A Barometer for Mormon Social Science

1. For anyone unfamiliar with the term, crowdfunding is soliciting donations from a large number of patrons usually via the internet.

2. Snowball sampling is a method by which a researcher asks respondents to recruit their friends or acquaintances as additional participants in the research project. Thus, by exponential growth, participation should “snowball.” Issues with generalizability arise by virtue of homophily. My friends likely have a lot in common just as do yours, therefore research participants recruited through networks likely differ systematically from the rest of the population. Because snowball sampling is nonrandom, we cannot be sure that we aren’t collecting data that is biased ab initio.

3. The exception of course are large-scale surveys like GSS and Pew among others, but these don’t focus exclusively on Mormonism.

4. Stephen Cranney, review of The Next Mormons: How Millenials Are Changing the LDS Church, by Jana Riess, BYU Studies, 58, no. 2 (2019): 177-83. For more on the issue of age, period, and cohort effects and the precarity these present Riess’s conclusions, see Cranney’s review of Riess’s book. Cranney is indeed correct that the only way to solve this issue would be to employ a longitudinal research design.

5. The debate rages on as to this point. As with many things in the social sciences, there are a host of studies on the topic, but a conclusive answer remains elusive. See James Tilley and Geoffrey Evans, “Ageing and Generational Effects on Vote Choice: Combining Cross-Sectional and Panel Data to Estimate APC Effects,” Electoral Studies 33 (March 2014): 19-27 for an example of both an age and a cohort effect on intergenerational conservatism.

6. Jana Riess, The Next Mormons: How Millennials Are Changing the LDS Church (New York: Oxford University Press, 2019), 46.

7. Benjamin Knoll and Jana Riess, “Infected with Doubt: An Empirical Overview of Belief and Non-Belief in Contemporary American Mormonism,” Dialogue: A Journal of Mormon Thought 50, no. 3 (Fall 2017): 1-38. In conjunction with Benjamin Knoll, the book’s author, Jana Riess, co-authored a paper in Dialogue, focusing on this portion of the Next Mormons Survey. It is from this paper which I draw details of their methodology for the analysis. The results were merely reported in the book with no real discussion on method.

8. Knoll and Riess, “Infected with Doubt,” 16. The authors briefly acknowledge the possibility of “dual-causation” in their article, however no satisfactory remedy is offered. In one place they state, “the fact that this analysis controls for other factors that are also correlated with strong activity growing up strongly suggests that attending seminary has at least some causative effect on the likelihood of being a Believer later in life.” These gymnastics are unnecessary and unproductive, measuring prior belief would have been a more direct remedy to the problem.

9. Judea Pearl, Causality: Models, Reasoning, and Inference, 2nd ed. (Cambridge: Cambridge Univeristy Press, 2009). It is helpful when designing a model to conceptualize it in a causal graph, sometimes called a directed acyclic graph (DAG), in order to visually tease out the logic of controls. Pearl is the household name for causal graphs.

10. Tyler J. VanderWeele and Ilya Shpitser, “A New Criterion for Confounder Selection,” Biometrics 67, no. 4 (2011): 1406-13.

11. The researchers could have directly measured an individual’s prior belief any number of ways and then conditioned or stratified on this. Simply conceived, such a strategy would be intended to calculate the average treatment effect among the control and the treated. The table following is to visualize the concept, I make no argument here as to ideal cutpoints:

Treatment

Prior State Treatment = 1 [Attended church] Treatment = 0 [Did not attend church] Total Outcome [Belief]

Prior Believer

Prior Doubter

Total Outcome [Belief]

Through standardization, we can also calculate the expected value of the observed outcome averaged over the distribution of the covariate of interest:

E(Y^a) = Σ_x E(Y|A = a, X = x)P(X = x) where a is the treatment and x is the covariate, in this case, prior belief.

12. Gary King, Christopher J. L Murray, Joshua Salomon, and Ajay Tandon, “Enhancing the Validity and Cross-cultural Comparability of Measurement in Survey Research,” American Political Science Review 98 (2004): 191-207. Given the Mormon penchant for testimony and “true conversion” I would argue that individuals would reliably recall and relate their belief through time. If concerns remained, the use of vignettes has been shown as a promising way to standardize survey responses open to subjectivity or interpretive bias. Not only could this be of help here, but would likely have been a help on questions later in the book in which Millennials overestimated their own religious behaviors compared to more mature generations. Vignettes could also have assisted in closing the generational gap on the subjective interpretation of survey questions.

13. John H. Goldthorpe, Sociology as a Population Science (Cambridge: Cambridge University Press, 2015).