Many people incorrectly assume that the number of interviews required for statistical significance is a function of the size of the population of interest—the larger the population, the larger the sample needed. In reality, relatively small samples, carefully selected from large populations, can yield precise estimates of how the total universe of potential respondents would respond. For example, political surveys of less than 1,000 respondents can predict the behavior of one hundred million voters with an accuracy of plus or minus 3%.
While the “confidence interval” (the “plus or minus 3%” quoted above) does get smaller and the precision of the answer improves as the sample size increases, appropriately chosen samples can provide statistically significant results with far fewer respondents. Aside from statistical significance, however, is the intangible factor of having a sample size large enough to appear credible to a judge or jury. This desire for “face validity” often leads to samples of 200 or 300 respondents. Another factor leading to larger sample sizes is the particular analytical techniques used, as some may require larger samples.