Experimental Design
Discrete Choice Experiment (DCE)
We will conduct an online DCE to elicit preferences for new GLP-1 WLM. Respondents will choose between Medicine A, Medicine B and an opt-out corresponding to their current approach to weight-management. Medicines A and B vary along six attributes. For each attribute, there are different levels, which vary across tasks. Each respondent will complete 15 choice tasks from one of four blocks. The order of the choice tasks within each block will be randomised across respondents. The opt-out option does not have attribute levels.
Randomised intervention
Respondents will be randomly allocated to a control or treatment group. The intervention in the treatment group aims to increase self-awareness of self-control limitations that tend to cause failure to stick to weight-loss plans. Treatment and control group respondents will each be asked to report their ideal weight in one year and their expected weight in one year. We measure sophistication by the respondent-level difference between expected and ideal weight. The naive individual has no reason to expect their weight to deviate from their reported future ideal weight. Reporting expected weight greater than ideal weight reveals some degree of sophistication. For a given degree of lack of self-control, the greater the gap between expected and ideal weight, the greater the degree of sophistication.
We inform the treatment group a) that many expect to achieve their ideal weight and most do not, and b) among those who do not expect to achieve their ideal weight, most do not even achieve their ideal weight.
To reduce the likelihood that optimism bias causes respondents to view information on others' incorrect expectations (about maintaining self-control) as irrelevant to the veracity of their own expectations, we also provide information about general proneness to optimism bias, without explicitly using that term. We tell them that evidence shows that most people believe that they are less likely to make mistakes than the average person, and give them an example.
We follow up with questions intended to prompt self-reflection on self-control limitations that can impede achievement of ideal weight and appreciation that, without overcoming these limitations, one should not expect to achieve one’s ideal weight. After providing the information and asking the self-reflection questions, we ask the treatment group to again report their expected weight in one year.
The control group is not given information nor asked self-reflection questions. It reports expected weight only once.
After these initial questions, both treatment and control complete the 15 choice tasks.
Survey questions
All respondents will be asked about a) socio-demographics (self-reported age, gender, education, household net income), b) current weight-loss management strategies (self-reported), and c) perceived difficulty of the survey.