Robert Moran, noting the Washington Post does need to do some clarifying, disagrees with some of the Post poll criticism from this morning:
I think the party ID criticism of the Post poll is incorrect.
If you look at the survey results, it appears that, as is usual in our business, the “leaners” question was asked of people not stating a party ID.
The Post’s data report doesn’t explain this, but the math seems to work out.
For example, the “leaners question”, is typically only asked of people who did not say Democrat or Republican. In this case, that would be 39% of the sample as 61% said either Democrat (30%) or Republican (31%).
Among this uncommitted 39%, 48% appear to have leaned Dem and 34% leaned Republican, with the rest as hard indies.
39% * .48 = 18.72% + 30% = 48.72% Democrat or Lean Democrat
39% * .34 = 13.26% + 31% = 44.26% Republican or Lean Republican
This comes very, very close to what the Post had as the “Leaned Party ID” — 48% Democrat / 45% Republican. I have a feeling the discrepancy is due to rounding.
Bottom line, I don’t think there is any trickery going on here in the party ID question on the survey — just poor reporting of the topline data.
But, I think detractors of this survey have a good point when they argue that political surveys of adults are a waste of time. This has been a bone of contention in the industry for a long time, but most Republican pollsters prefer only to survey people who are actually registered to vote for political surveys. My take on this is that if an organization wants to poll adults in its political surveys – which they do mostly as a way to keep costs lower – they should also report the data among the x% of the survey who are registered to vote. This would give us all a better read. As it is, it’s impossible to compare this survey’s demographic data to the 2004 exit poll demographics because we’re looking at two different universes.
Please keep in mind that most media produced polls are not very good, are done by the lowest bidder in the cheapest possible way, and are generally biased in their wording and flow. The best polling is the polling done for private clients that is never released, because this polling is done to find where voters really are and to make campaign decisions, not advocate a position or add juice to a story.