Abstract
To understand the characteristics and attitudes of the US Jewish population, researchers have increasingly relied on respondents drawn from established online consumer survey panels. In the absence of a large, nationally representative study of US Jews, online panels provide a faster and lower cost alternative to cross-sectional surveys. The present paper describes general issues associated with studies conducted with consumer panels that affect the validity and generalizability of their findings. The paper focuses on one of the largest and most often utilized probability based consumer panels, that of Knowledge Networks. The key question about the results of studies based on the Knowledge Networks panel, as well as studies based on nonprobability samples, is whether their findings represent the population. Along with considering the bias that may result from sampling designs, including the willingness to participate in consumer research, the paper also considers whether bias can be addressed by application of post-stratification weights. Analyses are reported of a survey conducted using the Knowledge Networks panel. Various alternative weighting protocols were applied. Relationships between variables appear robust regardless of survey weights, but precise estimates of the magnitude of such relationships as well as overall population proportions require current and accurate socio-demographic data about American Jews for development of weights. Although it is possible that there is systematic bias in regard to willingness to participate in a consumer panel, the availability of high quality demographic data makes a prerecruited probability sample a valuable source of information for ongoing study of the American Jewish community.