CLOSE OVERLAY

InTrans / Jul 01, 2022

Evaluating driver behavior at rural stop-controlled intersections

A rural stop-controlled intersection, left, and sample safety treatment employed at those locations, right

Nicole Oneyear, a research engineer at the Institute for Transportation (InTrans), recently led an Iowa Local Technical Assistance Program (LTAP) webinar that discussed the results of a study that evaluated driver behavior at rural stop-controlled intersections using the Second Strategic Highway Research Program (SHRP2) Naturalistic Driving Study (NDS).

But why this type of intersection? According to Oneyear, stop-controlled intersections make up approximately 30% of all crashes in rural areas. They also make up about 6% of all fatal crashes. One reason for such a high fatality rate is because of higher-than-average speeds (often 1.6x to 2x faster).

“Because of the higher speeds, when there is a crash, the crash is going to be more severe,” said Oneyear.

The goal was to use the SHRP2 NDS data to identify intersection or roadway characteristics that correlated to more risky behaviors to allow agencies to better target resources to address those issues.

The researchers first looked at the literature to identify rural intersection crash causes and found a few likely reasons: (1) inappropriate gap selection; (2) failure to stop on minor approach; and (3) various roadway characteristics (e.g., curves, offset intersection, lack of turn lanes, and intersection configuration).

“Drivers often take too small of a gap while leaving the intersection and then run into another vehicle,” said Oneyear. She notes that a study out of Minnesota found that about 56% of all right-angle crashes at rural Minnesota stop-controlled intersections were due to inappropriate gap selection.

The data used in this study—the SHRP2 NDS—were originally completed from 2011 to 2014. Drivers’ cars were instrumented with equipment to capture data as they drive. There were approximately 3,100 drivers of all genders and ages. It includes about 4,000 data years, including 5 million trip files and 30 million data miles in six states (Florida, Indiana, New York, North Carolina, Pennsylvania, and Washington).

“We decided to use the NDS data as opposed to other available data—like crash data, which only tells you what happened based on what was included in the crash report or crash narrative. So, you don’t get some of that bias from what the officer may have captured and what the crash participants said happened. We wanted to capture what drivers were actually doing in their vehicles.”

Analyses for this study included driver reaction points, stopping behaviors, and safety critical events (SCEs) for approximately 219 intersections (a total of 7,470 traces).

For the first analysis, a lineal mixed effect (LME) model was developed to determine most relevant variables related to driver reaction points. Signification results included the following: (1) the reaction point varied by state (23 to 51m) and (2) drivers reacted earlier (38 to 64m) when on pavement signing was present. Also, (3) drivers turning right reacted later than left turning and through vehicles.

For stopping behaviors, the researchers looked at three different stopping types: full, rolling, and no-stop. This analysis included 128 unique drivers on 81 unique approaches at 58 intersections. It was found that vehicles on major approach were 2.22x more likely to perform a full/rolling stop and speeding drivers (5 mph or more) upstream were 2.1x more likely to not stop.

“This isn’t surprising,” said Oneyear. “People who make risker decisions when it comes to their speed are more likely to make riskier decisions with their stopping behavior.”

For t-intersections (87 unique intersections with 157 unique drivers), the researchers found that if a vehicle was on the major approach, a full/rolling stop was 55.53x more likely. If lighting was present, a full/rolling stop was 2.32x more likely. Additionally, speeding upstream made a no-stop 2.23x more likely.

And for all-way stops (46 unique intersections with 276 unique drivers), the researchers found that a driver on another approach made a full/rolling stop 7.60x more likely and speeding drivers (10 mph or more) resulted in making a no-stop 1.85x more likely.

“Additionally, we did find that drivers who divert their attention away from the roadway multiple times are more likely to not stop.”

In looking at SCEs, the researchers identified 38 in the NDS data (i.e., crashes, near crashes, and crash-relevant events).

Some major SCE findings included the following: (1) More likely to have a no-stop, (2) 1.52x more likely to be distracted with 5 sec of intersection for major approach vehicles, and (3) 3.56x more likely to be distracted within 5 sec of intersection on minor approach.

Overall, the study did find that some behaviors, characteristics, and countermeasures were found to impact rural intersection safety more than others. For example, on-pavement signing tended to improve reaction time (30–64 m), beacons improved stopping behavior at all-way stops, and lighting increased the probability of coming to a full or rolling stop at t-intersections.

“We found that increasing that awareness of the intersection, especially at night, did make an impact,” said Oneyear.

This project was sponsored by the Federal Highway Administration (FHWA). To watch the Iowa LTAP webinar on this topic, visit this link. The published report is not yet available; however, a similar report about transverse rumble strips at rural stop-controlled intersections is available on the InTrans website at this link.

TOP