We are still more than a month from Election Day, so I’m not ready to make predictions about state and federal races yet. For the sake of argument, however, let’s imagine that the current polling leads for Barack Obama and Pat McCrory are predictive of the outcomes of their respective races.
Perhaps such a result would finally encourage pundits, professional and amateur alike, to abandon the notion that politics consists of hard-and-fast rules working within a mechanistic system. Politics is much messier than that. It resists simplistic analysis or precise statistical modeling.
If Barack Obama wins the presidential race despite the horrendous state of the economy and widespread public disaffection with the direction of the country, that will violate the “rule” that presidents don’t get reelected when the unemployment rate exceeds 8 percent. And if Pat McCrory is elected governor of North Carolina, that will violate the “rule” that politicians from the Charlotte area can’t win statewide elections.
The two outcomes together would also violate the “rule,” broadly assumed and acted on for decades by political pros of both parties, that North Carolina Republicans fare best when riding on the coattails of a strong national ticket, while North Carolina Democrats fare best when separating themselves from the national party and carving out their own identities.
It’s not that any of these rules was plucked out of thin air, or failed to fit the facts available. Jimmy Carter did lose his 1980 reelection bid in large part because of negative public perceptions of the economy. So did George H.W. Bush in 1992. Former Charlotte mayors Eddie Knox and Richard Vinroot did lose their bids for governor, while former Charlotte mayor Harvey Gantt lost two bids for U.S. senator. And the only two Republican governors elected in the 20th century, Jim Holshouser in 1972 and Jim Martin in 1984, did benefit immensely from being on the ballot with Republican presidents in the process of being reelected overwhelmingly.
The problem here is that, although it may not seem like it, elections are rare events. Presidential and gubernatorial elections only occur every four years. That’s not very many data points with which to build robust statistical models and from which to draw up “rules.” Keep in mind that other statistical efforts to predict human behavior, such as economic models or even sports forecasting, have vastly more events to work with.
Furthermore, it’s not clear that elections held decades in the past are really comparable to elections held today. It’s hard to do much analytically with elections held before the advent of true random-sample polling in the 1930s, for example. Arguably, recent changes in election practices, phone use, and media consumption are also making the lessons of past political campaigns less relevant to campaigns waged today.
I know that whatever happens in the 2012 election cycle, some journalists and talking heads won’t stop coming up with “rules” that perfectly predict the past, and then insisting that they also predict the future. (In this, they are at least wiser than those who use flawed climate-change models to forecast three-foot increases in sea level by 2100. Those models don’t even predict the past accurately.)
Politics is better understood in probabilities, not certainties. And every streak gets broken eventually.
Hood is president of the John Locke Foundation and author of Our Best Foot Forward: An Investment Plan for North Carolina’s Economic Recovery.