RTI uses cookies to offer you the best experience online. By clicking “accept” on this website, you opt in and you agree to the use of cookies. If you would like to know more about how RTI uses cookies and how to manage them please view our Privacy Policy here. You can “opt out” or change your mind by visiting: http://optout.aboutads.info/. Click “accept” to agree.
Nonresponse Bias for Univariate and Multivariate Estimates of Social Activities and Roles
Amaya, A., & Presser, S. (2017). Nonresponse Bias for Univariate and Multivariate Estimates of Social Activities and Roles. Public Opinion Quarterly, 81(1), 1-36. https://doi.org/10.1093/poq/nfw037
Nonresponse bias is a fundamental concern for survey researchers, as understanding when and to what extent it occurs is critical to producing accurate statistics. According to the social integration hypothesis, individuals who participate in a broad range of social activities and roles should be more likely to respond to surveys (Goyder 1987; Groves and Couper 1998). As a result, prevalence estimates of social activities and roles should be upwardly biased. By contrast, models predicting these activities and roles may be unbiased if the nonrespondents are missing at random, as the results of Abraham, Helms, and Presser 2009 suggest. Using the rich frame information available on the American Time Use Survey (ATUS) and the Survey of Health, Ageing, and Retirement in Europe (SHARE) Wave II, we compare the full sample to the respondent sample on 30 different social roles and activities. The results suggest that nonresponse bias was widespread and often large on univariate estimates, but was usually small in multivariate models and typically did not alter the inferences drawn from such models.