My impression is that Church rhetoric defines women by their roles more often than it does men. Women are wives and mothers. Even if they aren’t technically mothers, women are mothers, because that’s just who they are. Men, on the other hand, sure we’re admonished to be good husbands and fathers, but those roles are discussed as being much less central to who we are. I would be shocked, for example, if someone gave a talk titled “Are We Not All Fathers?” in General Conference.
When this difference in the centrality of women’s and men’s gender roles is discussed, one hope that is often held out is that the Church is changing. Women are coming to be defined less by their roles and more as people of worth even if they don’t take on those roles, and men are being reminded more often that our roles as husband and father should be central to our lives.
It occurred to me recently that I could easily test for whether such a change is actually occurring by looking at how often different words are used in articles archived at LDS.org. Specifically, I chose to look for how often role labels (wife and mother for women; husband and father for men) were used versus how often the general labels woman and man were used. So for each year from 1971 to 2007, I counted how many articles in the Gospel Library used the words woman, wife, mother, man, husband, or father. For each year, I found the ratio of articles using wife and mother to the number using woman, as well as the ratio of articles using husband and father to the number using man.
My first hypothesis was that the discussion of women’s roles relative to the discussion of women in general would be declining. This hypothesis would be supported if the wife to woman and mother to woman ratios were declining over time. My second hypothesis was that the discussion of men’s roles relative to the discussion of men in general would be increasing. This hypothesis would be supported if the husband to man and father to man ratios were increasing over time.
The results are in the figure below. The red squares plot the mother to woman ratio. The orange squares plot the wife to woman ratio. The blue circles plot the father to man ratio. The black circles plot the husband to man ratio. For each of the four ratios, the dashed line of the same color shows the best linear fit to the data.
Both the mother to woman and the wife to woman ratios declined with time. Only the decline of the wife to woman ratio was statistically significant, though1. You can see why the mother to woman ratio did not achieve significance: its values are much more scattered around the best fit line than are the values for the wife to woman ratio. These results offer partial support for my first hypothesis, that women’s roles are being emphasized less over time.
Both the father to man and the husband to man ratios increased with time. Although the father to man increase was much more dramatic, both increases were statistically significant2. These results are consistent with my second hypothesis, that men’s family roles are being emphasized more over time.
A few other random notes:
- These ratios are clearly extremely crude measures of how much women’s and men’s roles are being emphasized. Given that there are clear effects even using them, I suspect that a better measure might show the changes over time more clearly. Perhaps they would even show effects depending on who the prophet is. Unfortunately, I don’t have the time or energy to do a study with a more precise measure.
- It would be interesting to see how many articles refer to wife or mother versus woman instead of just looking at wife and mother separately, but unfortunately, the Gospel Library search tool does not support boolean “or” searching.
- I said at the beginning that I think Church rhetoric defines women more by their expected roles than it does men by theirs. Unfortunately, I don’t think these data can be used to test whether that’s true. My first thought, for example, was that I could compare the mother to woman ratio to the father to man ratio to see which was higher. The problem with this is that both father and man are used in many contexts that have nothing to do with fathers or men. Man is often used to refer to all of humanity, for example, and any talk that refers to Heavenly Father would show up as a match for father. If we assume (as I did here) that these kinds of usages are fairly constant over time, then we can take some meaning out of the change in the ratio over time. But without accounting for these usages, we can’t meaningfully compare the ratios for men’s labels to the ratios for women’s usages.
- Related to the previous point, if the use of man to refer to all of humanity is in decline in the Church, this would account for some of the increase in the husband to man and father to man ratios. If this effect accounted for the majority of the declines, though, we would expect that the slopes of the man lines would be more parallel. Given that they’re quite different (and it’s the woman ratios that are in fact nearly parallel) I think there’s clearly something else going on here. A change in how man is used might account for some of the decline in the ratios involving man, but certainly not all or most of it.
- I considered including partial year data for 2008, but ultimately decided against it. I was concerned that General Conference talks might be consistently different from other materials in the Gospel Library, so having data including only one General Conference might make the partial year data a bad proxy for the full-year data.
- Of course I began with 1971 because that’s as far back as LDS.org goes. Does anyone know where I can get electronic indexes of earlier Church publications?
1. For mother to woman, p = .102. For wife to woman, p = .002. These are two-tailed p-values based on a permutation tests: I shuffled each series of ratios 9,999 times and found the proportion of the 10,000 datasets (9,999 shuffled and one original) in which the absolute value of the slope of the best fit line was greater than or equal to the absolute value of the slope of the observed best fit line.