| Okay, so when I was in college, one of the classes I tutored and later T.A.'d was Statistics. One of the things I tried to impress upon every student during the unit on polling that a poll is only as good as the questions asked. After the 1992 presidential election, I made a transparency for the overhead projector (shut up right now with the old lady cracks, got it?) that said in big, bold, red letters "IT'S THE FRAMING, STUPID!!!"
It is because I speak pollster that I am absolutely going bonkers over the polls that keep showing confounding results, and have come perilously close to rendering myself unconscious in fits of banging my head on my desk.
It is like they are being purposefully obtuse when they persist in asking the questions the way they do.
National Survey of 1,000 Likely Voters
Conducted 14-15, 2009
By Rasmussen Reports
1* Suppose that Democrats agreed on a health care reform bill that is opposed by all Republicans in Congress. Should the Democrats pass that bill or should they change the bill to win support from a reasonable number of Republicans?
28% The Democrats should pass the bill
59% They should change the bill to win support from a reasonable number of
Republicans
13% Not sure
2* Will the health care reform plan that emerges from Congress truly be a bipartisan product, or will it mostly be what Democrats want?
18% Truly a bi-partisan product
65% Mostly what Democrats want
17% Not sure
3* Is Republican opposition due more to the actual contents of the health care reform plan or to partisan politics?
35% The actual contents of the healthcare reform plan
42% Partisan politics
23% Not sure
NOTE: Margin of Sampling Error, +/- 3 percentage points with a 95% level of confidence
Those are indeed the percentages that Rassmussen, which skews right, gets when they frame the questions the way they do. But if they were to ask the questions in a way that conveys the policy implications - i.e. If the question conveys that the price of bipartisan support in congress means sacrificing the public option, a whole different poll result is obtained.
QUESTION: Which of the following scenarios do you prefer/ do you prefer? (ROTATED): Getting a health care bill with the choice of a strong public health insurance option to compete with private insurance plans that's supported only by Democrats in Congress, OR Getting a health care bill with no public option that has the support of Democrats and a handful of Republicans? [ROTATED = the order the options were presented in - rotated]
When the pollster first asks if the respondent favors the public option, and the question is asked in a manner that conveys the implications of each position, 52% of those responding to the poll want the public option, even if it means the legislation is Democrats-only.
It infuriates the hell out of me when pollsters - who took a hell of a lot more inferential statistics than I did - frame their questions in a sloppy manner. Pollsters are the scientists of the political community, and they do a disservice to both their field of study and, frankly, their nation, when they don't ask responsible questions, but instead offer up head scratching, seemingly at-odds results that cast a pall on all polling.
They ought to be ashamed.
Post Script: Oooh. Glad I surfed a bit before posting. I see Greg Sargent has been irked by this stuff, too. I'm going to go ahead and declare a case of "great minds think alike" and get on with my day. |
***************
Best New SoapBlox - 2008
