Survey programs are operating in a difficult climate. Response rates for a number of major government surveys have declined. Among them is the Current Population Survey, where the response rate has fallen below 90 percent.

Research into this phenomenon has focused on micro-level influences, such as interviewer workloads, because survey programs are looking for data collection improvements that will maintain or increase response rates. Recently, survey methodologists in the Center for Survey Measurement and the Center for Adaptive Design began thinking about macro-level influences on response rates and asking the following questions: Can we identify large-scale influences on survey response? If so, what are the implications?

We begin to answer these questions by extending the work covered in the article Exploring the Relation of Economic and Political Conditions with Refusal Rates to a Government Survey (1999) by Brian Harris-Kojetin and Clyde Tucker. The authors used Current Population Survey data and a time-series regression approach to examine economic and political influences (unemployment rates, presidential approval ratings, inflation rates and consumer sentiment scores) on survey refusal rates from 1960 to 1988. The authors found relationships between refusal rates and some macro-level characteristics and refusal rates including presidential approval rating, consumer sentiment score and unemployment rate.

We are investigating whether the authors’ findings hold over the extended period of 1960 to 2015 and if there is additional information to add to their time-series models. Most of the data the authors used are still collected, allowing us to pick up where they left off. We successfully replicated their work up to 1988 and extended the work through 2015, and we are identifying covariates for additional analysis that the authors had not considered. We will be presenting our initial findings for the 2016 Annual Association for Public Opinion Research conference and proceedings.