Stephen Cranney’s BYU Studies article “Who Is Leaving the Church? Demographic Predictors of Ex–Latter-day Saint Status in the Pew Religious Landscape Survey” not only answers an interesting question, but also provides some basic lessons on interpreting statistics. In as close to a representative sample of the US population as any survey that provides the needed data, he compares 191 people who grew up Mormon who no longer self-identify as Mormons and 379 who grew up Mormon and still think of themselves as Mormons. (Let me say that I refuse to accommodate current Mormon Church President Russell Nelson’s disavowal of the nickname “Mormon.”
Miles Kimball considers the following as important:
This could be interesting, too:
Miles Kimball writes xkcd Comics on Statistics
Timothy Taylor writes Could Environmentalists Just Buy What They Want?
Scott Sumner writes Anti-ostrich
Stephen Cranney’s BYU Studies article “Who Is Leaving the Church? Demographic Predictors of Ex–Latter-day Saint Status in the Pew Religious Landscape Survey” not only answers an interesting question, but also provides some basic lessons on interpreting statistics. In as close to a representative sample of the US population as any survey that provides the needed data, he compares 191 people who grew up Mormon who no longer self-identify as Mormons and 379 who grew up Mormon and still think of themselves as Mormons. (Let me say that I refuse to accommodate current Mormon Church President Russell Nelson’s disavowal of the nickname “Mormon.” An earlier Mormon Church President Gordon B. Hinckley embraced the nickname “Mormon.”)
On overall numbers leaving, Stephen points out that it is roughly the same as the number of converts, so that growth of Mormonism in the US is largely from natural increase: births minus deaths. (Mormons still have relatively high fertility and are younger than population averages, so natural increase is substantial.)
Stephen asks the interesting question of what religious status ex-Mormons go to. The answer (with some sampling error in it) is:
58% went to “no religion” or unaffiliated
18% went to evangelical Protestant denominations
8% went to Mainline Protestant denominations
10% went to generic Christianity
That leaves about 6%, about equally split between becoming Buddhists, joining a Mormon splinter group or going to some other religion not listed above.
Stephen’s Table 2 in this article gives a logit for predictors of leaving Mormonism rather than staying, among this sample of individuals who all grew up in Mormonism. I favor the logit that includes all covariates at once: the rightmost column. Stephen makes no multiple hypothesis testing correction, but the hypotheses he is testing are natural enough that at least I am not worried about a lot of hidden hypotheses that are not reported.
In my own research with coauthors, I have found using a false-discovery-rate (FDR) threshold to be so convenient and practical that there isn’t much of an excuse for anyone to neglect making multiple-hypothesis-testing corrections any more. (I hope to do a blog post in the future on using the false discovery rate approach.) A false discovery rate of x% means that on average no more than x% of the claims made are likely to be false. The more claims made, the more total claims are likely to be false, but the percentage of claims that are false in expectation is below the threshold.
Doing the false-discovery-rate approach myself, I can see that in the rightmost column of Table 2, everything with *** is significant with a false-discovery-rate threshold of no more than 1.5%, everything with ** is significant with a false-discovery-rate threshold of no more than 3.75%, and everything with * is significant with a false-discovery-rate threshold of no more than 15%. (Mechanically, take the p-value for which the significance level reported in the table is a ceiling, multiply by the number of hypotheses tested—15—then divide by the number of claims made up to that point, including the current claim, when they are made starting from the lowest p-value and going up. One of the ***s is the first claim. The ** is the fourth claim. The * with the lowest p-value is the fifth claim. There are some caveats to this procedure, but it is generally quite accurate.)
Given that in this case am not worried about hidden hypotheses that were tested, found wanting and then put in a drawer, these significance levels are adequate.
Here are the results, from ones we are more sure of, to those we are less sure of, with some of Stephen’s interpretative comments.
FDR threshold of no more than 1.5%
Cohabiting and divorce are correlated with leaving Mormonism. Stephen is admirably cautious about interpreting this:
Because there is no information on when people left the Church, it is difficult to speculate about why ex–Latter-day Saints tend to be divorced more than those who remain in the Church. It is theoretically plausible that the trauma of undergoing a divorce led to a loss of faith, activity, and ultimately identification with the Church; it is also possible that a loss of faith led to intermarital strife with a member spouse; finally, it is possible that Latter-day Saint marriages tend to have lower divorce rates overall. Some incidental support exists for this last point in the fact that the Latter-day Saint sample here has a significantly lower chance of being in the divorced category than the general non–Latter-day Saint PRLS sample, whereas the ex–Latter-day Saint sample does not show a statistically significant difference with the general sample. This suggests that ex-members may simply lose whatever Latter-day Saint–specific protections against divorce that may exist.
Political liberalism is correlated with leaving Mormonism. Stephen is careful to put things in perspective:
Ex–Latter-day Saints do appear to be more liberal than those who stay (see tables 1 and 2). However, the political switch may be less of a switch from “conservative” to “liberal” than from “conservative” to “moderate.” Contrary to stereotypes about liberal ex–Latter-day Saints, many ex–Latter-day Saints (27 percent) still identify as politically conservative, with 39 percent identifying as political moderates, and only a minority (35 percent) identifying as politically liberal.
While on their face the political findings support the familiar narrative of liberal latter-day Saints leaving over social issues, the fact that only a minority of ex–Latter-day Saints identify as liberals and that hardly any of them switch to liberal Protestant denominations nuances this perspective. While social issues are undoubtedly salient for some people’s exodus from the Church, it is likely that this narrative receives a disproportion- ate amount of attention in informal and online discourse on this subject, and the size of the liberal Latter-day Saint exodus over social issues should not be exaggerated.
FDR threshold of no more than 3.75%
High education predicts a lower probability of leaving Mormonism. Again, Stephen is admirably cautious in interpreting this result:
In the summary statistics, ex–Latter-day Saints tend to be less educated, with lower income. While distinct, these findings conceptually support prior research that has shown that, unlike most religions, for Latter-day Saints education is positively associated with activity.4 However, when education is controlled for, income becomes insignificant, suggesting that those who stay in the Church are wealthier because they are more educated.
… there are a number of theoretically plausible stories for why ex–Latter-day Saints tend to be less educated and have lower incomes. It could be that there is a Latter-day Saint emphasis on education and occupational success that leads to higher incomes and more education, or it could be that people are more likely to stay in the Church if the lifestyle is working out for them socioeconomically.
FDR threshold of no more than 15%
Men are less likely to leave Mormonism. For most of those who grow up in Mormonism, gender can be taken as fairly exogenous. Given the controversies about how Mormonism treats women, one might have thought that women would be more likely to leave. Here is what Stephen says about that:
Related to the issue of leaving over social issues is the question of gender. For a religion with an all-male priesthood that treats the notion of gender seriously, it is worth investigating whether women are more likely to leave than men. In this sample, men are overrepresented among those who have left; these results comport with prior findings in the large American Religious Identification Survey that men tend to disproportionately leave the Church.2 This difference may be a Latter-day Saint– specific manifestation of the fact that in the United States men tend to be less religious than women.
Older people are more likely to have left Mormonism. Stephen tends to dismiss this finding because he focuses less than I do on the logit with everything in it. I think this is real, because a simple hazard model would suggest that if one is ever going to leave Mormonism, having more time pass makes that more likely to have happened already. (Even if some people return to Mormonism after having left, when starting with a group that are 100% Mormon in childhood, there is likely to be convergence—which is most likely to be monotonic—to a higher fraction non-Mormon.)
Those living in Utah are less likely to have left Mormonism. Stephen explains the interpretive issues:
Ex–Latter-day Saints also appear to be less likely to reside in Utah in the summary statistics (34 percent versus 26 percent, but this barely misses the cutoff for significance at p = .065), and the Utah effect is sporadically significant in the regression analysis, suggesting that, whether because they are more likely to leave when growing up outside of Utah or because they are more likely to move outside of Utah after they leave (or a combination of both), ex–Latter-day Saints are disproportionately found outside of Utah compared to Latter-day Saints who did not leave.
Being Hispanic is not associated with leaving Mormonism (in a sizeable enough way to be reliably detected in this size of sample), but being Black non-Hispanic or of a race other than White, Black or non-Black Hispanic is correlated with leaving Mormonism. Here is what Stephen writes about that:
Finally, the racial effects found here lend themselves to any number of interpretations, but perhaps the most reasonable is that being a racial minority in a predominantly white Church may cause its own stresses that make continued activity and identification with the Church less likely.
A bit of background here that Stephen doesn’t here is that the Mormon Church is thriving in Latin American countries and the Mormon Church has many congregations with church services conducted in Spanish in the US, so that Hispanics in the US may feel somewhat less like outsider minorities in the Mormon Church than their raw numbers (4%) suggest.
Don't miss these posts on Mormonism:
Also see the links in "Hal Boyd: The Ignorance of Mocking Mormonism."
Don’t miss these Unitarian-Universalist sermons by Miles:
By self-identification, I left Mormonism for Unitarian Universalism in 2000, at the age of 40. I have had the good fortune to be a lay preacher in Unitarian Universalism. I have posted many of my Unitarian-Universalist sermons on this blog.
Don’t miss these guest posts on Mormonism by my brother Chris Kimball:
In addition, Chris is my coauthor for