Possible aggregation biases in road safety research and a mechanism approach to accident modeling.

In accident reconstruction, individual road accidents are treated as essentially deterministic events, although incomplete information can leave one uncertain about how exactly an accident happened. In statistical studies, on the other hand, accidents are treated as individually random, although the parameters governing their probability distributions may be modeled deterministically. Here, a simple deterministic model of a vehicle/pedestrian encounter is used to illustrate how naïvely applying statistical methods to aggregated data could lead to an ecological fallacy and to Simpson's paradox. It is suggested that these problems occur because the statistical regularities observed in accident data have no independent status, but are simply the result of aggregating particular types and frequencies of mechanisms.

[1]  Ezra Hauer,et al.  Traffic conflicts and exposure , 1982 .

[2]  Van Fraassen,et al.  Topics in the Foundation of Statistics , 1997 .

[3]  A. Birnbaum.  On the Foundations of Statistical Inference , 1962 .

[4]  D Shinar,et al.  SPEED AND CRASHES: A CONTROVERSIAL TOPIC AND AN ELUSIVE RELATIONSHIP. APPENDIX B OF TRB SPECIAL REPORT 254 , 1998 .

[5]  J. Tanner,et al.  A PROBLEM IN THE COMBINATION OF ACCIDENT FREQUENCIES , 1958 .

[6]  M Salusjarvi,et al.  Road accidents: on-site investigations , 1989 .

[7]  Rudolf Kapustin.  Transportation Disaster Investigation , 2000 .

[8]  G A Davis,et al.  Accident reduction factors and causal inference in traffic safety studies: a review. , 2000, Accident; analysis and prevention.

[9]  K. Schaffner.  Clinical trials and causation: Bayesian perspectives. , 1993, Statistics in medicine.

[10]  Ezra Hauer,et al.  Screening the Road Network for Sites with Promise , 2002 .

[11]  John N. Ivan,et al.  Statistical challenges with modeling motor vehicle crashes: Understanding the implications of alternative approaches , 2004 .

[12]  Robert J. Beaver,et al.  An Introduction to Probability Theory and Mathematical Statistics , 1977 .

[13]  D. Gillies.  Philosophical Theories of Probability , 2000 .

[14]  E. H. Simpson,et al.  The Interpretation of Interaction in Contingency Tables , 1951 .

[15]  Kay Fitzpatrick,et al.  DETERMINATION OF STOPPING SIGHT DISTANCES , 1997 .

[16]  Ian Hacking.  Logic of Statistical Inference , 1965 .

[17]  D. Freedman.  From association to causation: some remarks on the history of statistics , 1999 .

[18]  P. Holland.  Statistics and Causal Inference , 1985 .

[19]  Stuart Glennan,et al.  Probable Causes and the Distinction between Subjective and Objective Chance , 1997 .

[20]  D. Freedman.  Ecological Inference and the Ecological Fallacy , 1999 .

[21]  Donald B. Rubin,et al.  Bayesian Inference for Causal Effects: The Role of Randomization , 1978 .

[22]  A. Dawid.  Causal Inference without Counterfactuals , 2000 .

[23]  C Lave,et al.  Did the 65 mph speed limit save lives? , 1994, Accident; analysis and prevention.

[24]  J. Pearl.  Causality: Models, Reasoning and Inference , 2000 .

[25]  S. Glennan.  Mechanisms and the nature of causation , 1996 .

[26]  Christopher R. Hitchcock.  Causal Generalizations and Good Advice , 2001 .

[27]  Judea Pearl,et al.  Probabilistic Evaluation of Counterfactual Queries , 1994, AAAI.

[28]  Hsin-Li Chang,et al.  MODELING THE RELATIONSHIP OF ACCIDENTS TO MILES TRAVELED , 1986 .

[29]  Gary A. Davis,et al.  Bayesian reconstruction of traffic accidents , 2003 .

[30]  Ezra Hauer,et al.  Bias-by-selection: Overestimation of the effectiveness of safety countermeasures caused by the process of selection for treatment , 1980 .

[31]  Ian Hacking.  ON THE FOUNDATIONS OF STATISTICS* , 1964, The British Journal for the Philosophy of Science.

[32]  Craig N. Kloeden,et al.  ALCOHOL, TRAVELLING SPEED AND THE RISK OF CRASH INVOLVEMENT , 2002 .

[33]  G. Davis,et al.  DEVELOPMENT AND TESTING OF A VEHICLE/PEDESTRIAN COLLISION MODEL FOR NEIGHBORHOOD TRAFFIC CONTROL , 2002 .

[34]  D. W. Ball,et al.  A Short History , 2001 .

[35]  L. J. Cohen,et al.  Applications of Inductive Logic , 1983 .

[36]  D. Rubin.  Estimating causal effects of treatments in randomized and nonrandomized studies. , 1974 .

[37]  C. Lave.  Speeding, Coordination, and the 55 MPH Limit , 1985 .

[38]  Stephen Turner,et al.  "Net Effects": A Short History , 1997 .

[39]  Neil W. Henry.  Thoughts on the Concept and Application of Statistical Models , 1997 .

[40]  D. Freedman.  From Association to Causation via Regression , 1997 .

[41]  Bhagwant Persaud,et al.  Calibration and Transferability of Accident Prediction Models for Urban Intersections , 2002 .

[42]  E Hauer,et al.  CHALLENGING THE OLD ORDER: TOWARDS NEW DIRECTIONS IN TRAFFIC SAFETY THEORY , 1990 .

[43]  David Jarrett.  ASSESSING THE SAFETY EFFECT OF TREATMENT USING DATA FROM A NUMBER OF SITES , 1998 .