In 2004 between the convention and the first debate, there were 42 national election polls, and George Bush led by an average of 6.1%, breaking the 50% barrier in 18 out of 41.
In a sample of 100 news stories by the national media, 68% of the time the race was referred to as "a toss up" or similar description.
In 2012, in the 86 national polls between the convention and the first debate, Obama lead by an average of 3.2%, and in a sample of 100 national news stories, 79 referred to Obama as leading.
It is easy to check this for yourself, go back to the 2004 poll archive on RealClearPolitics (or any other site) and verify for yourself.
The media reports what they want to see.... go ahead... check this out for yourself.... I dare you....
And guess what? It was a tossup that came down to one or two states. The national polls are useless. We vote by the electoral college if you remember and we all knew it was coming down to OH, FL, PA, etc, the same states we're battling over eight years later.
My point was not on the actual election, but on how the media reported it...
If Bush is leading by 6.1% is a "tossup", then how is a 3.2% lead "Obama leading".....?
It is very similar to the 2000 exit polls where states won by Bush and/or Gore by essentially identical margins would be called for Gore in minutes, while states won by Bush would often take hours, even though the actual margins and data were very similar....
Seriously, do think Florida getting called for Gore before the polls even closed in the Panhandle was an accident?
No, it wasn't an accident at all. It was based on accepted procedures at the time for calling states. If you are saying that it was wrong to call a state before all of the polls have closed in that state, I would agree with you, and the networks changed the procedure which up to that time only required 75% of the polls to be closed in a state in order for it to be called.
The data in the states called for Bush and Gore was not "very similar". Does that mean that exit polls are flawed as a method of calling states? I wouldn't necessarily disagree with that, though the overall track record of such calls is excellent, the multiple mistakes in Florida with it being called for both candidates before it should have been notwithstanding. I would point out that the mistaken call for Bush was far more damaging to Gore in the subsequent recount since it added to the public perception of Bush as the "rightful" winner of the election, given how many people went to bed on election night believing Bush to be the winner.
I would be interested in seeing a link to the study you mentioned. Did they take into consideration the timing of each of the stories throughout the time period listed, and the timing of the polls conducted during that time period? Simply looking at an average isn't necessarily a valid way to study the issue. Also, why only a "sample" of 100 stories? Why not study all stories? It's not as though it would be that difficult, using Google, to study more than a random sample of 100 stories (giving the study the benefit of the doubt and assuming the stories and their sources were chosen randomly).
In addition, "leading" and "tossup" are not necessarily mutually exclusive terms. I would be interested in seeing if the study controlled for this factor.
