Ever-increasing traffic volume on our roads has caused congestion and a rapid increase in the number vehicular crashes. As a direct result of these alarming safety statistics, there also has been a welcomed uptick in interest to enhance roadway safety through safety research and safety-conscious design.

The main goal of this study is to quantify the effects various safety treatments have on roadway operations and safety, specifically on urban collectors with access.

Urban collector road run through highly diversified areas, so various factors have to be considered in this study when making before-and-after comparisons of safety improvements. For 25-40 mph urban collectors with access, the safety treatments considered in this research are:

1. Increase in lane widths (10′ or 11′ to 12′)
2. Construction of 4-, 6-, 8-, or 10-foot shoulders
3. Removal of trees in median and border areas
4. Installation of guide rails and changes in vertical and horizontal geometry to improve sight distances.

A number of sites along 25- to 40-mph urban collectors with access at which the above safety improvements have been implemented were identified in close collaboration with New Jersey Department of Transportation (NJDOT).

The data sources include  Highway Safety Information System (HSIS), Ohio Department of Transportation, California Highway Patrol, and Caltrans. From these sources, the research team has identified seven different treatment sites in New Jersey, six different treatment sites in Ohio, and two treatment sites from California.

Once the site selection process was completed, historical crash data for each of these sites were collected. NJDOT’s crash database was the main source of data for this comparative evaluation study.

The impacts of improvements on safety were determined by an analysis of NJDOT crash database for three years before and three years after the safety treatment was implemented. In addition to the crash data, traffic and other relevant data were also collected because the technique selected for implementation  was based on its safety impact as well as traffic performance. Thus, the final determination of safety impact resulting from the potential techniques, if applied future candidate sites, was based on both traffic performance and safety impacts. While conducting before-and-after analysis for the sites, four different methodologies were considered:

1. The simple (or naive) before-and-after study method
2. The before-and-after study with comparison group method
3. The before-and-after study with Empirical Bayes (EB) method
4. The before-and-after study with Full Bayes (FB) method

After conducting before-and-after analysis, Crash Reduction Factors (CRF) and Accident Modification Factors (AMF) were estimated for each countermeasure. The analysis results reveal that the individual CRF values and their relative order among different countermeasures are similar to the values in the literature. In particular, improvements in vertical and horizontal alignment were found to affect the highest reductions in the accident rate, followed by adding shoulders, installing a median barrier, increasing lane widths, and installing guiderails. However, impacts of guiderail installation were mixed in that it did not show positive results at some sites.

It should be noted that the total benefit of implementing a countermeasure included cost savings resulting from a reduction in the number of crashes or crash severity, and the total cost of implementing a countermeasure included construction and a possible savings in maintenance costs.

The determination of benefits from countermeasures depends on projected crash reductions, which is calculated as the expected number of crashes without the countermeasures multiplied by a CRF. Thus, CRF is simply a quantitative statement of the percentage of crashes that a countermeasure is expected to reduce.

Moreover, when considering individual CRF values, transportation planners should keep in mind that the estimated values depend on specific characteristics of the treatment site, reference groups considered in the estimation process, time period included in the analysis, and the statistical tools used to calculate the CRF values.