Saturday, March 14, 2009

Total 08-B vote count is heavily skewed by late and early presbytery voting

I've been taking a look at the voting patterns on 08-B so far, and a few days ago noticed that it appeared that there were a large number of "safe for equality" presbyteries that hadn't voted on 08-B yet. A quick analysis confirmed that this was true, but I didn't fully run the numbers.

So today I finally got around to adding a separate analysis section to the bottom of the tracking spreadsheet linked from this blog. It's down there in the section that says "Late voting competitive disadvantage / early voting competitive advantage analysis".

The results are striking. They show a consistent pattern of likely anti-LGBT presbyteries voting early, and pro-LGBT presbyteries voting late. These vote scheduling disparities heavily skew the current presbytery vote count (which, as I've suggested earlier, isn't the critical number to watch -- what I care about are the voting percentage TRENDS) towards "no".

Here's a summary, and the full set of numbers are in the spreadsheet.

-- So far, 2/3 (67%) of all presbyteries have voted.
-- BUT, out of 36 "likely yes" presbyteries, only 53% have voted.
-- This causes a 5-vote "yes" disadvantage compared to what we'd see if these presbyteries were scheduling their votes at an average rate, i.e. if 67% had already voted.

But wait, there's more...

-- Also, out of 77 "likely no" presbyteries, 82% (!) have already voted.
-- This causes an 11-vote "no" advantage compared to what we'd see if
these presbyteries were scheduling their votes at an average rate of 67%.

In other words:
-- Anti-LGBT presbyteries are voting early.
-- Pro-LGBT presbyteries are voting late.
-- The result (as of right now) is a 16-vote skew towards "no".
-- This is why the current presbytery vote totals look so lopsided.

Of course, the goal of this marathon (not a sprint, it's a marathon) is to shift hearts and minds over time, and the best indicator that we have of that are the percentage vote trends over time.