How not to spot personality test fakers
How do you know when someone is answer honestly?
14 September 2011
Personality tests are an effective recruitment tool: higher scorers on conscientiousness and lower scorers on neuroticism tend to perform better in the job. But a major weakness of such tests is people’s tendency to answer dishonestly. A study now shows that a popular approach to spotting cheaters is likely to be ineffective.
This approach, which has gained momentum in the research literature, is to focus on applicants’ response times. Honest test-takers show an inverted U-shaped response profile, being fast when they strongly agree or disagree with test items (these come in the form of statements about the self, such as “I pay attention to details”), and slower when they answer more equivocally.
This is thought to reflect a process whereby test takers refer to their self-schema and find it easier to answer when statements clearly conform or contradict this schema.
At least two theories predict that fakers won’t show this inverted U-shape, and that response times therefore offer a way to expose those who are cheating. One theory has it that fakers refer to their self-schema and then exaggerate the truth on key statements. This has the effect of extending answer times for unequivocal answers, flattening out the inverted U-shape response time profile shown by honest answerers.
Another theory says that fakers don’t refer to a self-schema at all – they simply assess the social desirability of each item and exaggerate answers where necessary. This is a cognitively simpler task than referral to a self-schema, and again the inverted U-shaped response profile is predicted to flatten.
To test these predictions, Mindy Shoss and Michael Strube had 60 undergrads (38 women) complete a personality test (the Revised NEO Personality inventory) three times: once honestly, once to create a general good impression, and lastly, either to create a good impression specifically for a public relations role, or specifically for an accountant role.
The key finding is that participants showed the inverted U-shaped response time profile regardless of whether they were answering
onestly or not. Response times were faster overall for the fakery conditions, and the inverted U-shape was actually accentuated in the specific public relations fakery condition. Shoss and Strube said these results are consistent with the idea that fakers form, and refer to, an idealised personality schema in their mind when completing a personality test, and so their answers show a similar response time profile to an honest test-taker.
The accentuated inverted U-shape for the PR-role condition comes from the fact that the schema for such a role is like a caricature, making unequivocal answers for certain items even easier to provide than usual.
Digging deeper, the researchers found that when striving to make a good impression, participants scored higher on extraversion, agreeableness, openness and conscientiousness and lower on neuroticism. The inverted U-shape in response times was greater for agreeableness and conscientiousness in the fake conditions than when answering honestly.
“This study casts doubt on the validity of response times for detecting faking in general,” the researchers said. “… it seems that researchers and practitioners interested in detecting and reducing faking would do well to focus on other strategies.”
An alternative approach to reducing test fakery is to force applicants to choose between pairs of equally appealing statements about themselves, as reported previously on the Digest. Other recent research has shown that many recruitment measures might actually be testing applicants’ ability to discern what’s required of them, rather than anything more specific, as reported recently by the BPS Occupational Digest.
Shoss, M., and Strube, M. (2011). How do you fake a personality test? An investigation of cognitive models of impression-managed responding. Organizational Behavior and Human Decision Processes, 116 (1), 163-171 DOI: 10.1016/j.obhdp.2011.05.003