Correcting for “regression to the mean” in list selection decisions

In list selection decisions, direct marketers typically select, prior to the list purchase decision, a number of names to be sampled from the list. The number of responses from the sample is used to estimate the rollout response rate. If it is high enough, the list is purchased. Practitioners have voiced dismay that rollout response rates are often lower than sample response rates. This phenomenon has been described in the literature as “regression to the mean.” In this paper, correction for this bias is presented. Also, software is given that can readily assist the direct marketer correcting for this bias.