By Dr. Bruce Smith ——Bio and Archives--August 25, 2024
Cover Story | CFP Comments | Reader Friendly | Subscribe | Email Us
Public polling can be useful, but it often becomes a tool for manipulating public thinking.
Never having been a math person, my skepticism toward polling has only grown more pointed over the years. With polls and samples and surveys, the usual technique is to select a pool of people to sample, then send questionnaires or conduct interviews over the phone to determine results.
The results are then subjected to different kinds of mathematical formulas to allow for variables. Once the math is done, then the responses become legitimate plus or minus a percentage point or two. The math is used to make the survey “scientific” and build its credibility with the public, with the target audience, or with whoever paid for it.
I say this results in a lot of hogwash, and here are some reasons why.
Everything depends on whom you’re asking and how the questions are phrased. Polls must use a question of some kind, and that makes them unreliable and subject to manipulation. One of the few things I actually learned in a Big Ten university school of education many years ago is that test questions are notoriously unreliable as viable tools of testing. ‘Do you think Mayor Bloat should stop beating his wife?’ ‘Did the president know about the Watergate break-in ahead of time?’ Those are very different questions than something like this: ‘Do you prefer butter or margarine on your toast?’ The first two cannot be reliable while the last one isn’t one hundred percent reliable. What if people don’t want anyone to know they eat butter? What if the responder prefers dry toast? What if they don’t care either way? What language is used for the questions? Do people understand that language the same way as other people do? Do we rely on the math people to allow for the misunderstandings that come from language?
Polls are taken with a chosen pool of respondents. Somebody chooses the members of the pool. Choosing the pool makes all the difference because the respondents are supposed to represent a wider societal group whose views matter to someone. Polls are skewed to suit the pollster, so if a pollster wants to show that the public favors soy-based products over animal protein, there are ways to skew the sample to get a preferred result.
Selecting polling groups “randomly” doesn’t help, either. Where does the list come........