The Public Religion Research Institute (PRRI) released The 2012 American Values Study last week, which looks at religious affiliation and election issues among those of different affiliations. The second line of the Executive Summary was a surprise:
“While Catholics and white mainline Protestants remain two of the largest religious groups in the United States, they have each experienced significant declines in membership.” [emphasis added]
There are fewer Catholics in the United States now than in the past? When was the population ever higher? The claim in the PRRI report is incorrect. As I understand it, this may have just been a poor choice of words rather than something that the PRRI researchers literally intended to say or believed. Regardless, it is the kind of thing that gets picked up and then repeated in error by a religion reporter (example). Then your pastor reads it and puts it in a homily. Your great aunt posts it on Facebook. Before you know it this suddenly becomes "conventional wisdom."
It is true that Catholicism, like all religious affiliations, does not retain all of those raised in the faith (although the Church's retention rate is higher than many others) but that does not mean that the overall number of Americans self-identifying as Catholic is in decline. Just look at the unaffiliated, the fastest growing (non)affiliation which loses more of it's youth (as they later become religiously affiliated), as a percentage, than the Catholic Church or any other major religious group (i.e., retention rates reported on pg. 9 of the PRRI report). PRRI researchers seem to have similarly relied on only part of the "ledger" of affiliation in that second line of the Executive Summary. Your bank account would likely have "significant declines in its balance" if you similarly did not pay close attention to how and when all deposits are made. The PRRI researchers later in the report provide a bit of a qualification:
“[T]he Catholic share of the U.S. population has remained relatively stable, despite losing a significant number of adherents (12%) and attracting relatively few converts from other religions (2%). The primary explanation for this paradox is an influx of Catholic immigrants, mostly from central and South America.” [emphasis added]
This already has become a classic of "conventional wisdom" in commentary on Catholic affiliation. You know, the Catholic Church is "hemorrhaging" members and their places are being filled by immigrants. Yet researchers rarely present data or do the simple math that would provide any evidence for how much immigration plays a part in the population dynamics. It just "has to be true." I have responded to different aspects of claims of membership decline and the effects of immigration in the past (1, 2, 3, 4) but I'll try to be more clear and present new data to show that this notion just does not add up the way it is imagined. It won't stop people from claiming this but maybe your great aunt will discover this post in a Google search and I can save her from making a misleading Facebook post. I try to do my part.
The figure below includes three of the highest quality long-term series of surveys that we have to study religious change over time (...and you need a "moving picture" to do this. A one-survey snapshot is insufficient. Some knowledge of demography and attention to how religious affiliation varies over the life-cycle is also important). The first is Gallup’s religious affiliation data (details). In Gallup's 2011 surveys, Catholic affiliation among U.S. adults was estimated to be 23%. In its first observation of the series, in 1948, Gallup measured Catholic affiliation at 22%. For the series overall, this averages 25%. The second source is the American National Election Study (ANES). This survey’s sample (details) is based on citizens so if immigration is so important to maintaining the stability of the Catholic population percentage we should see a big drop in Catholic affiliation here. From 1948 to 2010 the Catholic share of respondents in this survey averaged 23% with 21% self-identifying as such in 1948 and 19% self-identifying as such in 2008 (the most recent data point for the ANES. Another will be released in a few months). The final series (details) has a shorter range. This is the General Social Survey (GSS) which began in 1972. Here, Catholic affiliation has averaged 26% for the series with an initial observation in 1972 of 25% and the most recent observation in 2010 of 25%.
Differences within all three series are generally within margins of error. You'll find recent survey estimates for the adult Catholic resident population percentage typically ranging anywhere from 20% to 26% with occasional outliers in either direction. Taking any one of these surveys within this range and assuming that it is showing a decline or an increase would be bad science. Margin of error really does matter (more on this below). Thus, each of the three historical trends represent the stability that you can see in the figure below. The black line running through the figure represents a triangulated aggregation (by average) of these three Catholic population estimates. This smooths out some of the random variations (i.e., sampling error and occasional outliers) in each source and provides a more reliable estimate than relying on any one source.
Of course a stable Catholic population percentage amidst a growing U.S. population means that the absolute number of adults self-identifying as Catholic over time is increasing. Even a decline in the population percentage does not always result in a decline in the actual population. A sub-group population can still be growing and adding net positive members but fail to grow as fast as the overall population and thus lose a percentage point or more. Below I will isolate changes since 1980, which coincides with the beginning of the most recent wave of immigration the U.S. has experienced. If immigration is the only thing "holding up" the Catholic population we should definitely see it clearly in the post-1980 period.
The U.S. Census Bureau provides population estimates by year and by age group for the United States (and other countries) in a publicly accessible database (details). I collected adult population totals (those ages 18 and older) from 1980 and 2011 and multiplied each annual total by the aggregated Catholic population percentage for each year from the figure above. This results in the trend you can see below. Don't make too much of the sharp waves in the figure as these represent fluctuations within the margin of error from the surveys measuring Catholic affiliation. With these data the adult Catholic population, as measured by self-affiliation, is estimated to have grown by 32 percent from 41.3 million in 1980 to 54.7 million in 2011. It is not possible to measure the change in the under-18 Catholic population. However, The Official Catholic Directory records 30.4 million infant and child baptisms between 1980 and 2011 with 17.8 million of these occurring since 1993 which are not yet "visible" to national polls that almost always use adult samples. Some of those baptized will no longer affiliate themselves as Catholic later in life as we know that a good chunk of those leaving the Church do so in their youth. Some return when they are older as "reverts." These reverts, as I will explain below, are an important part of the story of how the Catholic population percentage has maintained stability.
How much of this increase in affiliation is related to immigration? Another question frequently asked in surveys is country of birth and parents' country of birth. Below, I use these from the GSS and Pew's Social and Demographic Trends survey for 2011 to estimate the impact of immigration (Note: the GSS is not fielded every year and in recent years shifted to being done every other year. With the exception of 2011, I have interpolated any other missing year using data from the neighboring years. This is a reasonable method as the distribution for place of birth does not shift annually in the aggregate like attitudes can). The figure below breaks out the adult Catholic population into three stacked groups. The first are those born in the U.S. to native-born parents. The second are those born in the U.S. who have one or both parents born outside the U.S. The final group includes all those who are foreign-born.
The adult Catholic population born in the United States to native-born parents grew by 24% from 1980 to 2011 from 27.4 million to 34.1 million. This population is not in decline as many assume and continues to make up a majority of the U.S. adult Catholic population. The number of U.S. adult Catholics who are foreign-born or native-born children of at least one foreign-born parent grew from 13.9 million in 1980 to 20.6 million in 2011. This represents a 48% growth rate. Not all of this growth is directly related to immigration and some of this is linked to differences in fertility rates. For example, the 2011 birth rate (i.e., births per 1,000 women ages 15 to 44) for Hispanic females was 17.5 compared to 10.8 for non-Hispanic white females. It is clear from the figure below that the Catholic population has gained some membership through immigration and the children of immigrants throughout the period (it should also be noted this statement would be true for most of American history in regards to Catholics and for some other religions in recent decades as well—including Nones. With native-born fertility rates falling below replacement, there are a variety of groups that now grow primarily through a combination of immigration and generational replacement). However, it is also the case that this growth is insufficient to explain the whole "paradox." More interestingly the percentage of adult Catholics indicating that they were born outside the U.S. is now in decline (since 2007) and the adult Catholic population percentage has continued to remain stable. This has occurred as immigration rates from Mexico have collapsed (1, 2).
The figure below shows the breakdown of the adult Catholic population by place of birth in 2011. About a quarter of adult Catholics (24%) are foreign-born representing 13.1 million individuals. This is up from 10% of Catholics being foreign-born in 1980. This is certainly part of the Catholic population growth story but it is also clearly not the only story or perhaps even the biggest story in the numbers. In 1980, 66% of adult Catholics were born in the United States to native-born parents. In 2011, this segment consisted of 62% of adult Catholics.
Also of interest is the declining numbers of Catholic adults who report that they are a child of a foreign-born parent(s) (i.e., 23% in 1980 and 14% in 2011). This may represent those who were children of the last waves of European immigration from the early 20th century passing away. The figure below shows how the proportion each group adds up to the total Catholic population percentage (i.e., summing the numbers equates the U.S. adult Catholic population percentage that was 23% in 2011. Note that the PRRI measurement of 22% is not statistically different from 2011). You can see there is a declining percentage for native-born Catholics due to their slower growth rate, but again the numbers of these individuals is growing in the aggregate population as shown above.
Immigration has contributed to the stability of the Catholic population percentage, as shown above, but not nearly to the degree argued by many (...Pew's 2007 data shows specifically that there are not enough foreign-born Catholics in the population to make up for the numbers of native-born Catholics who are estimated to have left. These data also document that a sizeable number of those who have left Catholicism are foreign-born. Note there is a difference between the results in Pew's report and what one can find in the data that are publicly available. No researcher is ever able to include a completely comprehensive analysis of their data in a single report). Thus, on the second statement highlighted above, I believe PRRI is simply putting too much emphasis on foreign-born Catholics to solve the paradox (...and they certainly are not alone in doing so). At the same time, I will reiterate that the PRRI statement is absolutely factually incorrect in claiming that the Catholic Church in the United States has experienced a decline in members (...even if one wanted to stretch to a "decline" in parish registration this would still be inaccurate as many young Catholics and Hispanic Catholics do not register but are still attending Mass with some regularity. We know this at CARA because we have surveyed more than 385,000 Catholics in-pew, during Mass to date). It is also incorrect when others make the assumption that there are fewer native-born Catholic adults with native-born Catholic parents now than in the past and that this group is declining in number (...someone let Robert Putnam know).
So what makes up the difference? Most pollsters don't think much about demography and seem to forget that there is an important population that is not often in our view. Polls are generally done with adults ages 18 and older. Here lies some of the briefly "disappearing Catholics." Consistent with CARA's own surveys, Pew notes that "Almost half of Catholics who are now unaffiliated (48%) left Catholicism before reaching age 18" (i.e., before they ever enter the population we regularly survey). Notice that PRRI estimates that those raised Catholic are equivalent to 31% percent of the population (p. 9) and Pew similarly estimates this to be 31.4% (p. 26 of the Religious Landscape report). Yet, as shown previously, the Catholic population percentage has never actually been 31% since the 1940s (...or in any earlier period of U.S. history). Some insight into teen religious affiliation can be seen in a recent CARA study. We conducted a national survey including an over-sample of 677 teens, ages 14 to 17, in Catholic households. Their Catholic parent(s) provided permission for their children to respond. Of this group CARA found that 174 or 26% did not self-identify as Catholic even though their parent(s) do and presumably some are unaware of a change in their child's affiliation. Many of the teens who leave their childhood religion grow up and continue to have no affiliation or join another faith. But as noted previously some come back to the Catholic Church later in their lives as "reverts" or returned Catholics (...and this is nothing new as Hoge et al. wrote about the topic in 1981). The complete solution to the "paradox" of the stable Catholic population percentage is a combination of immigration and fertility effects along with minors raised in the faith who leave as teens and later return to the faith most often in their 30s and 40s.
To better understand the revert process in more depth one would need a panel study which includes children or teens who are interviewed in their youth and again at later points in their life as adults. These surveys exist but it is unlikely that many include a religious affiliation question. If the U.S. Census asked religious affiliation this would also help us understand these life-cycle changes as well.
We do know reverts are in the Catholic population and in the pews. They tell us so when we ask. CARA surveys estimate there are more than 5 million Catholic reverts nationally (i.e., equivalent to 9% of all adult Catholics). PRRI or Pew don't "see" them because the structure of their questions do not allow them to. Both have asked for a respondent's current affiliation and the affiliation they were raised in. If I grew up in a Catholic household, stopped affiliating as such for a time, and later came back I would look like a "cradle Catholic" in their polls even though I had "switched" out for a time. Also there are many who were raised Catholic but who are now unaffiliated (leaving as minors) who will return to the Church in the future contributing to growth of the adult population (i.e., if history continues to repeat itself... and it may not. If it doesn't the Catholic population will likely be in decline).
We can see reverts in other Catholic populations as well. For example, Ireland asks about religion in its census. This is broken down by cohort in the table below across 20 years. The first row of the table isolates the cohort born between 1987 and 1991 who were under 5 in 1991. We can follow this group ten years later seeing their numbers when they aged 10 to 14 and finally can examine their population total when they are 20 to 24. The first four cohorts including those under 40 are all showing a decline in population. But then we see growth in the next three cohorts among those who are currently ages 40 to 54. Each cohort is lowest in numbers when they are in the 20 to 29 age range. The same thing happening again and again in the same way (i.e., unlikely to be related to something like changing patterns of immigration). Each experiences growth from their 20s into their 30s. For example, the cohort born between 1972 and 1981 showed losses as they reached their 20s and growth moving into their 30s. It's hasn't yet been enough growth to wipe out the losses from the 20s, but if this group is anything like those born 1962 to 1971 they should be back in positive territory in 2021 in their 40s.
So if life-cycle effects are so important to explaining how the U.S. Catholic population percentage remains stable and how the U.S. Catholic population continues to grow, why isn't this a part of "conventional wisdom" yet? In part because I don't think there are many people who want to hear it. Most religion reporters ignore research that indicates growth in or satisfaction with the Catholic Church. For example, Pew did a study in August that showed "large majorities of Catholics are satisfied with Church leaders." Oh you didn't hear about that? Even within the Church many don't want to hear it. Both "progressive" and "traditional" Catholics want to be able to argue that the Church is losing members and can only regain them by urgently doing __________. Many would like to continue believing the Church is hemorrhaging members and that Mass attendance is declining even though neither of these claims can be found in the data (the Mass attendance trend can be seen here). Reality marches on whether we pay attention or not...
A final note... PRRI did not provide much information in the report about margin of error and statistical significance other than the overall N and margin of error. This N is printed under every figure even though this cannot be reflective of some of these (e.g., anything summarizing results for likely voters). I estimated my own sub-group margins of error for the report (here). There are several findings in the study where I really wonder about statistical significance. For example, the researchers really had to stretch the data thin to get to likely voters by religion and race and ethnicity. When a national random sample survey has 3,003 interviews it is difficult to drill down to the levels this report tries to get to without having to deal with very large margins of error. At the same time, I will note that this practice is not uncommon among some pollsters and social scientists (...it just makes it more difficult for readers to parse out the most meaningful findings). For example, the 2006 Faith Matters survey used by Robert Putnam and David Campbell in American Grace: How Religion Divides and Unites Us includes a sample of only 212 Hispanic Catholics (sub-group Ns and margins of error were not published in the book). There was a lot made of differences between Hispanic Catholics and non-Hispanic Catholics in that book based on very few interviews and readers were never made aware of this (...I have noted my skepticism of some of their findings previously).
I think any research organization that is releasing data that they intend to be read by the media and disseminated to the public needs to be very careful about what is put into print. Saying the the Catholic population has declined significantly is going to be something that catches the eye of religion reporters and regurgitated. When that statement is false and people are misled, even when it is unintentional, it can't effectively be "pulled back." Reporters don't write a new story and even if they issue corrections these receive little notice. Similarly I think it is important that researchers provide some context of differences and changes that are statistically significant and those that are not. Reporters and the public tend to take even the slightest differences within polling far too literally. Science is about helping people understand the realities of this world. We shouldn't cloud this with results that are incorrect or insignificant.
Photo above courtesy of Patrick Gage from Flickr Commons.
