As General Assembly Season tapers off and the voting on the amendments to the constitution of the Presbyterian Church (U.S.A.) finishes up, I have finally found time to start crunching the final numbers on amendment voting.  While I have several different analyses planned, let me begin with the summary statistics related to the passage of Amendment 10-A and we will see how many of the other ones I would like to do actually get finished.

As of today it appears that the voting on 10-A has finished with a vote by the Presbytery of Kiskiminentas against the amendment 33 to 58.  However, the passage of the amendment was secured back on May 10 when the 87th presbytery voted in favor of it.  For the purposes of this analysis I am going to use the data at about the time of the passage, including a few presbyteries that voted shortly after the unofficial passage.  My intent is to eliminate any change in voting patterns that may have occurred after the concurrence was a foregone conclusion.

As usual, my data comes from an aggregation of the usual sources: Covenant Network, Reclaim Biblical Teaching, Yes on Amendment A, and The Layman Online.  I have posted my spreadsheet with the data and the analysis.

While the unofficial vote count currently stands at 97 yes to 74 recorded no and two not voting, my “freeze-frame” was from the point the count was 92 yes and 68 no. (Not much of a difference I will admit.) The objective of this analysis is to look at the change in voting from 08-B to 10-A. Accordingly, there were 86 presbyteries voting yes this time and 62 presbyteries voting no on 10-A that have recorded votes on both amendments. I will focus my analysis on those 148 presbyteries.

In this population there were 19,607 total votes cast on 08-B and 18,705 on 10-A, a decrease of 4.6%.  For comparison, the membership of the PC(USA) decreased by 3.0% between 2008 and 2009, so the voting decline over two years is a bit less than the projected total decrease if you double the 08-09 decline.  The number of yes votes increased from 9711 to 10,301, an increase of 6.1%.  The no votes decreased from 9896 to 8404, a 15.1% drop.

If the number of yes and no votes decreased in proportion with the total decrease we would have expected 9264 yes votes and 9441 no votes. The discrepancy is 1037 votes more for yes and less that for no.  So, in the simplest analysis, the change in voting can be viewed at a uniform decline of 902 votes (447 yes and 455 no) and a shift of 1037 votes from no to yes. If you wish to attribute all decline to the no votes, it could also be modeled as a decline of 902 no votes and a shift of 590 votes from no to yes. The best answer probably lies somewhere between these two end-members.

Let us now break this down on a presbytery level.  This gets a bit trickier because presbyteries that have small numbers of yes or no votes can have extreme values for ratios when these quantities change by even a few votes.  I have not gone into the data on this analysis to eliminate extreme values but will take this into account by comparison of the mean and the median statistical measures. As it turns out, while the extreme values do stretch out the maximum and minimum values as well as the standard deviations, the differences between the means and medians are generally not exceptional.

On the spreadsheet you will find that I calculated summary statistics for the whole population and for a series of subsets.  There are also some frequency distributions listed.  I will only touch on the measures of the center of the data, the mean and median, for some of those categories in the discussion that follows.  The data are there if you want to see all the numbers.

For the yes votes in presbyteries, overall there was an increase with the ratio of 10-A to 08-B votes having a mean of 1.14 and a median of 1.08. Interestingly, for presbyteries that voted yes on both amendments, the number of votes was on average unchanged with the mean being a ratio of 1.00 and the median 0.99. In an interesting match, presbyteries that voted no both times and presbyteries that flipped from no to yes had means of their yes vote ratios around 1.27 and identical medians of 1.20.  The presbyteries that flipped from yes to no, not surprisingly, is the category that showed a decrease  in the ration with a mean of 0.93 and a median of 0.91.

Overall, presbyteries had a decrease in the number of no votes with the overall mean a ratio of 0.89 and the median a ration of 0.86.  Presbyteries voting yes saw a bit larger of a decrease, presbyteries voting no a slightly smaller one.  Again, the category that was the exception was the presbyteries that flipped from yes to no where on average more votes were seen with an increase shown in a mean of the ratio of 1.21 and a median of 1.26.

Probably the most telling is the total number of votes for each presbytery.  For the total and all of the sub-categories, the ratios are pretty constant around 0.96 in both the mean and the median.  The noticeable exceptions to the down side are the presbyteries that voted yes both times with a median of 0.92.  The only categories having increased ratios are the ones for the switched votes — Presbyteries that flipped from no to yes had a mean ratio of 1.03 and a median of 1.07; presbyteries that flipped from yes to no had a mean and median ratio of 1.06.

It would seem that the message is that change came through better commissioner turn-out, whether it be an organized “get out the vote campaign,” or just informal increased interest on a particular side. And I find it striking that this was true for presbyteries that flipped in either direction.

This of course is only on average and when you consider the details for each presbytery individually you find variability.  For example, for the four presbyteries that flipped from yes to no, two had an increase in the number of yes votes, three an increase in the number of no votes, and one had a decrease in both. So three of the four were changed by improved no vote turn-out, but the one with both declining may be attributed to who showed up for a close vote. (That was West Jersey Presbytery which tied.)

Similarly, for the 20 presbyteries that switched from no to yes, 14 showed an increase in the total number of votes, four a decrease, and two a decrease that was small enough I would consider it noise. All but one had an increase in the number of yes votes, some a very marked increase.  And only one showed a clear increase in the number of no votes, the rest remaining stable or decreasing.

I am painting with a bit of a broad bush here by looking only at the group averages while each presbytery has its own story.  This stands out when you look at the voting in these two groups as combined total.  In the four presbyteries that switched from yes to no total votes increased from 431 to 436, an increase of 1.2% that falls into the range I would consider “noise” or “random fluctuation.” Again, for the 20 presbyteries that switched from no to yes the total of vote counts increased from 2698 to 2726, an increase of 1.4%, again not substantial.  It is on the presbytery level where these changes are larger and become more influential statistically.

So there is a bit about the summary statistics for the Amendment 10-A voting. I hope to write about other details including trends over time, cross-issue correlations, more specifics on the presbyteries, and maybe look at some other variables so see if there are correlations.