If you need another reason to hold current and immediately upcoming presidential election polls in minimum high regard, check out this reminder from Larry Sabato’s Crystal Ball:
With all of the polls, models and history at their disposal, political analysts should be able to figure out who is going to win a November presidential election by June, right?
Well, not quite….
If anyone doubts that a reassessment — maybe several of them — will come as 2012 wears on, consider this: Over the past eight elections, Gallup — the most recognizable of polling organizations — has only identified the eventual popular vote winner twice in its early June horse race polling.
The two June “hits” by Gallup polling, it seems, were in elections that did not turn out to be close at all: 1984 and 1996. But in 1980, the June Gallup had Jimmy Carter with a seven-point lead over Ronald Reagan (and John Anderson at 21%!); in 1988, Dukakis led Poppy Bush by 14 points; at this point in 1992, believe it or not, Ross Perot was running first, 8 points ahead of Poppy and 14 points ahead of Clinton; in 2000, W. led Gore by 5 points; in 2004, Kerry led W. by 6 points. And lest we forget, in 2008, the June Gallup had John McCain up a point over Barack Obama.
You can make the case that today’s partisan divide is more stable than it was from 1980 through, say, 1996, but not so much more than from 2000 on. This does not mean that current polls have no meaning: the internals sometimes show important dynamics, and trends shown by individual polling operations typically reflect not only campaign events but reactions to external events and the candidates’ ability to consolidate and “energize” support.
But don’t put any money down on polling this early, or get too excited or discouraged, as the case may be. These are still early days.
