THE PREDICTION OF ACCIDENT LIABILITY THROUGH BIOGRAPHICAL DATA AND PSYCHOMETRIC TESTS

Abstract

A CONTRASTED SAMPLE OF ACCIDENT AND ACCIDENT-FREE DRIVERS WERE EVALUATED IN DETAIL IN ORDER TO DETERMINE FACTORS RELATED TO ACCIDENT INVOLVEMENT. COLLECTED INFORMATION REPRESENTED BIOGRAPHICAL AND DRIVING- RELATED DATA, PERSONALITY TRAITS AND ATTITUDES, PARENTAL RELATIONSHIPS, PERCEPTUAL STYLE, PERCEPTUAL MOTOR COORDINATION, AND DRIVING SIMULATOR PERFORMANCE. THE VARIABLES WHICH WERE SIGNIFICANT UPON CROSS- VALIDATION WERE MARITAL STATUS, MILEAGE, TRAFFIC CONVICTION RECORD, SOCIOECONOMIC FACTORS, RATING OF ONE'S DRIVING ABILITY IN COMPARISON TO ELDERLY DRIVERS, AND PERSONALITY AND ATTITUDINAL FACTORS DERIVED FROM A PSYCHOMETRIC INVENTORY CALLED THE CIDAO. NONE OF THE VAST ARRAY OF PERCEPTUAL MOTOR AND SIMULATOR PERFORMANCE MEASURES PROVED SIGNIFICANT, ALTHOUGH THERE WAS SOME SUGGESTIVE RELATIONSHIP BETWEEN SIMULATOR SPEED VARIABILITY, TWO PSYCHOMOTOR MEASURES AND FIELD DEPENDENCE AND ACCIDENTS. CLASSIFICATION OF DRIVERS THROUGH CLUSTER ANALYTICAL PROCEDURES REVEALED SEVERAL HIGH AND LOW ACCIDENT TYPES. THESE FINDINGS INDICATE THAT A COMBINATION OF CLUSTER ANALYSES AND MULTIPLE REGRESSION ANALYSES IS A MORE POWERFUL METHOD THAN EITHER ALONE, AND THAT CONVENTIONAL MULTIPLE REGRESSION PROCEDURES CAN OBSCURE COMPLEX RELATIONSHIPS. THE RESULTS FOR FEMALES CLOSELY PARALLELED THE FINDINGS FOR MALES.