Sullivan has just blogged that he thinks the presidential race won’t be as tight as the CW holds. He’s quoting an email on the track record:
“On 20 occasions, the incumbent President has been re-elected. Median margin of victory in the electoral college is 66%. That would require Bush to get 447 electoral votes. On 11 occasions, the incumbent President has lost his bid for re-election. Median margin of loss in the electoral college is 37%. That would require Kerry to get 366 electoral votes. If you want to define an electoral ‘landslide’ as being 10% or more difference, then of the 31 occasions described above 29 were landslides (Adams lost by 5 in 1800 and Wilson won by 5 in 1916). If you want to define it as 20% or more, then 27 were landslides (19 of the 20 wins and 8 of the 11 losses).”
Sullivan then adds: “See my point?”
Actually, I don’t–or rather, I don’t see his correspondent’s point. Because most presidential races with incumbents have not been close, we’re supposed to believe that this one is unlikely to be? Why can’t we just conclude that the electorate is more evenly divided than it has been for most presidential elections? I’ll go out on a limb and make a prediction: Neither Bush nor Kerry is going to get over 55 percent of the two-party vote. Clinton didn’t in 1996.
Or is Sullivan’s correspondent using “a 10% difference” to mean that one candidate will get 10 percent more votes than the other? In that case, the candidate would have to clear around 52.4 percent of the two-party vote and thus beat the other guy by 4.8 points. I’ll agree that one of the candidates could do that. That would probably mean an absolute percentage of the vote that qualified as a landslide by post-1988 standards (since nobody has gotten an absolute majority during this period). But it wouldn’t be a landslide by historical standards.