Spreading the Gain: How an Abbreviated Nonresponse Follow-Up Can Improve Overall Survey Accuracy
Iannacchione, V. G., Federman, E., & Ebert, L. (2005, August). Spreading the Gain: How an Abbreviated Nonresponse Follow-Up Can Improve Overall Survey Accuracy. Presented at Joint Statistical Meetings, Minneapolis, MN.
When cost considerations require that a nonresponse follow-up sample be small, the reduction in bias obtained from the follow-up may be negated by the increased variability in the sampling weights of the combined sample. In prior work (Singh et al, 2003), we demonstrated that dual-frame estimation together with weight calibration can be used to develop composite weights that improve the stability of survey estimates that incorporate a small follow-up. The composite weights combine the variance of a biased but stable estimator (based on the main survey without the follow-up) with the variance of an unbiased but unstable estimator (based on the main survey with the follow-up) to produce estimates that strike a balance between variance and bias. In this research, we show how the use of composite weights can extend the gains in efficiency to questionnaire items not included in the follow-up. The gains are commensurate with the correlation between the items included in the follow-up and those not included in the follow-up. These findings suggest that even a small follow-up that uses an abbreviated questionnaire can have a beneficial collateral effect on overall survey efficiency.