The quest for honest answers
Rik Pieters is the present incumbent of the Arie Kapteyn Chair (June 2016 – May 2019). This professorship is an honorable position that serves as a constant reminder of Arie Kapteyn’s signal contribution to research at the Tilburg School of Economics and Management (TiSEM). The professorship is awarded to a researcher who has released many high-quality publications, has prominent external visibility, and whose research has a high impact factor. The Arie Kapteyn Professor must almost have demonstrably contributed to the development of research at TiSEM. The chair rotates every three years and comes with a research budget.
Rik Pieters is now halfway through his term – an excellent opportunity for a midterm interview.
The interview takes place in Rik’s office. The first thing that strikes you about it is the holiday atmosphere it radiates. There are two beach chairs and a beach umbrella, and a matching table full of stuff typical of beach vacations. “If you spend a great deal of your time in your office, it is important to create a pleasant atmosphere there,” he explains.
Before talking about his research, Rik Pieters first wants to share something about Arie Kapteyn. “I am grateful and proud to have been given the honor of holding this chair. Arie Kapteyn was appointed by TiSEM at a time when the School was at the bottom of the rankings and Arie was tasked to press on and turn the tide. One of the things he did was to found the CentER research institute and he had an eye for picking the right people. Together with those he engaged it took him but a short while to guide the School to the top of the rankings. It was at that time that I joined and I think of Arie as my main inspirer, not least because we work in the same subject area.”
Just like Arie Kapteyn, Rik Pieters has done survey-based research all his life, and he is particularly interested in developing research methods that yield more reliable responses than does the direct-response method. “Sometimes people are unable or unmotivated to answer a question honestly. That may be because of shame or because a nontruthful response is considered strategically expedient. To overcome this drawback I use the revealed-preference method, and the Arie Kapteyn Chair has enabled us to further develop that type of research.”
Randomized-response methodOne example he gives of revealed-preference research is the randomized-response method: participants are given a coin that partly determines the answers they give to listed questions. Heads means giving the honest answer, tails means answering “yes” to a question, and the researcher is in the dark about whether the coin came up heads or tails. This method is applied in sensitive situations where people tend not to give honest answers, because they fear their privacy may be compromised. “My colleague Martijn de Jong of Erasmus University Rotterdam and I have used this method in a large-scale study of people’s wishes regarding erotic services. It’s a market worth billions, but our understanding of it is imperfect, because very few people will want to admit using these services, let alone sharing their individual preferences. Normally, then, we would never be able to reliably estimate the potential size of that market, nor recommend how the services offered could be made more widely accessible, better, and safer. Yet in a large sample our “indirect”-question method will yield useful information about honest answers, because the percentage of “yes” answers resulting from the coin landing tails can be statistically established.”
Answers to the listed questions were also harvested using the direct-response method and the results pointed to a surprising difference: not only did more men report using erotic services, the number of women reporting using such services was substantially higher than showed up when the direct-response method was used. “For women the social convention that this is improper is much stronger than it is for men, but women too, when asked about erotic services indirectly, had more wishes than when they were asked directly. The beauty of this result is that it shows that this method works.”
Rik Pieters continues by pointing to another method, one he helped develop in the context of his professorship, to obtain more honest answers to questions about sensitive issues, such as smoking during pregnancy. There are still quite a few women who smoke while pregnant and resulting complications for bother mother and child weigh heavily on healthcare budgets and cause a lot of distress and misery. “For this study we used another method: the list technique.” Each respondent receives a list of three questions, only one of which is a sensitive one (“Did you smoke during your pregnancy?”). All the respondents are asked to do is specify to how many of the three questions their answer would be “yes”. The two non-sensitive questions are inversely related: if the answer to one of these two questions is “yes”, odds are that the answer to the other question is “no”, which near effectively precludes the sum from ever being three (affirmative answers). A set of non-sensitive questions could for example be “Did you drive to work?” and “Did you go for a walk”?For one group the direct-response method was used. “The outcome came as a bit of a shock. When asked directly, 11 % admitted to smoking during pregnancy, but when the list technique was used, this percentage rose to about 18. We felt this to be a huge percentage. Another noticeable result was that women reported giving up smoking only towards the end of their pregnancy. By then it’s of course far too late (which obviously is not to say that there is ever a good time for smoking): the harmful effects on the unborn child occur at a much earlier stage in the pregnancy.”
Hair as a biomarker“Of course, what we’d really like to know is how people actually behave and what their actual preferences are,” Rik Pieters continues. “Ideally, we’d want to read over someone’s shoulder. One of the methods that enable us to do that without infringing on people’s privacy involves the use of biomarkers.” Rik Pieters consulted his mentor Arie Kapteyn, asking him which particular biomarker he could use for a particular study. Arie’s reply was as clear as it was concise: hair! “The tips of hairs contain the residues of stress hormones the body produced a few weeks previously.” These residues are very useful in a study he is conducting together with Martijn de Jong: ordinary addictions that follow a pattern – gambling online a little too often, drinking a little too much, using medication slightly too freely, smoking one cigarette too many. “We observe the stress people experience and patterns of mild addictions. When you’re stressed, the risk of becoming addicted increases, but being addicted causes stress.”
The research is carried out in collaboration with the Erasmus Medical Center, which analyzes the hair samples of the respondents. The respondents have agreed in advance to make hair samples available, and their privacy is guarded with the greatest possible care. What does Rik Pieters hope to achieve? “Eventually we hope to increase our understanding of the mechanisms underlying the development of certain behaviors and what the effects of these behaviors are. What are the carousels of cause and effect behind these patterns of common addictions?; how can they be predicted and how can they be stopped? This in turn may result in policies being developed, for the government as well as for companies: what are the tools that need to be developed to arrive at the best possible way to stimulate the behavior you want, and conversely to understand what the effects of this behavior are.” The statistical analyses of the research are currently being carried out.
Rik Pieters has already got plans for a new research project, in which he wants to employ yet another method. ”Some questions are so sensitive that people are reluctant to even think about them. For example, we want to research the escort services market. When people are asked whether they use such services, they might feel that is an impertinent question which they would rather not answer or which for religious or other reasons they will not even consider. You cannot put such a question before them, unless you give them the opportunity to indicate that they do not wish to answer it. If they choose this option, this constitutes an intentional “non-response.” We are currently developing statistical models that allow us to gain insight into what they would have answered if they had not declined to answer.”
Rik Pieters is an enthusiastic man who talks about his research with great passion. There is a clear unifying thread in his research: it is always about self-reporting, stated and revealed preferences, entirely in line with the survey-based research started by Arie Kapteyn. And he is constantly looking for methods that yield better results than direct questions, in order to arrive at a better understanding of the causes and effects of people’s consumption behavior, to test theories and to develop tools for policy development.
(For more information (in Dutch) on Arie Kapteyn’s track record, see the Medjudice website)Annemeike Tan
See also: Expert and Expertise