|

Statistical Forecasting of Traffic-Related Pedestrian Fatalities in the United States

Author(s): Pascua, Dubois, Bedard, Pernia

Slidedeck Presentation Only (no paper submitted):

4C - Pascua

Abstract:

Prior to 2009, pedestrian fatalities in the United States have been exhibiting a declining trend. Since then however, pedestrian fatalities have been increasing. In the last five years, American pedestrian fatalities have increased by 25%. Furthermore, between the years 2006 and 2015 the proportion of pedestrian traffic-related fatalities have risen from 11% to 15%). As municipalities are incorporating eco-friendliness and active transportation into their development plans, the importance of pedestrian safety has become more apparent. This study aims to predict the magnitude of American pedestrian fatalities to the year 2035, accounting for changes in walking behaviour. A secondary objective is to identify specific at-risk pedestrian age-sex groups susceptible to traffic-related injury. Annual pedestrian fatality counts, estimates by exposure, and projections by age (5-15, 16-19, 20-34, 35-54, 55-64, and 65+) and sex categories were computed using a combination of fatality and exposure data. Pedestrian fatality data were acquired from the United States' Fatality Analysis Reporting System (FARS). Three measures of pedestrian exposure (number of walk trips, person miles walked, person minutes walked) were derived using data from multiple instances of the National Household Travel Survey (NHTS). Survey years considered in the analysis include 1977, 1983, 1990, 1995, 2001, and 2009. Interpolation was employed to determine exposure for years in which a survey has not been conducted. Fatality projections by pedestrian age and sex were forecasted using the SPSS (V24) CURVEFIT procedure. Models from the fitting procedure were selected based on AIC and practicality. Figures containing the following results will be presented: a) Actual pedestrian fatalities between 1975-2015; b) estimated fatalities between 1977-2015 by the three measures of exposure; and c) projected fatalities drawn from the forecasting analysis. Cohorts with higher forecast curve slopes are considered to be more at risk of being involved in a pedestrian fatality, and will be ranked accordingly. These pedestrian fatality counts, estimates by exposure, and projections are intended to be used to advise transportation agencies and policy makers in setting traffic-related injury reduction targets in accordance with Vision Zero doctrine. The findings from this study may also be used to initiate the development of demographic-specific pedestrian fatality countermeasures from the transportation engineering and public health perspectives (i.e. educational improvements and resource allocation for at-risk demographics).