Gallup issued this interesting historical analysis, reviewing the predictive power of its July polls on November presidential results. Overall, their numbers project accurately in more than half of the elections. In nine out of the last 15 contests for the White House, the candidate leading in July goes on to win in November–not a bad batting average for early polls. Barack Obama has held a small lead of around four points for the past month in Gallup daily tracking. But as Gallup notes, some of those 15 contests were lopsided for the duration (like Eisenhower in 1952 and 1956, Johnson in 1964, Nixon in 1972 and Reagan in 1984). All of those candidates led by double digits in July and went on to win by double digits in November. Yet in the closer races, leading in July is more problematic. When Gallup narrows its analysis to the nine most competitive, the leader in July went on to lose in six of those elections. The survey trends suggest the convention periods for the two parties (late August for the Democrats and early September for the Republicans) could become the next pivotal moment. In several past elections (Nixon in 1968, Bush in 1988 and Clinton in1992), the eventual winner forged a lead immediately following his party’s convention and never lost it through the remainder of the contest. It other races, where neither party received a larger post-convention bump (Kennedy in 1960, Reagan in 1980, Bush in 2000 and 2004), the race remained tight until the end.
