A more frequent publishing of Rich Galen's take on politics, culture and general modern annoyances. This is in addition to MULLINGS which is published Mondays, Wednesdays & Fridays at

Friday, April 2, 2010

By the Numbers

From Newport Beach, California

  • For those of you who may have missed the class on political polling here's the basic rule:
    You treat poll numbers like we used to treat Olympic figure skating scores. You knew the French and the East Germans were cheating so you threw out the highest and the lowest scores and whatever was left was probably an accurate reflection of the skater's performance.

  • Polling is the same way. Not that anyone is cheating, but if you look at a list of polls on the same subject, taken at about the same time, throw out the high and the low you probably have a pretty good idea of what is going on.

  • The polling for President Obama since Healthcare passed the House on March 21 is instructive.

  • The Quinnipiac poll shows the President's job approval at 45-46 which is the lowest approval number among the seven polls listed on the RealClearPolitics poll summary page this morning.

  • The best approve-disapprove number is from the Washington Post poll which has the President at 53-43 (plus 10) telling me the poll was done among the reporters who happened to be hanging around the city room that day.

  • I know that isn't true, but it is just after 4 AM here in California and I needed a smile.

  • The worst approve-disapprove number is from the Rasmussen poll which as of Friday morning was at minus 6 (47-53). Although Democrats' eyes roll and their chests heave heavy sighs at the mention of Rasmussen (because they believe it tilts toward the GOP) it is the only poll among those listed which measured Likely Voters as opposed to Registered Voters (Marist, CNN and Quinnipiac) or Adults (Washington Post and Gallup).

  • Every polling company has its own method of determining whether someone is a likely voter or not. This is called the "screen" and as we get closer to the mid-term elections, which will be held exactly seven months from today (all of which save April and June hath 31), you will hear that word in a phrase like "they use a very tight screen" to determine who is likely to vote and who is not.

  • As long as we're doing a general overview of polling, other than choosing the respondents, there are several factors which can influence how a question will be answered. The first is its position in the poll.

  • A good polling firm may ask early in the poll, "Do you approve or disapprove of the way President Obama is handling his job?" (or some variant). Then they will ask about how he is handling specific issues: the economy, jobs, foreign policy, energy, and so on.

  • At the end of that series of questions they may ask the job approval question again to see if, after thinking about individual issues, there is a change in the overall approval of the President.

  • The major bit of skullduggery that can occur in polling is now the question is asked. Asking what a respondent thinks about how the President is handling his job is far different than asking: "On a scale of one-to-ten, do you think President Obama is doing a good job as President or do you agree with the Tea Party and Sarah Palin that he is doing a poor job?"

  • That is why almost all public polls done by any legitimate firm will not only publish the results of the poll, but will let you read the actual questions.

  • Finally, there is the matter of looking deep into the polling numbers (words like "cross-tabs" will be tossed about) and this is not for the faint of heart.

  • For an excellent example of how to do this properly look at the essay Newsweek's Howard Fineman published earlier this week on the dangers facing Democrats in November based upon his looking at a number of different polls on health care.

  • Last thing about polls today: When someone says to you: "How can they ask 500 people a question and know what the whole population is thinking?"

  • You might say, "Last time you went to the doctor for a blood test, did the technician take out all of your blood, or just a small vial or two which represented the rest of it?"

  • On the Secret Decoder Ring today: Links to the RealClearPolitics page, the Rasmussen page, and to the Fineman essay. Also another Mullfoto proving the arrival of Spring and a really interesting Catchy Caption of the Day.


    Post a Comment

    << Home