Despite its relevance to the real world, research on team incentives, especially field research, is still sparse, largely because of the complexities inherent in such an environment. Studies in this area mainly focus on team composition or on competition between teams, but relatively few focus on the design of team incentives. Maximum contracts, in which team compensation is highly dependent on the performance of the top performer, have shown promising results in boosting the performance of the most productive individuals while the underperformers are merely maintaining their performance levels. In healthcare, it is often more important to increase the performance of the low-performing individuals to reduce health risks while keeping costs low.
We aim to contribute to this challenge by implementing different designs of minimum and maximum contracts in a RCT using a stratified randomization method. Therefore, we developed a mixed design to test the effects of these mechanisms on participants' step counts. Approximately 450 participants between the ages of 18 and 85 will be randomly assigned to one of four experimental or control treatments. The treatment groups will form teams of two, while the control group will not be assigned a partner. Two groups will receive "minimum" contracts, one "pure" and one "mixed," while the other two groups will receive the equivalents of the "maximum" contract. We implement two control groups, one “pure” and one “fixed incentive” group. In addition, we introduce three stages for each treatment group. At first participants in treatment groups receive no information about their partner. Secondly, they receive daily feedback on their partner's step count, and from stage 1. Thirdly, they receive information about their partner's social characteristics.
We run the experiment within an ongoing 365-day long study with subjects who are all seeking to improve their physical activity. All participants have been positively health screened, are using a smartphone app (ActiVAtE Behavior) to transmit their steps (main performance measure) in a timely manner and have already provided extensive individual data at the time of the intervention. While all participants indicated in the application questionnaire that they were eager to walk more steps per day, there is a large variation in ex-ante daily steps submitted via a smartphone app over the past 14 months. Furthermore, we have a very rich dataset on each individual, including not only activity data before, during, and after the intervention, but also a wide variety of preferences (measured via economic laboratory experiments) as well as sociodemographic data and individual attitudes and self-reported behavior (measured via questionnaires).