“…In that case, many studies have shown that for ANOVAs and t tests, departures from normality are not critical (e.g., Glass, Peckham, & Sanders, 1972;Harwell, Rubinstein, Hayes, & Olds, 1992;Lix, Keselman, & Keselman, 1996;Schmider, Ziegler, Danay, Beyer, & Bühner, 2010). For t tests, the Type I errors still meet Bradley's liberal criterion even with extreme deviations from normality, at least if the sample size is higher than about 20, if there are (approximately) equal numbers of observations per group (balanced design) and if two-tailed tests are conducted (Kubinger, Rasch, & Moder, 2009;Sawilowsky & Blair, 1992).…”