<p>The change this study reports is enormous. Going from 42% binge drinking in the coed dorms to 18% binge drinking in the single-sex dorms? You don’t need a big sample size for that to be statistically significant. 500 students would be ample. The sample size is fine. And as for whether the results at these five campuses are applicable to other college campuses-- again, the reported change is so big that if the results are correct, there will be a big effect at other campuses too.</p>
<p>However, one mystery is why the researchers only got 500 students. The five colleges surely have many more students than that. Are the 500 students a good random sampling of the student bodies at these schools? At least at the four non-religious schools, most dorms are mixed sex, so how many single-sex dorm residents were polled from those schools? The sample size is fine, but the sample composition might not be.</p>