Editor’s note: Paul Johnson is a former international market researcher who worked in advertising agencies and as a market research supplier and was also an adjunct lecturer in marketing. He is currently retired.

The first time I read a newspaper article quoting some polling numbers related to a national election, I admit that I was quite uncomfortable. In fact, it immediately reminded me of James Joyce’s protagonist Stephen Daedalus in Portrait of the Artist as he divided art into proper and improper categories. Improper art he further divided into two classes: didactic and pornographic. Now, Joyce certainly has been associated with the latter but that is for theocrats to debate. However, the use of polling results distributed to media is tantamount to distributing pornography. Once the source becomes familiar to those who have sampled the material, they are wantonly waiting for the newest and most shocking factoids to be graphically depicted. This to me fits Joyce’s description of pornography: material that is meant to motivate someone to do what they would not normally do. From my vantage point as a researcher, this is diametrically opposed to the outcomes that research information is intended to provide.

This early 2016 election cycle is a perfect example. Numerous Republican Party candidates have been publically portrayed in articles using polling numbers to describe them as a fractured group in disarray, given the non-establishment entries who have “moved ahead” of candidates with more traditional political tracks on their resumes. Similarly, among Democratic Party candidates, a progressive candidate is challenging the presumptive leader in early primary and caucus states.

The two improper applications of research which have disturbed me are sampling and reporting. I will explore both of these issues with the intent of applying these observations to our clients’ projects in both the private and public sectors.

Let’s take a look brief look at the history of using polling data meant for public consumption in political campaigns. In the beginning, research campaigns conducted public polling “on the cheap” using random samples of adults. State and national campaigns began to publically question the sample and opposing campaigns began to cry foul (and rightly so) because these random samples were throwaway numbers. A political campaign does not care about the opinion of someone who is either not an active registered voter or not registered to vote.

Once campaign managers found that random samples would only bring further challenges from opposing campaigns, polling firms began using a registered voter sample. This added cost to each poll that screened out non-registered voters. Further, the samples had to be divided between voters of one or the other major party or neither. This created a further conundrum in how to deal with those not aligned with a major party.

Increased confidence levels

In the private sector, you can see the historical vestiges of samples used in taste tests by major beverage companies like Pepsi and Coca-Cola. Cola drinkers were used to sample new beverage offerings such as New Coke in blind taste tests. The publicly reported results showed that cola drinkers favored the new Coke offering about 1 percentage point better than Pepsi. Now, given that one percentage point in beverage share translates into huge numbers in sales and revenues, Coca-Cola saw an advantage to introducing a cola that gave it a greater share than Pepsi. However, we are all well aware of how brand can trump product characteristics. In the same line, if only your candidate’s party results are reported, the confidence levels will increase and so, public poll reporting became suspect and open to challenges from opposing campaigns.

For instance, in statewide election polling, samples are drawn from likely voters in precincts that have a stable predictive history of voting for specific parties. Samples from these precincts are polled at regular intervals during the period leading up to an election. These results are tracked to show the momentum a candidate or opponent might have leading up to Election Day, providing a portrait of how the election outcome should look like for good or bad for their candidate.

Sampling has gained a significant place in political polls distributed for public viewing. The press releases generally lead with “among likely voters” or “among registered voters” to blunt the opposition challenges. However, the general public does not discriminate on polling samples and use the numbers in a manner that the campaigns desire, which is to shape public opinion not reflect it. While there is little value for opposing campaigns to challenge political polls publically, it should be noted that no campaign manager pays any attention to public polls save for how their opposition is using the poll to craft a message for their candidate.

Reporting

While mudslinging is not a new political tool, poll-reporting is. During the 1976 presidential campaign, there was not an extensive use of polling data in newspaper reporting. However, knowing the public sentiment during that time in the wake of Watergate, there was a significant difference in the word choices reporters and commentators used in describing then President Ford and the challenger, Jimmy Carter. Suffice it to say, the differences in words used to describe the candidates painted a bleak portrait for Ford and one of an “outside the beltway” honest challenger, Carter.

Taking into account the four-decade difference, the news media narratives today are shaped – as well as confounded – by polling numbers that provide a snapshot of how the data are used to support a storyline. Now, this is significantly different from privately commissioned polling (or any other research) for decision-making since the results are not generally made public. The early polling among Republican candidates has put two “outsiders” in favorable positions relative to the establishment candidates. The reporting has taken two distinct stances on the poll leaders:

  1. The frontrunner enjoys an advantage of public awareness and name recognition. The commentary suggests that these advantages will fade as more candidates present themselves in public forums.
  2. Public distrust of beltway insiders has suppressed the polling numbers for candidates who are known to have political ties to Washington or governors are only known regionally and need time for voters to form an opinion of them and their stance on important issues.

To date, the Republican frontrunner has gained ground as have other outsider candidates so the poll report writers and news reporters have had to resort to speculating on why their observations have not seen the outcomes they were expecting.

Similarly, among the Democratic candidates, the frontrunner has seen much media attention but is bereft of any gains in polling numbers.

As a former research reporter, I have often taken a position of trying to find a story in the numbers but with an advantage, as every study has objectives. A writer can use those objectives and search the data to find the information that either shows the objectives have been met or where those objectives have fallen short. In two instances the significant findings were that there was little correlation between what consumers said they would do and what they actually did. Surprised? You needn’t be. This reflects the fluid nature of consumer attitudes. Thus what we see in political public polls today will likely be far different by November 8, 2016.

So, if polling sampling and reporting during this election cycle don’t confound you, don’t disturb you and don’t affect the way you intend to vote, that’s great! However, if you feel a little bit uneasy, be assured, this will pass and you can simply turn the channel or Google another topic such as, “What did Taylor Swift wear to the Golden Globe Awards?”