Pollster.com (now a part of Huffington Post) is my favorite site for watching political polls over time. Their tracker for the national polls on the 2012 Republican primary seems to me to be especially illuminating.
If Paul, Perry, and Cain aren't selected when you arrive on that page, select them. Then in the “Edit This Chart” area below the chart, select “Less Smoothing” and set the start of the date range to June 1, 2011.
The pattern is striking. First there's a little bump for Cain. Then Cain's popularity subsides, and there's a bump for Bachmann (who's for some reason no longer listed on this chart, but you can see her on the much harder-to-read realclearpolitics chart). Then Perry is ascendant. Wow, look at Perry go! He's way more popular than Romney. Surely he will be the Republican candidate! Romney can't seem to get above 25%!
. . . Oops, look at Perry fall. He has about six weeks of outpolling Romney, then he drops into obscurity again, and Romney's line goes up.
Aha, but that leaves room for Cain's comeback! Go, Cain! Doing better than Romney!
For a month.
But that's only because the base finally found their real candidate: Gingrich! Wow, look at him go! Way more popular than Romney!
For six weeks. At the end of which time Romney's polling goes up by about ten points.
Then there's another Gingrich surgelet, but this time he doesn't pass Romney at all. (Well, okay, if you look at the specific poll numbers at that time, some of them do show Gingrich over Romney, but not for long enough to appear on the trend-line graph.) Note that by this point (mid-January of 2012), Romney is consistently polling around 28%-32%, after having previously had a hard time staying above 20% or so.
Ah, but really all this time the true Republican candidate, Santorum, has just been biding his time. Finally, it is Santorum's turn to shine! He zooms to the top of the charts! Number one with a bullet!
For two weeks. And then his popularity starts to plummet.
If you scroll down on that page to the Latest Polls section, you can see the same thing in a different format. From about February 10 through about February 22, Santorum was showing in national polls as anywhere from slightly ahead of Romney to quite a bit ahead. Then, starting with the Gallup poll ending on February 27, the polls shifted back to Romney's favor. During the ten or eleven days since then, every national poll has shown Romney a minimum of six points ahead of Santorum, and in a couple cases as much as sixteen points ahead.
The narrative that I've been imposing on that graph for the past five months or so (not original to me; I picked it up from the media) is that there's a certain Anyone But Romney contingent of the Republican party. (Or rather, Anyone But Romney As Long As It Isn't Ron Paul; Paul has been steadily polling around 8% to 10% pretty much the whole time, with minor variations along the way.) They seize on whichever other candidate seems momentarily potentially able to beat Romney; that candidate enjoys a few weeks in an intense media spotlight, then makes a variety of mistakes and flames out (or just turns out not to be the Anti-Romney that the base was looking for) (or, as Jim points out in comments, gets slammed by negative ads from Romney and co), and the ABRALAIIRP wing looks around for another candidate they can focus on. It makes me want to write a kids'-book parody to be called “Are You My Not-Romney?”
So when Santorum jumped to the top of the charts, my expectation was that he would stay there for four to six weeks and then drop. I was wrong; it was more like two.
There are problems with my analysis. For example, that graph shows national polls, and what really matters in terms of getting delegates (and thus the nomination) is state polls. It may well be that Santorum is continuing to do well in those, and if he wins another couple of states in the next couple of primaries and caucuses (he's apparently likely to do quite well in today's Kansas caucuses), I can imagine he might come back in the national polls.
Also, as the pollster.com people used to explicitly say, their polling charts are more useful in looking at trends than at any particular data points. If you look at the same graph with moderate smoothing turned on, then Santorum's trend line is still upward. I don't think that'll continue to be true for long, but I could be wrong.
One scenario in which I could be particularly wrong is if Gingrich drops out and almost all of his supporters shift to Santorum. That would give Santorum a big boost, especially if Gingrich explicitly endorsed Santorum. But that would be, as Nate Silver of fivethirtyeight put it (after doing the delegate math), “at best, a necessary but not sufficient condition for a comeback” for Santorum.
Silver goes on in a separate post to explain that Santorum could still theoretically win the nomination. But he would have to do not just much better than currently expected, but phenomenally well, “generating some real momentum by sweeping just about everything in March.” I just don't see that happening.
So I agree with Vardibidian (who mentions a couple of other scenarios as well): Rick Santorum is not going to be the Republican nominee this year.
Some among my liberal friends are probably disappointed at that. I've seen several people say things like “We should support Santorum, because he's obviously unelectable, so if he gets the nomination, then Obama will win easily!”
I would like to think that would be true. But I'm wary of that sort of argument, because I remember 1988, when some of my liberal friends (possibly including me, I forget) were delighted that George H. W. Bush had won the Republican nomination, because there was obviously no possible way that he could get elected President.
Anyway, what I'm really saying here is that I think we can be pretty sure that, in the absence of some huge game-changing event (like Romney coming out as gay, or dying), Mitt Romney is going to be the Republican candidate for President this year.