Straightening the Post's skewed health care polling

In today's Washington Examiner, Byron York had the same question I did about The Washington Post's Sunday poll claiming Americans were "split" over ObamaCare instead of opposed -- why didn't the Post reveal how independents felt?

York finally got the partisan breakdowns from poll director Jon Cohen and they reveal independents bitterly oppose ObamaCare and believe Obama is lying on virtually every claim he made.

York didn't address the other problem with the poll -- how they deflated opposition by undersampling Republicans by a whopping eight percent.

The Post's sample was of 1,000 adults was composed of 34 percent Democrats, 24 percent Republicans and 38 percent independents (with four percent saying "other" or "don't know.") That conflicts sharply with current partisan identification trends, which put the country at 35.1 percent Democratic (and falling,) 32.1 percent Republican and 32.9 percent unaffiliated.

The Post undersampled Republicans by eight percent. Making up for it by oversampling Democrats would be blatant bias, but filling that gap in with independents is harder for some to spot.

Now, York writes, "On the support-or-oppose question, 82 percent of Republicans oppose the new law (73 percent strongly), while 76 percent of Democrats support it (56 percent strongly). Among independents, 56 percent oppose the new law (44 percent strongly), while 40 percent support it (26 percent strongly)."

While York gave full numbers for independents, but did not provide Republicans in support or Democrats opposed, an estimated breakdown of the Post's sample goes like this.

Democrats: 258.4 support, 81.6 oppose
Republicans: 43.2 support, 196.8 oppose
Independents: 152.0 support, 491.2 oppose

That comes out to 453.6 support (45.4 percent) and 491.2 oppose (49.1 percent.) Almost exactly the Post's final numbers.

So let's take the Post's numbers and figure out what the real number is when you use an accurate universe, and not the Post's light-on-Republicans universe.

Assuming support and opposition numbers aren't skewed to favor Obama, as the Post's sample is, by applying the Post's by-partisan-ID numbers to a partisan-accurate universe of voters you find:

Democrats: 266.76 support, 84.24 oppose
Republicans: 57.78 support, 263.22 oppose
Independents: 131.6 support, 184.24 oppose

That comes out to 456.1 support (45.6 percent) and 531.7 oppose (53.2 percent.)

So had the Post not undersampled Republicans, but used real numbers, the story would have been that 53 percent of Americans oppose ObamaCare, not the supposed "46 percent to 50 percent 'split'" the Post claims.

That 53 percent to 46 percent margin being, of course, the same margin of victory Obama got when the Post reported he won a "landslide" in 2008.

That's more on par with virtually all other polls, and an indicator of just how skewed the Washington Post's polls are.

Not a 50 percent to 46 percent "split," but a healthy 53 percent to 46 percent margin against. Or as the Post puts it, a "landslide."

No comments:

Post a Comment