Remember these polls on Bush losing mean nothing. Polls said in 1984 that Mondale was ahead of Reagan. Reagan won almost all 50 states.
Not only that, but if Bush does beat Kerry, it would mark the first time since 1948 that an incumbent president came from behind to win reelection after having trailed in ANY Gallup poll at any point over the course of the election year.
Not only Reagan in 1984, but also Clinton in 1996, Nixon in 1972, Johnson in 1964, and Eisenhower in 1956 all led their eventual November opponent in every single Gallup poll for the entire year of the election.
That's not necessarily all that relevant, but it's still something to ponder.
You are wrong. You might even be wrong in more than one of those cases, but you are certainly wrong regarding 1996.
Gallup (1,039 -- REGISTERED VOTERS; ± 3 PCT PTS)
96 Jan 12-15 (Clinton-Dole-Undecided): 48-49-3
96 Jan 5-7 (Clinton-Dole-Undecided): 46-49-5
Well, he used the data from a Gallup-maintained page. You can check his link, it shows Clinto ahead consistently. So you should back up your data.
It's possible that the Gallup graph shows a January average.
1936 Roosevelt continuously in the lead (but margin seriously underestimated)
1940 dito
1944 dito without the sentence in brackets
1948 Dewey overtakes Truman in June, leads right to the end (ie, wrong result predicted)
1952 Eisenhower consistently ahead
1956 dito
1960 lead swinging back and fro several times, final prediction a two point lead for Kennedy
1964 Gallup predicted an even worse shopwing for Goldwater (at one point they had him trailing by 54 points)
1968 Humphrey overtook Nixon for a while in May and June
1972 very accurate throughout the year
1976 Carter ahead from the beginning, 33% up in July, then comes down heavily. The very final figure shows Ford ahead (ie, wrong result predicted)
1980 Carter leads til june (at the beginning of the year by 29 points), Reagan overtakes and has a 16 point lead in August, the fall sees Carter narrowly overhead again, Reagan overtaking him at the last poll (and even then his margin is underestimated)
1984 Reagan leads throughout (at one point in august, the lead is down to one point. Soon after, however, the predicted results takes on the proportions of the final tally)
1988 Dukakis ahead from april to august
1992 Bush ahead into may, Perot in may and june, Bush again in late June. From early july to the end, Clinton leads. He looks headed for a landslide in August, but in October its very close.
1996 Clinton consistently ahead. Except for a few bumps in both directions, even the margin is quite consistently accurate.
2000 the lead changes about as often as in 1960, but there's a point in October when Bush leads by 13 points, and even at the end he's two points ahead
