In the spirit of my tag line – “I never met a data set I didn’t like” – I was thinking about how to drill down a bit further into the statistical results related to the size of churches being dismissed from the Presbyterian Church (U.S.A.).

In an ideal world there would either be statistical information about the theological leanings of every church in the PC(USA) or I could go to all their web sites and figure that out. Well, I don’t have the time to visit 10,000-plus web sites (and not every church has one) to divine their theological positions and I am not aware of a publicly available statistical study of all the PC(USA) churches. So I had to find something else.

What I will analyze as a related data set is the membership list of the Fellowship of Presbyterians that is posted on their web site. For the record, since this is a bit of a dynamic list, the version I will be using was copied on October 14, 2013.

The list has their member churches in two categories, those that are Fellowship members and those that are members of the related body, ECO: A Covenant Order of Evangelical Presbyterians. It also lists individuals who are members and on this pass that was not used but I will probably come back to that in a later post. For each of the churches in the list I searched a couple of sources for their 2012 membership. The primary source was the church directory for the PC(USA). For churches dismissed to ECO this source usually did not have the membership numbers so I turned instead to the presbytery statistics.

Problems arose for those dismissed before the 2012 numbers were recorded. In some cases I had their 2011 membership numbers. In other cases I had to use the 2007 number listed in the presbytery statistics. Out of 72 ECO congregations that are counted there were 16 that I used the 2007 number and 7 that I had the 2011 membership number. In addition, three congregations are fellowships with no data, one had no reported data and two are churches that divided with one group going to ECO and a continuing congregation in the PC(USA). None of these were counted. All of this and notes about naming differences are documented with the summary statistics in the comments on the spreadsheet.

So what are the summary statistics? Well, for all 256 churches counted the mean membership is 568.0 and the median is 301.5. For the subset of churches in ECO it is a mean of 417.9 members and a median of 170 for those 72 churches. For the 184 churches in the Fellowship only, the mean is 626.7 and the median is 343 members. We should consider the ECO numbers qualitative, or at best an upper limit, because of the heterogeneous nature of the data set, but even considering that the numbers are high.

For comparison, the numbers for all of the PC(USA) are a mean of 180 and a median of 89 members for 10,262 congregations. If you want to roll the statistics back, in 2007 the mean was 204 and the median was 103, still well below the statistics for the mixed data set.

Now, it would be nice to extrapolate these numbers to all of the churches that might be considered theologically conservative since, for example, the three churches in my presbytery that have requested dismissal to ECO, while statistically larger than the average PC(USA) congregation, are not now members of the Fellowship. While tempting, that does have its statistical pitfalls. What we can say is that these sub-groups which have self-identified together around particular tenets of conservative theology are statically larger, and apparently significantly larger, than average PC(USA) congregations and so there is at least the suggestion that this could hold true for conservative churches in the denomination as a whole.  We will come back to this after a bit more explanation.

I wanted to drill down into this idea a bit more so I found a couple of other data sets to test this against. The first is the membership of the More Light Presbyterians and the second is the membership of the Covenant Network. Here More Light chapters were removed from the data set as were fellowships and NCD’s. For a church that appears on both lists and was recently dismissed from the PC(USA) the 2007 membership numbers were used and the same for a church which did not report a number in 2012. There were seven churches who reported no data at all, one more I could not find but found what appears to be the continuing congregation in the same town, and one church that I could find no trace of it having existed outside this list. The two organizations are separate entities so in addition to their individual statistics I calculated the stats for the intersection of the data sets (i.e. those that are in both) as well as the union of the sets (i.e. for every church on both lists I had numbers for). The lists were copied from the internet on October 16, 2013. As with the other data set the numbers, comments and summary statistics are available on my spreadsheet. 

So, the 179 More Light churches have a mean membership of 213.7 and a median of 116. Close to, but still above the stats for the denomination as a whole. The Covenant Network churches have significantly higher numbers for their 359 churches – a mean of 322.6 and a median membership of 190. The group of 102 churches that are in both organizations has a mean membership of 217.1 and a median of 138. For all 436 churches the mean is 302.6 and the median is 172 members.

So it appears that having a leaning towards liberal theology is also good for membership. In fact the ECO data set and the MLP/Covenant combined data sets have similar numbers with ECO having a median of 170 and the MLP+Covenant having a median of 172.

Now, how well is the PC(USA) represented? These are groups that we have been and are hearing about all the time in the PC(USA). However, there are 256 congregations on the Fellowship list representing about 2.5% of the total congregations in the PC(USA). (Counting only the Fellowship exclusive congregations it is 184 or 1.8% of the congregations.) Similarly, on the combined MLP and Covenant Network list it is 436 congregations or 4.2%. As for membership the Fellowship list would represent 7.9% of the PC(USA) membership and the MLP+Covenant list is 7.1% of the total PC(USA) membership. In other words, this analysis covers 15% of the membership of the PC(USA) leaving 85% not represented. And if the congregations in these data sets are statistically larger it means the the remainder are on average smaller.

So far I have only discussed the summary statistics. To take this one step further let’s look at the distribution of congregation sizes using the binning that Research Services uses in its annual report.

	PC(USA)		ECO		Fellowship		MLP		Covenant
Number of Members	Number	%	Number	%	Number	%	Number	%	Number	%
1-50	3112	30.4	7	9.7	3	1.6	31	17.3	38	10.6
51-100	2394	23.4	15	20.8	17	9.2	50	27.9	68	18.9
101-150	1384	13.5	11	15.3	22	12.0	21	11.7	42	11.7
151-200	876	8.6	6	8.3	16	8.7	18	10.1	35	9.8
201-300	922	9.0	6	8.3	25	13.6	24	13.4	54	15.0
301-500	811	7.9	9	12.5	40	21.7	19	10.6	65	18.1
501-800	400	3.9	11	15.3	21	11.4	10	5.6	33	9.2
801-1200	186	1.8	4	5.6	20	10.9	4	2.2	12	3.3
1201-1600	71	0.7	0	0.0	7	3.8	1	0.6	5	1.4
>1600	78	0.8	3	4.2	13	7.1	1	0.6	7	2.0

And in graphical form (you can right click and view image for a better view).

It can be seen that for the PC(USA) as a whole the peak of the distribution is in the range for the smallest congregations. Interestingly, the four data sets considered here have, to varying degrees, a double peaked distribution. The most even of these peaks can be seen in the Covenant Network distribution with one peak in the 51-100 range and the second in the 301-500 range. For the More Light Presbyterians and ECO the lower peak in the 51-100 range is higher while for the Fellowship of Presbyterians the peak in the 301-500 range is higher. (I would note that I suspect that the double peak effect is artificially enhanced by the choice of ranges for the bins and that is something I might investigate more in the future.)

Remembering that correlation does not imply causation, there are three approaches to interpreting these results that shows there is something about having an association with  these affinity groups with theological leanings that is
good for church membership.

The first possibility is that churches with these declared affinity associations and therefore explicit theological leanings tend to attract people and so have larger memberships.

The second possibility is that when churches have larger memberships it provides something – possibility the resources of members’ time, talents, gifts and service – to be able to expand their ministry beyond just Word and Sacrament in that location so as to join and participate with various affinity groups.

The third possibility is to consider neither of these factors as the specific cause but to think of both of these as components of a larger picture. What I personally suspect is going on is that each of
these is simply one facet of a dynamic and defined church ministry with many different aspects that also probably includes living into a
mission statement (explicit or implicit), outreach and some form of
activism reflecting the church’s theological leaning and chosen affinity with a group.

In other words, these churches have higher membership because they are visible and
active and the affinity group membership is just one part of that
activity. So it is the sum total of this activity would tend to attract members to that church.

Based on my experience and observations I personally think the best interpretation is the third one, but there is nothing I see in the data itself to distinguish here. And yes, I am looking at a general trend in all of these churches and circumstances and therefore explanation will vary from one specific church to the next.

One tie-in here is the study from two years ago done by the PC(USA) Research Services called Fastest Growing Presbyterian Churches. My data above does not come anywhere close to testing all of the components listed in that report, but there are a couple of interesting points of correspondence. The first is with size and the report found, based on worship attendance not membership, that the median size of fast-growing congregations is 150 as opposed to a median worship size of 78 for the PC(USA) as a whole. The study also found that theologically liberal churches have a slight tendency to be faster-growing churches but the largest difference between the fast-growing churches and the rest of the PC(USA) was in the theologically moderate range. On the theologically conservative end the fastest-growing churches were not as well represented. While not specific to my data, the study does support the third interpretation above showing that the fastest-growing churches have more programs both within the church and for the community. (And yes, in this discussion there is an implicit association of “fastest-growing” with church size.)

One thing that should be noted, and may be reflected in the study of fastest-growing churches, is another study that showed that churches dismissed to the Evangelical Presbyterian Church actually had steeper membership declines than the average for the PC(USA). This study is by the Rev. Mike Cole titled “The Statistical Grass Is Not Greener on the Other Side“. In my own data gathering for this piece in looking at the presbytery reports I qualitatively noticed that before dismissal many of the churches that went to ECO had above average membership declines. It will be interesting to see longer-term if dismissal is better or worse for membership numbers.

Well that is enough geekiness for right now – I’ve got a bunch of other writing to get finished in the next couple of days. But there is plenty here to think about and plenty of jumping off points for future investigations as well as revisiting this list as the situation evolves. Stay tuned.