The wrong questions will give the wrong answers

People are often wrong when asked to state their preferences or predict what they'll do. William Hanmer-Lloyd explains how brands can understand human behaviour better.


In his fantastic book Dataclysm, Christian Rudder, one of the three founders of OKCupid, found that although straight men aged 25-50 stated that they desired women only a few years younger than themselves, when it came to rating the attractiveness of photos of prospective dates they consistently rated pictures of women aged 21-22 as the most attractive - regardless of their own age.

In contrast, women aged 25-50 stated that they’re most attracted to men a few years younger than themselves and their ratings agree.

When it comes to actual behaviours, for men the ages of the women they send messages to sits in-between their stated preferences and their real desires. Thus revealing a discrepancy between what men say they desire and what they actually do; capturing that there is often a difference between what people say they think and do and what they actually think and do.

In many areas where humans are asked to state their preferences or predict their behaviour, they are wrong. In a 2000 study by Epley & Dunning, 90% of respondents said they would vote in an upcoming US election, although the actual voting rate turned out to be 61%.

Similarly, 83% of students said they would buy a daffodil from the American Cancer Society, but in an upcoming charity push only 43% did.

This pattern, however, is not just true in our personal lives. We also don’t know how frequently we buy brands, what we think of brands, or what brands we will buy in the future.

What do I think of that brand again?

External and unconscious factors influence our thought processes, which is why when consumers have been asked to give their views on brands they often provide different answers depending on when they’re asked.

For instance, in a 2008 test by TNS, people were asked to decide on 12 attributes for 10 brands. A week later they were recontacted and asked the same questions on the same 10 brands, and they only gave around 50% of the same attributes to each brand.

Advertisers often don’t care about this because of mutually compensating errors; although individual respondent’s answers may change, the overall percentage stays the same as on average everyone is equally likely to change their answer.

Phrasing changes our response.

Not only do our answers to questions change each time we are asked, we also answer differently depending upon how the question is phrased.

An Ipsos Mori poll from 2014 asked one sample to vote on “reducing the voting age from 18 to 16?” and only 37% of people supported the motion. They asked a different sample to vote on “giving 16-17 year olds the right to vote” and the motion achieved 52% support.

We ask brains questions they don’t care about

Our brains are bad at remembering the detail of our past behaviour and forecasting our future behaviour. They are wired to remember things which will help us survive in a scarce environment, not capture a perfect record of what we buy.

They can remember what helps us take the right action in the right situation; whether we should eat the purple berries, not when we last ate them.

Yet as an industry, our research questions often focus on how frequently we consume products and which brands we will consume next. These kinds of questions do not play to our brain’s strengths, making it unlikely that a subject will be able to recall their behaviours with much accuracy.

Stop asking the wrong questions

Consumers can’t forecast their own broad behaviour. They change their answers to generic questions each time they respond, and change their answers to surveys based on the way the question is phrased.

Yet advertisers often rely on simple, broad questions to create audience segmentations, understand consumer wants, and know consumer media habits.


This can be overcome with two main focuses. Firstly, intelligently designed surveys which ask questions that can be accurately answered and can observe people’s actual behaviour.

For example, modern research into cognitive psychology shows that while we are not very good at calculating the likelihood that we will do something, we are much better at estimating the proportion of times we will do it.

Therefore, asking questions such as “How many of your next 10 purchases will be at shop X?” will produce a more accurate result than “Which shop do you buy at most frequently?” or “Where will you make your next purchase?”

Questions that ask about our actions in specific contexts also produce better results, so asking consumers what music they listen to at the gym gets more accurate responses than asking what music they listen to generally.

Finally, we can get a better picture of human behaviour by putting the time and effort into observing it; an area getting increasingly easier to quantify through growing data sets, even if the answers they provide are sometimes an indictment of our actual desires.

Contact Media Performance for more information about this topic.

Featured Posts
Recent Posts
Search By Tags
Follow Us
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Social Icon