Shifting Sands
Before I get started, I want to make one thing clear: this is not a Green fundraising essay, it is a data analysis essay! My last essay on this subject caused some unintended stress for a community member, and that was not at all what I intended.
This is part two of an analysis of the July FEC individual contributions data. I play with this data set at work because I'm a database researcher, and it has a lot of good stuff in it (time, space, money and politics) along with a lot of data (about 5 million records) which makes it a good research tool (being free helps too!) The particular topic I'm working on is called path analysis, which gets used for things like tracking how patients flow through hospital departments or how visitors flow through web sites. The paths in this data set are donor support for different candidates over time.
The first thing we learn, is that about 98% of presidential donors only back one candidate:
Notice that the total number of contributors is about a million. This is over 50 times larger than the number of "turncoats":
So the moral here is that most people - including heavy hitters - pick a candidate and leave the table when their candidate loses.
The third party candidates have a much smaller donor base than the duopoly's, so let's look at this a different way. Instead of totals, let's break down the source of each candidate's "fresh blood" by percentages:
Now we start seeing some interesting details. There are obvious things, like the Republican field splitting up among the two front runners, but we can also see that Johnson picked up the largest fraction of his fresh blood from that group. He also picked up some Berners. Another take-away is that multiple support trolling doesn't seem to happen at a meaningful level (at least with individual contributions.)
But the most interesting I think is the breakdown of where the switching Berners ended up, so I've broken this down with counts and percentages:
There has been a lot of arguing here and in the press about what Berners will do, and this is the first reasonable looking measurement I've seen: 75% Clinton, 20% Stein and 5% Conservatives of some kind. That last one is rather suspect because the numbers are so small and may even include a few trolls, but who knows. This is not a good sample of the electorate as a whole (its more like activists with some spare cash), but my experience with this data set is that it is probably in the right ball park.
The last chart I have is not from this data set, but from a sleazy little poll I set up over at TOP last weekend. I just asked people to rank the non-Clinton candidates in preference order. The results were...interesting:
There were only 65 votes tallied, so it's not a large data set, but it doesn't smell bad either (Stein's support is about 45% in spite of the moderately bitchy comments) so I think it is in the right ball park. It's also what we have come to suspect: a moderate majority of TOP are corporatists who would prefer a Libertarian - or even Trump - over a real progressive, and a large minority who just suffer in silence. Sort of like the electorate as a whole.
