Synthesis of Safety for Traffic Operations - Transports Canada
Synthesis of Safety for Traffic Operations - Transports Canada
Synthesis of Safety for Traffic Operations - Transports Canada
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Turn Lanes<br />
1998 one <strong>of</strong> their high crash locations was modified to include a left and right turn lane<br />
on one <strong>of</strong> the intersection approaches. Crashes were reduced from nine to four, <strong>for</strong> a<br />
CMF <strong>of</strong> 0.44.<br />
While impressive, the Waterloo analysis is a naïve be<strong>for</strong>e-after study <strong>of</strong> crash frequency<br />
using one-year <strong>of</strong> be<strong>for</strong>e and one year <strong>of</strong> after data. Furthermore, the countermeasure<br />
was implemented, at least in part, because this location had an aberrant crash record. The<br />
results are very unreliable due to a failure to account regression-to-the-mean, the limited<br />
sample size, and the failure to account <strong>for</strong> exposure.<br />
Harwood et al (2002)<br />
In a comprehensive analysis <strong>of</strong> the safety impacts <strong>of</strong> left- and right-turn lanes, Harwood<br />
et al (2002) conducted be<strong>for</strong>e-after studies <strong>for</strong> intersection improvements in eight states.<br />
A total <strong>of</strong> 280 sites were treated and 300 sites were used as comparison or reference sites.<br />
The sites were in urban and rural settings and were either two-way stop controlled, or<br />
signalized (i.e., all-way stop controlled sites were excluded from the analysis). All sites<br />
has either three or four approaches; all approaches were public streets (i.e., no private<br />
driveways were included in the sites).<br />
The Harwood et al study employed three different evaluation methods: the matched-pair<br />
approach, the be<strong>for</strong>e-after evaluation with a comparison group, and the be<strong>for</strong>e-after study<br />
using Empirical Bayes methods. Crashes that were included in the analysis are those that<br />
occurred within 250 feet <strong>of</strong> the intersection, and were coded as intersection-related.<br />
Be<strong>for</strong>e and after time periods varied between one and 10 years, with averages <strong>of</strong> 6.7<br />
years and 3.9 years, <strong>for</strong> the be<strong>for</strong>e and after periods, respectively.<br />
The results <strong>of</strong> the analysis and the recommendations <strong>of</strong> the researchers are shown in<br />
Table 9.8.<br />
TABLE 9.8: <strong>Safety</strong> Impacts <strong>of</strong> Left-turn Lanes on Major Road Approaches<br />
No. <strong>of</strong> Approaches on which<br />
Intersection Type Setting <strong>Traffic</strong> Control Left-turn Lanes are added<br />
One Two<br />
Three-leg<br />
Rural<br />
STOP control 0.56 --<br />
Signal 0.85 --<br />
Urban<br />
STOP control 0.67 --<br />
Signal 0.93 --<br />
Four-leg<br />
Rural<br />
STOP control 0.72 0.52<br />
Signal 0.82 0.67<br />
Urban<br />
STOP control 0.73 0.53<br />
Signal 0.90 0.81<br />
Page 114