By David Gopoian

As George W Bush’s approval ratings sink deeply in late 2005, Democrats can conveniently dismiss what we never fully accepted in 2004 – that Bush was as popular an incumbent among Republican followers in the last campaign as Ronald Reagan was at his zenith in 1984.

One consequence of this oversight is the now fashionable assumption among Democratic elites, bloggers and web-deprived alike, that any Democrat other than John Kerry could have defeated Bush handily in 2004. I am taking this occasion to dissent from that perspective.

There are any number of simple empirical arguments that may be made in defense of the presidential candidacy of John Kerry. Among them are these:

• He took a Party totally demoralized by 9-11 and Bush’s foreign policy dominance and came within one state of dethroning the incumbent war-time president.

• In that state of Ohio – with a Democratic Party organization in name only - he ran the best Democratic campaign in four election cycles, since Clinton carried the state in 1996. No Democratic candidate since 1996 got as much as the 48.7% Kerry gathered. The last statewide Ohio Democrat who carried the state was John Glenn in 1992.

• Unlike Gore in 2000, Kerry left no doubt by any objective, and most subjective, criteria about who won the 3 debates with Bush.

• According to candidate trait data from the 2004 American national Election Study data, Kerry matched Clinton’s 1992 performance on each attribute measured among Democratic identifiers (cares about people like me, provides strong leadership, knowledgeable). In short, Kerry appealed effectively to the Democratic base.

• Contrary to the conventional wisdom, my multivariate analysis of the 2004 election indicates that a sizeable chunk of Kerry voters voted for him precisely because they admired Kerry’s personal traits, not despite them. Controlling for all other voter predispositions, Kerry’s persona, no matter what the pundits suggest, was a plus – not a minus.

Having made those simple points, I wish to address one more complex argument in this space – the notion that any number of other Democrats would have defeated Bush handily in 2004. In the 2004 ANES study, the only direct candidate preference measure available is the item asking how respondents voted for president.

However, political scientists have long recognized “thermometer scores” as accurate proxies of vote intent. Fortunately, the study includes thermometer scores collected from respondents for numerous political figures. These scores ranged from 0 degrees to 100 degrees, where 100 represented extremely warm feelings toward the political figure and 0 represented extremely cold feelings. A mid-point score of 50 represented neutral feelings.

In general, stated candidate preferences correlate very strongly with differences observed between candidates’ thermometer scores. Voters typically vote for the candidate with
the higher relative thermometer score. For example, if Candidate A gets a thermometer score of 70 from a voter and candidate B gets a score of 50, there is an extremely strong likelihood that Candidate A will receive that voter’s vote.

This point may be illustrated using data for Kerry and Bush in 2004. Among voters who gave a higher thermometer score to Kerry than to Bush, Kerry received 96% of their votes; likewise, Bush received the votes of 94% of the respondents who gave him a higher thermometer score than Kerry.

Use of thermometers is required to make my points because the survey did not directly ask respondents how they would have voted if some candidate other than Kerry were the Democratic nominee. By applying thermometer score differences as a proxy for candidate choice, comparisons between Kerry and other Democrats may be made.

In 2004, for the weighted sample of 787 respondents who did vote for president, 49% gave Bush higher thermometer scores, 46% gave Kerry higher thermometer scores, and 5% gave identical thermometer scores to each candidate. [Note: that 5% split their actual votes 57%-43% in favor of Bush.] These findings illustrate that thermometer score comparisons do provide an accurate prediction of vote intentions. Overall, thermometer scores provided correct vote predictions for 93% of all respondents in 2004, including the 5% with no Kerry-Bush thermometer score differences.

Thermometer Score Comparison # 1

Bush Rated Higher Than Kerry: 49%
Both Rated the Same: 5%
Kerry Rated Higher Than Bush: 46%

To make the case that some other Democrat could have easily defeated George Bush, the data would have to show an advantage in thermometer score comparisons versus Bush for that Democrat. In other words, the 49%-46% advantage that Bush enjoyed in thermometer score comparisons to Kerry would have to shift at least a few points to demonstrate a Democratic advantage that eluded Kerry.

No eligible Democrat emerges from the 2004 data base with that kind of magic. The two we may compare are Hillary Clinton and John Edwards. Hillary stands as the odds-on favorite for the 2008 nomination and Edwards remains a viable contender. Let’s examine their numbers circa 2004.

A comparison of Bush’s thermometer scores with Hillary’s shows very little movement and virtually no change in Bush’s advantage. There is a shift of roughly one-half percentage point away from Bush in the aggregate. Overall, Bush gets higher scores from 48%, Hillary Clinton from 46%, and 5% rate both identically. The overall rounded percentage point difference nets a one percentage point gain for Hillary relative to Kerry.

Thermometer Score Comparison # 2:

Bush Rated Higher Than Hillary: 48%
Both Rated the Same: 5%
Hillary Rated Higher Than Bush: 46%

Details show that 43% preferred both Democrats over Bush and 44% preferred Bush over both Democrats. An additional 1% remained tied in both scenarios.

In short, about 12% of voters’ preferences would have shifted if Hillary, rather than Kerry, were the Democratic nominee. But as Comparison #2 demonstrates, the net effects of such movements would have been minimal. Here are the reasons why.

Among those neutral in a Kerry-Bush match-up (5% of all voters), 80% would have taken sides in a Bush-Hillary contest. But they would have divided evenly – 2% from neutrality to a preference for Hillary and 2% switching from neutrality to a preference for Bush.

Among those with a Kerry preference in a Kerry-Bush comparison, 7% would have shifted to a Bush preference in a Hillary-Bush match-up. And 10% of those with a Bush preference in a Kerry-Bush match-up would have shifted to a pro-Hillary position in a Hillary-Bush contest. The ultimate effects of all these shifts in sentiment would have generated a net gain of one percentage point for Hillary Clinton versus Bush compared to Kerry’s status versus Bush.

Edwards would have fared worse than Kerry. Only 42% rated Edwards higher than Bush, 53% rated Bush higher than Edwards, and 5% rated each candidate identically. Again, 88% would not have been affected at all, but the shifts that would have occurred demonstrate that Kerry held a stronger position than Edwards against Bush.

Thermometer Score Comparison # 3:

Bush Rated Higher Than Edwards: 53%
Both Rated the Same: 5%
Edwards Rated Higher Than Bush: 42%

There was only one Democrat who would have given his party the upper hand against Bush and that was the one live Democrat constitutionally prohibited from entering that fray. Bill Clinton could have flipped the table. He received higher scores than Bush from 49% while Bush got higher ratings from 46%.

Thermometer Score Comparison # 4:

Bush Rated Higher Than Bill Clinton: 46%
Both Rated the Same: 5%
Bill Clinton Rated Higher Than Bush: 49%

Again, most voters (87%) would not have been moved across party columns whether Kerry or Bill Clinton were the Democratic nominee. For the remaining 13%, mobility would have been active in both directions in a Bill Clinton-Bush contest compared to a Kerry-Bush contest. But President Clinton would have had most of it flowing in his direction, resulting in a net gain of about 6 percentage points relative to a Kerry-Bush match-up.

Rather than demonstrating that Kerry’s candidacy was a dud, these findings illustrate the limited potential personal impact of any Democratic nominee in 2004. Even the best Democratic campaigner of our time, Bill Clinton, would have moved the electorate only minutely from the division of the vote observed in 2004.

The unmentioned element in each of these assessments was the unwavering loyalty to George W Bush from the Republican base circa 2004. Bush’s average thermometer score among Republican identifiers was the highest recorded from a partisan following– besting Ronald Reagan’s status among Republican voters at the height of his popularity in 1984.