... for estimating single states because it depends on the margins of victory.

Here is why. Almost all primary states award about 65 percent of their delegates based on congressional districts(some use state districts but the math is still the same). Most districts have between 2 to 8 delegates. This means that the popular vote in the state will be tempered by the district allocation.

In Arkansas and Illinois the votes in the districts made it so that the rounding error in voting districts favors the winner. For example:

In AR district 1 Hillary got 78% of the vote and .78 x 6 works out to 4.7 giving: Obama 1.3 Clinton 4.7 which rounds to 1 and 5

So Obama is shorted .3 delegates due to rounding. Add it over the 4 districts and Obama lost 0.899 delegates to rounding. Add to that that he also lost out in the rounding error for at-large and PLEO delegates and you learn why it is advantageous to win 70% of the vote but not 83%.

So in general, the districts will bias toward the loser in a close election and toward the winner in a blowout, unless the blowout is too big.

In Illinois insult was added to injury for Clinton because she didn't clear 15% of the vote in two 8 delegate districts so she got a rounding error of 2.1 delegates from those districts alone. She actually only lost 0.9 delegates from rounding error in the other 17 districts. So the rounding affect by itself is only about 1% of the error. The other 1.5% of the error is due to the 155 criterion. In the end it doesn't really matter because the error cancel over all the states. We will be left with a 0.3% error in favor of one or the other but they will have won by 4% the error is insignificant.

correction: 15 percent criterion

not 155 criterion

In simple terms, it doesnt matter. If you play by the rules and just do the primary then everyone gets their say.

And no, FL and MI dont deserve to have their votes counted because after all they did ignore the rules. Im sure if the cons had a similar issue they wouldnt be playing fast and loose with the rules-after all, this is a democracy right? We dont just change the rules and say "best two out of three elections" like kids tossing a coin right?

I agree that using the popular vote to predict the distribution of delegates is a crude system, given all the contingencies that cridge details here. But our after-the-fact examination of the discrepancy between the predictions and the actual breakdown suggests that this is actually surprisingly accurate, even on an individual state level. For Clinton, the popular vote predicted the delegate total correctly in 7 of 29 primaries, and was one off in either direction in another 8; in Obama's case, it was exactly right 7 times and one off in 12.

Overall, using the popular vote to predict the delegate vote was off by more than 5 percent in only four primaries for Clinton and three for Obama. For both candidates, the median error is slightly lower than the average.

I'd be happy to share the raw numbers -- Email me at chris.wilson@slate.com.

I can think of no more accurate estimate of a future or pending delegate allocation then the popular vote. I think the slate's delegate calculator is probably the fastest and easiest
way get a picture of the Democratic primary.

I was just pointing out that for an individual state the estimate will err in the losers favor for close races because of the district allocation