Public relations practitioners, and others, received a reminder last week not to put too much faith in surveys. In the Wisconsin recall, pre-vote polls benchmarked the election as extremely close with a 50-47 margin in favor of Gov. Scott Walker – within the margin of assumed error.
The actual result? Walker won easily by a 9-point margin. The surveys missed it – big time. So is something wicked in Waukesha? Hardly. Some thoughts:
* Whom did you ask? To be accurate, a truly random sample has to be built very carefully. Even such assumptions as using only respondents who have landline telephones skew accuracy.
* As surely as chemistry or physics, there is an observer effect in surveying. Respondents may feel an obligation to give a certain answer, or may give an unexpected or incorrect answer just to be perverse. My daughter participated for a year in a nationally recognized television rating service. She admits that box on the TV made her want to skip Entertainment Tonight and watch PBS instead.
* Burnout. I heard a piece on radio a few days ago with workers who had done exit polling at Wisconsin precincts. One made the comment that it became obvious some voters were “avoiding people with clipboards” as they got in their cars. My hunch is Badger State voters may have already endured endless surveys and were just surveyed-out by the time they made it into voting booths. I live in a Texas state senate district that had a very intense Republican primary in May and I received daily phone calls, plus at least one visitor to my front door and mail, all asking my opinion. It became tiresome and I started hanging up on people, thereby skewing their results.
So are surveys worthwhile? Yes, but only if used properly. And always ask yourself if the results provided are prophecy or puffery.