The effect of speed limits on drivers' choice of speed: A random parameters seemingly unrelated equations approach

• Using a survey, drivers' choice of speed on US interstate highways is studied.
• Three distinct speed limits (55 mi/h, 65 mi/h and 70 mi/h) set the study basis, and the drivers' speed choices under the three speed limits are modeled simultaneously.
• A random parameters seemingly unrelated regression estimation approach is used.
• Interrelations among the speed choices and unobserved heterogeneity are accounted for.
• Various socio-demographic and perception factors affect speed-limit compliance.

Drivers’ choice of speed has long been known to be a critical factor in both the likelihood and severity of vehicle crashes. Given this, understanding drivers’ choice of speed and the possible effect that posted speed limits may have on this choice, is a critical element of safety research. This paper seeks to provide new insights on drivers’ speed-choice process by studying U.S. interstate highways (all of which are constructed to the same design-speed standard) under three distinct speed limits (55 mi/h, 65 mi/h and 70 mi/h). Using a survey of interstate drivers that asked respondents their normal operating speed on interstates posted with these speed limits (under light traffic conditions), a random parameters seemingly unrelated regression estimation approach is used to account for both the interrelation among the choices under the three speed limits and for the unobserved heterogeneity across respondents. The estimation results show that a wide variety of factors influence the choice of speed in the presence of speed limits, including driver age, gender, marital status, number of children, driver education level, household income, age when the driver was first licensed, and opinions about pavement quality. The findings in this paper have important implications relating to the factors that may affect speed-limit compliance, and also demonstrate the methodological potential of the random parameters seemingly unrelated regression estimation approach to address a number of safety-related problems involving a series of inter-related continuous dependent variables.