Why do people walk on this street?: comparing quantitative and qualitative measures of imageability and their association with pedestrian patterns

dc.contributor.authorMsingaphantsi, Mawabo
dc.descriptionA thesis submitted in fulfilment of the requirement for the degree Doctor of Philosophy to the Faculty of Engineering and the Built Environment, School of Architecture and Planning, University of the Witwatersrand, Johannesburg, 2022
dc.description.abstractImageability, the quality of a place that makes it distinct, recognisable, and memorable, is traditionally considered to make places pleasant and attractive to pedestrians. The use of primarily qualitative (rather than quantitative) descriptions to discuss the concept of imageability poses a challenge for the application of this concept in physical designs in practice, where designers work with environmental features that have specific dimensions and where they must decide how much of each feature (building height, number of courtyards, number of trees etc.) is necessary to make the environment imageable. There is also wide disagreement in urban design theory and practice about what factors contribute to imageability. Quantitative models attempt to address these challenges by producing operational definitions of imageability with strictly defined variables (factors) that are based on the mathematical relationships between physical environmental features (such as building shape or street length) and the occurrence of imageability. The chief benefit of these models for urban design is their potential use as a means to measure and describe the presence of imageability in a given place. However, the drawback of models is the limited number of verification studies to test their applicability in different contexts. The Ewing model is a street-based statistical model that uses a street audit to describe how imageable a place is from the point of view of a pedestrian on a street. The model identifies eight variables that have a statistically significant correlation (R>0.6) with imageability (Ewing, Clemente, Handy, Brownson, & Winston, 2005). In this study I apply the Ewing model to a low-density environment to measure the imageability of part of Diepkloof (Zone 5), a former black township in Johannesburg, South Africa. I use sketches and qualitative descriptions to validate measurements taken on 30 streets. The purpose of the study is to determine the extent to which environmental features such as imageability can explain pedestrian patterns in a neighbourhood. I tested for correlation between pedestrian activity and imageability and then created a linear regression model to predict pedestrian volume on a given street based on the level of imageability on that street. My conceptual framework, however, demonstrated that imageability has three key aspects (structure, identity and meaning) and that different quantitative models have in-built assumptions that privilege one or more of these aspects and may affect how the resultant measurements should be interpreted. I use mapping to illustrate other potential factors of imageability (as described in the literature and in other models) and argue that these represent conceptual gaps in the Ewing model that should be considered when interpreting the model’s outputs and their correlation to pedestrian patterns. The results of applying the Ewing model in Diepkloof Zone 5 show low levels of imageability, which is consistent with my qualitative assessment of the site, as lower densities reduce the potential for composition because the environment has fewer elements. Bivariate linear regression was found to be an inadequate measure of the correlation between imageability and pedestrian activity. These simple linear regression models had R2 values of less than 0.65 and had many outliers, which suggested that there were factors outside of the model that had a significant effect on pedestrian activity. When multiple regression is used to account for other neighbourhood conditions, correlation increased and the R2 value (which describes the models’ predictive capacity). There are three statistically significant variables (with p-values less than 0.05): street length, street integration and imageability
dc.description.librarianTL (2024)
dc.facultyFaculty of Engineering and the Built Environment
dc.schoolSchool of Architecture and Planning
dc.subjectEwing model
dc.subjectPedestrian patterns
dc.titleWhy do people walk on this street?: comparing quantitative and qualitative measures of imageability and their association with pedestrian patterns
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Msingaphantsi - PhD Submission - 2022-12-05.pdf
10.58 MB
Adobe Portable Document Format
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
2.43 KB
Item-specific license agreed upon to submission