Safety Impacts of "Road Diets" in Iowa

Institute of Transportation Engineers. ITE Journal, Dec 2006 by Stout, Thomas B, Pawlovich, Michael D, Souleyrette, Reginald R, Carriquiry, Alicia

ROAD DIETS ARE FREQUENTLY SUGGESTED SOLUTIONS TO PROBLEMS OP LEPT-TURN RELATED CRASHES. THIS FEATURE ANALYZES THE IMPACT OF THE CONVERSION OF 15 SITES USING A PULL BAYES APPROACH AS WILL AS A CLASSICAL BEFORE-AND-AFTER STUDY (WITH YOKED COMPARISON SITES). THE STUDY FOUND A REIDUCTION IN THE CRASH RATE, CRASH DENSITY, INJURY CRASHES, THE INVOLVEMENT RATE FOR DRIVERS OLDER THAN 65 AND LEFT-TURN RELATED CRASHES.

INTRODUCTION

The "road diet," generally converting a roadway from four lanes to three lanes (one through lane in each direction and a two-way, continuous left-turn lane), frequently is suggested as a traffic calming solution or to address left-turn related crashes on undivided four-lane urban roadways where widening may not be an option. In Iowa, 15 of these conversions have been completed. This feature analyzes the safety impacts of these road diets. Figure 1 presents an example of a road diet in Osceola, IA,

With the objective of assessing whether road diets result in a safety benefit on Iowa roads, the Iowa Department of Transportation's (DOT) Office of Traffic and Safety (TAS) funded two independent effectiveness evaluations.

The first utilized a classical before-and-after (B/A) study with "yoked pair" control sites and was conducted by researchers at the Center for Transportation Research and Education (CTRE), The second utilized a Full Bayes (FB) approach to implement a B/A analysis and was conducted by the Iowa State University Department of Statistics, in cooperation with TAS.

Both studies began with the same 15 treatment and 15 control sites, The B/A study examined 10 years of annual data (crashes, crash types and volumes) with comparisons to annual crash trends city-wide and to similar, unconverted roadways (such as yoked pair control sites), The FB study utilized monthly crash data and estimated volumes obtained from TAS for the 30 sites over 23 years (1982-2004).

The intervention and control sites had traffic volumes ranging from 2,000 to 17,400 annual daily traffic during that time span and were located mostly in smaller urbanized areas (ranging in population from 1,169 to 198,682, according to the 2000 U.S. Census), Table 1 presents a summary of the study sites, including annual average daily traffic (AADT), population, length of study segment and a brief description of the land use within the study corridor.

The results of the B/A study indicated reductions of 21 percent in crash frequency and 29 percent in crash rate. The results of the FB study largely agree with these numbers and indicated a 25-percent reduction in crash frequency per mile (density) and a 19-percent reduction in crash rate.

These results differ from a previous, much publicized study, which reported a 6-percent reduction in crash frequency per mile and an insignificant indication for crash rate effects. The results from the Iowa studies also fit practitioner experience with this type of safety countermeasure. Additionally, the B/A study explored crash type, severity and ape group effects.

The results of these explorations all are positive, with a reduction in crash types associated with left-turn and stopped traffics a 34-percent reduction in the number of injury crashes; and reductions in the involvement of at-risk age groups (25 and younger and 65 and older).

LITERATURE REVIEW

Huang, Stewart and Zegeer evaluated 12 road diets and 25 comparison sites in Washington and California using 1 to 3 years of data before and after.4 For their frequency analysis, they utilized 11 road diets and 24 comparison sites and found that the road diets had an average crash frequency that was only 6 percent lower than the corresponding comparison sites. For their rate analysis, they utilized 8 road diets and 14 comparison sites and found that crash rates did not change from before to after. They also found that crash severities were not affected and that crash types did not change significantly.

In 2001, CTRE published a final report by Knapp and Giese, which presented guidelines for the conversion of four-lane roadways to a three-lane configuration. It discussed case studies in Montana, Minnesota, California and Iowa. In all cases, there were reductions in crashes after the conversions-in absolute values, rates of crashes, or both. It also reported that levels of service were not "dramatically decreased" by the conversion.

RESEARCH PROCESS

The Iowa TAS maintains a rich statewide crash database that provides a consistent, readily available source of crash data for several Iowa countermeasures effectiveness studies, including these road diet studies.

For the classical analyses, the data covered 5 years before (in all but one case) and up to 5 years after the conversions. The classical (B/A) study compared changes in the crashes at the conversion sites to changes in the sites' cities, as well as to crash patterns at similar comparison sites. The analyses conducted in this study addressed crash frequency, rate, severity, type, driver age and major cause.

For the Bayesian analyses, the process used monthly crash data covering 23 years, as well as estimated monthly traffic volumes provided by TAS. Each treatment site had different known intervention dates; therefore, the number of before and after crash records varied from site to site. Individual control sites (the same ones used in the classical study) were matched to each treatment site to provide a control sample similar to the treatment sample.