The long-term effects of an intervention to exercise.
How to cite
Ekström, Mathias; Cappelen, Alexander; Charness, Gary; Gneezy, Uri; Tungodden, Bertil. 2023. "The long-term effects of an intervention to exercise.." AEA RCT Registry. AEARCTR-0010801. https://doi.org/10.1257/rct.10801-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
This project will evaluate the long-term effects of an exercise intervention among university students. The exercise intervention consisted of a free gym membership in the academic spring semester of 2016 (January – June), and was provided as part of a randomized-controlled trial, where a total of 778 university students in Bergen, Norway, participated: 398 students received the free gym card and were thus treated, whereas the remaining students did not receive free access to attend the student gym and thus constitute the control group. The immediate effects of the intervention – i.e., the effects in the spring semester 2016 – are documented in the published article “Exercise Improves Academic Performance” (forthcoming in the Journal of Political Economy), which in turn built on the pre-analysis plan AEARCTR-0001949. In the current project we will investigate the long-term effects of the intervention – i.e., the effects after the free gym card expired in June 2016 – addressing three distinct research questions.
First, speaking to literature on habit formation we want to understand whether the exercise intervention enabled the formation of an exercise habit in subsequent semesters (focusing on the fall 2016). More specifically, we will analyze whether treated students are more likely to buy a gym card at a cost (capturing people's exercise intentions) as well as actual gym attendance using administrative scanner data.
Second, speaking to the literature on the dynamics of academic achievements, we will analyze whether the boost in completed study points observed in the spring 2016 (caused by the exercise intervention) promotes sustained academic success or whether it crowds out academic performance in subsequent semesters (focusing on the fall 2016). Importantly, the long-term effects on academic performance may be completely independent of whether exercise is habitual or not.
Third, using comprehensive data from an endline survey conducted two years after the intervention (spring 2018, 92% response rate) we will adress the effects of the exercise intervention on a broader set of life outcomes (outside the sphere of exercise and academic performance), including a health index, BMI, life satisfaction, a self-control index, and occupational status.
We will consider whether to write three separate papers based on this project or bundle the findings into one paper, but for complete overview we specify all the analysis in one single pre-plan.
Intervention
The intervention consisted of (some) students at the University of Bergen and the Bergen City College receiving a free membership to the student gym SIB during the spring semester 2016. This membership costed 1,100NOK (≈122USD) at the time of the intervention.
As described in the published paper on the immediate effects, and the accompanying pre-plan, we had three subtreatments in our study design, with some treated students being assigned a personal trainer or a bonus scheme in addition to the free gym card. We did, however, pre-specify that we would focus on the combined treatment group in the main analysis to gain maximum power, and we will do this also in the current study on long-term effects.
Experimental design
The study is based on a between-subject design. Treated subjects receive a free gym card to the student gym for the spring semester 2016, whereas control group subjects do not get the free gym card. Due to random assignment of treatment we can evaluate how the free gym card affects gym exercise and other outcomes both in the intervention period (the published paper) and afterwards (the current project).
We will use the same OLS estimation strategy as in the analysis on the immediate effects, and control for the same set of background variables to improve precision in the statistical tests, namely: age, gender, year of study, institution, and dummies for being above the median in terms of i) study hours, ii) life style index, iii) self-control index, iv) overall life satisfaction (as well as Registered Study Points in the spring 2016 for analysis on academic outcomes).
Design details
Cited by
No citation data available for this registration yet.