Skip to main content
. 2006 May 10;114(10):1479–1485. doi: 10.1289/ehp.9101

Figure 3.

Figure 3

Sample size required to demonstrate a statistically significant effect (p = 0.01) 90% of the time for different magnitudes of response (effect size) and different proportions contributed by sensitive subpopulations. The ability to detect a small overall change when only a small proportion of the total population consists of responders requires extremely large sample sizes.