Special report: The Pillar survey - By the numbers and under the hood
The Pillar Survey on Religious Attitudes and Practice
Twenty-four percent of Americans identify as Catholic, and two percent of Americans say they are converts to Catholicism. About 10% of previously churchgoing Americans say that since the pandemic, they never go to church. And rates of belief in the real presence of Christ in the Eucharist decline among weekly Massgoers by birth decade — except that those born in the 2000s show surprisingly high rates of belief.
That data and more is derived from The Pillar 2021 Survey Religious Attitudes and Practices.
Our survey received responses from 2653 Americans about their beliefs, religious affiliation, and practices. It’s a wealth of data we’ll continue to mine in future reporting, and which we hope to follow up with future research in the years to come.
We also hope that other interested people will do their own original analysis using our data, so we’re sharing the full raw data set, and we break down the demographics of our data sample below.
These are just a few of the top findings from our first Religious Attitudes and Practices survey. To learn more, read our series of in-depth reports:
A snapshot of faith
Reading the results from such a large survey can feel like drinking from a fire hose. So here are a few basic facts and insights.
Twenty-four percent of Americans identify as Catholic; another 10% were raised Catholic but now identify themselves as part of some other religion (or none at all.) Two percent of Americans are converts to Catholicism, while 22% are Catholics who were raised in the faith.
The percentage of people who describe their current religious affiliation as Catholic is smaller among younger generations, dropping from nearly 30% among older respondents to twelve percent among those born since 2000. The share of Americans who describe themselves as Protestant or Christian has remained fairly stable, while the number who describe their religion as “none” or as agnostic/atheist is growing.
About 40% of Catholics and other Christians said they were going to church on a weekly basis before the pandemic, while fewer than 20% said they never went. But when asked how often they go to church now, about five percent fewer said they were still going to church on a weekly basis. About 10% more say that they ‘never’ go to church now than before COVID.
It is worth noting that this rate of weekly church attendance is higher than has been found in some other surveys. The Center for Applied Research in the Apostolate (CARA) has found 20% to 25% of Catholics attend Mass on a weekly basis in their surveys, while about 50% say they attend at least monthly. Sometimes small differences in the way that questions are worded result in significant shifts in the way that respondents answer. It is possible that a portion of those in our survey saying they attend weekly intend to do so, but only make it to church a few times a month.
With many people who are raised Catholic leaving the church or not going to church regularly, an important question many have asked is what factors cause people to remain Catholic. Two of the most important factors in whether people remained active Catholics as adults were whether they were taken to Mass on a weekly basis by their parents and whether they attended Mass regularly while in college (a time of life during which many people fall away from the Church.)
Catholics and leadership
To read the news, you might get the impression that Catholics are riven by political divisions and increasingly do not trust Church leaders. But, respondents to our survey across the political spectrum showed high levels of trust in their pastors, in their local bishops, in the US Catholic bishops, and most of all in Pope Francis.
What we believe
Results also confirmed alarming trends which have been identified by other surveys. Only half of Catholics who go to Mass on a weekly basis stated they believed in the Real Presence.
Even among those who go to Mass on a weekly basis, belief in the Real Presence was lower among those born since 1970 than those before.
If you want to get really into the details, access the full dataset here.
Who we sampled
Here we summarize the demographic profile of the respondents within our sample groups and compare them with US Census data and other major data sources where possible to provide you with context for the survey results.
Our survey consists of three groups of respondents.
The nationally representative sample, with 1564 respondents of all faiths. This group was draw collected from research firm Centiment’s national survey panel, which is intended to be representative of the US in terms of age, sex, and region.
An oversample of an additional 1021 respondents who were raised Catholic for a total of 1525 people raised Catholic in both this and the national sample. No filters were applied based on whether the respondents was still Catholic. This group is intended to be a representative sample of people raised Catholic.
An oversample of an additional 68 respondents who were raised Catholic and are no longer Catholic to provide additional detail on people who have left the Church. This sample, plus the ex-Catholics from the other two sample groups, provide a total of 494 people raised Catholic who no longer consider themselves part of the Church.
The national sample had slightly more males than females, while according the US Census the US population has 1% more females than males. The results were within the error margin of +/- 3%
In the raised-Catholic oversample, we had significantly more completed responses from females than males. The only filter on the over-sample was a question on whether or not the respondent was raised Catholic. However, it’s possible that there was more willingness to complete a ten minute survey focused on people raised Catholic among women than among men.
Among both those in the original national sample who were raised Catholic (which was a male-heavy group of respondents) and the subsequent oversample, women were 6% more likely to leave the Church than men.
The fact that the raised-Catholic oversample contained significantly more women than men could suggest that women who were raised in nominally Catholic homes (many of whom no longer consider themselves Catholic) are more likely to list their upbringing as Catholic than men in similar circumstances, or it could suggest that women are more likely to leave the Church than men.
Perhaps unsurprisingly for an online panel, we were lighter on responses from the very youngest and the very oldest groups of adults: those born after 2000 and those born before 1950. Those born in the 1980s were also somewhat over-represented.
There was not any significant difference in age distribution between the national sample and the raised-Catholic oversample, and we segmented a number of key findings by birth decade in order to avoid any bias resulting from the over-representation of people born in the 1980s.
Our question on race and ethnicity allowed respondents to check all which they believed applied, so we compared the results to the US Census findings for race “alone or in combination”. (This also means that the results total to more than 100% since some respondents checked more than one box.)
Our national survey sample slightly overrepresented white and black Americans, and had 6% fewer respondents of Hispanic origin than the US Census. Although within our national sample we saw higher representation of Hispanics among Catholics than in the U.S. population as a whole, the overall survey is somewhat light on Hispanic respondents.
We based our household income question on the 2018 US household income quintiles and the threshold for the top 5% of households. This means that we would expect a representative sample to have 20% of respondents fall into each quintile.
Our results found more respondents than we would have expected in the bottom two quintiles with the top two quintiles and the top 5% somewhat under-represented. This is fairly typical of survey samples, in which it is hard to get responses from the most affluent Americans.
Overall, we believe that our survey sample provides a representative view into the religious beliefs and practices of Americans. Key issues that others should consider when reviewing the results or conducting their own analysis are: the sample is more female, more white, more middle aged, and more low income than the US as a whole.
Of those issues, we believe the one most likely to bias results was age, so we showed many of our key results broken out by birth decade in order to remove the effect of age bias. The next most important issue in terms of analyzing Catholics is the underrepresentation of Hispanics in the overall sample. While we did not find Hispanics to be significantly different from other Catholics in most of our analysis, capturing more responses from Hispanics would be desirable in future survey work.