: Wait and see…Dr. Mayer responds…
Here is Dr. William Mayer’s response (received by e-mail this morning) to my inquiry about the recent Gallup article that appears to challenge his predictive model of primary campaigns.
1. My model only applies to the race that took place after 1976. It strikes me as highly misleading to lump together the races that occurred up through 1968 with those that have occurred since 1976. Literally EVERYBODY who studies the pres. nomination process recognizes that the rules of the process were completely rewritten in the early 1970s, and that generalizations that apply to the 1980s and 1990s are frequently inapplicable to the 1950s and 1960s (and vice versa).
2. Where did these people learn their history? By what possible logic do they describe the 1980 race (where the early January poll was predictively accurate) as one in which a Democratic incumbent president ran for re-election with “little or no opposition”? I’ll bet that’s news to Jimmy Carter and Ted Kennedy.
3. As I note in my article, 1988 clearly was an exception to the rule: Dukakis did not get a lead in the national polls until March.
4. As for 1992, Clinton may not have had a lead in the early January poll, but he had a very large one in the late January poll — which was the last one before the Iowa caucuses.
We have an interesting development today: Howard Dean has lost his MoE lead in the most recent USA TODAY/CNN/Gallup Poll.
Kos has this to say:
So is this the famous last one before Iowa? The one that is supposed to be the best indicator of who wins the nomination? For the record, I don’t buy any argument based on the inevitability of history repeating itself. Period. So just because Gallup has pegged it with its last national poll before, doesn’t mean it’ll happen this time.
Predicting the future is not the point of an academic predictive model; understanding how processes work is the point. Will the Mayer model work in 2004? Who knows? If it works, and I think it will, that gives us more data with which to make positive statements about how the primary process works and why it works that way. If the model fails, that gives us more data with which to make positive statements about how the primary process works and why it works that way. (link to Kos via Political Wire)
