Abramowitz predicts Clinton re-election, `time' a factor

Forget about Whitewater, Hillary, the character issue. President Bill Clinton will win re-election with a decisive 56 percent of the majority party vote because it's simply not his "time" to go, according to political science professor Alan Abramowitz, who bases his prediction on a statistical forecasting model that accurately called the 1992 Clinton-Bush contest within one-tenth of a percentage point.

Abramowitz's model, which produced the most accurate prediction made by any academic forecaster in 1992, uses an indicator unique from other formulas: a dummy variable that takes on the value of "0" if a party has been in power for four years and "1" if a party has been in power eight years or longer. This dummy variable is intended to reflect the presence of "time for a change" sentiment in the electorate. The other two indicators Abramowitz uses---the U.S. Commerce Department's estimate of the change in real gross domestic product (GDP) between the first and second quarter of the election year and the president's Gallup Poll approval rating---are commonly used in other statistical models.

"I estimate that the time factor reduced Bush's share of the major party vote by about five percentage points in 1992, enough to lose the election," said Abramowitz. "The same factor suggests that Clinton should have a good chance of winning because the Democrats have controlled the White House for only four years." In addition to the time factor, his 1996 projection takes into account the president's approval rating in mid-May (55 percent) and the rate of economic growth during the first quarter of 1996 (2.8 percent).

Identifying a pattern of oddball characteristics supposedly linked to election wins has become somewhat of an American sport---no left-handed president ever won re-election (Clinton is a lefty), the candidate with the longest name wins (Dukakis shattered that theory), Democrats win when Bordeaux wines have a good vintage year, Republicans when hemlines remain the same. But election outcomes are not random events, say researchers; patterns of electoral behavior can be identified long before Election Day.

"The forces that determine the outcome of presidential elections---the state of the economy, the public's evaluation of the incumbent president and the length of time that the president's party has held the White House---are largely set before the start of the fall campaign, and last-minute shifts are usually too small or too late to have much of an impact on election outcome," he said. "Despite all of the time, money and effort devoted to campaigning, there is very little that the candidates can do during September and October to alter the outcome of a presidential election."

But don't worry, Republicans---time will be on your side in the year 2000.

--Nancy Seideman

Return to the September 16, 1996 contents page