Special to Veritas — The following commentary by Douglas Kelly provides important insights into how much stock to put into “pollsters” and mainstream media commentaries regarding the poll results they choose to publish. Doug has been relied upon for his in depth analysis by organizations in which I have had a high degree of involvement. Doug started off his communication to me with a link to an article he came across which he presented as a good primer for “why be wary.” Doug then expands on the article from his professional experience.
Almost Everyone Misinterprets The Polling On The Presidential Election — Here’s Why
Having been in the marketing business all my life (so far) and written, polled and analysed the results of many, many polls by cross tabulation, which is a science in itself, I can attest that there are huge credibility and veracity problems when the biased media or some “spokesperson” tells you about poll results.
Polls are not to be read as a “horse race”. Polls are not meant to be read one line at a time and then the most pleasing line (response) plucked out to support the agenda of the the advocate (a trick constantly employed by the news media). They are meant to be read in whole and seen as a whole view of a subject being polled. To do otherwise, is to completely miss the point of polling.
In my position as one who has interpreted polling results for clients, I assure you that no one can make any kind of judgement or evaluation of polling results without having access to and analysing the cross-tabulations of answers to every question to discount consistently inconsistent responses based on other responses by the same respondent. So if a respondent says in answer to one question, I hate Obama, yet in a series of following questions, is found to agree entirely with his political agenda, then there is lack of veracity that undoubtedly pervades the whole respondent’s questionnaire, and it is tossed out.
Then, in addition, there are simply bad pollsters. And most, with the exception of Gallop, Rasmussen, and Zogby, are not providing good polling data. The inferior or unqualified pollsters will have systemic errors in their survey questionnaires which leads to their results not being worth the time spent to read them. As is mentioned here, one of the most common systemic errors is oversampling of a particular and significant group, which irreparably skews all the results of a poll. This error can be compounded by disregard of age, kind of work or career, income, neighborhood, proclivity to vote in this election, and quite importantly, how the questions are structured and the order in which they are asked to best elicit the most accurate answers from respondents.
A poll can be constructed by a skilled but unethical, unscrupulous “pollster” to provide any result that is wanted. Fake or false polling has actually been used as a form of propaganda by leading the respondent to the answers desired and implying that the respondent’s attitudes are the desired response. For some strange reason, interviewers find that respondents seem to want to please them.
By and large, good polling by ethical professionals can be trusted within the margin of error and the confidence factor of the random distribution of the respondent list. I suggest you be wary of all polling results announced in the news media, or by the client for whom the polling is done. There are only about three pollsters that can be believed, the rest are charlatans or incompetents. And never trust the results of an online poll, a newspaper poll or a poll mailed to you, since the responses to those are voluntary and will result in only the most extreme ends of the opinion spectrum. Meaningless.
Be sure to notice the link at the bottom of the linked article regarding Romney’s October surprise.
Douglas Kelly
