Assessment of the ACT WorkKeys Job Skills Test

ACT WorkKeys^® is a proprietary job skills assessment system for employers to help select, hire and train workers. These tests can also help job seekers and educators in assessing worker skills. Indiana has provided these tests for employers and has multiple sites where individuals can take these tests, such as Ivy Tech campuses and regional planning offices.

This article provides a preliminary analysis to determine the predictive value of ACT WorkKeys tests using tests given from 2008 through 2011. The data were matched with information from the Indiana Workforce Intelligence System (IWIS) to infer outcomes regarding work and wages.

The Data

Sufficient data existed to analyze six specific tests: Reading for Information, Locating Information, Observation, Teamwork, Applied Mathematics and Applied Technology. We controlled for gender, age at the time of testing and highest educational attainment. Education was added both because of its availability and the desire to help control for this probable source of selection bias within the data. Age was controlled as a partial proxy for experience.

The Methodology

The analysis examined the predictive potential of ACT WorkKeys tests with regard to two elements. The first was “time to employment.” People taking WorkKeys tests were either completing schooling or actively looking for employment. The time to employment was measured as the number of weeks from the date of the test until the individual appeared on a wage record. The observation count for this was 200,044. Not everyone taking the exam had a wage record, as some individuals may have never found work, retired or moved out of state.

Wages were also of interest. Ideally, if this were a longitudinal dataset, an assessment could have been made examining the score on the test versus wage growth over a five- or 10-year horizon. Unfortunately, such a longitudinal series was not available so two different wage measures were used. View pdf with more detail about the methodology and regression results.

Time to Employment Results

The time to employment analysis indicates a general reduction of unemployment time with higher ACT WorkKeys scores. While these results are encouraging and show significant promise, they are not conclusive. The ACT WorkKeys test is provided at the behest of the potential employer. The result could be simply consistent with the employer providing “employment decision weight” to a test it requires of applicants.

Wages after Test Results

The wage after test results also showed a strong relationship between scoring well and higher wages. However, this analysis is also subject to a source of selection bias. Applicants scoring well on the test likely had higher wages prior to entering the reemployment market, creating a potential for bias. It is difficult to parse the weight of prior wages from WorkKeys scores when determining the influence on subsequent post reemployment wages. One attempt to eliminate the potential bias  was to examine the impact of the scores on differences in salary of the individual before and after entry into the reemployment market. When examining the wage difference results, the link between test score and wage attainment dissipated. However, the results are encouraging and show some aptitude in predicting wages. The results will become more conclusive in the future with fewer data limitations and a longitudinal dataset.

Conclusion

These preliminary results suggest benefits to the ACT WorkKeys tests and their ability to predict worker productivity. Future and more robust research is warranted with the inclusion of enhanced wage data and increased explanatory variables to strengthen conclusions. More information about ACT WorkKeys is available at http://www.in.gov/dwd/2418.htm.

Timothy E. Zimmer, Ph.D.
Manager, Research and Analysis Division of the Indiana Department of Workforce Development