Frequent Survey Requests and Declining Response Rates: Evidence from the 2020 Census and Household Surveys

Abstract One hypothesis to explain declining survey response rates over time has been that individuals are receiving more and more survey requests. However, there has been little prior investigation of this hypothesis, largely due to difficulty in knowing whether nonrespondents were recently sampled for a different survey. This article investigates the frequent survey request hypothesis by analyzing self-response in the 2020 Census for the United States. Specifically, do households that were sampled in the American Community Survey (ACS) or the Current Population Survey (CPS) from 2015 to 2019 have a lower self-response rate to the 2020 Census? By leveraging two large, nationally representative surveys with monthly data collections, these analyses have statistical power that smaller surveys may not provide. This allows for precise estimates of how the frequent survey effect varies by the time between two data collections, and how the effect varies depending on the similarities between the two data collections. Households recently sampled for the ACS had lower self-response rates to the 2020 Census, with the decrease varying from 1.65 percentage points for households sampled in January 2019 to 15.23 percentage points for households sampled in December 2019. Smaller effect sizes are found for the CPS, which has more dissimilarities to the decennial census than the ACS. In summary, these results provide additional evidence that the proliferation of surveys may lead to lower response rates.