WHY DON’T PEOPLE SAY WHAT THEY WILL REALLY DO?

We field multiple questions related to why we don’t directly ask respondents their likelihood of doing something in the future. The simple answer is that we can predict their future behavior more accurately by modeling it than by asking the question directly. How can this be? Why don’t we get an “honest” answer when we ask what someone will do in the future? For the researcher, this is the problem of measuring Sentiment*, or the strength of commitment a respondent assigns to a statement about their commitment to a future behavior such as voting for a particular candidate or buying one brand instead of another. 

Market researchers, pollsters and economists have worked for decades to come up with “adjustments” to ”likelihood-to-buy” and “voter preference” questions to make the results they obtain more accurate. Most of these adjustments decrease the actual likelihood of doing something in the future compared to that which respondents said they would do. In other words, the researcher, economist or pollster did not “believe” the stated future behavior would occur as often as that given by the respondent. So, the real question is not, “Why don’t people do what they say they will” but “Why do people overestimate their likelihood of doing what they are asked?”  

The answer is one of four biases based on a behavioral economics framework:

Pleasing the Interviewer - The most obvious explanation is that the respondent wants to give the “right” answer to the firm or interviewer asking the question so that they can be interviewed and paid again. Often there are subtle clues in the questionnaire or respondent screening to which the respondent is sensitive for this very purpose. What appears to be a wonderful research result becomes, in reality, a false positive as in the Brexit and 2016 presidential election polling.

Picking the Less Risky Alternative – Why let the interviewer know your true feelings if there is social risk in doing so. For example, saying that you will vote for staying in the EU eliminates the possibility of explaining a vote for Brexit. A similar effect is observed in higher stated likelihood to buy the more expensive market leader by not having to admit to the interviewer that “I’m too cheap to buy the more expensive brand.”

Don’t Want to Tell – Telling the truth may be perceived as “giving something up” which could be “used against you” for those who do not have a history of being rewarded for transparency. In this case, anything other than the truth has more utility for the respondent.

Don’t Know – Since the direct question is being asked, the expectation is that the respondent will have an answer ready. Therefore, if the respondent doesn’t know, there is more utility in giving any answer rather than admitting the lack of an answer. This makes “not knowing” similar to that of Picking the Less Risky Alternative.

Behavioral Design from Behavioral Science Lab uses a unique proprietary methodology which models the buyer’s or voter’s decision system. This approach produces a more precise estimate of future behavior than by asking a direct question and allowing the biases described above to degrade the result. Let us know if you would like to learn more about how Behavioral Design can help you predict future behavior in this utterly new business environment.   

Tim Gohmann Co-founder and Chief Science Officer 805.405.5420 | tim@behavioralsciencelab.com Behavioral Science Lab, LLC 500 West Second Street, 19th Floor, Austin, Texas 78701 USA  www.behavioralsciencelab.com 

* The Thorny Problem of Measuring Sentiment, BSL 2020.