Validation of a Trainability Test for Young Apprentices

Abstract

Predictive test validation data were collected to analyze the usefulness of a content-valid trainability test of carpentry for predicting success at a full-scale course. Results supported the effectiveness of the standardized test CAR-1 to predict the individual job-learning potential in a sample of 109 young unemployed persons. Several advantages of structured trainability tests as predictors of job learning are indicated. These include greater standardization, the content and predictive validity, and the possibility of assessing the potential of candidates with no prior experience. The high and significant coefficients obtained in this study suggest that the trainability testing methodology may provide useful guidelines for developing job learning predictors in other areas of vocational training.

Updating the trainability tests literature on Black–White subgroup differences and reconsidering criterion-related validity.

“…The mean validity for predicting training performance was .79 ( k = 6, N = 523) for the six samples in the archival data. We also added the two additional trainability tests noted above (i.e., Aramburu-Zabala & Casals, 2003; Aramburu-Zabala Higuera & Casals Riera, 2004). 5 The mean correlation corrected for criterion unreliability is .70 ( k = 8, N = 820).…”

“…First, trainability tests use a work sample to measure work-related behaviors (Robertson & Downs, 1989, see also Sackett, 2000). Second, trainability tests involve a period of time when an applicant is expected to learn key principles, and the learning typically occurs before applicant task performance is evaluated (e.g., Aramburu-Zabala Higuera & Casals Riera, 2004; Sackett, 2000). Third, trainability tests are useful for testing untrained applicants (Aramburu-Zabala & Casals, 2003; Robertson & Downs, 1989).…”

“…In perhaps the earliest journal article reporting adverse-impact-related information on this test type, Siegel and Bergman (1975) reported, “No statistically significant differences were found between races” on scores on a trainability test (p. 335). More recent journal articles summarized the general beliefs about trainability tests by noting trainability tests had “small or null adverse impact” (Aramburu-Zabala & Casals, 2003, p. 233) or trainability test “adverse impact on minorities is comparatively low” (Aramburu-Zabala Higuera & Casals Riera, 2004, p. 56). In a well-known book chapter reviewing predictors of job performance, Reilly and Warech (1993) noted trainability tests were a suitable alternative to tests of cognitive ability because they had similar validity and were likely to have less adverse impact.…”

“…More such data would be helpful. To our knowledge, the only additional validity data in the literature come from two studies by Aramburu-Zabala and Casals (2003; Aramburu-Zabala Higuera & Casals Riera, 2004). Those authors reported applicant-level, trainability test validities of .59 and .61 with end-of-training scores as a criterion (corrected for criterion unreliability).…”

“…5The tests described by Aramburu-Zabala and Casals (2003) and Aramburu-Zabala Higuera and Casals Riera (2004) covered several days of learning (e.g., 3–5 days), and, thus, they may be construed as a stretch of our definition of trainability tests occurring in a day or two. However, we report these results to allow the reader to see the cumulative results in multiple ways.…”