Replies

Not sure anyone is lying. One of the biggest assumptions in sampling for an election is turnout. A turnout of 2x larger than assumed does not merely double the raw vote for each candidate - there are major shifts in the votes cast by each half decile of additional turnout. The campaigns and their pollsters try to gauge turnout and voter intensity, and then forecast a turnout level. That is what they use to adjust the raw numbers coming in from the polling survey. It is difficult to do in normal years, but I think Trump truly represents something that political pollsters have never seen. In my view he is activating voters who may not vote very often. The usual stratification of turnout estimates is done on the basis of an answer to questions like "Of the past 4 presidential elections, in how many did you cast a ballot?" [recent] or "In all the presidential elections you have been eligible to vote, in how many have you cast ballots?" [lifetime]

My gut tells me that Trump is getting voters that are off the radar of the typical set of assumptions that pollsters use. That can lead to very inaccurate pre-election polling forecasts.

The other element is self service by campaigns. Bush can commission a poll for his campaign and dial in the turnout model applied to the results, and when he gets what he likes, he can release it through back channels to the press for whatever effect he would like.

Unless you have someone who gets to see the sampling plan, the weighting for turnout and the exact text of the polling questions, as well as the rejection rate of the pool of respondents - it's really hard to know how much faith to place in any given poll. The press today is naive enough to simply regurgitate polling results leaked to them, so, my first thought on this was to discount it heavily because it was the message the Bush campaign wanted to float, not necessary a reflection of real voter sentiment.