Abstract
A viewpoint is presented concerning "accident" causes essentially, that they are really accident, and not the results of wilful negligence or other motivated behaviour Accidents result from confusion. A means is presented for measuring average degrees of driver confusion, and comparing these average accros different highway and traffic condition. Suggested cross-validational procedures are discussed. The argument is set forth that "confusion" as operationally defined by the procedure, will furnish valuable intermediate criterion to investigators in the accident prevention field.