Choose the Method for Aggregating Religious Identities that Is Most Appropriate for Your Research

Abstract

Some U.S. surveys measure many religious identities, including detailed Protestant affiliations. Researchers must decide how to aggregate these diverse identities. There are now a variety of options for aggregating religious groups into categories. Depending on the research question, it may be appropriate to use one of the existing options or to develop an aggregation strategy tailored to the project in question.

Supporting Evidence: Why Social Scientists Should Sometimes Consider Data Other than the General Social Survey

“…As we pointed out in our response (Hackett et al. ), the GSS does not have a stringent measure of this criterion, and, as a result, Lehman and Sherkat provide no original evidence of the theological exclusivism of respondents in their proposed taxonomy. Other data sources, however, can profitably be used to understand both the properties of the Lehman‐Sherkat taxonomy and the multitude of empirical questions not measured in the GSS.…”

Identification with Religion: Cross‐National Evidence from a Social Psychological Perspective

“…The dominant approach (particularly among sociologists and political scientists) is to conceive of “identification” as affiliation (e.g., Djupe, Neiheisel, and Conger ; Hackett et al. ; Hayes ; Lehman and Sherkat ) or to operationalize religious identity using one or more measures of religiosity (Hirschl, Booth, and Glenna ; Kim and Wilcox ; Lim, MacGregor, and Putnam ). Although Neitz () rightly cautions against conflating religious affiliation and religious beliefs, it has been all too common for scholars to conflate the terms identity and identification as well (e.g., Alwin et al.…”

Specificity and Conceptual Clarity in the Measurement of Religious Identification

“…We view it as Quantitative Methods 101, and we are not the only nor the first researchers to note this issue with RELTRAD (see Dougherty, Johnson, and Polson 2007: 488;Hackett et al 2018). To be sure, any treatment of nondenominational Protestants will miscategorize some respondents, but the "considerable and nonrandom loss of data" (Hackett et al 2018) stemming from dropping purportedly nominal nondenominational respondents in RELTRAD is not best practice by any standard. 1 Ideally, this sizeable group can be analyzed separately, as we do in the scheme delineating seven Protestant identifications.…”

“…We view it as Quantitative Methods 101, and we are not the only nor the first researchers to note this issue with RELTRAD (see Dougherty, Johnson, and Polson : 488; Hackett et al. ). To be sure, any treatment of nondenominational Protestants will miscategorize some respondents, but the “considerable and nonrandom loss of data” (Hackett et al.…”

“…To be sure, any treatment of nondenominational Protestants will miscategorize some respondents, but the “considerable and nonrandom loss of data” (Hackett et al. ) stemming from dropping purportedly nominal nondenominational respondents in RELTRAD is not best practice by any standard . Ideally, this sizeable group can be analyzed separately, as we do in the scheme delineating seven Protestant identifications.…”