Objective. List experiment respondents may misreport the number of list items that they associate with in order to associate themselves with a socially desirable test item or to disassociate themselves from a socially undesirable test item. Tests for such misreporting were conducted.
Methods. List experiments from the 1991 National Race and Politics Survey, the 2006 Cooperative Congressional Election Study, and the 2008 Cooperative Campaign Analysis Project were analyzed or reviewed.
Results. Evidence suggested that some respondents deflated their report more than necessary to avoid association with a socially undesirable test item.
Conclusions. List experiments may provide inaccurate estimates of the percentage of the population to which the test item applies, but the direction of bias is predictable.