From LA Times Poll: Of those who already voted, Bernie 48, Hillary 43.

is wrong.

Or the early voting demographic might be tough to sample.

This is the second poll in a row where Bernie leads among Latinos.

If she gets the nomination, I bet half the Bernie delegates don't even show up at the convention.

LA Times poll conducted May 16-31,
WSJ conducted May 29-31.

Guess we'll find out later.

I wonder if conducting a poll over a longer span allows us to get a more diverse sample and thus more accurate?

that wouldn't be subject to opinions fluttering around with the latest stories, and the shorter poll at the end of the period would be more reliable because it would include all people who have already voted in equal proportions, while the longer period poll would over represent people who voted early (because they would answer that they already voted whether they were called on May 16 or May 31).

but no poll is really that reliable, as we learn repeatedly. so we shall see.

One of the big things he's had to deal with all year is that, in a given state, by the time he starts holding rallies, advertising, participating in local town halls/debates, firing people up and moving polls in his direction, often a bunch of people have already voted, and those early votes that were cast before all that activity would tend to favor Hillary, especially earlier in the process when he wasn't well known at all.

2 percent for another candidate
2 percent undecided? But they already voted?
5 percent refuse to answer
1 percent will not vote (But they already voted?

for 101 percent