Research

This paper examines whether wildlife crossing structures reduce the number of wildlife–vehicle collisions. Using Washington state crash data from 2011 to 2020, I employed a difference-in-differences methodology at the year level on each of 13 observed wildlife crossing structures in Washington. The treatment area consisted of wildlife–vehicle collisions within 10 mi of a wildlife crossing structure, and the control area included wildlife–vehicle collisions that were 60 to 70 mi from the same wildlife crossing structure. I found evidence that wildlife crossing structures resulted in one to three fewer wildlife–vehicle collisions on average per mile per year. The marginal treatment effect also held within a 5-mi treatment area, a 15-mi treatment area, and when controlling for the presence of other structures within the baseline of a 10-mi treatment area. However, the collision reductions were more consistent among wildlife bridges than culverts, suggesting that not all wildlife crossing structures have the same effect in reducing accidents involving wildlife. Using a back-of-the-envelope approach, each wildlife crossing structure yielded annual benefits of $235,000 to 443,000 in 2021 U.S. dollars. 

The tire byproduct 6PPD-quinone is lethal to several salmonid species when it is washed off roads into waterways. We study private consumer demand for a hypothetical, reformulated “Salmon Safe” tire with an online survey of n=695 vehicle owners in California, Oregon, and Washington. Using a pre-registered study design, we also test whether consumers comprehend information differently when the science component of the survey is presented as narrated video or static text and images, and whether forcing consumers to deliberate for 45 – 60 seconds affects their decision and choice certainty. We find that the median respondent would be willing to pay approximately 10% more, or $76, for a set of four reformulated tires, though approximately 25% of subjects were willing to pay $225 or more.  Information presentation and forced deliberation had no effect on willingness to pay for reformulated tires or comprehension, but weak (and negative) effects on choice certainty.

In March of 2017 Utah became the first state in the United States to lower the legal blood alcohol content (BAC) for driving from 0.08 to 0.05 g/dL. We take advantage of the fact that enforcement was delayed until 2019 to test whether this policy change significantly affected the number of traffic accidents or the severity of those accidents. We employ a difference-in-differences strategy on Utah counties using neighboring states as controls. Results show the policy appears to decrease the total number of accidents, limited primarily to property damage-only accidents. Surprisingly, we find strong effects during the day on weekdays and weekends. We confirm our findings using a doubly robust difference-in-differences estimator. We believe these results may be partially explained by drivers who, after the policy is enacted, avoid reporting property damage-only accidents if possible. Using insurance claims data, we show there is no corresponding fall in insurance claims or payouts suggesting that the fall in total accidents likely comes from under-reporting.

Governments and environmental NGOs in the United States and around the world implement behavior change campaigns to prevent and manage stormwater runoff and nonpoint source pollution. What is the evidence that these programs succeed in modifying behavior? In this paper, we use a systematic review approach to identify, evaluate, and summarize evaluations of the effect of behavior change campaigns in water quality or stormwater management. We collected 47 studies from peer-reviewed journals and the gray literature and assessed the quality of each study based on nine metrics. All studies reported that behavior change campaigns had some degree of successes in persuading people to adopt stormwater runoff mitigation practices. However, we found that very few studies (13%) used a control group and 36% did not report or use baseline data. Furthermore, most evaluations (81%) measured behavior change based on self-reported information from surveys and 36% measured long-term changes, one year or longer. Given the large and growing staff and financial investments in these campaigns in the US and around the world, it is important that future evaluations produce higher-quality evidence of whether these investments produce sufficient benefit.