Publication: International Journal of Selection and Assessment
Article: Two step testing in employee selection: Is score inflation a problem?
Blogger: Benjamin Granger
Many organizations are moving from traditional proctored selection tests to unproctored web-based tests. Why? Among other advantages, unproctored web-based selection tests are often cheaper (no need for physical testing facilities or proctors) than traditional proctored tests (e.g., paper-pencil tests), and they can be completed by job applicants in any geographical location. But, as always, there’s a catch. One of the most problematic disadvantages of using unproctored selection tests is its susceptibility to cheating! In high-stakes selection contexts, does it seem reasonable that at least some job applicants will misrepresent themselves (cheat, fake) if given the opportunity? There is quite a bit of evidence suggesting that faking does occur, especially on personality tests. But what about ability tests that can’t be faked? Does cheating occur in these contexts (e.g., use supplementary material, have a particularly smart friend or family member complete the test)?
To address this potential problem, some organizations are using what is known as two-step testing . This method utilizes both unproctored online tests and traditional proctored tests. More specifically, job applicants complete a screening test online (unproctored) in the first step. The organization then selects the top applicants and tests them again (using a parallel form of the first test) in a proctored setting. But the key is this: organizations inform applicants ahead of time that they will be tested twice (the second time in a proctored setting).So does this two-step testing method discourage job applicants from cheating? Researchers Nye, Do, Drasgow, and Fine (2008) attempted to shed light on this issue by utilizing an ability test in an actual employee selection context. They were ultimately interested in finding out if two-step testing provides a solution to the potential problem of cheating on unproctored selection tests.
Although cheating on unproctored selection tests is a real possibility, Nye et al. found that when two-step testing was used, job applicants’ scores on the unproctored version of the test were not inflated. Thus, cheating did not occur (and everyone sighs in relief)! In fact, the authors found that scores tended to be higher for the second test of the two-step process. Nye et al. suggest that this may have been due to familiarity with the tests since they were highly similar (parallel).Nye et al.’s results suggest that two-step testing is a viable approach to maximizing the benefits of unproctored web-based selection tests while ensuring that selection decisions are not basedon distorted information.