Abstract

Researchers have begun to explore new selection variables in search of measures to complement a long industry tradition of IQ-type intelligence or ability tests as predictors of job performance. Among the most promising variables is practical intelligence(PI), the construct supposed and measured by Sternberg. A PI inventory for entry-level job applicants was developed and known to have predictive validity with the criterion measure of supervisor's ratings in Kim, Kim, Lee(1996)'s study. The purpose of the present study was to examine the construct (particularly discriminant) validity of this inventory in a large corporation. More specifically, the relationships between the PI inventory and other common selection measures such as general or academic intelligence(AI) test, personality test, interview, TOEIC, education level, etc. to examine whether the PI inventory has the potential to make a unique contribution to the field of selection testing. The inventory was which comprised of 25 job-relevant situations or scenarios with 4 or 5 response alternatives was administered to approximately 20,000 entry-level job applicants each time in May, 1995, December, 1995, and May, 1996. It was found that the PI inventory showed weak correlations with AI inventory ranging from .22 to .29 across five job groups. It also correlated trivially with personality test(about -.14~.14), interview(about .06), and TOEIC(about .09). In addition, the correlation between the PI test and education level was .26 which was much smaller than that between AI inventory and education level. Consequently, PI is not a proxy for measures of AI, personality or interview, and the test of PI could increase the variance not accounted for in the real-world criterion of job performance by existing selection measures. Lastly, the implications of these findings and limitation of this study were discussed. Also additional research tasks on PI were suggested.