Being asked “what if” questions in a poll survey can affect our behavior if we are not aware of it, conclude US researchers in a paper published recently in the journal Organizational Behavior and Human Decision Processes. The researchers describe such hypothetical questions as “wolves in sheep’s clothing”, and make reference to how “push polling” uses them as a tactical tool during election campaigning.

In their study, researchers from Duke University’s Fuqua School of Business, the University of Alberta School of Business, the University of Southern California and Stanford University found that:

  • Hypothetical questions (for instance those starting with “what if …”) increase our brain’s accessibility to the positive or negative information in the question. For example, if we are asked, “what if you knew that Mr Smith was a convicted criminal, would this make you less likely to vote for him?” then we would retain that “negative”, albeit hypothetical description of Mr Smith, and readily access it when we next consider the subject of Mr Smith.
  • Hypothetical questions influence behavior according to the “valence” of the question (eg whether it comes across as saying something positive or negative about the subject).
  • Hypotheticals exert a stronger influence when they are consistent with our existing knowledge about the subject, but, their influence is weaker if we are aware of how they affect us. Their effect also weakens with time.

Gavan Fitzsimons, professor of marketing and psychology, who among other things teaches consumer behavior at Fuqua, told the press:

“Hypothetical questions are essentially wolves in sheep’s clothing.”

“Seemingly innocuous questions can make positive knowledge accessible while negative questions can make negative knowledge accessible. In other words, being asked hypothetical questions that are consistent with our existing knowledge or our preconceived notions has a biasing effect on us — without our knowledge and without our consent,” he added.

First author Sarah Moore, assistant professor of marketing at the University of Alberta, said:

“Political pollsters are very aware that the language they use in so-called ‘push polling’ can not only influence the way questions are answered, they can also push people away from one candidate and pull them toward another.”

For the study, Moore and colleagues carried out a series of surveys to explore voting choices, legal decision-making and consumer behavior.

One survey asked respondents a series of hypothetical questions about politicians: some of the questions sounded positive and some sounded negative. The other surveys asked questions about the nutritional value of sugar substitutes, snack food preferences, and the defendants in a mock court trial.

In the survey about politicians the researchers asked respondents if learning a politician had been convicted of accepting a bribe would make them less likely to vote for that person. And they asked another group of respondents if learning a politician had refused to accept a bribe would make them more likely to vote for that person.

37% of respondents who answered the “negative” question said they would be less likely to vote for the politician, while 83% of those who answered the “positive” question said they were more likely to vote for the politician.

“In both cases, the bribe context was purely hypothetical, and yet the two groups’ voting intentions differed dramatically,” said Moore.

In another survey, a group of prospective jurors waiting to serve on a case were invited to take part in a a simulated jury selection and imagine they were being screened by attorneys at a criminal trial.

A subgroup of jurors answered a series of hypothetical questions that alluded to the defendant being a member of a gang. Another subgroup had a similar set of questions that excluded reference to the alluded gang membership. The group exposed to the questions that included hypothetical references to gang membership was more likely to find the defendant guilty and recommend sentences harsher than the group that was not exposed to those questions.

The researchers then carried a similar experiment with a different group of prospective jurors. This time, the jurors were briefed: they were informed that the questions had been put together by attorneys for the defense and the prosecution, and in their role as potential jurors, they should not draw conclusions about the case from these questions.

It appears that the respondents in this second experiment were able to correct for the influence of the negative connotations in the hypothetical questions, because their responses showed they were less likely to find the defendant guilty and recommended shorter jail terms.

The researchers conclude that by manipulating factors “known to enhance (consistency, elaboration) and attenuate (awareness, delay) accessibility”, they showed “these factors moderate the influence of hypothetical questions on individuals’ voting choices, legal decision-making, and consumption behavior”.

Co-author Baba Shiv, professor of marketing at Stanford, said their study shows:

“… hypothetical questions can influence our actions, but we’ve also confirmed that when we are aware of the effects of hypothetical questions, we can correct our biases.”

“Public education is needed to raise awareness of how hypothetical questions can sway our actions,” said Shiv.

Written by Catharine Paddock PhD