ESTIMATING THE REDUCTION IN NONRESPONSE BIAS FROM USING A MAIL SURVEYAS A BACKUP FOR NONRESPONDENTS TO A TELEPHONE INTERVIEW SURVEY

Because low-income population groups are more likely than higher incom e groups to live in nontelephone households, the use of telephone inte rviews to survey low-income groups may result in relatively low contac t rates. Other sample members may refuse an interview. Such nonrespons e typically has an unknown effect on sample estimates. This article at tempts to gauge the relative effects on sample estimates from using a mail survey as a supplemental data collection strategy for increasing the response rate of sample members from a low-income population group who either could not be reached by telephone or who refused to partic ipate in a telephone interview. The assessment is conducted by using v ariables from the sampling frame and an unbiased comparison sample. Al though the use of the mail survey increased response rates by 56%, thi s resulting increase had only a marginal effect on sample estimates. I mplications of these findings for social work researchers and practiti oners are discussed in the larger context of estimating and adjusting for nonresponse bias in survey research, and recommendations are sugge sted for further research.