Jeffrey P Ebert, Ruiqi Yan, Yingying Lu, Kelsey Bartolozzi, Michael O Harhay, Catherine C McDonald, M Kit Delgado
Estimates of variance explained are sample-dependent, and the present sample underrepresented certain groups (ages 18-24, races other than white). Further research is needed with more representative samples. Nevertheless, reorienting away from demographic and regional factors, which combined explained <0.5% of the variance in buckling behavior in the present sample, and toward individuals and situations for which buckling is less common, which combined explained >50%, may prove fruitful for researchers and seat belt campaigns. Findings point to a need to encourage buckling on drives that are short, high-speed, or after sundown.
OBJECTIVE: To quantify the relative strength of association of situational (e.g., trip distance), demographic (e.g., age), and regional (e.g., rurality) factors with trip-level seat belt use in a diverse sample of adult drivers.
METHODS: We conducted a secondary analysis of baseline data from a nationwide trial that used vehicle telematics to unobtrusively measure buckling behavior. Data were collected in 2022 and included 1,000 participants residing in 48 U.S. states with 129,494 single-segment driving trips. Logistic mixed-effect regression quantified variance explained.
RESULTS: Participant age ranged from 18 to 88 (mean = 51.3); 59.6% were male, 40.4% female; 84.4% identified as white, 9.4% as Black, and 6.2% as another race. Mean (SD) seat belt use was 83.6% (19.1%) of trips. In the primary analysis, situational factors explained 29.9% of the variability in seat belt use, whereas demographic and regional factors explained only 0.4% and <0.1%, respectively; 23.2% was attributable to individual differences not explained by fixed factors. Drivers were more likely to buckle on longer-distance trips, but unexpectedly less likely to buckle on trips with greater average speed. Drivers buckled less after sunset and late at night. None of the demographic or regional factors significantly predicted buckling.
CONCLUSIONS: Estimates of variance explained are sample-dependent, and the present sample underrepresented certain groups (ages 18-24, races other than white). Further research is needed with more representative samples. Nevertheless, reorienting away from demographic and regional factors, which combined explained <0.5% of the variance in buckling behavior in the present sample, and toward individuals and situations for which buckling is less common, which combined explained >50%, may prove fruitful for researchers and seat belt campaigns. Findings point to a need to encourage buckling on drives that are short, high-speed, or after sundown.