(vanity) In 1980 two weeks from election polls had Jimmy Carter 47 -- Ronald Reagan 39

Lubbock online ^

[yefragetuwrabrumuy]

A good way to reduce this perfidy is to create pollster ratings. Once both parties nominate their candidates, then track the pollsters on about a monthly basis for how accurate they were *based* on the actual vote. Four months, five poll checks: August, September, and twice in October, then the beginning of November.

Say the Republican beats the Democrat 52% to 47%.

A pollster in August says the Democrat is ahead by 20%. In September by 15%. At the start of October by 12%. The second week of October by 5%. Then the first of November tied, because the media loves a “horse race”, and the pollsters want to sell them one.

But with the *next* presidential race, that pollster is then “handicapped” for their poor performance. And it can be put next to their “margin of error” number as an offset.

So that August if they say the Democrat is ahead by 20%, it effectively means that the candidates are tied. September, they say the Democrat is ahead by 15%, it means the Republican is ahead by 5%, etc.

Granted this would have to be tweaked a lot by a good statistician, but it would be a good way of judging the credibility of pollsters.

[SMGFan]

But then came the only debate on Tuesday October 28

[pangaea6]

Does anyone know the party breakdown of the 1980 or 2012 polls folks on here are talking about. I think right now most msm has dems at +7 or higher