I would think that even the casual observer would notice that over the last couple of weeks there has been an explosion of articles in the news, and a corresponding explosion of discussion in the blogosphere, about the changes in membership patterns in American churches, particularly the mainline denominations.  While I think that there are a number of reasons for this, it appears that what catalyzed the reaction was the publication of the American Religious Identification Survey (ARIS) by the researchers from Trinity College in Connecticut.  I have done a first read of the survey and will be studying it in detail and crunching some of their numbers over the next couple of weeks.  More details when I have had a chance to study it, but I want to make two comments related to the survey and the media response now:

1)  This survey (full report PDF) is the third in a series tracking the American religious landscape over time.  The first was in 1990, the second in 2001, and the present one in 2008.  Reading through the tables and main text, nowhere do they give error-bounds with their results.  In the Research Design section on page 2 they do say that their high number of samples (>50,000 in 2008 and 2001, and >110,000 in 1990) gives “a standard error of under 0.5 percent for the full population in 2008.”

The patterns in Table 1 showing changes in the broad categories of religious affiliation showed significant (and I mean that technically) changes in almost every category between the 1990 and 2001 survey.  (Other Religions only grows from 3.3% in 1990 to 3.9% in 2008 so that change is completely within the uncertainties.)  However, from 2001 to 2008 the changes in most categories are much smaller than in the earlier time period.  While “Other Christian” is a statistically significant decrease from 52.2 to 50.9% in seven years, no other category changes by more than 0.8% which the Research Design section leads us to believe is statistically indistinguishable. (Two numbers would need to be more than 1% different if each has an error bound of +/-0.5%)  Future publications with error bounds on each number, or more detailed error limits, may show that these changes are significant.

2)  In a CNN article on the study Mr. Mark Silk, of Trinity College and the author of Spiritual Politics, makes the following statement:

The rise in evangelical Christianity is contributing to the
rejection of religion altogether by some Americans, said Mark Silk of
Trinity College.

“In the 1990s, it really sunk in on the
American public generally that there was a long-lasting ‘religious
right’ connected to a political party, and that turned a lot of people
the other way,” he said of the link between the Republican Party and
groups such as the Moral Majority and Focus on the Family.

Now, this is an interesting statement and it may even be true.  The problem is that CNN does not provide Mr. Silk’s data or source so it is left as the expert assertion of Mr. Silk that this is the case.

(It is important to clarify that Mr. Silk is listed as being on the Research Team for the ARIS survey, but is not one of the Principal Investigators.)

Since there are three types of lies… For the sake of argument let me present some data and argue the opposite — Liberal Christianity is driving people away from the church, or at least certain denominations.

But first, I find it interesting that I can’t show that within a denomination based on presbyteries in the PC(USA).

The link between theological perspective and membership changes has been of great interest to me as I follow the voting patterns on Amendment 08-B in the Presbyterian Church (U.S.A.)  At the present time, I have found no clear association.

Here are the data from voting to date for all 97 presbyteries that have voted and reported their numbers in 2001-2 and 2008-9.  This first graph is the change in presbytery membership as a ratio of the 2007 membership to the 2000 membership (the most recent membership numbers available from Research Services), versus the percentage of “Yes” votes on 08-B.  Using the “Yes” votes as a proxy for the degree of liberal political leanings of the presbytery, the theory is that if one theological perspective is more attractive to the general membership of the PC(USA) we should see higher membership declines for opposite leaning presbyteries.

See the correlation?  I don’t either.  In fact the correlation coefficient is -0.08, very close to zero for no correlation.  So there does not appear to be an association between the presbytery leanings on this theological issue and membership decline.

What if we were to ask if the change in membership favored one theological perspective.  Here is the data with the same membership change measure but with the change in the percentage of commissioners voting yes.  Based on one theory, if the conservatives are leaving, presbyteries with higher membership losses should see an increase in the number of yes votes.

No, I don’t see it either and the correlation of 0.04 is even worse.  In fact, these two graphs are typical of what I have been finding for the PC(USA), that on a presbytery basis there is virtually no correlation between membership changes and voting patterns.  I’ll continue with this another time.

(Technical note:  I did not include Eastern Korean Presbytery in the data above because it has a remarkable growth of 2.68X from 2000 to 2007.  It is an extreme outlier and it would have leveraged the correlation.  However, as a single point it supports the theory that the fewer yes votes the more attractive the church:  It has always been a 100% vote for “fidelity and chastity” (“No” votes) and is in a class by itself in growth.  And looking at the statistics it appears to be fairly constant growth over those seven years.)

Let me turn to another set of data, the membership numbers reported by the National Council of Churches.  A quick glance at the list shows that four member denominations had increases in membership:  Church of Jesus Christ of Latter Day Saints, Assemblies of God, Jehovah’s Witnesses, and the Church of God (Cleveland, Tenn.).  The denominations that reported losses of more than 2% were the UCC (-6.01%), AME Zion Church (-3.01%), and the PC(USA) (-2.79%).  Looking at the list of churches, the general perception is that churches with more “liberal” leanings have greater membership losses than those identified as “conservative.”  Is this perception true?

To measure “liberal” versus “conservative” let’s turn first to the Public Religion Research‘s new report Clergy Voices: Findings from the 2008 Mainline Protestant Clergy Voices Survey.  The first problem here is that they look at Protestant churches so only seven of the NCC churches are in the study.  But it is current and it is specific to denomination.

To supplement this I turn to a study done for the Southern Baptist Convention by Ellison Research.  In this study, from 2006, there is a table of “Political Views of Clergy” but besides the SBC the other denominational categories are more general.  I will use the number for “Other Baptists” for the second largest Baptist denomination, the National Baptist Convention (NBC).  For the “Pentecostal” category I will use the largest Pentecostal denomination, the Assemblies of God (AOG).

Finally, a 2003 article provides a final source: “The Political Attitudes and Activities of Mainline Protestant Clergy in the Election of 2000: A Study of Six Denominations,” by Smidt and others (the last of the others is Beau Weston) in the Journal for the Scientific Study of Religion, v. 42, p. 515-552.  This survey is older but specific to denominations and adds the Reformed Church in America (RCA) to the list.  Getting the RCA membership change was not straight-forward since they are not on the NCC top 25 list, but a PC(USA) comparison reports a 12% decline in the decade ending in 2005 so I’ll use an annual change of -1.25%.

Denomination	Ann. member. change	% Progressive Clergy voices	%Liberal Ellison	%Liberal Political Attitudes	median %
SBC	-0.24%		4%		4%
UMC	-0.80%	28%	35%	38%	35%
NBC	0%		5%		5%
ELCA	-1.35%	27%	29%	61%	29%
PC(USA)	-2.79%	34%	35%	46%	35%
AOG	+0.96%		7%		7%
Episcopal	-1.76%	43%			43%
ABCUSA	-0.94	24%		23%	24%
UCC	-6.01%	55%			55%
DOC	0%	39%		61%	50% (avg)
RCA	-1.2%			26%	26%

Just a technical note that since the Political Attitudes study was significantly higher in a couple of cases I chose to average when I had two values and use the median if there were three values.  Also, the clergy voices study had both a “Progressive” measure and a “Liberal” measure and I chose to use the Progressive because of the better agreement with the Ellison numbers.

Is there information in this noise?  There sure seems to be a correlation.  The R-squared is 0.44 which gives a correlation of -0.66. That is pretty good.  If you are worried about the high UCC number leveraging the correlation, the correlation is -0.52 without it, so it has an effect, but not an overwhelming one.  And yes, some non-linear relationship would fit the data better, but for a preliminary exploration the linear works well.  Clearly, a detailed analysis of this would want to use a greater quantity of data that is consistent in source and acquisition.  And the decline numbers are for just one year and a multi-year trend might show something completely different. But this preliminary exploration is interesting.

So there is a correlation between the political leanings of the clergy and the change in membership such that membership losses and liberal leanings increase together.  I do admit that I have not established a cause and effect here:  Does the more liberal clergy encourage the loss of membership, or is the loss of membership preferential to the conservatives leaving the liberal side behind in the Mainstream churches?  Or for that matter, are these two similarly correlated to another causal factor I am not considering.  But for some reason people seem to be leaving the more liberal churches preferentially.  The interesting thing is that if Mr. Silk is correct the rise of evangelicals did not uniformly drive members from the Mainstream churches but it was preferential to the more liberal churches.  (Note: We may be talking about two different time periods here as well, Mr. Silk in the 90’s and I in this decade.  The other major difference is that I am using reported membership numbers while he is looking at the larger population of those who self-identify with the denominations. And there are a host of other factors involved in the protocols of the surveys that might invalidate any comparison.)

Enough statistics and number crunching for today.  More to come.  But let me just say that even after my first read of this, and several of the documents that have come out recently, I don’t agree with every conclusion I read.  Your mileage may vary.

Programming note:  For the next week or so I will be intensely occupied with some meetings.  I have a couple of unrelated posts in the pipeline but more of this and other intensive analysis will be on hold for a week or more.  Sorry about that.  Thanks for understanding.