Equivalence and predictive validity of paper-and-pencil and computerized adaptive formats of the Differential Aptitude Tests

This study evaluated two main issues associated with the use of computerized adaptive tests (CATs) for selection and assessment. The first was the equivalence of paper-and-pencil (P&P) and CAT formats of the Differential Aptitude Tests (DAT). The second was the assessment of the predictive utility of CATs to forecast training programme performance. Three DAT subtests were used: numerical ability (NA), abstract reasoning (AR) and mechanical reasoning (MR). A total of 122 refinery operator trainees (aged 18 to 20 years) took part in the study. The trainees were randomly assigned to two groups. The first group took the P&P format first, then the CAT format. The second group took the formats in reverse order. The study showed equivalence between the two modes for the AR and MR tests, but not for the NA test. Significant positive relationships were found between the NA and MR tests and the overall evaluation of trainee performance (range r=.27 to r=.41). However, this result was not found with the AR test. The CAT format displayed a marginally higher correlation with trainee performance than its P&P counterpart. The study also showed that high school performance is a valid and important predictor of training performance (r =.41).