Abstract:
|
Declining participation in government surveys is a concern for data users who rely on such programs for high-quality population and household statistics. In recent years, a rising trend in sampled persons who refuse to participate in government surveys is alarming, since increased unit non-response requires more reliance upon adjustment methods that - when used excessively - can generate bias and reduce data quality. While there are many factors that may affect one's decision to participate in a survey, the authors are concerned with the influence of economic and political conditions. In a prior analysis, the authors built upon an earlier paper (Harris-Kojetin and Tucker, 1999) to explore the relationships of these external factors with refusal rates of the monthly Current Population Survey (CPS). The time series regression model of the 1999 paper was replicated, and the scope of that model was expanded to 2015. However, analysis of the more recent data indicated that the model was not ideal for studying modern refusal patterns. In this follow-up paper, the authors consider new covariates for the model and devise an alternate construction of the autocorrelated error.
|