Over the past 15 years, a considerable body of literature has built up concerning the automation of psychological tests. Most of this research has been in the area of clinical testing, and may be unfamiliar to many of those involved in occupational assessment. However, the growth of office automation over the past few years has provided the hardware support needed for the development of automated personnel assessment techniques in industry and commerce. The present paper: (1) reviews the major developments in the field of automated testing, bringing together the literature from the clinical and occupational fields; (2) describes the potential provided by microcomputers for the development of new forms of testing; (3) outlines certain problems peculiar to automated testing; and (4) describes the possible future development of ‘expert’ personnel assessment and selection decision support systems.

The study was carried out to assess the validity of the Eysenck Personality Inventory (EPI) and Cattell's 16 Personality Factor Questionnaire (16PF) as predictors of flying training outcome. In addition, it examines differences in profile between self-selected applicants for flying training and the general population; the effects of test-taking conditions on scale scores; incidental selection effects related to personality differences and the reliability of the personality data. The EPI and l6PF inventories were administered to samples of men during selection testing at the RAF Officer and Aircrew Selection Centre, Biggin Hill. Further samples were tested at the Army Air Corps Centre at Middle Wallop prior to their Selection Board interviews. In addition, data were obtained for non-enlisted applicants tested at Biggin Hill and amateur aviators tested at various flying clubs.The results confirmed previous findings that applicants for pilot training are highly 'self-selected', being much more emotionally stable and more extraverted than the general population. Furthermore, the 16PF profile for the unselected sample was found to be very similar to that for US airline pilots. The pattern ofdifferences between those who succeeded and those who failed in training was as expected. The magnitude of these correlations (in the region o f r = .20) was also at the level expected. The results support the findings of previous work and indicate that there are small but potentially valuable increments in validity to be obtained by considering personality factors in selection for pilot training. The problems associated with the use of self-report measures in selection are discussed.Given the ongoing increases in costs associated with military and commercial flying, it is not surprising that the search for better predictions of success in training continue to receive high priority. Aptitude test batteries tend to have predictive validities of around .20 to .30 (Hunter & Burke, 1993). If we are to make any substantive improvements on these levels of validity, it is likely that we will need to look in domains other than ability. The emphases on crew resource management, safety and the growing complexity of operational flight requirements have tended to focus attention on the potential role of personality variables.Guilford (1 947) examined correlations between flying training outcome and measures from a wide range of personality inventories and found very little evidence of predictive *Requests for reprints.