Street Pattern and Traffic Safety

Over the last fifty years, the lollipop design has become the basic building block of many suburban road networks in most North American cities. Although developed to improve the social living environment, this combination of cul-de-sac and loop streets has the support of many traffic engineers because of its traffic calming effects. Perhaps due to its intuitive appeal, few studies have been conducted to examine the impact of this design on road crashes. While the social benefits and drawbacks of different types of street patterns and urban forms have been discussed extensively in the literature, little attention has been devoted so far to the traffic safety effects of the different street network designs. The main objective of this program of research is to explore the effect of neighbourhood street pattern on the frequency and severity of crashes.;Our study found that limited-access street patterns were associated with fewer traffic collisions. In addition, although limited access design increased the injury risks for pedestrians and cyclists involved in a crash, it reduced the crash severity in vehicle-vehicle collisions and vehicle-fixed object collisions. When the results of all crash frequency and crash severity analyses were summarized, this study suggested that limited-access street patterns might be safer than the traditional gridiron pattern. These traffic safety benefits of limited-access street patterns should be considered together with the social benefits of increased connectivity in the traditional grid pattern when deciding on the optimal urban form for a community.;In addition to examining the safety effects of different neighbourhood street patterns, this program of research also examined two important methodological issues that had received little attention in the literature. First, our study estimated a variety of statistical models that incorporated different theoretical assumptions about the effects of contributing factors on the severity of crashes, including the multinomial logit model, logistic regression model, ordered logit model, the heterogeneous choice model, generalized ordered logit model and partially constrained generalized ordered logit model. The results from the different models showed that our main conclusion was robust with respect to different statistical methods used.;Second, this research investigated different methods to aggregate crashes with different severity levels to examine the robustness of our policy recommendations. Many risk factors and countermeasures in road safety are expected to have differential effects on crashes of different severity levels. Since placing relative values on the different injury levels, especially fatal injury, is inherently contentious, most traffic safety analyses chose to examine these effects separately. In this study, however, crashes of different severity level were aggregated using different methods to obtain an overall conclusion about the safety effect of different neighbourhood street patterns. The results from the different models showed that our main conclusion was robust with respect to different aggregation schemes used.;Using data from the City of Calgary, traffic safety performance functions using the count data models, ordered and unordered response models were estimated to provide policy makers with evidence based recommendations on the type of subdivision that would be safer for Calgary and other rapidly expanding cities. Since the focus of this research was on the effects of different neighbourhood road design on crashes, it used the neighbourhood as the unit of analysis which enabled the model to capture some of the effects of neighbourhood road designs. Our study also controlled for the confounding effects of many external variables including road engineering, environmental conditions, crash characteristics, vehicle features and driver attributes.