I've written in the past about the huge differences in polls on the same question taken at about the same time. Well, it's happening again. In the last week, four pollsters asked voters which party they will vote for in the next congressional election. This so called "generic congressional vote" question is a mainstay of the political pollsters. Three of the four polls questioned registered voters while one questioned adults. The results were all over the place. The Democrats were ahead in all polls, but the margins were by 5, 7, 9 and 17 percent. Normally, polls of adults show a higher performance by Democrats, but in that poll, the margin was 9 percent.
So how can this be? There is no way to square the poll that found a 5% margin with the one at 17%. The difference is much more than the margin of error. Quinnipiac at 17% seems to clearly be wrong. The two polls of registered voters at 5 and 7% are reasonably close. They line up with the poll of adults at 9%, because polls of adults normally show better results for the Democrats. It's just the one at 17% that seems way out of line.
And what does this mean for the election itself? The real question if how a poll of likely voters would come out. At this point, none of the polls question likely voters. We do know, however, that the normal result, especially in midterm elections, is that likely voters are more Republican than registered voters and much more Republican than polls of adults. That would put the actual margin among likely voters at 3 or 4% for the Democrats. In other words, as of now, the outcome of the elections is still far from decided.
That leaves the question of how Quinnipiac could put out a 17% poll. If this were a one-time aberration, we could chalk it up to a poor sample or some mistake. This, however, is a consistent result in the Quinnipiac polling. For the last four months, Quinnipiac has been showing a double digit lead for the Democrats in this generic congressional poll, while all other polls save one were in single digits. Quinnipiac is clearly showing a bias or else there is a massive mistake by every other polling organization for the last half year. My guess is that Quinnipiac is skewing its polls.
So how can this be? There is no way to square the poll that found a 5% margin with the one at 17%. The difference is much more than the margin of error. Quinnipiac at 17% seems to clearly be wrong. The two polls of registered voters at 5 and 7% are reasonably close. They line up with the poll of adults at 9%, because polls of adults normally show better results for the Democrats. It's just the one at 17% that seems way out of line.
And what does this mean for the election itself? The real question if how a poll of likely voters would come out. At this point, none of the polls question likely voters. We do know, however, that the normal result, especially in midterm elections, is that likely voters are more Republican than registered voters and much more Republican than polls of adults. That would put the actual margin among likely voters at 3 or 4% for the Democrats. In other words, as of now, the outcome of the elections is still far from decided.
That leaves the question of how Quinnipiac could put out a 17% poll. If this were a one-time aberration, we could chalk it up to a poor sample or some mistake. This, however, is a consistent result in the Quinnipiac polling. For the last four months, Quinnipiac has been showing a double digit lead for the Democrats in this generic congressional poll, while all other polls save one were in single digits. Quinnipiac is clearly showing a bias or else there is a massive mistake by every other polling organization for the last half year. My guess is that Quinnipiac is skewing its polls.
No comments:
Post a Comment