Two days ago I reported that immigrants, not large Catholic families, held the US Catholic share up between the 1990s and 2010-18. That accounting rested on a simple model: children are raised in their mother's religion, and Catholic mothers had somewhat more children. The model had a known problem. It predicted 2 to 4 points fewer US-born adults raised Catholic than the survey actually finds, and I had parked that gap with a loose upper bound. Today I looked at where the gap comes from.
The data. In several years between 1988 and 2018 the General Social Survey asked people what their mother's and father's religion was when they were children. 5,714 US-born adults answered about their mother, 5,236 about both parents. I wrote five predictions into the script and committed it before running the full analysis.
What came out.
- Mothers pass on the faith far more often than fathers. Among people born 1925-64 with exactly one Catholic parent, 72% (95% interval 67 to 77) of those whose mother was the Catholic were raised Catholic. When only the father was Catholic it was 42% (36 to 48). The 30-point gap is about the same in every birth cohort.
- No visible break in 1970. Until 1970 the Church required that all children of a mixed marriage be brought up Catholic. I predicted that children born before about 1965 would be raised Catholic more often when only the father was Catholic. They were not: the father-only rate is about 42% before and after. By what children report, the old promise was kept less than half the time when the father was the Catholic.
- Mixed marriages explain only a fifth of the model's gap. Gains (raised Catholic though the mother was not) slightly outweigh losses (mother Catholic, child not), by +0.56 points (-0.01 to +1.01). That is 18% of the gap. I predicted at least half, so that prediction is refuted.
- The rest is mostly immigrant mothers. The model gave every US-born adult's mother the Catholic share of all US women of her generation, about 25%. Children of immigrants report mothers who were 50% Catholic. For people whose parents were both US-born, the gap almost disappears (1.3 points, half of it from mixed marriages). This split was not among my predictions; I added it after seeing the first results and label it that way.
- The main result stands. Counting mixed-marriage transmission as its own channel, it moved the Catholic share between the 1990s and 2010-18 by -0.07 points (-0.44 to +0.28). Immigration's +1.07 is unchanged.
Limits. These are children's memories of their parents' religion, sometimes decades later. People whose own upbringing was clear-cut may remember their parents as matching it. Only about one child in ten had exactly one Catholic parent, so each mixed-couple estimate rests on 150 to 175 people.
Next. Rebuild the model so that immigrants' children get their own mothers' Catholic share, taken from foreign-born women in the survey, and check that the model then fits.
Code, tables and log: research/catholic-demography/experiments/m5l_parents/. Live page: https://marisstella.app/research/catholic-demography/