Odds and Evens

[silverscreenselect]

There's something interesting at play tonight in the various Democratic primaries and caucuses. All of them have proportional selection of delegates statewide and by Congressional district.

Depending on how the districts voted for Democrats in the past, almost all districts are allotted between three and six delegates. Now that there are only two viable candidates left (BTW, Biden, Dodd, Edwards, Kucinich, Richardson and Gravel were all listed on my Democratic ballot, which was also listed in alphabetical order), whichever candidate does better in the districts that have an odd number of delegates will do considerably better in terms of total delegates than his or her percentage of the vote would seemingly dictate.

In a district with three or five delegates winning the district by one vote gives the candidate a 2-1 or 3-2 split. In a district with four delegates, a candidate has to win by at least 62-38 to get a 3-1 split. In a district with six delegates, a candidate has to win by 58-42 to get a 4-2 split.

So a candidate who wins a bunch of even districts by 10-12 points gains nothing by it (they still split delegates 2-2 or 3-3), while a candidate who wins a bunch of odd districts by a fraction builds up delegates. The state winner will get a few extra delegates but not that many unless a particular state is a landslide. Based on the anecdotal evidence I've heard, Obama rates to do well on a district basis in CA, since many of the heavily Hispanic districts have four delegates, and a lot of SF and black districts have five delegates. I'm sure the same situation happens in reverse in some states.

What that means is that a candidate can win a state by 5-10 percentage points and depending upon how the Congressional districts break, either split the total delegates or win by what appears a landslide.