The electability question revisited, with new general election polls

The electability debate has been raging for months now and has served as an important subtext of the Democratic primaries. In fact, electabilty arguments were invoked far less in the GOP primaries, surprisingly so given that the polls were much more consistent in who would be the strongest Republican in a general election. And now that McCain is the presumptive nominee, we will probably no longer see general election polls from very red states in which Democrats manage a tie or even lead (a poll even had Clinton competitive in Texas some time in the fall!).

In the Democratic race, however, it has been much more difficult to find a pattern of electability. I have addressed this question many times in the past few months, the first time at the beginning of September when an Ohio poll had Clinton running significantly better than Obama and a Colorado poll had her much weaker than her rival.

Over the next few months, I attempted to find some pattern, some consistency and the only conclusion I have been able to reach is that (1) Clinton polls better than Obama in red Southern states (which Clinton often looked competitive in) and in the Northeast (Obama often struggles against McCain in polls from New York and Massachusetts) and that (2) Obama is stronger in the Southwest and Northwest. That hypothesis is usually confirmed by surveys and, as we will see, this week's SUSA surveys give more evidence for that. I first stated this theory in November, after a poll of Alabama showed Clinton surprisingly strong and a poll of Oregon showed Obama outpacing her.

In the rest of the country (especially in the Midwest) it has been seemingly impossible to find a pattern of electability. That same day, I also wrote , "Hillary looked more electable in this week's MN poll, while Barack looked more solid in the Iowa poll from the day before."

Today's new polls show that it remains very difficult to judge the electability case of the two Democratic front-runners. First, there is Time's national poll that will likely be often cited in the coming days:

• Time has Obama leading McCain by 7%, 48% to 41%, while Hillary is tied at 46%.
• Commenting on these numbers, Time's pollster says, "Independents tilt toward McCain when he is matched up against Clinton But they tilt toward Obama when he is matched up against the Illinois Senator."

We also got this week a series of three state polls from SUSA. They all tested Romney, whose results I am including as an indication of McCain's relative strength:

• A Kentucky poll has McCain leading Clinton 53% to 43% and crushing Obama 61% to 32%. Clinton manages a lead against Romney, 48% to 46%, while Romney trounces Obama 55% to 38%.


• In Indiana, the two Democrats poll equally. Clinton trails McCain 51% to 42% and Obama is behind 50% to 40%. Both also lose against Romney, albeit by smaller margins: 49% to 44% and 47% to 44%.


• Finally, in Washington state, it is Obama who does much better, as he destroys McCain 55% to 38% while Clinton is forced into a tie at 46%. Against Romney, Clinton leads 51% to 41%, but that's far weaker than Obama's 61% to 31%.

These results do not contradict my theory of which regions Obama and Clinton are strong in; Kentucky in particular has been a state in which Clinton has been consistently competitive. Despite this, the media inexplicably keeps running with the story of Clinton hurting down-the-ballot Democrats in those states.

The internals of the SUSA polls have a few other interesting clues relating to the electability question. First, on the gender gap:

• In Kentucky, Clinton runs 20% better then Obama overall. And that margin entirely comes from the female vote. Among men, both candidates are running roughly equal (Obama trails 62-30 and Clinton trails 60-36); among women, Obama trails 60% to 34% (no gender gap whatsoever) while Clinton actually leads 49% to 46%, 29% better than Obama.
• In Washington, on the other hand, it is the male vote that makes the difference between Clinton and Obama. Both candidates have a slight lead againt McCain among women, but look at the male vote: Clinton trails 47% to 43% while Obama leads 62% to 32%. That's a 34% swing.

A third characteristic that holds across those polls is that Clinton is much stronger among registered Democrats than Obama, who gets more cross-over votes from Republicans:

• In Kentucky a full 44% of Democrats vote McCain if Obama is their nominee, versus only 28% if the nominee is Clinton.
• In Indiana, McCain takes 18% of the Democratic vote against Clinton, and 30% against Obama. On the other hand, Clinton takes 9% of the GOP vote and Obama gets 19%. (Clinton does a bit better among indies).
• In Washington, Clinton does better among registered Democrats but she only takes 6% of Republican votes, while Obama gets 25%.

Obama and Clinton both will have significant strength and weaknesses in a general election. Obama is stronger among men, but Clinton causes female voters to massively vote to Democrat. Obama attracts the center and can get votes from the right, but what use is that if he cannot hold is his base? Clinton does that holding very well, but she needs to appeal to independents to get to the White House.

The Latino vote will also be very interesting to look at. They have gotten Clinton out of trouble many times over the past few weeks, starting with Nevada and ending with New Mexico and California. On the GOP side, John McCain is the only one who potentially could appeal to Hispanics given that he never fully followed the lead of fellow candidates towards nativism and immigrant-bashing. Will a McCain-Obama showdown help the GOP attract more Latino votes than they were expecting to this cycle? Can Clinton offset McCain's strength among the Hispanic community? Or will the GOP's increasingly harsh and nativist language on immigration doom any chance McCain has of winning the Latino vote? Given that immigration will be a big issue in November, it is unlikely that McCain will be able to distance himself from those in the party who Latinos justifiably will reject.