Do Planning Prompts Increase Gym Attendance? A Randomized Field Experiment

Planning prompts are potentially an attractive behavioral health intervention due to their low cost and ease of implementation. Previous research has shown that a prompt asking subjects to choose multiple days and times when they would visit the gym over the next two weeks was ineffective despite engagement with the act of planning (Carrera et al forthcoming). Other research, however, finds that planning interventions may be effective when individuals are presented with a limited set of opportunities for action. For example, giving workers with a single opportunity to get a flu shot was more effective at raising vaccination rates than presenting them with multiple opportunities (Milkman et al, 2011). It is also possible for a planning prompt to change the timing of a behavior without affecting the number of times it is carried out. This paper investigates these issues in a field experiment with 780 members of a fitness gym. Our first treatment sought to replicate the results of Carrera et al and asked individuals to choose up to 14 dates and times (one per day maximum) when they would like to visit the gym over a two week window. Our second treatment asked individuals to choose a single day and time when they would like to visit the gym. Our third treatment asked individuals to select up to three days when they were available to go to the gym. We then randomly selected one of these days and prompted them to go to the gym on that day. Using electronic records of gym attendance, we compare gym visits under treatments one, two, and three with visits under the control group to establish whether these planning prompts were effective at increasing gym attendance. We test the effect of planning on the timing of visits by creating an (unbalanced) panel data set that includes only individuals in the third treatment who said they were able to go to the gym either two or three days, and for each individual we include only the two or three chosen days they said they were able to go to the gym. We then test whether our prompt to visit on one of their chosen days increased the probability of attendance on that day over the probability of attendance on their other chosen days.