Investigating selection bias of online surveys on coronavirus-related behavioral outcomes

Abstract

The coronavirus SARS-CoV-2 outbreak has stimulated numerous online surveys that are mainly based on online convenience samples or commercial online access panels where participants select themselves. The results are, nevertheless, often generalized to the general population. In our paper we investigate the potential bias that is introduced by respondents' self-selection. The analysis is based on survey data of the "GESIS Panel Special Survey on the Coronavirus SARS-CoV-2 Outbreak in Germany", together with background information of the GESIS Panel. Our analyses show indication of a nonignorable amount of selection bias for measures of personality traits among online survey respondents. This provides some evidence that participating in an online survey and complying with measures that can minimize the risk of being infected with the SARS-CoV-2 virus are confounded. Hence, generalizing these results to the general population bears the risk of over or underestimating the share of the population that complies with specific measures.