EMPIRICALLY KEYING PERSONALITY MEASURES TO MITIGATE FAKING EFFECTS AND IMPROVE VALIDITY: A MONTE CARLO INVESTIGATION

Personality-type measures should be viable tools to use for selection. They have incremental validity over cognitive measures and they add this incremental validity while decreasing adverse impact (Hough, 1998; Ones, Viswesvaran & Schmidt, 1993; Ones & Viswesvaran, 1998a). However, personality measures are susceptible to faking; individual’s instructed to fake on personality measures are able to increase their scores (Barrick & Mount, 1996; Ellingson, Sackett & Hough, 1999; Hough, Eaton, Dunnette, Kamp, & McCloy, 1990). Further, personality measures often reveal less than optimal validity estimates as research continually finds meta-analytic coefficients near .2 (e.g., Morgeson, Campion, Dipboye, Hollenbeck, Murphy, & Schmitt, 2007). Some researchers have suggested that these two problems are linked as faking on personality measure may reduce their ability to predict job performance (e.g., Tett & Christansen, 2007). Empirically keyed instruments traditionally enhance prediction and have been found to mitigate the effects of faking (Kluger, Reilly & Russell, 1991; Scott & Sinar, 2011). Recently suggested as a means to key to personality measures (e.g., Tawney & Mead, In Prep), this dissertation further investigates empirical keying methods as a means to both mitigate faking effects and as a means to increase validity of personality-type measures. A Monte Carlo methodology is used due to the difficulties in obtaining accurate measures of faking. As such, this dissertation investigates faking issues under controlled and known parameters, allowing for more robust conclusions as compared to prior faking research.PH.D in Psychology, December 201