With Obama riding high in the polls, comparisons between John McCain and Bob Dole are gaining ever more currency. Former Hillary campaign flack Howard Wolfson has gone as far as declaring it’s over:

Perpetually fretting Democrats will not want to accept it. The campaigns themselves can’t afford to believe it. Many journalists know it but can’t say it. And there will certainly be some twists and turns along the way. But take it to a well capitalized bank: Bill Ayers isn’t going to save John McCain. The race is over.

Last Saturday, Wolfson went as far as penning a “premortem for the McCain campaign“, and was promptly criticized for it by ABC News’ Jake Tapper. On his blog, Tapper warned Obama backers to “Check Yourself Before You Wreck Yourself“:

Buoyed by encouraging poll numbers [..] lots of Obama backers out there seem to think this thing is over. [But that’s] not what lots of smart folks in the Obama campaign think. They believe Obama’s poll numbers are artificially high, McCain’s are artificially low, this race will come down to two or three points, and anything could happen.

As Anne Kornblut and Jon Cohen in the Washington Post today remind us, “recent history suggests that mid-October leads are vulnerable […]”.

Is Tapper right? Should we check ourselves before rejoicing too early, only to find ourselves flat-footed when the numbers unavoidably narrow again? That’s always a piece of advice that’s close to my heart, but it bears mentioning that Tapper cherry-picks his data to make the point.

Tapper mentions several historical examples, taking his cue from Kornblut and Cohen in the WaPo in citing 1976 and 1992. The most recent example, however, he adds himself:

CNN on October 5, 2000 reported that the “CNN/USA Today/Gallup tracking poll indicates that Vice President Al Gore may be opening a solid lead over Texas Gov. George W. Bush, after nearly two weeks of neck-and-neck competition. Today’s figures — 51 percent for Gore to 40 percent for Bush — represents a significant margin for the vice president.”

Obviously the numbers narrowed a little bit.

As will these as well.

Now Tapper’s post is from yesterday, 13 October. So why cite an October 5 poll from 2000? Because it’s the only one that would make his point. That 11-point lead for Gore was something of an outlying data point. On October 3, Gore had led by just 2 points. On October 5, he led by 11. On October 6, he was tied with Bush. At no point between then and the elections a month later did Gore ever enjoy a lead of more than a single percentage point in the daily Gallup poll again.

What happened? Well, Gore’s support appeared to be spiking right on that day – though it may as well have been something of a statistical outlier. In any case, by the day after and the day following that, the impact of the first presidential debate of that year, which took place on the 3rd, got to properly weigh into the polls. And Bush apparently did well enough in that first debate to undo any gains Gore was making, and establish an advantage that he would not relinquish again until the elections.

Historical leads of presidential candidates in the last month of campaigning

This, of course, has been the point of many of those making Dole comparisons in the first place. In order to come from behind under unfavourable circumstances, a candidate must make use of only a handful opportunities to change the game. The convention and nomination speech, for example. And the first debate. It is the point at which McCain demonstratively threatened not to come to the first debate on September 26 and then caved in and came after all, only to perform poorly, that’s been pinpointed by many pundits as the moment on which McCain lost. It was his last opportunity, barring something like a terrorist attack hitting America, to substantively change the game – and he missed it.

Obama’s put in two solid performances in the debates and it’s unlikely that the last debate will suddenly be very different. As of that week in late September with the outbreak of the financial crisis and the McCain campaign’s unsuccessful theatrics, Obama seems to have established a structural advantage, much like Bush apparently settled the game in 2000 at the time of the first debate.

Jake Tapper

Now, you can take all of this is as an occasion to tsk a bit at Tapper. After all, in order to make his case that leads like the one Obama is now enjoying can and have often still melted in the last month out, he cherry-picked an outlier of a poll, from a qualitatively different moment in the campaign. But the interesting question isn’t about Tapper, but about his thesis. How safe is Obama’s lead? How does his current lead match up with the volatility in recent election races?

The graph above shows that he isn’t quite in 1996 territory. Clinton systematically enjoyed 10+% leads, while Obama is only breaching that level now. On the other hand, you can see where the comparison is coming from. Obama’s current fortunes are certainly far removed from Gore’s and Kerry’s at this time. Both Gore and Kerry never once enjoyed a statistically significant lead in the Gallup poll in the last month before the elections. Obama on the other hand has had a lead of at least 5% for twelve October days now.

Obama’s lead is also comparatively steady. You can see in the 1996 and 2000 numbers that most of the volatility peters out with the end of the election debates. The last debate in 1996 was on 16 October, which resulted in a last spike in Clinton’s lead. After that, his lead remained in a band of 10-20% (he was to win by 9%). The Gore/Bush race had a last bout of volatility between 21-27 October, but there too the basic situation was never again at risk: Bush captured a lead of about 5% in early October and he didn’t surrender it again.

Is time running out for McCain? (Image used under CC license from Flickr user olkin11)

Will there still be volatility in the polling to come this year? Probably. But even if the wildest October swings from 1996 and 2000 are repeated, Obama’s lead can swing up or down 7 points and never be seriously endangered. In that sense it’s roughly like 1992, when, Tapper quotes the WaPo article, “Clinton held a 14-point advantage over incumbent George H.W. Bush in Post-ABC polling, and it was as high as 19 points before the election, which he won by six points.” Swings of five or eight points up or down, but a lead where those dont make a difference anymore. Obama’s lead is a couple of points less comfortable, but still solid enough to whether such swings.

Does that mean it’s over? Well, when the fat lady sings and all that. Can anything still happen? Course it can. Especially with something of a wild card on the effect of race still in play. Is writing premortems for the McCain campaign tempting fate? Maybe. Tapper’s general point of not getting overconfident is always good. But none of the concrete polling examples he cites from after 1976 show the kind of volatility that would still overturn the 10-point lead Obama now has according to Gallup. So from the perspective of recent history, Wolfson and co. don’t need to feel particularly bad about their confident assertions.