Precarious polling playthings

Few spectacles look sillier than adults engaged in serious play with children’s toys. It’s an apt description of the current discussion of presidential general election polls.

Polls at this point are often inaccurate because people are only mediocre predictors of their own future behavior.

A sampling of faulty results from the past reveals the temerity of analyses that assume today’s results are laden with meaning. In the spring and summer of 1960, Richard Nixon led John Kennedy. Humphrey bested Nixon by  five points in June. In the same period, Carter led Reagan and Dukakis had Bush down by nine — a lead that expanded, until it was reversed. President Clinton was behind in the summer of 1992 and Bush led Gore by seven points, though he indisputably lost the popular vote to the Democrat.

Indeed, in one-third of the last 15 presidential elections, the June leader lost in November. Dropping from consideration contests that were landslides throughout the cycle, like 1964 and 1996, makes the record of the spring/summer polls look even worse.

It’s not the polls’ fault. It’s ours. We do a less than a stellar job of predicting our future voting behavior, just as we do in predicting our future actions in a variety of realms.

Just 14 percent of those who told pollsters in one survey that they intended to purchase a computer actually did, while 60 percent of those who were “certain” to buy a car in another study did not. Forty percent of sailors who told researchers they were “very likely” to remain in the Navy left the service within five years; nearly one in five was gone within three. At the same time, half of those who claimed they were unlikely to stay nevertheless remained sailors.

In forecasting our own behavior, we fall prey to “fundamental attribution error,” underweighting situations and over-weighting our current feelings. Some voters shake their heads at the Democrats, stuck in the muck of a primary campaign, while John McCain talks to world leaders. But will it look that way in November?

Certainly not. The picture will be radically different, but we naturally have trouble imagining a reality different than the one facing us today.

We also ignore the influence of habit in predicting our own behavior. Today, significant numbers of Democrats claim they will defect if their favored candidate fails to win the nomination. Gallup reports that 28 percent of Clinton voters and 19 percent of Obama supporters will turn out for McCain if their candidate fails to secure the nomination. Balderdash.

In the midst of the hotly contested Republican primaries of 2000, voters were substantially more likely to predict defection. In March that year, 51 percent of McCain’s supporters asserted they would vote for Gore if Bush were the nominee. Did they? The data are imperfect, but the number of defectors was no more than half that, and probably much less. Outside his home state, McCain’s best primary performance was in Rhode Island, where he beat Bush with 60 percent of the vote. By November, just 11 percent of Ocean State Republicans defected to Gore. In New Hampshire, the home of McCainamania, Bush lost just 8 percent of Republicans. McCain voters vastly over-predicted their likelihood of defection in 2000, just as Democrats are doing in 2008.

These folks are ignoring the pull of their strong Democratic habit. Most Democrats will find a way back to the eventual nominee, regardless of who it is.

If we can’t accurately forecast our computers, cars or careers seven months out, why should we be able to predict our vote?

It’s fun to look at the plethora of horserace polls, but used as forecasts, they are more toys than tools.

Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982, including Sen. John Kerry (D-Mass.) in 2004.