Key points are not available for this paper at this time.
The authors evaluate the effectiveness of two models often used to measure the extent of nonparticipation bias in survey estimates. The first model establishes a "continuum of resistance" to being surveyed, placing people who were interviewed after one phone call on one end and nonparticipants on the other. The second assumes that there are "classes" of nonparticipants and that similar classes can be found among participants; it identifies groups of participants thought to be like nonparticipants and uses them as "proxies" to estimate the characteristics of nonparticipants. The authors use these models to examine how accurately they estimate the characteristics of nonparticipants and the impact of nonparticipation on survey estimates of means of child support awards and payments in Wisconsin. They find that neither model detects the true extent of nonparticipation bias.
Lin et al. (Sun,) studied this question.