Job applicants might try to cheat a per-hire assessment.
Personality assessments offer many advantages, but they also seem like they could be easy to cheat. These assessments allow you to freely select your responses—there is nothing to prevent you from selecting responses that might make you look good, even if they don’t accurately describe you.
But how often does cheating actually happen, and how does it affect the reliability and validity of assessments?
Research on cheating
What does it look like when someone fakes a personality assessment?
If you want to find out, just ask people to do it.
Dozens of studies have done this. The most common result from is that you get more extreme responses. Ask people to cheat an assessment, and they tend to agree or disagree more strongly in their responses. Thus, they appear to be higher or lower on most traits than they actually are.
Researchers have also tried asking participants to imagine that they are a job applicant. Interestingly, this has a much smaller effect than asking people to cheat.
There have also been dozens of studies comparing assessment results from job applicants to assessment results from job incumbents. The difference between them is even smaller.
In short, there are some differences, but it’s also clear that job applicants don’t even try to cheat nearly as much as they could.
What people think during a personality assessment
A deep dive into the underlying process reveals 3 main types of thinking.
- Semantic processing: Thinking about the possible implications of an item (Example: “If I agree with this statement, it suggests that I’m a nice person.”)
- Trait processing: Thinking about the trait represented by an item (Example: “I agree with this statement because I tend to keep calm.”)
- Conditional processing: Thinking about how the item might sometimes apply differently to you (Example: “I am really social and outgoing, but only when I’m with my friends.”)
By far the largest effect of asking people to fake an assessment is on semantic processing. When people are trying to look good, they spend much more time thinking about what each item might imply about them.
Semantic processing has almost no effect on assessment results.
Why? The implications of many items are ambiguous.
For example: “I enjoy learning about history.”
What’s the desirable response to that? Is it a good thing or a bad thing?
You might get a more universal response to an item such as: “I am a moral person.”
But personality assessments do not contain many items that are so transparent, and obviously “good” or “bad.”
You can only cheat an assessment if you know what the “right answers” are supposed to look like.
What are the “right answers?”
There are basically two ways this can be determined.
1. Use a theory to determine what you’re looking for. This is usually not your best option.
2. Use a data-driven approach to determine statistical relationships between assessment results and important outcomes. Not only is this approach much more difficult to cheat, it also provides more valuable and defensible results.