So do you think the LA Times polling methodology would

Posted on: August 22, 2016 at 11:02:09 CT

more accurately reflect how the voters will actually vote than a random sampling?

I think that a large enough random sampling of voters would be more accurate than trying to guess at sub categories and then trying to select the panels based upon the guesstimates of how many should be in each sub-panel.

I do think that skewing the random sampling in favor of registered voters who have voted more frequently in the past would make it a little more accurate yet.

I also believe that doing large enough samples in each particular jurisdiction would be more accurate yet than just a large national sampling.

The way things are set up in the electoral college the overall winner of votes may not be the winner of the election. It is really 52 or whatever individual elections in each jurisdiction added together. Only a couple of those jurisdictions divide the delegates up proportionally. Most are "winner take all". If Trump got 51% of Missouri's votes & Hellary got 49%, Trump would get all 10 of Missouri's delegates, for instance.