Since I posted the analysis last week eight more presbyteries have voted with the unofficial tally at 77-94.  There are only two more presbyteries left to vote.  This past week there were two more presbyteries that switched from “no” to “yes” bringing the total of presbyteries to switch in that direction to 33.

In the first post I looked at the total votes without regard to presbytery groupings.  In the second part I discussed the distribution of voting patterns for the presbyteries.  In this post I want to focus on the groupings of presbyteries and a couple of interesting features that appear.

From the usual sources 154 of the 171 presbyteries that have voted have numbers reported for both the 01-A voting and the 08-B voting. 

In the following frequency distribution plots the vertical and horizontal axes are the same in all the plots (except the All Presbyteries/Total vote plot has an extended vertical axis) and the horizontal axes are aligned with a reference line through the 1.0 (no change) point.  Data are binned and counted on intervals of 0.05 with the number on the horizontal axis the upper inclusive limit of the bin.

To look at the details the presbyteries have been grouped by those that voted “Yes” on 08-B and those that voted “No” on 08-B.  There are also subgroups of each of these for the presbyteries that switched their votes from the previous round of voting.  Since the “Yes” to “No” subgroup has only two presbyteries those are briefly discussed but not plotted.

Total Presbytery Votes

First the note that the top chart has a vertical axis from 0-30 while the upper limit on all of the other vertical axes is 20.

Looking at these distributions it can be seen that the changes in the total number of votes cast was very similar whether you are looking at the total population or the split-out groups.  Total votes are slightly higher in “Yes” presbyteries but it is not much.  All have averages and medians in the 0.86 – 0.90 range and while the standard deviations show a bit more variation ranging from 0.15 to 0.24, the difference is not extreme.

Presbytery “Yes” Votes

Here is where the division into groups and subgroups shows the most interesting results.  Just splitting the population into “yes” and “no” presbyteries shows no significant changes in the population.  The total, “yes” group and “no” group all have averages a bit above 1.00, medians very close to 1.00, and standard deviations in the 0.35 – 0.47 range.  It is tough to make a case that much is different between the “yes” and “no” presbyteries.

But if we split out the “yes” presbyteries that previously voted “no” it is clear that these presbyteries had a clear increase in the number of “yes” votes.  Of the 29 presbyteries, 7 had no change or a decrease and the other 22 had in increase in the “yes” vote.  I’ll return to this group at the end and take a detailed look at the behavior.

Presbytery “No” Votes

While the patterns in the “Yes” vote were not seen and the differences in the Total was slight, there is a bit more difference to be seen in the break out of the “No” vote.  All the presbyteries together had an average no-vote ratio of 0.76 while the average in presbyteries that voted “No” on 08-B was 0.83 and the average for “Yes” presbyteries was 0.68.  The numbers for the presbyteries that switched were statistically close to those for all the “Yes” presbyteries.

So presbyteries that voted “No” on 08-B generally had a lower decline in “No” votes, presbyteries that voted “Yes” had a higher decline, and no appreciable difference from that was seen for the subgroup that switched from “No” to “Yes.”  In other words, as a group a “No” to “Yes” switch was characterized by a statistically greater increase in “Yes” votes with a “No” vote decrease characteristic of the other “Yes” vote presbyteries.  This in contrast to a possible switch due to no increase in “Yes” votes but a statistically greater decrease in “No” votes.

Details of the “No” to “Yes” Switch
Taking a look at the 29 presbyteries that switched votes, two (6.9%) appear to be pure swing with almost equal numbers of lost “no” votes and gained “yes” votes.  (In this discussion “almost equal numbers” means a difference of usually zero or one, but no more than two votes.)  Six (20.7%) show little to no change in the number of “yes” votes and only a decrease in “no” votes, and five (17.2%) show a notable decrease in both “yes” and “no” votes with a larger “no” vote decrease.  These 11 (37.9%) appear to be more related to differential losses.  One (3.4%) shows a significant increase in both “yes” and “no” with a more pronounced increase in “yes,” but looking at previous votes 01-A has a significantly lower vote total and this is probably a special circumstance for 01-A.  Half the presbyteries, 15, show a more complex behavior with a gain in “yes” votes and decrease in “no” votes.  Five of those have a “yes” gain greater than the “no” loss and ten of those had a larger “no” loss than “yes” gain.  These, plus the two pure swing, suggest that 17 (58.6% of the switches and 11.0% of the total) presbyteries changed their vote from “no” to “yes” at least in part by a significant switch of voters between those positions.

Details of the “Yes” to “No” Switch
With only two presbyteries making the switch in this direction it is impossible to make generalizations, especially since their patterns of change are totally different.  In the case of San Francisco Presbytery the vote went from 216-186 on 01-A to 167-177 on 08-B.  There was a significant preferential decrease in the number of “Yes” votes attributed variously to complacency or attendance at conferences.  The case with Sierra Blanca is exactly opposite with the number of both “Yes” and “No” votes increasing, but the “No” vote increasing dramatically and preferentially.  On 01-A Sierra Blanca voted 18-17 while on 08-B they voted 23-30.  Again, special cases, but when you look at the details of many of the presbytery votes you begin to think that there is a back story to the voting.

Changes Relative to Strength of Voting
I will do a lot more with multi-variant statistics later, but this one jumped out at me and I thought it appropriate to include here.  I have previously commented that looking for correlations between various factors has yielded little, but here is a case where something of interest does appear.

I hope that this graph is not too confusing.  On the x-axis I have the “yes” vote on 08-B in percent.  All of the blue squares represent presbyteries that voted yes and so are above the 50% line, and all the red squares are presbyteries that voted no and so are on or below the 50% line.  On the top plot I show the change in the number voting in opposition from one vote to the n
ext as a ratio of 08-B votes to 01-A votes.  So on the left is the change in the number of “Yes” votes in presbyteries voting “No.”  And on the right are the change in the number of “No” votes in presbyteries voting “Yes.”  For the subgroup of presbyteries that switched from “No” to “Yes” the plot did not differ significantly so I did not include that data as a separate plot.

In the upper plot the trend for “No” votes to decline in presbyteries voting in the affirmative is strong with an R-squared=0.32 for the correlation.  The trend for the other half is not as strong and while visually suggestive the higher scatter results in an R-squared=0.02.  But based on the grouping of points in the down-to-the-left trend an argument could be made for some presbyteries with similar behavior, but a closer look at the outlying points for special cases would be necessary to really verify that.  It should be recognized that changes in small numbers of votes as is found near the ends of the X axis are amplified more than similar changes near the middle of the axis.

In the lower plot the change in concurring votes is plotted and for both the trend is statistically indistinguishable from flat.  In the “No” votes in “No” presbyteries there is a slight, but statistically insignificant, upward trend to the lower percentage votes that if true, and combined with the decreasing “Yes” vote in the upper plot, would actually suggest a swing from “Yes” to “No” in the presbyteries with the strongest “No” votes.  It is clear, both visually and statistically, that no such conclusion is even hinted at in the “Yes” presbyteries.  So there is a trend seen in “Yes” voting presbyteries, and suggested in “No” voting presbyteries, for the greater the strength of vote is the fewer opponents showed up, or were still around, for the vote on 08-B.

Well, enough of this for now.
One of the things I keep getting asked about all of this is something like “Wasn’t the vote on 01-A ‘different.'” There have been several ways that people have suggested the last vote was different but the most often mentioned one is that presbyteries voted “No” because the PUP Task Force was beginning work and they wanted to let that process play out.

Well, in multiple respects the voting on 01-A was different and in my next installment in this series I will look at that quantitatively and show, well, that every presbytery is different.  Actually, I’ll show that there are several different sets of behaviors seen for 01-A voting of which a shift to vote “No” is just one of them.  Sometimes that “No” shift came with no change in total vote, a true swing.  And sometimes that shift in percentages came with a significant increase in the total number of commissioners voting, a behavior that looks like a “get out the vote” campaign for those favoring the retention of the “fidelity and chastity” requirement.  My point right now is that an “undoing” of either of these would support some of the behavior seen in the data for 08-B voting.  So next time I’ll lay out those numbers.