I'm taking some time off work next week, and one of the things I'll do is review the 100 data points I'm collecting. That is, 100 Sunday services for the four weeks of October 2015 from 25 congregations, five each from the five new UUA regions, distributed by size. This is my method.
I downloaded this year's certification numbers. Now, this rules out the 41 congregations that did not certify, but since these tend to be overseas congregations (and not in any region anyway) and I've noticed before that non-registration is sometimes an early tell for congregations disbanding… well I think the certification list is a suitable list to take a sample from. The Church of the Larger Fellowship, also in no region, is an outlier and reasonably excluded.
So, I filtered by region, sorted by size, numbered the congregations and calculated how many congregation would be in the smallest 40%, the next larger 40% and the largest 20%. Then I generated a list of random numbers (YouTube video) from those three ranges. Looked up the congregations by the assigned number. Checked its website or Facebook page to see if I could find the Sunday services in October. No? Then I moved to the next on the list. Once I got two from the first group, two from the second group and one from the third group, I would move to the next region.
I collected as much information as was practical, including whether the congregation had its own minister and if the minister led the service, or a guest or affiliated minister, or a lay speaker or speaker. Plus the title or theme of the service, of course, and if there was a theme for the month.
So that's the collection method. I'm almost done. Next, putting some numbers on the framework.