Development of an automated performance test system for environmental and behavioral toxicology studies

In developing a menu of computerized performance tests for repeated-measures applications the metric properties of selected tests have been examined. Factors of chief concern have been stability and reliability, as well as the practical issue of the length of time it takes to achieve high levels of both. In this study, these factors, as well as predictive validity, are examined. 25 subjects were tested repeatedly (10 sessions) with 11 tests previously identified as "good" candidates for repeated-measures research in paper-and-pencil (marker test) versions. The 11 tests were administered concurrently in their traditional paper-and-pencil modes and newly implemented microcomputer-based versions, along with the Wechsler Adult Intelligence Scale (WAIS). Nine of the 11 microcomputer-based tests achieved stability. Reliabilities were generally high, with r greater than or equal to .77 for 3 min. of testing. Cross-correlations of microbased tests with traditional paper-and-pencil versions suggest equivalency between the test constructs in the different media. Correlations between six of the microbased subtests and the WAIS identified common variance, and these might comprise an efficient short (6 min.) battery of tests.