Characteristics of the Population of Internet Panel Members

Abstract

The proportion of the U.S. population with Internet access has risen dramatically from one in five households (18%) in 1997 to nearly four out of five households (78.5%) in 2013. Consequently, the majority of market research and public opinion surveys are now conducted by Internet. Non-probability panels of volunteer respondents who agree to receive invitations to conduct surveys account for the largest share of online research. Despite the major role of non-probability online panels in surveys, we know little about the size and characteristics of their panelists. This paper investigates the potential coverage of Internet panels among American adults. Since online panels frequently partner with other panel organizations in order to generate samples that are larger or more diverse than their own panel, the population who participate in any online panels is more critical to evaluating potential coverage error and bias in this type of survey than the population of individual panels. The size and characteristics of the population of online panelists is estimated from a national dual frame random digit dialing survey conducted in spring 2015. We believe this is the first published estimate of the size of the total population engaged in online panels based upon a national probability survey. The total sample size for the survey was approximately 500, which is adequate to estimate that size of the population. We have also presented the characteristics of the online panelists in the sample since this is the first survey of that population. However, due to the small size of the subsample of online panelists, these findings should be treated as preliminary.