Getting the poll results you want...

By David Montgomery
Published/Last Modified on Monday, Dec 21, 2009 - 04:44:53 pm CST

Polling is a really useful tool for understanding politics. But polls can also be misleading — especially when you move away from simple favorability ratings and political horse race questions to asking people about issues they might not understand.

A case in point was recently highlighted by Nate Silver, a center-left statistical guru who became famous last year for his models predicting election results on fivethirtyeight.com. Silver highlighted two poll questions about the same issue:

1. Would you favor or oppose requiring all Americans to buy health insurance — the so-called mandate — even if they find insurance too expensive or do not want it?

FAVOR: 38%, OPPOSE: 51%, NOT SURE: 11%

2. Would you favor or oppose requiring all Americans to have health insurance, either from their employer or from another source, with financial help for those who can't afford it?

FAVOR: 66%, OPPOSE: 31%.

That's a swing from -13 against to +35 in favor — a 48-point swing or 24 percent of respondents changing their mind.

The British comedy show "Yes, Prime Minister" had a good take on this:

Print this story Post A Comment

WARNING:

• Comments are not verified for accuracy nor have we verified the identity of any person’s name accompanying a comment, so please consider this as you weigh any statements made or opinions offered.

NOTE:

• Comments will not be posted if submitted in all caps, or all lower case, or do not show a reasonable attempt to follow basic rules of grammar and punctuation, nor will text-message like abbreviations be accepted, like ur for your.
• Postings must stay on the issues related to the thread and not veer off into personal comments about others. The tone of the post must be civil. For example, you can say someone’s ideas are stupid but you cannot say another poster is stupid.
• Postings must deal with matters of public concern.

Comments

1 comment(s)

caheidelberger wrote on Dec 19, 2009 6:29 AM:

" Excellent point! We can't say you can't trust statistics, but we always have to be able to look under the hood at the questions and the methods. "

WRITE A COMMENT

Use the form below to post a brief comment to this story, or respond to other readers. Please use the word count tool to assist you in keeping your remarks to 100 words or fewer.

Comments must be approved by an editor before appearing on the Web site. Editors review submitted comments periodically during the day for offensive or off-topic content before posting. Your thoughtful contribution to the online discussion is appreciated.