Consumer Self-report Data: You Can Ask What But Not Why
When is self-report data trustworthy (and when is it not)?
Posted November 17, 2013
This is a guest blog entry by Kyle Thomas by at TipTap Lab
Marketers are always seeking to understand how consumers perceive and connect with their product. In doing this, they are trying to find ways to improve the product, packaging, messaging, and promotion. Improvements in these areas lead to better sales. The question is, what is the best way to get this information?
In my previous blog post, I argued consumer self-reports can’t be trusted because of a funny little cognitive system known as the left-brain interpreter, the confabulator, or the press secretary of the mind. However, a lot of very good psychology research relies on self-reports, so clearly they can’t be completely worthless. This raises the question, what is the difference between self-reports that are trustworthy and those that are not? Understanding the cognitive processes involved in different kinds of self-reports sheds light on this question. Here I will argue self-reports about what people do or have done are generally reliable (as long as they have no incentive to lie), and they only become unreliable when people are asked to report on why they did something.
When someone answers a question about what they did, or generally do, it simply requires them to access their episodic - and possibly semantic - memory. As Elizabeth Loftus, and many other memory researchers have shown, our memories are not infallible, and it is possible to create false memories. However, such false memories tend to be the result of biased external influences, such as the leading questions of an interrogating police officer or prosecutor. Surely our memories are far from perfect, but importantly for consumer research, the memory errors people generally make tend to be random, unless they are systematically manipulated by the introduction of some kind of misinformation. The fact that misremembering generally just introduces random error is crucially important, because as any statistician can tell you, random error simply adds noise to a set of measurements, and while it may make it harder to detect a true signal, it will not bias the results and yield a false signal.
In contrast to the generally random error introduced by the cognitive processes of memory recall, the confabulator often introduces systematic error when people are asked why they did something. We often have no introspective conscious awareness of the decision-making processes that lead us to behave in certain ways, and this is when the confabulator kicks in to provide a reasonable sounding explanation to fill this knowledge gap. The problem with this is we all have similar conceptions of why people might behave in certain ways, even though these conceptions are often completely wrong. So, if you ask a group of people why they did what they did, they may all give you the same reasonable - but completely wrong - explanation. This bias towards socially acceptable, salient, and reasonable explanations for something that people actually have no conscious awareness of can introduce systematic error into a set of measurements, which would yield a false signal, rather than just introducing noise. Many people may say they bought a Mercedes because they appreciate superior precision engineering (a reasonable and socially acceptable explanation), when in reality they bought the car simply because it is a status symbol (not so socially acceptable, so better to confabulate). If this is true of enough people, then such systematic error introduced by the confabulator clearly gives a false impression of consumers’ true motivation.
This perspective suggests that as long as a researcher does not ask their participants leading questions, or give them an incentive to deceive, people’s self-reports about what they do should be relatively reliable, especially if multiple measurements can be averaged (one of the major advantages of using averages when analyzing data is that random error cancels out). But, asking people why they made a specific decision may or may not yield an accurate answer, and the problem is that because of the way the confabulator is designed, it is almost impossible to tell the difference between true answers and systematic confabulations. In other words, when taken in aggregate, consumer self-reports about what they do should generally be accurate, and the problems with self-reports tend to result from asking people why they do what they do. Stay tuned for future blog posts, where we will suggest ways to leverage this insight to understand why people buy through questions that only ask them about what they buy.
At BeyondThePurchase.Org we want to know what you spend your money on. To learn about what might be influencing what you spend your money on, Login or Register with Beyond The Purchase, then take a few of our spending habits quizzes:
How materialistic are you? Find out by taking the Materialistic Values Scale.
Are you a compulsive buyer? Take the Compulsive Buying Scale and learn about your spending habits.
How hedonistic are you? Take the Hedonism Value Scale and learn if you chase pleasure to an unhealthy extreme.
This blog was written by Kyle Thomas, VIP of research at TipTap lab and a Ph.D. candidate, Evolutionary Social Psychology, at Harvard University.