2004 User Predictions - Discussion

[dazzleman]

Gustaf, you've hit on a good point about polls and voter turnout.

In a close election, the party that best mobilizes its base will probably win.  Each party has certain segments that would never vote for the other party, but might stay home if they are not happy with the candidates, or even go over to a third party.

In 2000, both parties had problems with their base, with many Christian conservatives staying home, and thereby costing Bush the popular vote.  On the Democratic side, there was Ralph Nader, who siphoned votes away from Gore in certain critical states, most notably Florida.

Polls that don't take likelihood of voting into account can be seriously flawed.  This is a very hard thing to gauge, which is why the polls are sometimes unreliable in predicting the winner.

Nixon had his 40-40-20 theory, that each party would receive 40% of the voters regardless, and that the 20% of the swing voters would decide the election.  There are some theories out now saying that the percentage of swing voters is down to 10%, and that therefore makes it more important for a party to mobilize its base than to go for swing voters.  Maybe this is the theory that Dean is using.

The 31-31-31 theory goes the other way.

Different presidential candidates have used different strategies.  Both Nixon and Clinton used the "last vote" strategy, meaning that they would push as far as they could to the "frontier" of their support (left for Nixon, right for Clinton) as they could without endangering their base.  Others, like Reagan, have mainly concentrated on keeping their base happy, and following through on several core ideas.

Bush seems to be emulating the Clinton strategy, with his massive increases in government spending and the prescription drug care plan.  Dean seems to be going for the base.  We'll see which approach is more successful.

[dazzleman]

Well, thank you! Nice words are always appreciated! Now I get what the 31-31-31 thing is all about. But I am wondering if anyone know how the polls are made in this respect? If we suppose that turnout is higher among Reps and Dems than among undecided (which one would suppose) then the undecided wouldn't matter so much. If the polls are actually based on the electorate, and not on likely voters, then 35-40% would actually be well enough, if you just get all of them to vote. That seems wrong since the polls then would make no sense. But if you try to exclude people from polls it can easily backfire. Anyone got insights in what polling institutes do here?

Accurate polls must gauge the probability that a person will actually vote, and discount the opinions of those unlikely to vote.  People can be asked if they plan to vote, but they may not answer truthfully.  They can also be asked if they voted in the last election, or if they are even registered.  Statistically, adjustments can be made to determine voting likelihood, and discounting the opinions of those unlikely to vote.

As far as the undecideds go, they are also a problem, so certain assumptions have to be made there too, depending upon how far away the election is.  Generally speaking, voters who are undecided close to the election will probably break in favor of the challenger rather than the incumbent, since being undecided that late implies reservations about the incumbent.  In addition, other questions can be asked to determine the direction in which the voter is likely to lean.

It's true that Republicans generally have a better turnout than Democrats because their voters are generally more motivated.  But it may not be as true as it used to be.

I think I get your point about needing 35-40% of the electorate to win.  The actual number is lower, given our low voter turnout.  But I don't think it can be looked at that way for the reason you mentioned -- the people you count on to vote for your candidate must turn out.  So I think the only way to look at it is in terms of likely voters.

All these complications point out why it's so difficult to accurately predict winners in elections.

[dazzleman]

I think that this discussion centers too much around the assumption that the 2004 election will be reasonably similar to the 2000 election, with the 2000 election therefore serving as the baseline for predictions about 2004.

It may turn out that way, but I don't think so.

Think about the differences between the 1968 and 1972 elections.  In 1968, Republican Nixon barely beat traditional Democrat Humphrey, who was serving as VP to then-Pres. Lyndon Johnson.  The Democrats had been in power 8 years, and the economy was doing well, but the country was deeply divided over the Vietnam War, among other things.

There was also a third party candidacy that year, with George Wallace winning several southern states with his quasi-segregationist message.  It is questionable who would have won those states in the absence of the Wallace candidacy, since the south then was still very reluctant to vote Republican, and in some states, Nixon came in third, behind Wallace and Humphrey.

Come 1972, the situation was radically different.  The economy was doing well, having gone through a recession in the 1969-71 period.  The Vietnam War was all but over for the United States, although tragically this was not the case for the Vietnamese.  In his re-election bid, Nixon faced an opponent at the far left of the Democratic Party, rather than a traditional Democrat as in 1968.

In short, the political landscape had radically changed, and it would have been foolish to base 1972 calculations on the 1968 political landscape.  Nixon went on to comfortably win states that he couldn't have dreamed of carrying in 1968.

There are some similarities between the 1968-72 period and the period since 2000.  The political landscape was radically changed by the Sept. 11th attacks, and national security is a much more prominent issue than it was in 2000.  Will the American people entrust their national security to somebody like Howard Dean in 2004?  I'd say a lot fewer than would have trusted it to Al Gore in 2000.

I don't think Bush will win on the scale that Nixon did in 1972, and it's probably better if he didn't (look what happened to Nixon!) but I do think that the political landscape has changed radically against the type of Democrat that Dean is, and if he is the nominee, I would not expect his performance to be comparable with that of Gore.

[dazzleman]

You're assuming that Bush will win in a landslide, and that is possible, maybe even likely. I am basing all my predictions on a fairly close race. If Bush wins in a landslide, there isn't much to predict anyway.

I do think that the difference towards the national average will be roughly similar to 2000 in most states, but the national average can be very different. Do you think otherwise?

No, I think that states with a higher percentage voting Democratic, or Republican, will remain largely the same as 2000.  I don't think Massachusetts is suddenly going to turn strongly Republican, or Texas is going to turn Democratic.

But I do think the dynamics are very different this time than in 2000, and that Bush will win much more comfortably than he did last time.  I don't for example believe that Florida will be the ultimate swing state; I think it will be safely Republican this time.  But we'll see.  I never make bold predictions this far out, especially when we don't know the Democratic nominee.  Primaries can bring a lot of surprises.

[dazzleman]

Yes, I agree on Florida, if you read my posts carefully you will see that I am keeping Florida hanging as a tossup. Certain stated have changed, NV and WV are examples of states that might lean Dem this time, Florida would be an example of the opposite.

I think it's more likely that Pennsylvania and Michigan will be toss-ups than Florida.  I think that Bush will hold onto all the states he won in 2000, with a bigger margin of victory, and pick up some states that Gore carried that year.  The only question is how many.

The first "tier" of states that he could pick up are states like Wisconsin, Minnesota, Iowa and New Mexico, all of which he lost narrowly to Gore.  The next tier would be states like Pennsylvania and Michigan, and the third tier would be states like Illinois, New Jersey and California.

I don't know how far he'll go, but I don't think we'll be quibbling in 2004 over the same states that we quibbled over in 2000.

The election is ten months away and that's a long time, so I have to qualify my predictions.  I also have to say that they are based on the assumption that Dean will get the Democratic nomination.  It would be significantly different with Clark, Gephardt or Lieberman.

But if Dean is nominated, it's hard to imagine him picking up any state that Bush won in 2000.  Dean is the candidate for a nasty vocal minority.  They may be loud and obnoxious, but they can't carry a general election, and they'll drive away moderate voters in droves.  That is my prediction, so I would say forget Florida, Nevada and West Virginia too.  Worry about Dean winning Michigan, Pennsyvania, California and New Jersey.