Hacker News Reader

Comment by tspike | Hacker News Reader

I agree they could pick a better question. It's not interesting to know how many people agree with that statement.

However, it is very interesting that Democrats actually thought that Republicans would not agree with that statement

Perhaps, but there's two ways that this can be interpreted. One is that Democrats have a misperception about Republicans' beliefs on this issue. I'm sure to some extent this is true. The other, though, is that Democrats and Republicans interpret this statement in different ways. So Democrats believe the statement means X, and Republicans believe it means Y. Republicans tend to agree with X and disagree with Y. Democrats may have an accurate view of how Republicans feel about beliefs X and Y, but still underestimate how many Republicans agree with the statement.

Interviewer: "Will Republicans say the sky is blue?"

Democrat: "No."

Republican: "Of course it's blue. It's hot pink, which is a shade of blue."

Democrat: "... wait, what? You think... huh?"

Interviewer: "Looks like the Democrat was wrong!"

Which is true... ish. Truthy.

Not sure where the negative reaction is coming from for this comment. It's a perfect analogy. You can swap Republican and Democrat in every place if you'd like; the problem is the same. It's presented as if we have a predicate like "PARTISAN believes X" and another predicate like "OTHER_PARTISAN estimates that for all PARTISAN P, (P believes X)".

But it's not as clean as this. Because of the messiness of natural language -- which goes double for political language -- there's no guaranteed equality between the two Xs.

Yeah, the main question under discussion here when I made the post seemed to be the racism one, which appears to have played out the way I depicted. The general problem is that it's really, really hard to design complex social science surveys that even maybe measure what you're tying to measure. This one's complex, and I'm pretty sure doesn't measure what they were trying to measure. A big chunk of the two major parties don't even agree on the terms these questions are about, meaning they may well be aware of why their counterparts would answer "yes" where they said they'd say "no", but consider their "yes" to be fundamentally about something else entirely.

I wish they'd included "do democrats support murder" and "do republicans support murder". That'd have been fun. Some would certainly have said yes on either side (abortion, the death penalty) and they'd be right from their point of view, even though "do you support murder?" would have been near-zero on both sides.

Yep. Though I don't think the two ways are incompatible- if one side actually has an accurate understanding of the other side's point of view, they would also understand the different ways in which the question might be interpreted.

Why is it less interesting that about 20% of Republicans do agree with that statement (according to the article), than that Democrats think 50% do?

Reply on news.ycombinator.com