Bob Mercer’s lament about the absence of reliable polling data on South Dakota’s statewide Republican primary contests for governor and U.S. House gets me wondering: do pre-election polls matter?
Mercer notes that as we shift to cell phones and screen our calls, good polling data has become more costly, and South Dakota media aren’t willing to pay:
We don’t have any data-driven reply to how Dusty Johnson, Shantel Krebs and Neal Tapio stack up, or who’s gaining, or who’s not, as they scrap for the nomination.
In the Republican contest for governor, there have been numbers. But they come from previously unknown political firms, based outside South Dakota, who lack a record here.
…In the House contest, we have no idea other than what we see and hear and feel.
And in the governor’s race, we don’t know the motivations. We also don’t know whether people with vested interests paid for the surveys.
We are in a different time [Bob Mercer, “Surveys Became Unreliable with Arrival of Cell Phones,” Black Hills Pioneer, 2018.05.19].
I’ve covered plenty of polling results (including the two bogus polls released so far this year in the Noem/Jackley tilt). They make good blog fodder and fortune-telling. But do campaign-season polls serve any practical public good? Does knowing how 500 or 1,000 of our anonymous neighbors plan to vote help any of us decide how we should vote?
I can see how poll results help candidates and hangers-on make decisions. If a Noem/Jackley poll shows that Marty will beat Kristi on June 5 58% to 42%, then Marty’s people know they have jobs for at least five more months, Kristi’s people know they should start applying for jobs elsewhere, and Billie Sutton knows he can pre-order 100,000 postcards on EB-5 and GEAR UP. Less cynically, good polling data can help campaigns understand which messages are working with which segments of the electorate. But do the current polling numbers matter to anyone outside the campaigns? If they don’t, I can see why KELO-TV and that Sioux Falls paper might not want to spend money gathering data that only benefits campaign insiders.
But I’m open to argument and evidence to the contrary. How does polling data about campaigns help you, Mr. and Mrs. Voter? Do we learn important information about the electorate and the issues from reliable, scientific surveys of public opinion about candidates before the election?
Just as science and the defining and verification of facts has been dismissed by many in American life. it has been eliminated as a requisite in polling. Many polls are scams. They have little in the ways of scientific protocol.
When I was a working newspaper editor, I was assigned the role of “polling coordinator.” As a poll was a very expensive and time-consuming proposition, a number of news organizations would join together to sponsor a poll. The coordinator’s job was to haggle with other media representatives and the polling organization over what questions to ask and how the results would be published so that no sponsor received an advantage over others and to insure an accurate reporting of the results..
The polls I was involved with were done by a organization at the University of Iowa that grew out of George Gallup’s work there. One of the hardest aspects was defining the sample of people polled. The term “random sample” is misunderstood in that regard. First an overall sample is created which conforms in relevant characteristics to the general population. For example, if religion was a relevant factor, 1,000 people in a poll group would reflect the proportions of the religious distribution of the general population (and this would further correlate with other factors). From that group, 500 or some other lesser figure would be randomly selected to be the actual respondents in the poll. The statistical
analysis of the poll is dependent upon knowing how the randomly-chosen sample deviates from the characteristics of the general population being polled.
Back then, the polls were preferably administered by the poll-takers in person. When telephones were used, an appointment was made in advance so the subjects could arrange the time in order to respond completely to the questions. Often, a preliminary poll would be made in order to test the efficacy of the questions, which could be refined and sharpened to eliminate any ambiguities they might have for the poll respondents.
Bob Mercer is right that phones have affected the reliability of the polls, but the biggest factor is that the scientific controls that make polls “scientific” have been eliminated. They are too expensive and time-consuming to provide the factual basis upon which legitimate statistical inference depends.
I think a very recent factor that’s diminished respect for political polls is election cheating.
Polls will show an expected result, but the votes turn out to be very different. That’s happened since the no-paper-trail-Diebold-machines went into use. That and hacking, Russian and otherwise, have resulted in stolen elections. People have made video recordings of their votes changing.
Attacking poll integrity is another part of the GOP attack on truth and facts,
Good one, Debbo.
Debbo,
Please link us to this election cheating.
Google is a wonderful tool.
The majority of polls are boguser than most. I get calls almost daily for these pollings and they are invariably skewed with questions like “Shantelle clubs baby seals when they’re not even looking. Does this information make you more or less likely to vote for her?” “Dusty once threw a snowball at a nun. Does this information make you more or less likely to vote for him?” The best thing these polls do is entertain those who answer the questions, line the pockets of the pollsters, and make one candidate or the other feel better about themselves, if only temporarily.
Interesting, David, that polls were worth the expense to news organizations in the past but not now. One could argue that it would be easier and cheaper to collect information from a well-selected representative sample of citizens than it was 30–60 years ago, yet the decline in ad revenue has pushed polling below the cost-effectiveness threshold for most media.
Debbo, do you think the media have backed away from polls because election cheating has foiled their predicted results too often? Could we argue that suspicions of election cheating are all the more reason to conduct multiple rigorous polls to provide evidence of possible ballot manipulation? But if we make that argument, do we open the door for wealthy candidates to spend money concocting rigged polls that they can wave as evidence that their honest losses at the polls were really results of election-rigging by their opponents?
Good questions Cory.
First to the last one: Haven’t candidates done their best to rig polls to their advantage since there have been polls? Yes, the election cheating could make the crooked polls even more prolific.
I think election cheating and the far right/GOP attacks on the press that have been ongoing for 30 years and more have had a very deleterious effect on unbiased media. The media has steadily backed away from presenting facts and sticking by those facts. Instead, they often present “both sides” as if each uses equally valid sources, and as if there are only 2 sides, 2 extreme opposites. Polls are one of the items news media has backed off from. Rather than posit that something may have been wrong with the election itself, the media seeks to solely blame the polling.
I’m all in favor of “multiple, rigorous polls to provide evidence of possible ballot manipulation.” As I stated earlier, I think it already has.
I agree that the media needs to be more aggressive in its discovery, presentation, and defense of facts. I got on all three Sioux Falls TV stations yesterday, in addition to two print interviews in the past week, because they were producing debate pieces on Amendment Y. There are multiple parties advocating Y. I’m the only person they can find in the entire state, apparently, willing to go on camera/mic and offer reasons to vote NO on Y. On the one hand, those reporters are seeking facts that they may not be getting from the proponents, which is good. On the other, one could argue that they should be emphasizing that Heidelberger isn’t “leading the opposition”, as if I had some equal movement against Y, but that I am the only person making noise about it, which suggests some sort of imbalance.
Dang, maybe the false equivalency and safe neutrality are simply allowing one lone campaigning in Aberdeen to get an inordinate amount of press.
Don’t get me wrong: I want the media to present both sides of the Y debate and of every other major public policy debate. But much of the coverage, of Y and every other election, is “Pro said, Con said,” without deeper analysis to come to a conclusion about who is actually right.
Now I’m still not sure that pre-election polls on candidate races contribute to that understanding of who is actually right. When we’re polling candidates, the press simply expands the “Pro said, Con said,” to “60% said, 40% said,” which tells us who may win the election (useful, as Debbo notes, to check for election fraud) but not who should the election. Polling on issues may guide policymaking—”80% of Americans support background checks, so Congress ought to support background checks”—but the majority can still be wrong, and legislators of principle can still justify voting against what the polls say is popular at the moment.
After all, the media coverage suggests that I’m the only vocal opponent of Amendment Y, which otherwise enjoys the support of numerous public figures. In this case, those public figures all happen to be wrong.