« DLC Urges Dems to Fight for Center | Main | Seniors, the Prescription Drug Benefit and the 2006 Election »

Partisan Benchmarks and Political Performance in the 2004 Presidential Election

By David Gopoian

Evaluations of campaign performance should be tied to meaningful benchmarks. A classic measure of political expectations is one based on the partisan composition of specific populations. The attainment of 51% of the vote may be evaluated differently, for example, if earned in a jurisdiction that typically supports one's own party in contrast to a jurisdiction that typically does not.

In this post, I compare Kerry's share of the two-party vote in the 50 states, plus DC, to a measure of the expected Democratic vote in each. The expected vote measure is derived from an analysis of the voting behavior of national populations across five previous presidential elections. Estimates of the likelihood of a Democratic vote for specific categories of partisan identifiers are then applied to the state-level exit samples of voters in each state.

The first column of numbers shows the Expected Democratic Vote for each state. This percentage represents a partisan benchmark for Democratic candidates - the percentage he should get given the partisan makeup of a state's electorate.

An interpretation might flow as follows: given the sample of voters who actually voted in Massachusetts in 2004, the typical Democratic candidate should have expected to have gotten 62.9% of the vote. Kerry's actual vote of 62.5% in the exit poll indicates that Kerry came very close to matching how a Democratic candidate should have performed in his home state.

Examining the 19 states that Kerry won, plus DC, it is apparent that his actual vote percentages were within one rounded percentage point of the expected Democratic vote in all but 6 of them. The exceptions were DC, Vermont, and Illinois where he ran above expectations and Hawaii, Rhode Island, and New Jersey where he ran below expectations. In the remaining 14 states, Kerry ran about as well as should have been expected of the Democratic nominee..

The problem, of course, is that these 20 states left Kerry 19 electoral votes shy of the White House. There were seven other states a Democrat should have won based on partisan inclinations of voters. These included West Virginia, Arkansas, Kentucky, New Mexico, Missouri, Iowa, and Louisiana. Granted, in many Southern and Border States, partisan attachments do not necessarily carry much clout for the evaluation of political figures associated with the national Democratic Party.

However, a case could be made that the partisan benchmarks for these states are not unrealistic. In the 2004 election, West Virginia elected a Democratic Governor with 65% of the vote; Arkansas re-elected Senator Lincoln; Kentucky gave its Democratic Senate challenger 49% of the vote. Missouri had just elected a dead Democrat over a live Republican to the Senate four years earlier. And Gore won both Iowa and New Mexico in 2000.

In fact, there are 12 states where the expected Democratic vote ranged from 49% to 51% -- the very definition of swing states. Kerry won only two of them - - New Hampshire and Wisconsin. He lost four others where he finished within one rounded percentage point of meeting expectations for a Democratic candidate (Iowa, Florida, Ohio, and Nevada).

Three of the remaining six swing states were southern states where Kerry ran at least 6 points behind expectations for a Democratic candidate (North Carolina (-6), Louisiana (-8), and Kentucky (-11). He also ran below expectations in Montana by 8 points. The last two, potentially winnable swing states were lost without a battle - Virginia, where Kerry ran 3 points worse than expected and Missouri, where he ran 4 points below expectations.

I'll leave it to readers to draw inferences about the past campaign and about future ones as well. In the graph below, blue is used to highlight pro-Democratic inclinations or performance, pink to illustrate pro-Republican tendencies or performance, and yellow to show differences that did not deviate markedly from partisan expectations. The Kerry Vote column is based on weighted exit poll data, not official election returns.