[xienze]

You understand how statistics work, don't you? When you have 64% of the population voting, that's a pretty big sample size, enough so that you can reasonably extrapolate that the 49.8% share _probably_ holds across the rest of the population, give or take.

Put another way, if someone asked you to estimate what the split between the one third of the population that didn't vote was, what would you use as a reference point? Social media posts? Vibes? Or maybe polls leading up to the vote that showed the same roughly 50-50 split found in the actual results?

[larkost]

I don't think that you understand how that part of statistics works. You are said that we have sampled 64% of the population, so we can extrapolate to the rest. That works if the sampling is sufficiently random. But in this case the "sampling" is people who voted, so an entirely (self) selected population, and pretty much not random at all (i.e.: people who were mostly less decisive about their opinions/vote).

So I don't think we can extrapolate confidently at all. So we really don't know whether it holds for the rest of the population at all.

[xienze]

> So I don't think we can extrapolate confidently at all

But the parent was — the implication being that if the other third had voted, surely they would have voted in greater proportion for Democrats than Republicans. That’s based on nothing but vibes and assumptions. I argue that the 64% share of people that actually did vote give you a lot more confidence in how the remaining third probably would have voted than whatever the parent suggested. It’s at least a starting point for extrapolation.