Correlations

[mmcirvin]

nexstarman mentions that CNN has been harping on how since 1980 the winner of the South Carolina Republican Presidential Primary has always gone on to win his party's nomination.

How impressive is that, really? From 1980 to 2004, there have been seven US presidential elections. Of those, 1984, 1992 and 2004 had an incumbent Republican president running for reelection who was the party's obvious standard-bearer. That leaves four seriously contested nominations.

In 1988, Vice-President George H. W. Bush was the party's heir apparent; he unexpectedly lost Iowa and New Hampshire, but he had vast advantages in money and backing and came back quickly. Similarly, Bush himself had beaten Ronald Reagan in Iowa in 1980, but the primaries consolidated behind Reagan (who had had a significant following in 1976) pretty quickly. Since in both cases South Carolina was an early place where the general trend revealed itself, I suppose you could call it a useful indicator in that sense, but you probably could have called it without looking at South Carolina. (The real lesson here is that the Iowa caucus is weird.)

So... I guess we're basically looking at 1996 and 2000 as races where the nomination was seriously in doubt and South Carolina picked the eventual nominee (I'd guess its reputation is largely the result of its being McCain's Waterloo in 2000). Pretty amazing, eh? I should go into this business. It's kind of like spouting football stats.

Comments

[mmcirvin.livejournal.com]

...And, really, in hindsight 2000 was a lot like 1988: the establishment favorite (named George Bush) eventually got ahead.

So if there's a lesson to draw from all this, it's that the Republicans usually rally behind a clear frontrunner early on. But that might just be because the Republicans have been pretty successful in presidential elections since 1980, so obvious frontrunners tend to exist. The oddball case was 1996, when they had the unusual experience of running against a fairly popular Democratic president (despite the massive Republican wins of 1994). 2008 might be another oddball, when they're running with a massively unpopular Republican president with no designated successor.

[sunburn.livejournal.com]

I went back a little just now-- not on South Carolina as much as whether the GOP nominee was an early frontrunner as a candidate. In '68, Nixon was the strongest candidate in the first vote, but he was only at 60% (challenged by Nelson Rockefeller and Ronald Reagan). He was incumbent in '72, so that's a no brainer, but in 1976, the incumbent Ford only narrowly beat Reagan for the nomination. You may recall-- I don't-- whether Reagan had any serious challengers in 1980, but I know he had one who wasn't small potatoes-- George HW Bush. Once nominated for Veep, Bush Sr had to do a bit of back-tracking and comment-qualifying because he'd gone negative against the Gipper during the primaries. Whoops!

NPR likes to constantly reiterate South Carolina's amazing predictive powers as well, and is so very glued to the common media storyline for the season that I'm pretty irritated with its coverage. If I wanted third-rate analysis, I'd get cable.

One of the hosts (Morning Edition?) recently interviewed Romney and said something like "you're pretty well expected to capture the evangelical vote; are you surprised that other segments were voting for you as well?" Way to put your cards on the table, NPR.

I don't think you're right about the GOP in general, but the GOP since 1980? Sure. In a way, we're waiting for our next Reagan. I mean, Reagan's a guy who seemed to be the first step in the agenda of everyone in the party. Step 2 was not the same for everyone.

I'm glad we're having a real fight for both seats this year. Ever since I read HST's "Fear and Loathing on the Campaign Trail, '72," I've yet to witness any actual call for strategery [sic] at either party's conventions, and I've been looking very much forward to some.

[jwgh]

I heard that figure being tossed around earlier and thought: 1980 wasn't that long ago. That doesn't sound meaningful.

On the Slate podcast they were talking about why the South Carolina primary has a reputation for dirty politics. After everyone said they didn't know, much speculation followed, some of which made more sense than others. One thing that sounded sort of plausible is that at this point whoever didn't win New Hampshire or Iowa is getting kind of desperate and is more likely to be willing to go negative. Another that had a certain ring of truth to it is that people (and, in particular, journalists) remember the campaign waged against John McCain in 2000 and that looms large enough to color their perceptions of what goes on there.

[mmcirvin.livejournal.com]

I think a lot of this is recency bias; the SC primary in 2000 was pretty notorious and people just remember that.

[mmcirvin.livejournal.com]

As further evidence: "All Things Considered" on NPR interviewed John McCain recently, and just about all they talked about was the 2000 SC primary and whether or not the experience was applicable to the upcoming SC primary. No matter what they asked McCain about, his response always circled back to his vast optimism about South Carolina and the difference from 2000.

[gareth-wilson.livejournal.com]

I don't think it's possible to possible to find any predictive trends in presidential races - by the time you've assembled enough cases you're too far back to be relevent. Take the idea that the voters like electing governors, for example. Go back to 1976 and you've only got seven data points in your favour, and 1976 is hugely different from today.

[mmcirvin.livejournal.com]

So much of this is mining large data sets over short time series for spurious correlations. And you're right—the structure of national party politics in the US changed particularly hugely between 1964 and 1980 (as the Dixiecrats became Republican), which puts a limit on what you can really consider the modern era. Something like the modern red state/blue state division only begins to manifest in presidential elections around 1992, though that in turn is largely because the previous three elections were Republican blowouts.

[mmcirvin.livejournal.com]

...The stupidest thing of this sort is the supposed 36-year realigning election cycle (http://en.wikipedia.org/wiki/Realigning_election), which strikes me as a notion about on the level of numerology or biorhythms.