Hardly new.
Presidential campaign donations from government workers are heavily weighted towards (D). But there's the problem that we don't know what percentage of workers are represented by the number of $ involved, the proportion of large versus small donations and the distribution of each among federal workers, or whether they did that to curry favor with bosses. In other words, the skew doesn't say much about the actual cohort dubbed "federal workers." It's widely used in (R) circles to show bias among federal workers, however--hence Trump. They find it convincing, either because they're stupid to begin with or the desire to show a particular point made them stupid.
A survey of federal workers was done a few years back in which 40% said (R), 44% said (D), and the rest were undecided with a plurality leaning conservative. All said and done, it looks like a wash. Except that the poll, which had all the trappings of a real poll, gave the number of respondents--it didn't say how many had been asked, it didn't say if there was any skew or attempt to adjust the numbers to account for respondent bias. For example, if they sent out 100,000 surveys randomly and got 985 back, with 900 from federal law enforcement agencies, that's important to know. If they randomly sent out 1000 surveys and got 985 back, with the distribution of returns being the same as those who received them, that's important to know. Since it didn't give any clue--just number returned--that poll also doesn't say much about the actual cohort of "federal workers." This is widely used among the small number of (D) that bother to engage the conservative view that "federal workers are overwhelmingly Democrats." They find it convincing, either because they're stupid to begin with or the desire to show a particular point made them stupid.
A third approach looks at government employee unions. They donate to and support pretty much only Democratic nominees. Then again, they're unions. I know from observation and first-hand experience that unions don't always represent their local constituencies or even their national constituencies. And that an overwhelming majority of their constituencies don't really care, as long as the unions either get them stuff or stay out of the way.
A fourth view is that most people tend to vote like their neighbors. Now, this is true as a description at certain levels of generality, but lousy as a prescription. I mean, if I live in a neighborhood that votes 80% (R), I can't go house to house and say, "Gee, your two neighbors vote (R), so you must vote (R)." There's still that 20% (D), where most of those voters will have neighbors on either side that vote (R). You iterate over the entire neighborhood enough with that prescription and assuming you start with a fairly even distribution you'll find that the prediction is the neighborhood can't possibly vote 20% (D). It has to vote 100% (R), except maybe for a few people on the edges. (There was a game like this, which struck me as a dumbed down version of Go.) However, given what I assume is the distribution of federal workers, that would make them lean (D)--coastal states and cities. I suspect that if the granular your take on what "neighbors" means, though, the less clear predictions that assumption would make. I don't know what this particular assumption gets you, but everybody seems to think it necessarily provides overwhelming and unambiguous support for their point of view.
The survey would have been the way to go. But it was either run shoddily or reported on irresponsibility, and I don't see how those spending money on it could possibly think the published results provided anything useful.
= new reply since forum marked as read
