Back in 1998, as the dot com boom was heating up, pre-employment assessment providers began to migrate their tests to the Internet. Up to that point, virtually all pre-employment tests were conducted using paper booklets. However, there were many aspects of testing that made it a perfect fit for the Internet. Scoring paper tests is inconvenient at best. Booklets and scoring keys needed to be managed across locations. Test response data, a key to improving the performance of the test, was difficult to collect.

The Internet changed all that.

Internet-based pre-employment testing can be either proctored or unproctored. During proctored testing someone is watching (either in-person or virtually) to ensure the person taking the test is who they say they are, and to make sure they are not cheating during the test. Proctored testing takes time and money. As a result unproctored Internet testing (UIT), has become popular.

As soon as UIT became popular, psychologists and other testing professionals started to worry about the likelihood of cheating. Almost as quickly, a verification strategy was introduced by many test vendors to offset this risk. The process involves administration of an unproctored test, followed by a proctored verification test for candidates who advanced to a later stage of the hiring process. The idea was that anyone who cheated up front would be exposed (and embarrassed) during the verification stage.

However, the question remained. Specifically, psychologists and employers wondered what fraction of applicants cheat on their unproctored pre-employment test? Recently, a paper published by Tracy M. Kantrowitz & Amanda M. Dainis in December 2014 (1) provided some long sought after answers. The authors analyzed over 4,000 instances of UIT followed by proctored verification test to estimate the percentage of test takers who exhibit cheating behavior.

So, do you need to worry about people cheating on your unproctored Internet-Based pre-employment test? The results of this study say “probably not.”

Of course, this is for a certain type of test and a population of about 4,000 test takers and additional research is always a good idea. Certainly, some tests are easier to cheat on than others. Kantrowitz and Dainis based their research on an adaptive ability test, which may be more difficult to cheat on than some other types of tests. For instance, a skills or knowledge test is fairly easy to cheat on, since you can google for the answers in many cases, so it may have a higher rate of cheating. Also, some pre-employment tests may be perceived as “higher stakes” than others. For instance, if your dream is to work for the US Postal Service, when you take the USPS pre-employment test, the outcome will be very important to you - and you may be more motivated to go through the trouble of cheating on a test like that.

However, “one percent” is an important observation that should provide reassurance to companies using using UIT in their hiring process. All tests have errors associated with them. Sometimes they provide a high score to a someone who should have scored lower - a “false positive.” Other times they give a low score to someone who deserves a higher score - a “false negative.” The error rate for any test certainly depends on the test itself, but it is almost always higher than one percent - often much higher. The average error rate for an in-person interview, for instance is around 40 percent, for instance.

While a low cheating rate is reassuring for users of UIT, it’s also a reminder that it’s important to use multiple data collection points before you make a hiring decision. No method is ever 100% accurate, but when you collect information from multiple methods - tests, interviews, references - the real story starts to find its way out, and you will likely avoid making that terrible hiring mistake and might even hire that magical person who will make your life a dream!

Notes: (1) Tracy M. Kantrowitz & Amanda M. Dainis, Journal of Business and Psychology, Volume 29, Number 4, December 2014