefstroom  
University for Christian Higher Education. 
 
Lahmers, A.G. &  Zulauf, C.R. (2000).  Factors associated with academic  
time use and Academic performance of college students:  a recursive  
approach.  Journal of College Student Development, 41(5), 544-556. 
 
LASSI ? Background.  Retrieved April 14, 2003, from  
http://www.hhpublishing.com/_assessments/LASSI. 
 
Lavin, D.E.  (1965).  The prediction of academic performance.  New York:   
Russel Sage Foundation.   
 
Learning and Study Strategies Inventory (LASSI).  Retrieved April 14,  
2003, from http://www.esit.uc.edu/Learning/LASSI.aspx. 
 
Lea, S. (1997). PSY6003 Advanced statistics: Multivariate analysis II:  
Manifest variables analyses.  Topic 1: Multiple regression:  
Revision/Introduction.  Retrieved October, 15, 2004, from the  
University of Exeter, Department of Psychology website:  
http://info.ex.ac.uk/~SEGLea/multvar2. 
 
Louw,D.A.  (1990). Inleiding tot die psigologie. Cape Town:  Duncan Wulf. 
 
 
 
134  
Manalo, E. (1996).  Effectiveness of an Intensive Learning Skills Course  
for University Students on Restricted Enrolment.  Higher Education  
Research and Development, 15(2), 189-199. 
 
McCombs, B.L. & Dohikshanic, J.L. (1980, April).  The Study Skills  
Questionnaire (SSQUES):  Preliminary validation of a measure for  
assessing students? perceived area of weakness.  Paper presented  
at the Annual Meeting of the American Educational Research  
Association, Boston, MA. 
 
McCrae, R. & Costa, P.T., Jr. (1987).  Validation of the five-factor model of  
Personality across instruments and observers.  Journal of Personality  
and Social Psychology, 52(1), 81-90. 
 
McMillan, J.H. & Schumacher, S. (1993),  Research in education ? a  
conceptual introduction.  New York:  Harper Collins. 
 
Meece, J.L. & Eccles, J.S. (1993).  Introduction:  Recent trends in  
research on gender and Education.  Educational Psychologist, 28(4),  
313-319. 
 
Messick, S. (1989).  Validity.  In R.L. Linn (Ed.).  Educational  
measurement (3rd ed.).  New  York:  American Council on  
Education. 
 
Miller, R. (1992).  Double, double, toil and trouble:  the problem of student  
selection.  South African Journal of Higher Education, 6(1), 98-104. 
 
Monteith, J.L. de K. (1988).  Faktore wat die akademiese prestasie van  
eerstejaars beinvloed.  RAU Bulletin vir dosente, 20(1), 23-24. 
 
Muffo, J.A., Dickey, J.W. & Bodo, B.J. (1999, May).  Using multivatiate  
logistic regression and quantitative cyber quest to predict freshman  
retention.  Paper presented at the Annual Forum of the Association  
for Institutional Research, Seattle, Washington. 
 
Motulsky, H. (2002).  Multicollinearity in multiple regression.  Retrieved  
October 4, 2002, from http://www.graphpad.com/multicollinearity in  
multiple regression.htm. 
 
Murphy, K.R. & Davidshofer, C.O. (1996).  Psychological testing:   
Principles and applications (4th Ed.).  New Jersey:  Prentice Hall. 
 
Mutua, R.W., Omulando, S.J. & Otiende, J.E. (1992).  Study and  
communication skills.  Hong Kong:  The Macmillan Press Ltd. 
 
 
 
135  
Niebuhr, K.E. & Niebuhr, R.E. (1999).  An empirical study of student  
relationships and academic performance.  Education, 119(4), 679- 
682. 
 
Norton, L.S. & Crowley, C.M. (1995).  Can students be helped to learn  
 how to learn?  an evaluation of an approach to a learning programme  
 for first-year students .   higher education, 29(3), 307-328. 
 
Nunally, J.C. (1978).  Psychometric Theory.  New York: McGraw-Hill. 
 
Nunally, J.C. & Bernstein, I.H.  (1994).  Psychometric Theory (3rd ed.).   
 New York:  McGraw-Hill. 
 
Owen, K. &  Taljaard, J.J. (1996).  Handbook for the use of psychological  
and scholastic tests of the HSRC.  Pretoria:  Human Sciences  
Research Council. 
 
Padma, M.S. Research in Correlates of Achievement.  A Trend Report.   
 Retrieved April14,  2003 from  
 http://shikshanic.nic.in/cd50years/g/Z/91/0Z910J01.htm. 
 
Pavlich, G. & Orkin, M. (1993). Diversity and quality:  academic  
 development at South African tertiary institutions. Johannesburg:  
 Westro Parktown Publishers. 
 
Pienaar, J.L. (1991).  Die voorspelling van akademiese sukses met behulp  
van die AAT.  Unpublished Magister Artium (Psychology) thesis,  
VISTA University, Pretoria. 
 
Plake, B. (1981, April). A validity investigation of the Achievement Anxiety  
Test [Abstract].  Paper presented at the Annual Meeting of the  
National Council on Meaasurement in Education, Los Angelos, CA. 
 
Plake, B. (1981).  A validity investigation of the Achievement Anxiety Test.   
Educational and Psychological Measurement, 1(4), 215-222. 
 
Plug,  C. (1996).  An evaluation of psychometric test construction and  
psychological services supported by the HSRC.  Unpublished  
manuscript, University of South  Africa. 
 
Potter, C. (1998).  Honours course in research design and analysis.   
Reading Pack for catch up tutorials.  Johannesburg:  University of  
Witwatersrand. 
 
Potter, C.S. and Jamotte, A.N. (1985). African matric results: Dubious  
indicators of academic merit. Indicator SA, 3 (1), 10-13. 
 
 
136  
 
Rademeyer, M.M. & Scheepers, J.M.  (1998).  Voorspelling van die  
akademiese sukses van eerstejaarstudente.  Journal of Industrial  
Psychology, 24(1), 33-40. 
 
Rosenthal, R & Rosnow, R.L. (1991).  Essentials of behavioral research  
(2nd ed.).  Boston:  McGraw Hill. 
 
Ross, K.N., Farish, S., & Plunkett, M.  (1988).  Indicators of Socio- 
economic disadvantage for Australian schools.  Victoria:  Deakin  
University. 
 
Rothman, K.G. & Greenland, S. (1998).  Modern epidemiology.  
Philidelphia: Lippincott Williams & Wilkins. 
 
Rousseaux, K.  (1993).  Rurality and participation in schooling. 
 Queensland:  Department of Education. 
 
Rugsaken, K.T., Robertson, J.A. & Jones, J.A. (1998).  Using the Learning  
and Study Strategies Inventory Scores as Additional Predictors of  
Student Academic Performance.  NACADA Journal, 18(1), 20-26. 
 
Rutherford,M.  (1997).   Working Memory Revisited:  Can we use it to  
  select students for science courses?  International Journal of Science  
  Education, 19(8), 939-955. 
 
Samkin, J.G. (1996).  Should matriculation results represent the sole  
admission criterion to the first-year Accounting programme ?  
preliminary evidence from the University of Durban-Westville.  South  
African Journal of Education, 16(2), 117-122. 
 
SAS Institute Inc. (1995).  Logistic regression examples using SAS  
system.  Version 6, 1st Ed.  Cary, NC:  SAS Institute. 
 
Saville-Troike, M. (1991).  Teaching and testing for academic  
performance:  The role of language development.  Focus, 4, 17. 
 
Shephard, J. (1987).  College study skills (3rd ed.).  Boston:  Houghton  
Mifflin. 
 
Skuy, M., Zolezzi, S., Mentis, M., Frijdhorn, P. & Cockroft, K. (1996).   
Selection of Advantaged and disdvantaged South African students  
for university admission.  South African Journal for Higher Education,  
10(1), 110-117. 
 
 
 
 
137  
Smit, E.H.  (1992).  An outline for implementing or improving selection  
procedures for prospective technikon students, with specific  
reference to the engineering courses.  Paper presented at the  
SSCSA conference, South Africa. 
 
Smit, G.J. (1996).  Psychometrics ? aspects of measurement.  Pretoria:  
Kagiso Publishers. 
 
Smith, D.P.J.  (1979).  Taalgebruiksvaardigheid en universiteitsprestasie.   
Publikasiereeks van die RAU, C(20), 41-47 
 
Smith, M. & Smith, G. (1988).  A study skills handbook.  Australia:  Oxford  
University Press. 
 
Snow, R.E. (1992).  Aptitude theory:  yesterday, today and tomorrow.   
Educational Psychologist, 27(1), 5-32. 
 
Stanley, J.C. (1971).  Reliability.  In R.L. Thorndike (Ed.).  Educational  
measurement.  Washington, D.C.:  American Council on Education. 
 
Stallworth-Clark, R. (1996, April).  The teaching-learning process and  
postsecondary at-risk Reading students:  cognitive, metacognitive,  
affective, and instructional variables explaining academic  
performance [Abstract].  Paper presented at the Annual Meeting of  
the American Educational Research Association, New York. 
 
Study Process Questionnaire.  Retrieved July 8, 2002, from University of  
South Africa, Bureau for Student Counselling and Career  
Development Web site:  http://www.unisa.ac.za/dept/bsccd/spq.html. 
 
StatSoft, Inc. (2001). Statistica [Computer software].  Tulsa:  StatSoft, Inc. 
 
Swanepoel, S.  (2002).  The design of a differential selection model for  
specific study disciplines at a technikon. Unpublished Doctor  
Commercii (Human Resources Management) thesis, University of  
Pretoria. 
 
Tabachnick, B.G. & Fidell, L.S. (1996).  Using multivariate statistics.  New  
York:  Harper Collins. 
 
Taylor, T. (1978, February).  Relationship between oral English language  
proficiency and social acceptance, and scholastic achievement  
among Mexican-American and Anglo-American fifth grade children  
[Abstract].  Paper presented at the Annual Meeting of the Texas Joint  
Council of Teachers of English, San Antonio. 
 
Tinklin, T. (2003).  Gender differences and high attainment.  British  
Educational Research Journal, 29(3), 307-326. 
138  
Thorndike, R.L. (1951).  Reliability.  In E.F. Lindquist (Ed.), Educational  
Measurement (p.609).  Washington, DC:  American Council on  
Education. 
 
Van Zyl, M.A., Terblanche, S. & Jacobs, G.J. (1992).  Die profiel van ?n  
potensieel-suksesvolle eerstejaarstudent in Maatskaplike Werk:  ?n  
Benadering tot keuring.  Maatskaplike Werk, 28(1), 33-43. 
 
Venter, J. & Krause, C. (2000).  Lector 2000 [Computer software].   
Ecoltech:  Pretoria. 
 
Vogiatzi, E.  (2002).  Problems in regression analysis and their corrections.   
Retrieved October 15, 2004, from  
http://www.geocities.com/qecon2002/founda10.html. 
 
Walter, P.  (1973).  Two essential study skills for the community college  
student.  Reading World, 12(4), p.239-245. 
 
Watson, A.K. & Munroe, E.E. (1990).  Academic performance:  A study of  
relationships of IQ, comunication apprehension, and teacher  
perception.  Communication Reports, 3(1), 28-37. 
 
Watson, J. (1988).  Achievement Anxiety Test:  Dimensionality and utility.   
Journal of Educational Psychology, 40(4), 585-591. 
 
Weiten, W. (1989).  Psychology ?Themes and variations.  California:   
Brooks/Cole Publishing Company. 
 
Weinstein, C.E. (1987).  LASSI user?s manual.  Austin:  University of  
Texas. 
 
Wince, M.H., & Borden, V.M. (1995).  When does student satisfaction  
matter? Paper presented at the Annual Forum of the Association for  
Institutional Research, Boston, MA. 
 
Williams, K. (1989).  Study skills.  Basingstoke : Macmillan Education. 
 
Young, J.W. (1991)  Gender bias in predicting college academic  
performance:  A new approach using item response theory.  Journal  
of Educational Measurement, 28(1), 37-47. 
 
Zaaiman, H. (1998).  Selecting students for mathematics and science.   
Pretoria:  HSRC Publiishers. 
 
Zhang, Z. & Richarde, R.S. (1999, April).  Freshman academic  
performance:  A structural Equation model [Abstract].  Paper  
presented at the Annual Meeting of the American Educational  
Research Association, Montreal. 
139  
 
Zietsman, A. & Gering, M.  (1985).  Kriteria vir die keuring van  
eerstejaarstudente.  Bulletin vir dosente ? RAU, 17(3), 1-11. 
 
Zolezzi,S.A.  (1995). The effectiveness of dynamic assessment as an  
  alternative aptitude testing strategy.  Unpublished Doctor of  
  Education thesis, University of South Africa,  Pretoria. 
140  
 
 
Appendix A 
FIGURES 
 
  
 
 
 
 
 
               
  
 
 
            
   
 
 
               
 
 
 
 
 
 
 
 
 
 
 
 
_____________________________________________________________________  
Figure 3.  Study Process Questionnaire (SPQ). 
 
SPQ 
Surface 
domain 
Achieving 
domain 
Deep 
domain 
  
Motives 
  
Motives 
  
Motives 
  
Strategies 
  
Strategies 
  
Strategies 
? Extrinsic 
motives 
? Fear of 
failing 
 
? Reproductive 
or rote 
learning 
? Minimalistic 
learning 
 
Develop 
personal 
meaning for 
subject 
material 
 
 
Understand 
and relate 
understanding 
to other 
subjects 
 
 
  
Get high marks 
 
 
 
? Strategic 
learners 
? May use 
surface or 
deep 
approach 
 
 
Poor 
academic 
performance 
 
Academic 
performance 
may be 
adversely 
affected 
 
Maximises 
academic 
performance 
141  
   
Appendix B 
TABLES 
 
 
Table 3 
 McCrae and Costa?s Five-Factor Model of Personality  
Factor Description 
Neuroticism Anxious, insecure, guilt-prone, self-conscious 
Extraversion Talkative, sociable, fun-loving, affectionate 
Openness to experience Daring, nonconforming, showing unusually broad 
interests, imaginative 
Agreeableness Sympathetic, warm, trusting, co-operative 
Conscientiousness Ethical, dependable, productive, purposeful 
From ?Psychology ? Themes and Variations? by W. Weiten, 1989, p.422.  Copyright 1989 by Wadsworth, Inc. 
 
142  
Table 4  
LASSI Scales  
Scale 
Coeffecient 
Alpha 
Test-retest 
Correlation 
Coefficient 
Attitude .72 .75 
Motivation .81 .84 
Time Management .86 .85 
Anxiety .81 .83 
Concentration .84 .85 
Information Processing .83 .72 
Selecting main ideas .74 .78 
Study Aids .68 .75 
Self Testing .75 .78 
Test Strategies .75 .78 
 From  ?LASSI User?s Manual? by C.E. Weinstein, 1987, p.13.  Copyright 1987 by H&H Publishing Company, Inc.