Are product evaluations with a human interviewer accurate?

In sensory and consumer research, interviews can be used to ask panellists about their opinion of a product. Panellists can evaluate the product with minimal interruption of this evaluation process. Using a human interviewer can however create a bias in the results, referred to as interviewer bias. There are different factors that influence panellists’ answers. The panellists might, for example, say that they like a product while in reality they don’t. They might do this out of courtesy or because they think the interviewer will judge them. As a consequence of this bias, the evaluation of a product might not be accurate.


One type of interviewer bias is social desirability bias, which is a tendency to give answers that are socially desirable. When presented with a question, people analyse the gains and losses of giving a certain answer and will answer the question in such a way that they will gain social approval and avoid social disapproval. As an example, the interviewer asks whether or not the panellist eats lot of junk food. Let’s assume that the panellist indeed eats a lot of junk food. The panellist will then analyse his/her options. If the panellist admits to eating a lot of junk food, losses can for example be defined as feelings of embarrassment and the fear of being judged. Gains can for example be defined as receiving approval from the interviewer for being honest. The panellist then weighs the losses against the gains and decides what to answer. In this example, the panellist might choose to not admit to eating a lot of junk food, since they place more weight on the losses than on the gains. 


Another type of interviewer bias is acquiescence bias, which is a tendency to give a positive evaluation of a product. Some people have a natural tendency for this bias because they are predisposed to agreeableness during their childhood or because they are born with it. Another explanation is that panellists see the interviewer as an authority figure with a higher status than themselves. They give positive answers because they want to show respect and courtesy to this authority figure. Another explanation is referred to as satisficing. Answering a question can bring great effort with it: you must carefully listen and analyze the question. You then have to consider and review all the answer options and their potential consequences. This requires a lot of mental effort. Instead, people might give an answer that they think is “good enough” or “appropriate”, and this answer is often the answer that leads to a positive evaluation of a product.


A potential solution for interviewer bias is the dehumanization of an interview, for example by using Text-to-Speech techniques. We conducted a study with our panellists to test this possibility. For a summary of the study, see our blog AI Powered Speech-to-Text and Text-to-Speech techniquest limit the interviewer bias in sensory and consumer research.