ABSTRACT
Adverse impact (AI) has very often been presented as a dilemma (e.g., Arthur & Edwards, 2002; Campion et al., 2001; De Corte, 1999; De Corte & Lievens, 2003; Huffcut & Roth, 1998; Kehoe, 2002; Ployhart & Holts, 2008; Pyburn, Ployhart, & Kravitz, 2008; Sackett, Schmitt, Ellingson, & Kabin, 2001), a trade-off (e.g., Bobko, Roth, & Potosky, 1999; De Corte & Lievens, 2003; Roth & Bobko, 2000; Ryan, Ployhart, & Friedel, 1998; Sackett & Roth, 1996; Schmitt, Rogers, Chan, Sheppard, & Jennings, 1997), or a “perplexing problem” to solve (Campion et al., 2001, p. 150). For example, a study by De Corte, Lievens, and Sackett (2007) attempted to address a “selection quality-adverse impact problem” (p. 150), and a series of recent articles sought to describe a “diversity-validity dilemma” (Pyburn et al., 2008, p. 143). Indeed, according to Sackett et al. (2001), “this dilemma is well-known” (p. 303).
