Related Experiment Videos
Is the claim that 'variance kills' an ecological fallacy?
1Department of Civil Engineering, University of Minnesota, Minneapolis 55455, USA. drtrips@tc.umn.edu
Accident; Analysis and Prevention
|April 10, 2002
Summary
Traffic crash rates and vehicle speed dispersion show correlations, but these do not prove that speed variance causes individual crash risk. Further analysis is needed to understand the complex relationship between speed, variance, and crash risk.
Area of Science:
- Traffic Safety Research
- Transportation Engineering
- Accident Analysis
Background:
- Studies over 15 years show positive correlations between estimated traffic crash rates and vehicle speed dispersion.
- These correlations are often interpreted as evidence that speed variance is a causal factor in individual crash risk, or that extreme speeds increase risk.
Purpose of the Study:
- To critically evaluate the interpretation of correlations between traffic crash rates and speed dispersion.
- To determine if observed correlations provide evidence for the causal role of speed variance in individual crash risk.
Main Methods:
- Theoretical analysis of correlation in relation to different individual risk functions (increasing, decreasing, U-shaped).
- Examination of conditions under which correlations can arise irrespective of speed variance's causal impact.
Main Results:
- Positive correlations between crash rate and speed dispersion are expected regardless of whether individual crash risk increases, decreases, or is U-shaped with speed.
- Observed correlations do not support the hypothesis that increased speed variance causally increases individual crash risk.
Conclusions:
- The presence of positive correlations between traffic crash rates and speed dispersion does not, in itself, indicate a causal relationship.
- Further research is required to disentangle the complex interplay between vehicle speed, speed variance, and individual crash risk.