Youth and the Automobile in America

RETURNING now to the United States for a consideration of research developments there during the late fifties and early sixties. In reviewing the major publications of the time, one can see that the whole approach to the question of accident proneness was still one of openminded inquiry. Despite a certain amount of despondency over the lack of statistical evidence to support the concept (especially in American statistics) there seemed to be a general feeling that there was a lot of psychological truth behind the proneness concept, and that it was important to go on exploring the psychological factors associated with accident involvement. Conger's study (1957, 1959) was one of the most important of the period. Apart from the fact that his work is always noted for its thoroughness and for the impartiality with which he presents his results, certain aspects of his 1959 study are worthy of a detailed description because this was the first combination of the clinical and the experimental approach—clinical in the sense that a full clinical examination was given to each subject, but experimental in the sense that the results were assessed by means of a statistical analysis. The report on this study is reproduced in Haddon's Accident Research, so I will only describe the project briefly here. Conger felt that despite the technical difficulties which beset projects on accident proneness, the weight of the scientific evidence nevertheless suggested that significant differences did exist between the accident repeaters and the accident free. And he persisted with this belief despite the fact that an earlier project (1957) had produced few significant results. Here a group of 264 subjects were tested on a wide variety of psychological and psychophysical tests and the results were compared with a criterion of high-, middle-, and low-accident involvement. Most of the measures failed to discriminate consistently, for while a number of tests appeared promising initially, they failed to survive a crossvalidation. Only one test proved consistently stable, namely a measure of the individual's value system. In sample after sample this test made it possible to differentiate between highand low-accident samples with an accuracy of some 73 %. On three scales of a modified Allport-Vernon-Lindzey assessment of values, the high-accident subjects were found to be consistently less orientated towards religious values and consistently more orientated towards aesthetic and theoretical values than the non-accident group. However, as Conger felt that this result was very difficult to interpret, he decided that further investigations should be made to establish just what personality functions were producing these significantly different values. So, taking the extremes as being more likely