Iowa Caucus Vs. New Hampshire Primary: Which is the Better Predictor?

Great post, very interesting. However, I don't think that with 8 results we can draw any kind of conclusions from this data.

IMO, it's better for each year to *only* make comparisons with the party that ultimately wins the presidency. Was the ultimate electee properly predicted by his party?

Here's what I mean, using 1984 as an example. In this year, there was no way in hell anyone but Reagan would ultimately win -- it was a complete blowout. The Democrats had to pick *some* candidate in Iowa and New Hampshire, but it wasn't really important who it was since he would ultimately be crushed. So did it matter that Iowans liked Gary Hart over the ulimate nominee Mondale? No, not really. Using this point for prediction analysis distorts the reliability of the estimator -- each state will need to come up with one Democrat & one Republican, one of which will ultimately lose.

What is more interesting is when something like 1992 happens... In this case, Iowan Democrats went with Tsongas, but ultimately Clinton won the party nomination and the election. The party won, but Iowans came up with the wrong nominee. The same thing happened in 2000 with Republicans in NH when they went with McCain yet GWB won both the party nomination and the Presidency. In both of these examples, the respective state missed soundly.

BTW, in the titles for the tables, "Republication"is used instead of "Republican".

There is really not enough data to draw any kinds of general conclusions about the importance of Iowa and NH. Each election has been very different, you have to look at it case by case. Useless article.

Certainly it would also be more meaningful to throw out the years when the candidates were unopposed. These 'no-choice' years add favorable results that are not really indicators of these states' 'abilities' to 'choose' the winners.

With that in mind, Iowa's rankings are:
primary + nom (D) = 4/7 = 57%
primary + nom (R) = 3/6 = 50%

and New Hampshire's are:
primary + nom (D) = 5/7 = 71%
primary + nom (R) = 4/6 = 67%

It's also not really fair to straight-count the election results, as there can only be one winner, so that effectively forces a negative on one party's counts. The question for the general election should be whether either state predicted the winner at all, not whether both parties did in each state, which is impossible. Again, uncontested 'choices' don't really count. That leaves 13 total valid choices for each state through the years with both parties. Alternatively, we can count each of the 8 election years as one 'choice' and give incontesteds a half-point:

Iowa:
7 times a candidate won the primary of their party and then went on to win the general election = 7/13 = 54%
5 years with one winner chosen, but 4 of those were uncontested choices = 5/8 (63%) or 3/8 (38%) on points

New Hampshire:
9 times a candidate won the primary of their party and then went on to win the general election = 9/13 = 69%
6 years with one winner chosen, but 3 of those were uncontested choices = 6/8 (75%) or 4.5/8 (56%) on points

NH gets slightly better numbers overall, with the caveat that all these numbers are really too small to mean much.

this doesn't relate to the predictability of the caucus event, but does relate to its process itself... and is kind of a "whine"...

i stood up for edwards last night in my first caucus (moved across the river from IL last year). after weeks reading Daily Kos and others rip on the state's status as first in the nation, and on the caucus process itself, i felt a little angered at the anti-*iowan* (the *people*, not the process) sentiments, but also understood some of the frustrations. and after last night, i'm pretty much still in that same boat - the taste of my first caucus was a little sweet but also a little sour.

our precinct's numbers meant we needed 69 people to be rooting for a "viable" candidate (you have to have 15% of the total vote, which was 458 people in our precinct - GREAT turnout). obama's crowd had 260 or so folks, hillary had 80-something, and we/edwards had... yup... 69 exactly. the way this whole process works means that if we were one person fewer than this number, we would have been now seen as equal to the group of 3 kucinich supporters - meaning, equal to ZILCH. what a crock of shit. i read a couple similar other stories last night. does this happen quite often? scary.

So, if we try to make a conclusion based on the Iowa results:

Huckabee has a 75% chance of winning the Republican nomination.
Huckabee has a 37.5% chance of winning the presidency.

Obama has a 62.5% chance of winning the Democratic nomination.
Obama has a 12.5% chance of winning the presdidency.

That leaves 50% chance that a third party will win the presidency.

The only one I hear talked about seriously for that is Michael Bloomberg.

The Iowa results makes Bloomberg the most likely winner of the presidency in the general election with 50% chance.

Another problem with these statistics is that they show George W Bush winning the elections in 2000 and 2005. Al Gore and John Kerry, respectively, won these elections but were cheated out of taking office through manipulation of vote counting in Florida and Ohio and elsewhere. Which raises several questions--will 2008 also be rigged? And if so, what the hell difference does Iowa or New Hampshire make anyway?
We go on pretending that Bush won and is our legitimate president because facing the truth is just too overwhelming.

To repeat the meme, this just proves that the repugs stole the last two elections, in which case it would have been 7 of 6 instead of 5 of 8 previous elections.

@Rob-
1. There was no presidential election in 2005. Check your wayback machine, Peabody.

2. Manipulation of vote counting? You are just making yourself out to be a kook. The 2000 much-maligned Florida ballot (apparently not idiot-proof enough) was designed by a Democrat. There was no conspiracy to favor Republicans.

3. Florida recounted its votes a number of times and Bush came out the winner in each case. In no count did Gore win Florida.

4. Bush won outright on the first count, but because the margin was tight, there was (per state law) an automatic recount. Bush won that as well. Several of the counties wanted to AGAIN recount, but Florida law required that election results be certified within 7 days and a few said they couldn't finish in time. The Florida Supreme Court voted to allow recounting to continue, but did not establish standards on when to stop or how to uniformly count ambiguous ballots (these were punchcards and idiot voters could have trouble punching the holes completely).

4. By a vote of 7-2 (a pretty fair majority!) the US Supreme Court rejected the Florida Court's indefinite plan to continue recounting as unconsitutional. A subsequent decision said "game over, you've recounted enough times". Note again that on NONE of the counts did Gore wind up winning Florida. The Democratic plan was to keep counting the ballots manually until per chance (after enough handling and wear on the paper punches) Gore could win one of the counts. THEN they could sue to stop counting. Nice scheme, guys.

5. In the end, Gore had more popular votes, but fewer electoral votes. The US elects its presidents on the basis of electoral votes, though, so this is ultimately what matters. It's always been this way, was done for a reason, and is nothing new. The rules apply to all candidates of all parties and are fair, so stop whining.

6. Fast forward to 2004. Here, Bush carried BOTH the popular and electoral vote. Now what are you whining about? The results in Ohio you are crying about were not even disputed by Kerry! Four states wound up being tighter races than Ohio, in fact -- Wisconsin & New Hampshire (14 electoral votes total) went narrowly to Kerry and Iowa & New Mexico (12 electoral votes total) went narrowly to Bush. Bush won Ohio by over 2% of the vote -- it was close, but not a real nailbiter.

7. If liberals like you want anyone to blame for Democratic narrow losses in 2000 & 2004, blame Ralph Nader. The votes he pulled DIRECTLY from Gore in Florida in 2000 ensured Gore lost that state and the election. In 2004, Nader's impact was a little less clear cut, but Iowa & New Mexico (the 2 narrow Bush wins) would have been even tighter had not Nader sucked away votes from Kerry.

8. Stop whining about the past. If you don't like who is in office now, I can promise you somebody new in a year. Vote for the one you like this November and stop being a sore loser. Even Gore and Kerry are not the crybabies their supporters have turned out to be...

Hi, Sid.

You make valid arguments. However, you miss one crucial point. There is a concerted effort by Republicans to purge thousands (even tens of thousands) of legitimate, lawful voters from the rosters, or to deny such citizens the opportunity to vote outright. Such strategic efforts, unsurprisingly, are targeted to districts where voters are disenfranchised (i.e., poor, and therefore without realistic legal recourse) and where they would likely vote for the Democratic candidate. Specifically, I'm referring to African American, Latino and Native American communities.

This says nothing of the electronic voting machines that do not have an accompanying paper trail for purposes of recounts. The proprietary software of voting machines is not allowed to be investigated, and programmers have testified that, in some cases, it was rigged to deliver a tally result (by a narrow margin) for whatever candidate they chose. Doesn't that warm the cockles of your heart?

There are other voting abuses, like willfully neglecting to count absentee ballots (again, in specific districts), but I won't bother to go into details, as this is merely a 'comment' on an otherwise happy-go-lucky blog. One can do his own research, if he happens to care about the legitimacy of our democracy.

In an election that is decided by hundreds of votes, or even one single vote, you can see how voting irregularities add up and have an effect. Third party candidates, like Ralph Nader or Ross Perot, don't cause the "major" candidates to lose an election. In fact, our democracy would benefit from more third party candidates. That's another story, for another time.

You can wave your hands all day, spouting appellations of "kook, whiner, cry baby, sore loser," but it doesn't change the reality. People are standing up for a healthy democracy, not merely the appearance of one that goes through the motions.

Regardless of either party's machinations, though, a greater voter turnout -- and greater civic participation, in general -- will begin to reclaim the democratic ideal of a government of, by and for the people.

I can't believe people are still whining about losing elections 4 and 8 years ago. Bush won fairly both times, no matter what you want to believe.

Do you really think that only half of the politicians would try to do anything they could to win elections. I'd be willing to bet there are quite a few from both sides. I doubt we'll ever hear about most of the unscrupulous activities.

Dogrun,

Bush won, but not fairly. Are these voting abuses new to the election process, or unique to one party? Doubtful. But with low voter turnout (civic apathy), it does matter. Ultimately, we as citizens are to blame. We need fundamental changes throughout the election process, such as making election day a national holiday (or half work day), so more people can get to the polls. Or using "instant runoff voting" so candidates can win with a real and viable majority. And so on...

Yes, it's a small data set (only 8 data points), but to paraphrase Rumsfeld, we work with the data we have, not with the data we wished we had.

Originally, I tossed out the years where a candidate ran unopposed, but the result is almost the same (New Hampshire was the better predictor), so I kept all of the data points in.

The usual caveats about this little quick and dirty exercise in statistics apply: There are three kinds of lies: lies, damned lies, and statistics. (Mark Twain/Benjamin Disraeli)

According also to http://www.wmbriggs.com/, a candidate needs to win EITHER the Iowa caucuses OR the New Hampshire primary to have a shot at the nomination.