Just the other day I heard a pollster explaining what happened in the recent national election polls that were almost as wrong as the polls four years ago when Hillary was a “walk off winner over President Trump”.
For any poll to be accurate there are two big things that have to be considered: 1) representative sample pool; and 2) accurate responses. There are a lot of other things to worry about that are a little less important. Like how do we get responses? Polls nowadays use mostly landline telephones although there are a few other ways.
With landline phones there are questions like, “Who do you call?” and you can’t use cellphones, even if you can get the numbers to those phones because, I understand, some people have to pay for every minute they are being used and most would not want to pay for the privilege of talking with a pollster. Are the ZIP codes of the people being polled important? What about the Area Codes?
What about socio-economic strata? Age? Political affiliation? Eye color? Height? Do they like long walks on the beach? So, pollsters control for all these factors that may, or may not, affect a political leaning? Or are these controls only used to promote the desired responses of the entity financing the poll?
What statistical tools will be used to select the target subject pool and to evaluate the responses? What protection does the poll have from certain responders that typically lie to telephone pollsters as a matter of conscience? All important questions, but even more important is the composition of the poll questions: “Do you favor Ms. Sally Doakes’s program to refocus our government to look after our citizens OR do you favor Mr. John Doakes’s poorly conceptualized and grossly overbudgeted program to limit democracy?” (I’m going for Sally.)
One use of polls is to tell campaigns how their candidate and their issues are performing in the eyes of the voters and what adjustments should or can be made to attract more votes. If this means more clearly delineating the goals, experience, dedication, and so on of a candidate, or more clearly explaining the issues, excellent. Although this seldom seems to be the result of polling.
Even if polls were 100% accurate, should we attend to them when they tell us the percentage that favor or disfavor a candidate or issue but never share the reasoning of the responded if it was reported? In my estimation, these polls are not only deceptive, but they cheapen the electoral process because they report elections simply as popularity contests. The cascade of polls presented to us seems to me to be nothing more than “band wagoning” – telling us who is leading in popularity so we can join up with them. Are we that interested in being on the winning side regardless of the candidate or the issues? Not being “LOSERS”?
In the long run it might be better for our elective process if campaigns were to expend more resources on educating the electorate as to what the candidate stands for and less on polling to see what the candidate has to say to get elected. Perhaps explain the issue more carefully rather than hide its intent in endless verbiage or inundate us with mindless celebrity endorsements.
Would that make for better elections? What do you think? Let’s take a poll.