Mickey Kaus was talking about the “incumbent rule” which is basically that undecideds tend to break against the incumbent. Lots of nice analysis at the second link.
But something struck me about Mystery Pollster’s table, so I sorted the table by the difference between Gallup’s final projection and the actual result:
Incumbent | Final Gallup Projection | Actual Result | Difference | Party | |
1992 | Bush | 37% | 37.70% | -0.70% | R |
1984 | Reagan | 59% | 59.20% | -0.20% | R |
1972 | Nixon | 62% | 61.80% | 0.20% | R |
1976 | Ford | 49% | 48.10% | 0.90% | R |
1956 | Eisenhower | 59.50% | 57.80% | 1.70% | R |
1964 | Johnson | 64% | 61.30% | 2.70% | D |
1996 | Clinton | 52% | 49.20% | 2.80% | D |
1980 | Carter | 44% | 41.00% | 3.00% | D |
Gallup overpredicts an average of 2.8% for a Democratic incumbent but only 0.4% for a Republican incumbent. So is Bush actually hurt by the “incumbent rule”? Or is the “incumbent rule” only applicable to Democratic incumbents?