|
Abstract:
|
The COVID-19 pandemic greatly impacted data collection for household surveys, curtailing in-person interviewing across the world. For example, the U.S. Census Bureau conducted numerous in-person surveys before the pandemic, but temporarily switched to phone data collection for these surveys in March of 2020. To contact newly-sampled households, the Census Bureau relied on vendor-provided phone numbers matched to addresses. This sudden change in interviewing protocol and generally the effects of the pandemic on respondent behavior may have profound effects on who is able to be contacted for Census Bureau surveys (noncontact bias) and who is willing to respond conditional on being contacted (refusal bias). To investigate these biases, we use the Current Population Survey and the American Community Survey. We first use administrative data matched to sampled addresses to see how the characteristics of respondents and nonrespondent in 2020 compare to prior years. Next, we focus specifically on the extent to which phone number match quality can explain changes in noncontact bias, as these phone numbers were the primary method of contact.
|
