This past Monday the National Council of Churches USA announced the release of their 2011 Yearbook, a press release that traditionally includes the membership data for the 25 largest denominations in the country.
My first reaction, after a quick look at the data, was “nothing new here — move along to something else.”
My second thought was “why don’t I just take that part of the post from last year, copy and paste it for this year, strike out the old numbers and fill in the new ones.” In all honesty, the two sets of numbers look a lot alike and I was wondering if there was anything new worth saying about it.
Well, I finally came to my senses, remembered that my motto is “I never met a data set I didn’t like,” and on my commute home I thought about what I could do with it. I then spent my lunch hours the rest of the week crunching data. Yup, that’s the way I roll.
Now, a couple of years ago I correlated the NCC data against surveys about political opinions and found that for the mainline churches the degree of membership decline correlated with stronger liberal political opinions. But, based on reading I have done in the last couple of years I have modified this hypothesis and now think that part of the problem of decline is not the political opinions of the churches per se, but rather that the problem is a lack of clear and well defined beliefs and expectations, particularly in the mainline. That is to say that trying to be too broad in doctrine leaves those looking for a church uncertain about that church and no need to be committed to anything in particular. It is the hot and cold of Laodicea and shown on a small scale by the division of the Londonderry Presbyterian Church which split and, at least when I wrote about it a year and a half ago, the combined membership of the two churches had nearly doubled over what it was before the split. (Now, when I get the the end of this post I won’t necessarily have proven that thesis, but I think it will support it.)
Now, to give credit where credit is due, this is not something I pulled out of thin air but, as I said, saw in the studies and essays I was reading. Prominent among these, in the chronological order in which I read them: Beau Weston, Rebuilding the Presbyterian Establishment; Dean Hoge, Donald Luidens, and Benton Johnson, Vanishing Boundaries; Bradley Wright, Christians are hate-filled hypocrites… and other myths you’ve been told; and Kenda Creasy Dean, Almost Christian.
So, I set about seeing if I could find correlations between indicators of strength of faith and the NCC data. Thanks to Brad Wright’s book I knew that the Religious Landscape Survey by the Pew Forum On Religion & Public Life was a wealth of information. The data is split into two reports, the Religious Affiliation Report (full report ) and the Religious Beliefs and Practices Report (full report ). To tinker a little more, I downloaded The U.S. Religious Landscape Survey dataset from the Pew Research Center. The resulting analysis and data manipulation is mine and it should be kept in mind that “The Pew Research Center and the Pew Forum on Religion & Public Life bear no responsibility for the analyses or interpretations of the data presented here.” OK, you got the required disclaimer.
I was fun look at the raw data because there are some interesting details in there although they are generally not related to this present discussion. For example, of the 480 participants who identified themselves with the Assemblies of God, 420 said there was a heaven but 432 said there was a hell. While that may say something interesting about the theology, in fairness I would have trouble with the wording of the study’s questions because they were base on merit, that is if someone led “a good life” and not on Christ’s free gift of eternal life. Since individuals could self-identify the denomination they were with it is interesting to note that there is one who said Emerging Church, one each who identified as Liberal Presbyterian and Conservative Presbyterian. But my favorite has to be the two individuals who identified themselves as an Electronic Ministries Baptist and Electronic Ministries Pentecostal. Can I now call myself a Virtual Ministries Presbyterian? We will have to wait to see when the Open Source Church appears. 
I am going to keep playing with the dataset and see what other interesting details I can find.
Anyway, some additional interpretation details: The survey was conducted in 2007 so technically a bit of a time offset from the 2010 NCC data. In addition, the data package comes with one database for the continental U.S. and another for Alaska and Hawai’i. I only number-crunched the former which contains a bit over 35000 records. For the first set of correlations with the demographic data I have taken the numbers from Appendix 3 of the Religious Affiliation report which lists results as percentages with no decimal places. Results for religious behavior that I calculated from the provided dataset are reported as percentages with one decimal place. And for those interested in trying it themselves at home, the data is provided in SPSS format which you can also read with the open source package PSPP. I will talk about correlation coefficients which test only for a linear correlation and the data is supplied with a weighting scheme designed to reflect reliability, which I did not use for this initial exploration.
For the NCC data, of the 25 churches on the list only 14 provided numbers for membership change. Of these, we saw notable growth in the Church of Jesus Christ of Latter Day Saints (+1.42%), and significant growth in the Jehovah’s Witnesses (+4.37%) and the Seventh-Day Adventists (+4.31%). There was small growth in the Roman Catholic church (+0.57%), the Assemblies of God (+0.52%), and the Church of God (Cleveland, Tenn.) (+0.38%). The mainline/oldline churches had typical declines including the United Methodist Church (-1.01%), the Evangelical Lutheran Church in America (-1.96%), Presbyterian Church (U.S.A.) (-2.61%), the Episcopal Church (-2.48%), the American Baptist Churches in the U.S.A. (-1.55%), and the United Church of Christ (-2.83%). Slightly smaller declines were experienced by more evangelical churches, such as the Southern Baptist Convention (-0.42%) and the Lutheran Church – Missouri Synod (-1.08%).
All of the correlations I ran are available in a web-published Google spreadsheet and the sheet also contains the 2009 membership changes and correlations with those as well. For this discussion I only use the 2010 membership changes. As always, use at your own risk. For those who don’t regularly work with correlations a quick introduction: If the number is positive the correlation is direct and if negative it is inverse. Correlation statistics range in absolute value from 1, which is perfect, to 0 (zero) when there is no correlation. Values of 0.8 and greater are generally considered strong correlations and values below 0.5 have weak to no correlation and need to be looked at carefully. Also, this analysis assumes that the correlation is linear and I have not run tests for leverage effects by extreme values. (But as you will see in the graphs below there are a pair of high values that usually cluster nicely.)
The first demographic data I looked at was for members’ marital status and there was little to no correlation between that and a denominations growth rate. However, looking at the extremes of age distribution we find that growing churches have a higher percentage of younger members (18-20 years old) than declining members and the declining churches have more older members (>65) than growing churches.
These correlations are good with 0.77 and -0.78. The question is whether there is a cause and effect relationship. Are growing denominations growing because they have more young people, or are more young people there because they are growing. We can probably safely conjecture that the relationship is complex and mutual and there is a bit of each going on probably establishing a positive or negative feedback loop.
The correlation with number of children is somewhat predictable based on this preceding relationship. While families with no children are more likely at declining churches (correlation -0.63), it surprised me that the strongest correlation in the children categories was the relationship of families with one child to be at growing churches (correlation 0.81) and then to have families with two children to be completely uncorrelated (-0.03). The correlation returns with moderate strength for three children (0.63) and for four or more not quite as strong at (0.50). Like above, assigning dependency is problematic and there is probably a complex relationship. (Maybe something to crunch the numbers around a bit for.)
There is one other demographic relationship and that has a moderate correlation — college grads are more common at declining denominations (correlation = -0.55).
Now, what about the idea I really wanted to test – that patterns of behavior and belief that indicate more intense or dedicated religious practice are correlated with denominational growth. The survey provides us with several of these.
First, again taking a lead from Brad Wright’s book, I look at church attendance, as self-reported. I have combined six categories down to three with the frequent attenders (once a week or more than once a week) in one group, the occasional (less than once a week but still multiple times a year) in the second group, and the seldom to none in the last group.
In the first two cases there is a strong correlation with the frequent attenders (weekly or better) to be members of growing denominations (correlation=0.76) and the less-than-weekly to be members of declining denominations (correlation=-0.82). For the seldom to none, they are more likely in declining denominations, but the correlation is weaker (correlation=-0.40). For comparison purposes, the Presbyterian Panel asks a similar question and found that for members 26% responded that they attend weekly and another 38% said they attended “nearly every week.” That total of 64% is a bit higher than the 56.9% in the RLS data, but seems a reasonable match in light of the different wording of the questions.
The survey has two ways of looking at the importance of religion to the participants. The first is a direct question if their religion is very important, somewhat important, not too important, or not at all important. The percentages that answered very important and somewhat important are both well correlated with the growth/decline numbers, but in opposite senses. For those who said their religion was very important there was a correlation of 0.74 indicating they are more likely to be in growing churches. For those who answered somewhat important, the correlation is -0.74 and they are more likely in declining denominations.
The second is a question that asks “When it comes to questions of right and wrong, which of the following do you look to most for guidance.” Of the four choices, two were substantially preferred by respondents. “Guided by religious teachings and beliefs” is shown with the red squares in the graph below and has a 0.77 correlation with denominational growth. On the chart you can see the outlier to the trend at 0.57% growth which is the data point for the Roman Catholic church. Removing that data point the correlation jumps to a strong 0.83. As you can see, the other strong answer is “Practical experience and common sense”, shown in green, and that has an inverse correlation at -0.77. So in growing churches the members rely more on church teaching and in declining churches the members are guided more by their own experience. It is interesting, and somewhat surprising to this scientist, how far below the first two the reliance on philosophy and on science fall. And both of those have almost as strong inverse correlations.
You can have a look at the spreadsheet for a bunch of the other correlations I ran. It pretty much holds up that strong religious beliefs, certainty in those beliefs, and practices correlate with denominational growth while the moderate to weak responses for these things are inversely correlated and are more likely in declining denominations.
Well, crunching the numbers is the easy part. What does this all mean and can it be applied to reverse mainline decline?
First, let me say that I think it is difficult to separate what should be the neutral practices from the doctrine. As I said, correlation coefficients for the relationship between beliefs and growth/decline are pretty much identical to correlations between practices and growth/decline. To put it another way, at what point does regular weekly attendance at church change from being just a religious practice to being a matter of doctrine or belief?
Another tricky point here is that for most of the indicators measured, while the doctrinal ones may be teachings of the church, what the statistics show is not the effectiveness of the churches teachings directly, but the ethos of the church and the expectation for accepting those teachings. In other words, almost every church would want a member to be guided by the church’s teachings to determine right and wrong, but the growing denominations pass along not just the teaching, but the expectation that members take it seriously.
Finally, it has to be remembered that a denomination is composed of particular churches and in most cases we are measuring one of these on the level of the individual member and the other on the level of the denomination. Lost in the middle are the different congregations where this is actually implemented.
So by way of conclusion here are two things that surprised me in this analysis:
The first was the uniformity of the correlations. Yes, there were some variations but in general there were a lot that fell in the 0.7 to 0.8 range or the -0.7 to -0.8 range. This suggests to me that you should not be looking through this to find the “silver bullet.” Instead, these measures show broad patterns that probably reflect the overall nature of the denominations rather than where to improve on one or two specific practices.
The second thing that surprised me was how high the bar was. In looking at this data we are not seeing the line between growing and declining as being in heresy or apostasy. We are seeing the difference in whether members attend once a week or once a month. We are seeing the difference in whether someone is certain or God, or fairly certain of God.
Now, I welcome you to stare at the data and draw your own conclusions. My number one take-away is that “Being Christian” is not about what you do for one-hour on Sunday morning (OK, one and a half hours if the sermon goes long and you stay for a cup of coffee.) Rather, it is about how you live your life the other 167 hours out of the week. It is about whether that hour influences the other 167. It is about how your Christian faith affects the rest of your life. To me, these data show that the indicator of a growing denomination is a pattern of faithfulness in many areas of our lives.
Your mileage may vary. OK, now what do I do with my lunch hour next week?
Technical note: I think it is important to note that for questions with only two choices any correlations with a third variable will be of the same magnitude and opposite sign for the two choices. For the Guidance question above, while there were four choices, the Philosophy option and the Science option were selected by so few respondents that there are effectively only two answers, the Religion option and the Experience option. That is not the case with the demographic graph since substantial numbers of respondents fell into the age ranges between these two end groups. Combined, the two end members represent no more than 40% of the sampled population.