With a sample of 12 men from PSYC-2005, we tested the hypothesis that mens explicit gender stereotype about language arts correlates with their implicit gender stereotype about science. We found a correlation of .49, indicating a pretty strong positive relationship (as you know, Cohens rule-of-thumb for a large correlation is .5). But we couldnt reject the null hypothesis of no correlation because of our small sample size (low statistical power). Sure, we could have gotten this strong relationship just by chance, because of sampling error, but we bet its a case of Type II error, you know, a false negative! Were tired of these low-powered class examples! What if sample size for men was same as for women (N=56)? Would r = .49 be statistically significant at p < .05?

Critical t = 2.004