An elementary ambiguity in accident theory

MANY DIFFERENT MODELS HAVE BEEN SUGGESTED FOR THE ANALYSIS OF ACCIDENT DATA. AMBIGUITY OF INTERPRETATION ARISES BECAUSE MANY DISTRIBUTIONS CAN RESULT FROM MORE THAN ONE TYPE OF MODEL. PARTIAL ASCERTAINMENT OF ACCIDENT DATA, WHICH MAY OCCUR FOR VARIOUS REASONS, E.G., (I) CERTAIN EVENTS MAY NOT IN FACT BE OBSERVABLE; (II) A PARTICULAR OBSERVATION MAY BE PARTIALLY RECORDED ACCORDING TO A PROBABILITY DISTRIBUTION; (III) DIFFERENT EVENTS MAY HAVE DIFFERENT PROBABILITIES OF BEING RECORDED, PRODUCES A MODIFIED DISTRIBUTION (OF RECORDED OBSERVATIONS) WHICH IS, IN GENERAL, DIFFERENT FROM THAT OF THE ORIGINAL EVENTS. IN THIS PAPER, THE KIND OF AMBIGUITY RESULTING FROM PARTIAL ASCERTAINMENT OF ACCIDENT DATA IS CONSIDERED.