Presidential approval around election day seems a very strong predictor of election outcomes in reelection campaigns:
| President | Approval/disapproval rating | Approval/disapproval minus undecideds | Election outcome |
| Obama 2012 | 49.2% - 46.4% | 51.4% - 48.6% | 51.1% - 47.2% |
| Bush 2004 | 49.1% - 46.8% | 51.2% - 48.8% | 50.7% - 48.3% |
| Clinton 1996 | 54.6% - 38.6% | 58.6% - 41.4% | 49.2% - 49.1%* / 57.6% - 40.7%* |
| Bush 1992 | 32.6% - 55.5% | 37.0% - 63.0% | 37.4% - 61.9%* |
| Reagan 1984 | 57.9% - 33.0% | 63.7% - 36.3% | 58.8% - 40.6% |
| Carter 1980 | 37.9% - 54.8% | 40.9% - 59.1% | 41.0% - 57.3%* |
Years with * include major third-party candidates, whose numbers were added to the challenger. Note that presidents and their adversaries tend to follow very closely, respectively, their approval and disapproval ratings excluding undecideds. The only major deviance 1996, when one would need to add Perot's numbers to Clinton instead of Dole to closely match the graph in the previous column. Where does that leave Trump this year? Well...
| Trump 2020 | 42.6% - 53.6% | 44.3% - 55.7% |
That's a rather tight spot to be, to say the least. Even assuming all undecideds are hidden Trump supporters and that all ~3% third party votes will come from people who disapprove of Trump, he still loses by about 4.2 points, slightly above the predicted partisan lean of the tipping point state (that is to say, he still loses). Of course, his approval rating might change somewhat until election day.
Last edited by haxxiy - on 25 October 2020
