Estimating pedestrian and cyclist activity at the neighborhood scale
Introduction
Good estimates of the total amount of bicycle and pedestrian activity on our roads are needed for two main purposes. First, knowing how much cyclists and pedestrians are using roadways can inform where investments in bicycle and pedestrian infrastructure are needed. Second, estimates of total cyclist and pedestrian activity can serve as the denominator for calculation of cyclist and pedestrian crash rates, which, in turn, help to identify locations for road safety investment. While estimates of vehicle activity are readily available from routinely collected traffic counts as well as travel demand forecasting models, spatially detailed estimates of bicycle and pedestrian activity rarely are, as few communities conduct regular counts of pedestrians or bicyclists and few models generate estimates of the use of these modes.
This paper describes and implements a simple small-area estimation method for estimating cyclist and pedestrian activity in census tracts based on a combination of travel survey, census, and land use data. Cluster analysis is used to categorize census tracts into neighborhood types, and these neighborhood types are used to aggregate spatially sparse travel survey observations in a meaningful way to obtain estimates of travel activity for each tract. Two sets of activity estimates are calculated based on two different household-based travel surveys recently conducted in California, providing a robustness check on the results. The results are a substantial improvement over fixed per-capita estimates of activity based only on regional or statewide averages.
These tract-level activity estimates then are used to calculate two important policy indicators: intensity of road use by cyclists and pedestrians, and crash rates for these road users. The results show that roads are used most intensively for cycling and walking in the most densely populated neighborhoods of the state. The intensity of pedestrian and cyclist road use in urban census tracts is double that found in suburban tracts, which is again double that found in rural tracts. On the safety side, although non-severe crashes involving cyclists and pedestrians are much more likely in more urban areas, severe crash rates for the non-motorized modes exhibit no clear spatial pattern. The method presented is purposefully simple, and could be implemented by pedestrian and bicycle planners themselves.
Section snippets
Background
Estimation of total bicycle and pedestrian activity is hampered by a lack of basic data. The main sources of bicycle and pedestrian data are household-based travel surveys. One problem with these surveys is that they lack full spatial coverage. For example, at the geographic resolution of the census tract, there are more than 2500 tracts in California that were not sampled at all by the 2009 National Household Travel Survey (NHTS), and only 15 of the sampled tracts include more than 30
Method
In the absence of comprehensive counts of bicyclists and pedestrians, the method presented in this paper relies on data for bicycle and pedestrian activity from two household-based travel diary surveys: the 2009 National Household Travel Survey (NHTS) and the 2010–2012 California Household Travel Survey (USDOT, 2011, CDT, 2013). Reliance on household-based surveys means that this method produces estimates of walking and biking by the residents of each census tract, regardless of where these
Limitations
There are three limitations that bear mention. First and foremost, there is no way to ground truth the estimates. This is the nature of all small-area estimation, but it is worth emphasizing. Travel survey samples are not designed to be representative of the population at small geographies. This means that these survey data do not provide reliable estimates of biking and walking at the geographic level of the census tract, or even at the level of the city. The largest of these surveys has only
Quality of the estimates
There is not a straightforward way to test the accuracy of the estimates produced using small area estimation techniques because valid data at the small area level simply are not available. To get an idea of how reliable the estimates are, I take two approaches. First, I compare the small area estimates (SAE) to direct estimates of the average miles walked and biked per person at the regional level for each of the surveys. Second, I calculate the correlation between NHTS- and CHTS-based
Application examples
The estimates of miles walked and biked by census tract can be combined with other data to produce metrics useful for planning. The first metric divides estimated miles of walking and biking by walkable (i.e., non-highway) road miles. The results provide information about where roads are being used most intensively by pedestrians and cyclists, which can then be used to prioritize non-motorized infrastructure needs. The second metric divides pedestrian and cyclist crash data by estimates of
Conclusion
An average of 835 million dollars were invested in bicycle and pedestrian facilities in each of the past five years in the U.S. (FHWA, 2015a). Approximately 5000 pedestrians and bicyclists die and more than 100,000 are injured on U.S. streets in a typical year (NHTSA, 2010–2015). Knowing how much and where bicycle and pedestrian activity occurs is critical as cities prioritize future investments and consider new safety measures.
This paper has introduced a simple small-area estimation technique
Acknowledgements
This research was part of a project entitled Non-Motorized Travel: Analysis of the National Household Travel Survey California Add-On Data, completed with funding from the California Department of Transportation (Contract No. 65A0404). The author would also like to thank Susan Handy, who was the Principal Investigator on the project and provided useful comments on earlier drafts of this manuscript. The contents of this report reflect the views of the author who is responsible for the facts and
References (33)
- et al.
Health burden of road traffic crashes, an analysis of clinical data on disability and mortality exposure rates in Flanders and Brussels
Crash Analysis and Prevention
(2013) - et al.
How the built environment affects physical activity: views from urban planning
Am. J. Prev. Med.
(2002) - et al.
Estimating use of non-motorized infrastructure: models of bicycle and pedestrian traffic in Minneapolis, MN
Landsc. Urban Plan.
(2012) - et al.
Mapping ridership using crowdsourced cycling data
J. Transp. Geogr.
(2016) - et al.
The 3Ds + R: quantifying land use and urban form correlates of walking
Transp. Res. Part D: Transp. Environ.
(2006) - et al.
Revisiting exposure: fatal and non-fatal traffic injury risk across different populations of travelers in Wisconsin, 2001–2009
Crash Analysis and Prevention
(2013) - et al.
The link between built environment, pedestrian activity and pedestrian–vehicle collision occurrence at signalized intersections
Crash Analysis and Prevention
(2011) A Guide to Small Area Estimation – Version 1.1.
(2006)- et al.
Motor vehicle crash injury rates by mode of travel, United States: using exposure-based methods to quantify differences
Am. J. Epidemiol.
(2007) 2010–2012 California Household Travel Survey Final Report, Version 1.0
(2013)
The utility of the NHTS in understanding bicycle and pedestrian travel
Bicycle commuting and facilities in major US cities: if you build them, commuters will use them
Transp. Res. Rec.
Federal-Aid Highway Program Funding for Pedestrian and Bicycle Facilities and Programs
Federal Highway Administration Pedestrian and Bicycle Activities, January
Pilot models for estimating bicycle intersection volumes
Transp. Res. Rec.
Gender differences in bicycling behavior and facility preferences
Cited by (14)
Stplanpy: A sustainable transportation planner for Python
2023, SoftwareXWishful thinking? Addressing the long-term implications of COVID-19 for transport in Nigeria
2022, Transportation Research Part D: Transport and EnvironmentCitation Excerpt :However, Internet access, mobile technology adoption, commuters' purchasing power, and existing transport infrastructure are factors to be considered. People should be educated and supported to use non-motorised transportation, especially bicycles (Salon, 2016). This education is essential as findings from this study revealed that some adults in Nigeria do not know how to ride bicycles and do not even know where and how to try them.
Cycling diversity, accessibility, and equality: An analysis of cycling commuting in Bogotá
2020, Transportation Research Part D: Transport and EnvironmentCitation Excerpt :Also, statistical clustering methods allow researchers to estimate independent configuration of clusters relying only on data (Kaufman and Rousseeuw, 1990). Clustering individuals has been applied in transport studies to analyze specific groups users of non-motorized modes in relation to their mobility, access and experiences of public space, including mental health (Ahmed et al., 2017; Arranz-López et al., 2019; Huang et al., 2009; Huertas et al., 2020; Li et al., 2013; Pritchard et al., 2019; Salon, 2016). To achieve more comprehensive accessibility results and equality analyses, we classified the data using both numerical (trip distance, vehicles per household1, household size, age) and categorical (SES, gender, job role, education level, driver license) variables describing individual, trip and household characteristics.
The fiscal trade-off: Sprawl, the conversion of land, and wage decline in California's metropolitan regions
2018, Landscape and Urban PlanningCitation Excerpt :In the other one-third of the random samples, variables were insignificant, likely because the sample was drawn disproportionately from regions where the phenomena of interest (conversion of land, municipal reliance on sales tax revenue, and/or wage declines) were less pronounced. Ultimately, to take advantage of the richer data while still representing the diversity of California’s regions (as is typical in California studies such as those by Kahn, 2007; Salon, 2016), the analysis employs all census tracts with complete data, along with dummies representing regional location (based on OMB definitions for metropolitan area). The constructed dataset facilitates the exploration of how fiscal considerations and urban form shape both the conversion of land use and changes in wages, controlling for factors such as location, demographics, labor market characteristics, and industry mix (Fig. 1).
Measuring, understanding and modelling the Walking Neighborhood as a function of built environment and socioeconomic variables
2018, Journal of Transport GeographyCitation Excerpt :Another way to understand walkability comes from the Walking Neighborhood idea, understood as the relationship between walking intensity and neighborhood attributes, computed within boundaries. Some authors have studied this relationship using predefined zones as neighborhoods (Saelens et al., 2003a, 2003b; Wen et al., 2007; Salon, 2016; Neatt et al., 2017) finding significant effects. However, these definitions of neighborhood may be arbitrary, inadequate, or not flexible enough to understand walkability in a more detailed and spatially disaggregate way.