A Multi-Procedural Approach to Evaluating Walkability and Pedestrian Safety

Abstract

Walking has sound health benefits and tends to be a pleasurable experience requiring no fuel, fare, license, nor registration. Whilst walking is recommended as part of physical activity, it is necessary to provide a conducive and safe walking environment. In an effort to determine an optimum combination of infrastructure that would create walkable, transit-oriented neighborhoods eliminating unnecessary motorized trips; various approaches evaluating an assortment of features in the walking environment have been implemented. However, some factors such as crash risk which have an essential contribution to the suitability of the walking environment have yet to be considered. Therefore the objective of this study was to quantify the walking environment, by developing a comprehensive walkability index which reflects the condition of the walking environment as well as pedestrians' perceptions of the walking environment. Developing the walkability index included three sub-objectives as follows: 1. Incorporate crash risk in the development of walkability indices which has not been done in previous walkability studies. An overall safety index was designed to estimate safety in the built environment in a more complete form. 2. Analyze the impact of features in the built walking environment on walking for recreational or utilitarian purposes. The analyses also determined whether sampled residents' perception of their walking facilities is comparable to the objective audit observations in various categories. 3. Identify features in the built environment that influence resident perception of their walking environment. This involved analyzing patterns and relationships between features in the walking environment and resident perceptions. Results would relate resident perceptions and walking environment features using calibrated statistical models. The study methodology included conducting a resident survey, an audit of objectively measured features in the walking environment and a pedestrian safety analysis. The survey collected residents' perceptions of their walking environment expressed using natural language. A perception quality grade of walkability based on resident perceptions was developed from the survey data. An audit of survey neighborhoods was performed by a trained auditor using Google earth, maps, and site visits. Features in the walking environment such as driveways, signals, and crosswalks among others, were measured on a segment by segment basis. Using the various features measured on a segment, an audit quality index for walkability was developed for each neighborhood. A crash index was also developed as a function of population and commercial land use within the survey neighborhoods. The findings of this study are expected to enhance evaluation of walking environments. The safety index incorporating crash risk and objectively measured safety elements provides a more representative indicator of safety levels within the walking environment. In addition, crash data increases objectivity to neighborhood audits depending on how audit scores are estimated. The fuzzy logic approach to estimating resident perceptions of the walking environment enables analysis of imprecise information to obtain logical output through computing with words. As such, residents' opinions which are analyzed in an approximate framework similar to the human ability to manipulate and reason with perceptions are more consistent with initial resident evaluations. With improved walkability estimation, decision makers are better equipped during planning to select appropriate strategies that encourage walking in a safe environment for recreational and utilitarian purposes. Comparison of developed audit quality walkability indices, with and without crash data indicates significant differences in walkability indexes. Neighborhoods with initial high walkability indexes ranked much lower after crash data integration. Even without statistical significance, crash data provides more objectivity to audit quality indexes based on depending on data collection and reduction. The study used multinomial logit to identify parameters that influence walking frequency. Results indicate that land use, and aesthetic and amenities perceptions have a significant relationship with walking frequency. This is intuitive because more varied land uses not only attract more pedestrians, but also provides opportunities for trip chaining. As expected, better aesthetics and amenities and infrastructure are associated with higher walking frequencies. Both aesthetics and amenities and land use perception were correlated with safety, directness and continuity perceptions, implying improving the perception of one category was bound to have an impact on another perception category. The study also used mixed models to identify features in the built environment that influence the multinomial model perceptions that in turn influence walking frequency. Results from the continuity, directness, land use, aesthetics and amenities perception models are as expected. For example, neighborhoods with initial low land use perception are likely to be more sensitive to the presence of new commercial premises (e.g. small convenient store) nearby. However, directness-audit parameter serves as both a disincentive and incentive. To land use perception, increasing directness features results in uninhibited access to land uses which increases walking frequency. Conversely, increase in uninhibited access results in lower safety perception. Intuitively, enclosed communities have lower traffic flows as well as speed limits that are conducive for pedestrian activity as well as providing buffers from traffic. Overall, results indicate the need for a transactional evaluation approach, in which pedestrian behavior is multiply influenced by environmental features, perception of the walking environment, as well as social and cultural aspects.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 6 citations exist upstream.