“…The first studies to informally include benchmarking (e.g., Wade, Treat, & Stuart, 1998;Merrill, Tolbert, & Wade, 2003;Hunsley & Lee, 2007;Forand, Evans, Haglin, & Fishman, 2001) lacked statistical comparisons between routine care and benchmark studies: they provided the effect sizes for each, and because the two are close to each other (and in some cases the routine care actually yielded a larger effect size), the authors conclude that treatments were generally effective. Weersing and Weisz (2002)'s formal approach, which provided a statistical framework using meta-analytic methods to compare naturalistic and RCT samples, has two advantages over less formal approaches.…”