|
| |
A Dead Heat: The 2000 US Presidential Election
From: Columbia University
| By:
Robert S. Erikson |
EDITOR'S INTRODUCTION |
Some elections are hard to predict. Columbia University professors Robert S. Erikson (right) and Karl Sigman's "dead heat election model" calculated that, based on state polls, if Al Gore were to win the popular vote by even a slim margin he would also win the Electoral College and become the 43rd president of the United States. In this interview with Fathom, Erikson discusses the tense 2000 US presidential election and how it compares with a previous hotly contested election, the presidential election of 1876. |
Fathom: When was the last time a presidential race was this close? Over a century ago? |
Robert S. Erikson: We're all learning about the 1876 election, which is certainly the most disputed. We had an inversion of the popular vote and the Electoral College vote in 1888, which was when Cleveland lost to Harrison, but that wasn't as remarkable as 1876, because there you had states where electoral votes were clearly in dispute. |
It involved three southern states, Louisiana, South Carolina and Florida. And in each of those states the Democrats and Republicans each claimed to have the popular vote.
Each claimed to have won the state's vote and therefore the state's electoral delegates. So this, as we know from the history books, presented a real quandary for Congress to figure out. |
They set up a commission with five senators, five House members, and five Supreme Court justices. It was very hotly contested at the time, like now, and was eventually decided because of the partisan division of the court members, and Hayes the Republican was elected. |
There were reports of side deals that helped this along. For example, the end of Reconstruction in the South was a by-product of this election, so, in effect, the deal was perhaps that Republicans got the election, and the Southern Democrats got the Northern troops removed from the South. A great controversy achieved a very late resolution that way. |
 | |
| Robert Erikson describes his predictions for the 2000 US presidential election. | |
Fathom: Were there any recount controversies during that election? |
Erikson: As I understand it, in the South in three different states, each party claimed to have the benefit of the count. How it was done exactly in those primitive days I don't know, but it was a real dispute and obviously a real division, particularly in the Southern states where the two parties were very much at odds. Republicans in the South were the party that represented black voters, who were voting heavily at that time because they were enfranchised, and the Democrats tended to represent the white Southerners. It was a really tense conflict. |
Fathom: Based on your models for a dead-heat popular vote scenario, which candidate did you predict would win the 2000 presidential election, and how did you come to this conclusion? |
 | |
| Robert Erikson discusses the role of third-party candidates in the 2000 election. | |
Erikson: Well, Carl Sigman and I, just like everyone else, were saying that the election was going to be very close. The question was obviously who was going to win, how do we predict it, and what's the relationship between the Electoral College and the popular vote. So we developed a couple of models using mainly state polls. |
We had many state polls this year--every state has been surveyed at least once during the fall campaign, some as many as 10, 20 times, developing what we call huge state meta-samples of public opinion. So we looked at the state polls and relations, adjusting for the national trend at the time of the state poll, and then had an estimate of what these state polls would do. We also made some "what if" kind of estimates, if the vote was actually going to be a real dead heat or a statistical tie in the popular vote, and as it turns out it was. |
 | |
| Robert Erikson describes his predictions for the electoral and popular vote split. | |
Interestingly, we got most of the states right. We predicted Florida was going to be very, very close, which of course we all know is correct. And in addition to that, we made one error where we predicted that with the vote very close we thought that Gore was clearly more likely to win it. |
As I speak right now, we don't know what the outcome is, but our prediction was at the time that Gore had a little bit of an edge in the Electoral College and we thought the reason for that was the so-called battleground states, where Gore always polled a little better than he did nationally. So the general argument was that if the national polls were going to show a dead heat, then Gore would poll better than average in the battleground states, win most of those, and that would give him a good chance of winning the election. |
Somewhere that went wrong. I think where it went wrong was that, in the final analysis, Gore surged to gain an even split in the vote, and most of this gain was in the states that Gore already had, like New York, California and New Jersey, where it didn't do him any good. That's where we think our prediction was wrong. |
But in terms of predicting the vote of the key battleground states, as a set we predicted those very well, and if Florida were to go to Gore, then we have a very good prediction based on those state polls. |
Fathom: One of your key assumptions was that support for minor candidates would not affect the two-party verdict. Is this still true? |
Erikson: Well, we made some conservative assumptions, which were basically done because when you're doing research on the fly like this you have to make some simplifying assumptions. Basically, we looked at the two-party vote between Gore and Bush as if Nader and Buchanan did not exist. Our projection was that if things stayed the same, then we would have a good expectation of how the state would vote in the general election. |
As Gore's and Bush's percentage of the two-party vote increased, we looked at what the Nader voters were going to do. And the conventional wisdom was, if anything, that Nader voters were going to break toward Gore at the end, but we didn't know that and we couldn't take that into account. We basically assumed that if Nader voters were going to change, they'd already done that by the time of the late polls. |
In any case, those late polls in the battleground states as a group were very accurate, so we predicted that if you took those states as a group, the state polls were within one hundredth of a percent of what the states actually did in the tally of the popular vote. |
So the lesson of that is that state polls, particularly toward the end, do reflect the popular vote. It also indicated that, in those battleground states we were looking at, probably public opinion didn't move at the end. In those battleground states, the candidates had been campaigning there day after day, but the states were saturated with television ads for the candidates, and so by the time the last week came probably everybody in those states knew how they were going to vote. They had already sorted it out and so there was no change. Gore's very big surge toward the end came in states where there was not much advertising, not much attention to the campaign. |
In those states, people could take a look at the election at the very last minute, saying, gee, how are we going to vote, and there was a certain surge to Gore because people thought, well, the economy is good, and they liked the way Gore was campaigning or whatever. He got a surge in those states that ironically didn't help him at the end. |
Fathom: So the basic relationship between the electoral and popular vote that you based your models on still holds true? |
Erikson: Yes. The relationship between the electoral and the popular vote is one where, almost always--and of course there can be exceptions--the popular vote and the electoral vote go to the same candidate. |
Interestingly, in the year 2000 almost all observers, including Carl Sigman and myself in our work, thought that the likely scenario was going to be that the electoral vote might go to Gore even though the popular vote went to Bush. There's even a scenario out there for a major difference we could have seen maybe in advance, depending on how the states broke. Based on information everybody had, maybe the battleground states could break for Gore, but enough of them would give a narrow Electoral College win and yet Bush could have had a sizable popular vote advantage, maybe a couple of percentage points. Clearly he might have been the popular vote winner as the polls at the time were showing, and yet Gore could have won.
Now we know that didn't happen. Ironically, now the Electoral College-popular vote inversion scenario was opposite of how it may turn out, assuming that Bush were to win this election. And we don't know yet which way it's going to be. |
|
| |
