Critics of personality measurement make two claims. The first is that personality measures yield only modest to non-significant validity coefficients. To support this claim, critics typically cite Guion & Gottier’s view that “there is no generalizable evidence that personality measures can be recommended as good or practical tools for employee selection.” But what did Guion really think? I spent two days with him in 1984 talking about this. He said that personality is the most important factor influencing occupational performance. He also said that, despite its obvious importance, the data justifying the use of personality measures were weak and that was his point. Hogan and Holland show that validity coefficients for well constructed personality measures are only slightly smaller than those for measures of cognitive ability.
The second criticism is that personality measures are “…vulnerable to faking.” One problem with this criticism is that the definition of faking is incoherent. There is an ordinary language definition, and there is a psychometric definition. Philosophers repeatedly point out that the ordinary language definition makes no sense. For example, child-rearing concerns teaching children to balance their natural impulses against cultural norms, and to fake when appropriate. Successful adults know how to fake appropriately—because you can’t have a career by consistently telling people what you really think. Good manners and hypocrisy are the same; as Jean-Paul Sartre famously noted: “Sincerity is a very carefully constructed performance.”
Psychologists have studied faking for over 60 years using psychometric methods. That research converges on two robust conclusions. The first is that there are two distinct forms of psychometric faking. One is usually called “self-deception”—a tendency to make unlikely claims about oneself (“I have a good sense of humor”). Virtually everyone says “True” to this item, but it is not true for 66% of the population. Are they faking? The other form of psychometric faking is usually called “impression management”—a tendency to endorse items that can’t be true (“I have never told a lie”). When people endorse items like this, they must be faking. As Uziel notes, faking research should focus on “…social desirability scales that aim to measure conscious lying and other deception.” I will come back to this point shortly.
The second conclusion from 60 years of faking research is that the two dimensions of faking—self-deception and impression management—are trait like. They show the same temporal stability as any other trait measure; in twin studies, these measures have the same heritability coefficients as any other trait measure; and they predict substantive performance variance. Consider Table One, which presents meta-analytically derived estimates of the correlations between the two dimensions of “faking” and the standard dimensions of the Five-Factor Model. The correlates of self deception look like a “getting ahead” profile, and the correlates of impression management look like a “getting along” profile. Uziel provides a detailed review of the literature surrounding measures of impression management, and shows conclusively that these scales predict social effectiveness: greater life satisfaction, less aggressive behavior, stable and lasting marriages, and overall favorable interpersonal relations. Measures of impression management predict social effectiveness and not faking.
| Dimension | SD | IM |
|---|---|---|
| Openness | .13 | .03 |
| Extraversion | .26 | .05 |
| Emotional Stability | .46 | .23 |
| Conscientiousness | .32 | .28 |
| Agreeableness | .10 | .32 |
K=11; N=2304
There is a consistent empirical literature which shows that, when asked to fake, people can alter their scores compared to a “non-fake” condition. The real question, however, is whether people fake when completing personality measures as part of a pre-employment screening process. J. Hogan, Barrett, and Hogan show that on a second trial, with incentives to fake, the scores for 95% of real job applicants stay the same. For the 5% whose scores change, 2.5% increase their scores beyond the standard error of measurement, and 2.5% decrease their scores. In addition, the former are more socially effective than the latter.
The ordinary language definition of faking is incoherent. Psychometric measures of faking predict important social outcomes. And the data show quite clearly that real job applicants don’t change their scores when retested in high stakes employment situations where they have every incentive to fake.
