“…Instead of modeling skill level as a moderator variable, studies have typically controlled for skill level in order to test whether the effect of stereotype threat is statistically significant beyond that of skill level or to increase the power to detect the effects of stereotype threat on performance (Steele & Aronson, 2004). Many studies examining the effect of the stereotype about women's math abilities have sampled those with relatively high SAT math scores (e.g., Martens, Johns, Greenberg, & Schimel, 2006;Marx & Roman, 2002;Quinn & Spencer, 2001;Schmader, 2002;Schmader & Johns, 2003;Shih, Pittinsky, & Ambady, 1999;Spencer et al, 1999) or have controlled for SAT math scores for these reasons (e.g., Gonzales, Blanton, & Williams, 2002;Hollis-Sawyer & Sawyer, 2008;Inzlicht & Ben-Zeev, 2000Keller, TESTING HYPOTHESIZED MODERATORS OF STEREOTYPE THREAT 2002, 2007Lesko & Corpus, 2006;Marx & Roman, 2002;Schmader, 2002;Schmader & Johns, 2003;Steele & Aronson, 1995;Vick, Seery, Blascovich, & Weisbuch, 2008;Wout, Danso, Jackson, & Spencer, 2008;Wout, Shih, Jackson, & Sellers, 2009). As Steele (1997) noted, it is those with greatest skill who are expected to suffer the most from stereotype threat, implying those with less skill in the domain will suffer less or not at all.…”