Consider these two articles from Business Week:

Bush in ’04, Even with This Economy, AUGUST 4, 2003
Suddenly, He’s Vulnerable, MARCH 24, 2003 (posted ~MARCH 17)

That’s pretty amusing: two articles from the same source, a mere few months apart. They point out the danger in drawing too much of a conclusion from either a single data point, or too short of a data series.

Let’s get specific:

Suddenly, He’s Vulnerable, MARCH 24, 2003 (posted ~MARCH 17)

Two intertwined forces have tempered support for Bush. A weak economy has been further damaged by the uncertainty accompanying a potential war and a possible wave of terrorist reprisals. With gasoline prices topping $2 a gallon in some spots and the stock market swooning amid growing concerns about the health of the expansion, Bush has lost many post-September 11 supporters.

What conclusions can one draw on the eve of War? Its difficult to pull together a substantial perspective with staying power, at least based only upon the gestalt of that single moment . . . Since that article, the war went better than expected (but the peace went worse), oil prices dropped (but soon rose over $32 a barrel).

In other words, the situation is fluid and very subject to change. It continues to evolve. As the post war environment reveals a lack of adequate planning for the rebuild/rehabilitation of Iraq, the trend of weakening in the polls may continue.

Something important to remember when evaluating data: Most economic and political issues are “moving pictures;” Relying upon any single snap shot is potentially fatal to accurate forecasting.

Let’s look at what happens in a short data series:

Bush in ’04, Even with This Economy, AUGUST 4, 2003

Despite slow growth and lost jobs, two proven political prognosticators see omens pointing to a GOP victory — maybe even a landslide.

[Yale University economist Ray Fair] looks at overall growth, inflation, and what he calls good news — the number of quarters in which the economy grows at 4% or higher. Bush benefits from very low inflation, but GDP has risen at only 1.7% since he took office, and he has had only one good news quarter. Still, Fair figures if the economy expands at just its current modest rate of 2.4% for the first three quarters of ’04, Bush will be in good shape. If growth next year hits 3% — below most economists’ expectations — he could win in a landslide.

Ray Fair is hypothethesizing 3 more quarters of 2.4% GDP growth. I’m not saying that happens, or doesn’t, but it’s quite a stretch to predict a winner based upon one quarter’s GDP numbers. So actually, 2 of his 3 conditions for a President to comfortably be re-elected do not exist; Either he doesn’t follow his own methodology, or doesn’t trust it very much. Or, he’s a idealogue pursuing a politcal agenda (See The Danger of Dogma on Stanford economist — and former chairman of the Council of Economic Advisers under the first President George Bush — Michael J. Boskin’s embarrassing foray into “faith based economics”); I don’t know Fair’s work, so I don’t know how fair Fair is (sorry . . . couldn’t help it).

But maybe he’s extrapolating the change in GDP to from last Q to this Q and projecting forward?

Maybe, except for one thing: the quarter he’s referring to had the biggest increase in Defense sepnding since the Korean War. Its aberrational at best. Now if Businesses start to spend, hiring improves, and GDP picks up on sustainable basis, I surmise Bush will be difficult to beat. Not impossible, just unlikely. But those economic events have yet to occur.

The lesson here is that a single point does not make a trend, and 3 months is an eternity in politics . . .

POSTSCRIPT:
I just came across this recent Poll a few hours after the above post: White House 2004: General Election
Pew Research Center for the People & the Press, survey conducted by Princeton Survey Research Associates, July 14-Aug. 5, 2003
(N=1,866 registered voters nationwide. MoE ± 2.5).

“Looking ahead, would you like to see George W. Bush reelected president in 2004 or would you prefer that a Democratic candidate win the election?”
























Date Bush Democrat Other/Unsure
7/14 – 8/5/03 43% 38% 19%
6/24 – 7/8/03 47% 37% 16%
4/03 48% 34% 18%

George W. Bush’s Job Approval Rating
pr030808iii.gif
Source: Gallup Poll Analyses

Here’s the Gallup commentary accompanying the graph above:

“This period of leveling off for a president’s job approval rating is not uncommon. One important comparison point for Bush’s job approval is the trend line of George H.W. Bush’s job approval ratings in 1991. The elder Bush’s rating dropped rapidly from a high point of 89%, measured just as the Persian Gulf War was ending, to 72% in June 1991. At that point, however, Bush’s ratings stabilized, and stayed at about that level through the next three months, even rising slightly to 74% by late August. As the fall began, however, his ratings resumed their downward trajectory, ending up at 50% by the end of 1991.”

Puts into a little better context . . . But let me remind you that the election is still 15 months or so away. A lot can happen between now and then.

Category: Media, Politics

Please use the comments to demonstrate your own ignorance, unfamiliarity with empirical data and lack of respect for scientific knowledge. Be sure to create straw men and argue against things I have neither said nor implied. If you could repeat previously discredited memes or steer the conversation into irrelevant, off topic discussions, it would be appreciated. Lastly, kindly forgo all civility in your discourse . . . you are, after all, anonymous.

One Response to “The Value of Political Polls and trend analysis”

  1. K Harris says:

    Ray Fair, in addition to predicting election outcomes, predicts economic outcomes. He is the author of the now venerable “Fair model”. That model predicted as early as April that Q2 GDP would rise by 2.7% (annualized), far closer to the outcome in the first release of the data than just about anybody else. The model predicts growth for several quarters with each run. The most recent run puts growth comfortably above the pace he thinks necessary for Bush to win. That might explain his willingness make assumptions about the effect of the economy on the next presidential election. There is no element of judgement in the Fair model’s outcomes – it depends entirely on coefficients arrived at when the model is estimated. Fair may have lowballed his estimates of the growth Bush needs to win. I don’t know much about his political estimates, though I have a vague recollection that they are done pretty much the same way his economic model is – long on correlations, short on “judgement”. I know nothing about his personal political beliefs.