4 ms·
The point is that at the semantic level these statements are random noise. Sure you can see a signal in it if you look hard enough, but what you are actually do
by paulajohnson 10y ago
The point is that at the semantic level these statements are random noise. Sure you can see a signal in it if you look hard enough, but what you are actually doing is over-fitting. Some of these statements look like they might be meaningful, but that's just like generating 5 random points and finding that they are roughly in a line.
The finding here is that people on the conservative end are more prone to doing this semantic over-fitting than people on the liberal end. Sure, the sample is self-selected and not representative, but the results are still intriguing. The next step is to replicate this with a better sample, or at least a different one.
- metaphorm 10y agoI think the research is of such dubious quality in both experimental design, and unfounded underlying assumptions, that I wouldn't trust its results further than I could spit. The next step is not to replicate this with a better sample, but to seriously question the underlying assumptions that lead the research team to even attempt a study of this kind in the first place. I can think of better ways to "validate" your own in-group preferred belief system, language, politics, and whatever other kinds of tribal identifiers they want. The most dubious part of this exercise is not that its possible for people to do line-fitting on the bullshit sentences. Its that the research team has such a poor understanding of language and communication that they would think their experiment was anything more than an exercise in tribal piss-marking.