Do econometric models explain presidential elections, or do their authors simply play with figures until they stumble onto a formula that fits the curve of a handful of election results? Consider the Nelson Model, below, which relies on the same equation-generating techniques used by academic election forecasters. It posits that two major sporting events, the Super Bowl and the Summer Olympics, determine the outcome of presidential elections.

In the case of the Super Bowl, it is the performance of the losing team that's important. The more points the loser scores, the better the Democratic candidate will do. The theory is that when voters witness a valiant effort unrewarded by victory, they are more inclined to support activist government to redress imbalances created by competition.

The Olympic variable captures the effect of internationalism. Without an inclusive Olympics and its accompanying internationalist spirit, voters become more hawkish and suspicious of international agreements, to the benefit of the GOP.

The election equation is:

DV=a₁ + a₂SBLP + a₃OB

Where:

DV is the Democratic candidate's share of the two-party vote.

SBLP is the number of points scored by losing team in election year Super Bowl.

OB equals 1 if a superpower boycotts the Olympics during election year; 0 if no boycott.

The estimated model is:

DV=0.3874 + 0.0071SBLP - 0.0594OB

The model correctly predicts the winner of every election since the first Super Bowl, including the squeakers of 1968 and 1976.

Predicted Democratic Share of the Two-Party Vote (1968-1996)

In the unlikely event of an Olympic boycott by Russia, China, or the United States, the model predicts Gore will get 44.2 percent of the two-party vote in November. Otherwise, Gore gets 50.1 percent of the vote. Since the actual result could vary a fraction of a point either way from the predicted result, the model yields a scientifically embarrassing conclusion: Your vote might actually matter.