The Role of Regret Minimization in Choice of HPV Vaccination and the Impact of Time Preference and Risk Aversion

Open seminar with Morten Raun Mørbak, Department of Business and Economics, Econometrics and Management Science, University of Southern Denmark (SDU) on The role of regret minimization in choice of HPV vaccination and the impact of time preference and risk aversion.



The choice on whether to be HPV vaccinated depends on factors associated with the vaccination program, including, but not limited to, safety issues and side effects, the effect and duration of the vaccine, and social and societal norms. The preferences for vaccination also depends on behavioral characteristics of the decision makers, e.g. time preferences and risk aversion. Most existing DCE studies on the topic assume that individuals make decisions on the HPV vaccine from a utility maximizing perspective. However, evidence from previous studies suggests that regret can be an important factor in medical decision making. De Bekker-Grob and Chorus (2013) explored this in relation to the HPV vaccine and found that in some cases the regret minimizing model had a better fit than the utility maximizing model, and in other cases a hybrid model (assuming regret minimization on some attributes and utility maximizing on other) outperformed the pure utility maximizing model.


The aim of this paper is to extend the current research in this field by showing how to combine the utility maximizing and regret minimizing models in a latent class framework that also takes preference heterogeneity into account. We hereby allow for differences in decision making strategies as well as preferences between respondents. We also examine whether relevant behavioral traits, such as time and risk preferences, can explain the heterogeneity in decision paradigms.


Data was collected using an electronic questionnaire with physical supervision. The questionnaire contained a DCE with 12 choice sets on preferences for HPV vaccination. The target group was a representative sample of girls aged 12-18. They were approached at randomly selected schools (lower secondary schools, independent lower secondary schools, independent boarding schools for lower secondary students, upper secondary schools, upper secondary business schools, and upper secondary technical colleges) in Denmark stratified by region.


Data are analyzed using different behavioral models: 1) A mixed logit model assuming all respondents are utility maximizers, 2) A random regret minimizing model assuming all respondents are regret minimizers, 3) A latent class model allowing for decision paradigm heterogeneity as well as for preference heterogeneity, 4) Latent class models with different explanatory class variables such as current vaccination status, time preferences, present biasness, and risk aversion.

Results and conclusions

Preliminary results show that the regret minimizing model fits data better than the utility maximizing model but that the latent class model, taking the two decision paradigms into account, outperforms both. In the latent class model, we see considerable heterogeneity in decision paradigm. Interestingly, the results show that present biased respondents are more likely to adopt a regret minimization strategy than to apply the commonly assumed utility maximizing approach. We also find that current vaccination status, risk aversion, as well as long term time preferences do not explain the probability of using one decision paradigm over the other.