Situational Judgment Test

An SJT measures practical judgment about how to act effectively in work situations, but it is construct-heterogeneous rather than a measure of a single trait. McDaniel, Hartman, Whetzel, and Grubb showed that what it taps depends heavily on the response instruction: knowledge instructions, asking what one should do, correlate more with cognitive ability, while behavioral-tendency instructions, asking what one would do, correlate more with personality traits such as conscientiousness and agreeableness. In practice an SJT score blends ability, job knowledge, and dispositional tendencies, which is part of why it predicts performance incrementally over any one of those measures alone, but it also means SJT scores should not be interpreted as pure indicators of a single construct.

Responses are compared to a scoring key that assigns values to the options. Keys are developed rationally, by having subject-matter experts rate the effectiveness of each option and credit choices accordingly; empirically, by relating options to an external performance criterion and keying those that distinguish high from low performers; or by combining both. As the Weekley and Ployhart volume discusses, the keying method is consequential because different approaches can yield different scores and validities, and the credibility of the experts or the quality of the criterion data directly determines how meaningful the scores are. Transparent, well-justified keying is therefore central to a defensible SJT.

To a degree, and it depends on the response instruction. Behavioral-tendency SJTs, which ask what a candidate would do, are more susceptible to faking because respondents can report what they think the employer wants rather than their genuine inclination, especially in high-stakes selection. Knowledge-based SJTs, which ask what should be done, are harder to fake because there are more and less correct answers, though they shade toward measuring ability. Research suggests faking inflates scores somewhat but does not eliminate validity. Designers manage this by choosing instructions deliberately, using less transparent keys, and combining SJTs with other measures rather than relying on them alone.