A SHORT BRIEF ON THE NOVEMBER (2004) PRESIDENTIAL ELECTIONS

Comment: As we enter the Presidential electoral season, I am posting this brief on the 2004 Presidential Elections. This was written on October 5, 2004

Summary: Completely independent of my preferences, I predict that Kerry will win the Presidential elections on November 2, 2004. It is likely that Senator will get about 49% of the popular vote, and President will get about 47% of the popular vote. Further, I expect that Senator Kerry to secure at least about 280 electoral votes.

This has been forecast since June 2004. There have been good times and bad times (politically) for both Senator Kerry and President Bush but my prediction is still the same as it was in June 2004.

Reasons: There are three robust and stubborn structural reasons -- largely anchored in Iraq and to a lesser degree in the Economy -- for this conclusion. The reasons are:

(1) Right vs. Wrong Direction for the country: On average about 50% of the Americans believe that the country is in the wrong direction, and about 40-42% of the Americans believe that the country is moving in the right direction. That is a deficit of about 8-10 points for the President, the incumbent on this critical element.

(2) President Bush vs. Some One New: When asked President Bush vs. Some One New, about 45% of Americans want to reelect the President and about 50% want to elect Some One New (John Kerry is not evoked.) That is a deficit of about 5 points on average for the President.

(3) President Bush’s Job Approval Rating: Excluding outliers, the average of job approval ratings of the President for the last ten months (the average of the election year is what the model requires) is about 47%. That is not enough for reelection --- the President falls short by about 4-5 points.

The President’s job approval rating among the limited percentage (about 8-10%) of the undecideds is about 31%. The President is likely to secure only about one-third of the undecided votes, the other two-thirds going to the challenger Senator John Kerry.

There is one algebraic/tactical reason to be pointed out. If Kerry won all the states that Gore won in 2000, then Kerry would secure 260 electoral votes (it takes 270 to win the Presidency.) Except for Wisconsin, Kerry has been leading or marginally ahead or marginally (but statistically within the margin of error) behind in all the states that Gore won in 2000. In addition to the states that Gore won in 2000 (I expect Kerry to win Wisconsin too on the coattails of Senator Russ Feingold and with the greater consolidation of the support of the Democratic constituencies), I expect Kerry to win ten electoral votes from the portfolio of Colorado, New Hampshire, Nevada, and/or West Virginia.

Note that Colorado’s nine electoral votes may be apportioned in proportion to the popular vote if the current proposal for such an apportionment effective this election is approved by the Colorado voters on November 2^nd. (At this point, the said referendum is leading by 47% to 37%.) There are only two other states – Maine and Nebraska – which apportion the electoral votes in some manner other than winner-takes-it-all. In any case, if the election of the Presidency come to the votes from Colorado, there is likely to be challenge at the US Supreme Court which, in my opinion, will affirm the state’s right in this matter.

In my calculations, I have awarded Florida and Ohio to the President but he still will fall short.

However, please note that in spite of serious structural issues the President is likely to be holding his own in the opinion polls or even leading Senator John Kerry by a few points till the last week of the elections. History shows that Americans are very reluctant to retire their incumbent presidents – it takes a lot.

Remember Jimmy Carter – he was holding his own against Reagan till late October but in the last one week the floor caved.

Opinion Polls, Polls and Polls: There are so many polls and so many pollsters that it is difficult to make sense. But based on my analyses, the following polls and pollsters have demonstrated predictive-ability.

National Polls

First-Tier

(1) Zogby poll: Has been very good, and continues to be good; and

(2) CBS News/New York Times poll: Has been very good, and more lately has been somewhat questionable but still reliable, I think.

Second-Tier

(1) LA Times poll: Okay poll – best of the second-tier polls;

(2) NBC-Wall Street Journal poll: Okay but always slightly overstates the republican support; and

(3) ABC-Washington Post poll: Okay poll – just adequately above the threshold.

Pew Research, Harris poll, Annenberg poll and the AP poll are all respectable and non-partisan but I don’t think that they have the correct predictive models. They are better descriptors of the behavior and patterns.

CNN/USA Today/Gall Up poll, Newsweek poll, and Time poll are not reliable predictor polls for a variety of reasons.

State Electoral Polls

(1) Mason-Dixon poll: Only for Southern States;

(2) Field poll: Only for California – and okay for western states only; and

(3) Quinnipiac poll: Only for the Eastern sea-board states.

What makes a poll a good poll or a bad poll in a Presidential Race? In my opinion, the following criteria represent some of the important elements:

(1) Track Record: Zogby and CBS/NY Times score high on this; Gall Up, Newsweek and others score relatively low. See 1996 and 2000 presidential elections. On November 6, 2000... the final CNN/Gallup tracking poll showed Bush over Gore 47-45. Wall St. Journal: Bush over Gore 47-44. ABC/Washington Post: Bush over Gore 48-45. Tarrance: Bush over Gore 46-41. Christian Science Monitor: Bush over Gore 48-46. Only CBS (Gore over Bush by 1) and Zogby (Gore 47-Bush 46) got it correct. (Gore received 500,000 more votes than Bush... though Bush, with Florida, won the "electoral college" and the White House.)

(2) Sampling: How many democrats and republicans and independents in the sample? I think that Zogby’s model of basing the proportions on the recent past (1996 and 2000) voting behavior and attitude is most robust because the empirics show that basic consumer behaviors and attitudes such as party affiliation are stable, persistent and in equilibrium. Gall Up, News week, Pew Research and Battleground polls do not subscribe to this model – they subscribe to the belief that the voter attitudes regarding basic attribute of party affiliation can change from week to week (and I disagree with them.)

(3) Determination of Likely Voters: There are many models but I best like CBS/NY Times probabilistic model. See the methodology of October 4^th poll of CBS/NY Times. Zogby is also okay.

(4) Undecideds: Except for Zogby who does not force the undecideds to decide till the last week before the actual vote, others do. I agree with Zogby’s approach – undecideds should not be forced into a decision earlier in the cycle.

(5) The Measuring Scale(s): I disagree with Zogby’s scale to measure the job approval rating but on other metrics, the Zogby scales are quite clear and consistent. Not necessarily so with all the other polls on various metrics.

(6) Voter turnout: The strength and the mix of the voter turnout on November 2 matters. Different models make different assumptions. The general assumption is that the turnout would be about 55% but it could potentially be higher this time. Another assumption is regarding the turn out of the mix of various constituencies – for example, models assumed about 13% minority turnout in 2000 but the minority turn out was about 19% (Zogby got this mix correctly, not many others.)

Final Thoughts:

Watch for the Zogby’s daily tracking national poll beginning on Thursday, October 7^th, and Zogby’s battleground states’ daily tracking polls beginning on October 21^st.

Of course, watch for Zogby’s and CBS/NY Times’ polls released on the evening of November 1^st. If Senator John Kerry is at a tie or within a point as he enters the elections, Senator Kerry wins the elections because the undecideds will break in his favor -- also don’t forget that lots and lots of new voters coming into the fold, and the new voters tend to vote more democratic.