Using Crowdflower to Study the Relationship between Self-Reported Violations and Traffic Accidents

Abstract

Crowdsourcing is a promising approach for Human Factors survey research. We explored the use of a relatively new crowdsourcing platform called CrowdFlower. Our survey focused on the relationship between self-reported traffic accidents and violations measured with the Driver Behaviour Questionnaire (DBQ). We obtained 1,862 responses within 20 hours at a cost of $247. The demographic correlates of DBQ violations were consistent with those of traditionally recruited samples. The correlation between DBQ violations and self-reported accidents was ρ = .28. Self-reported accidents at the national level (N = 18 countries) correlated strongly (ρ = .68/.79) with accident statistics published by the World Health Organization. Large international differences were observed, with horn honking being relatively common in India and Indonesia and speeding being common in some Western countries. We conclude that CrowdFlower is an efficient tool for conducting international surveys.