While the strong consensus is to use probability samples as standard practice in survey research, interest in nonprobability samples has been around for decades (Brick, 2014). We thank our RTI colleagues, Jill Dever and Y. RTI’s work for the studies described in this article was funded through contracts with Duke University, with funding from the Ford and Annie E. We thank William Darity, Duke University, and Darrick Hamilton, Ohio State University, the co-Principal Investigators of the National Asset Scorecard for Communities of Color research team, which sponsored the two studies reported on in this article. KeywordsĬonvenience sampling, empirical comparisons, nonprobability samples, probability samples, rare populations, social media We suggest that nonprobability samples may be particularly appropriate for low-incidence populations we also suggest that similar techniques may be useful for other researchers as they investigate the utility of nonprobability samples. We conclude that combining cases from the two types of samples may be appropriate for analyses in these studies.
Analysis shows that while demographic characteristics were not consistent across the two types of samples, the source of the data-the probability sample or the nonprobability sample-was not significant in predicting the primary research variables of interest. Both studies collected rich survey data, particularly regarding household finances, enabling comparisons between respondents in the two types of samples. We discuss the use of a convenience sample in one study and social media recruitment in another when probability-based samples fell short of reaching target sample sizes for low-incidence populations. While probability samples are generally the preferred approach in survey research, nonprobability samples continue to be of interest and are used for multiple purposes.
This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0)
0 kommentar(er)
