Time-series intervention analysis of pedestrian countdown timer effects

Accid Anal Prev. 2014 Nov:72:23-31. doi: 10.1016/j.aap.2014.05.025. Epub 2014 Jul 5.


Pedestrians account for 40-50% of traffic fatalities in large cities. Several previous studies based on relatively small samples have concluded that Pedestrian Countdown Timers (PCT) may reduce pedestrian crashes at signalized intersections, but other studies report no reduction. The purposes of the present article are to (1) describe a new methodology to evaluate the effectiveness of introducing PCT signals and (2) to present results of applying this methodology to pedestrian crash data collected in a large study carried out in Detroit, Michigan. The study design incorporated within-unit as well as between-unit components. The main focus was on dynamic effects that occurred within the PCT unit of 362 treated sites during the 120 months of the study. An interrupted time-series analysis was developed to evaluate whether change in crash frequency depended upon of the degree to which the countdown timers penetrated the treatment unit. The between-unit component involved comparisons between the treatment unit and a control unit. The overall conclusion is that the introduction of PCT signals in Detroit reduced pedestrian crashes to approximately one-third of the preintervention level. The evidence for this reductionis strong and the change over time was shown to be a function of the extent to which the timers were introduced during the intervention period. There was no general drop-off in crash frequency throughout the baseline interval of over five years; only when the PCT signals were introduced in large numbers was consistent and convincing crash reduction observed. Correspondingly, there was little evidence of change in the control unit.

Keywords: Countdown timers; Crash analysis; Interrupted time-series design; Intervention analysis; Pedestrian safety; Time-series regression models.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Accidents, Traffic / prevention & control*
  • Cities*
  • Environment Design / statistics & numerical data*
  • Humans
  • Interrupted Time Series Analysis
  • Michigan
  • Regression Analysis
  • Walking / injuries*