Prediction of Dropout in Physical Activity: The Effect of Time

Leisure-time physical activity is the preferred domain by the population to follow physical activity recommendations, helping to maintain a healthy lifestyle. However, the main issue in this type of physical activity intervention is massive dropout.

Using monthly data from 20,908 leisure/fitness center members (2011-2016), we estimate a discrete-time duration model to examine how membership duration influences dropout probability, addressing cross-sectional limitations. Additionally, we assess the explanatory power of different covariate groups.

Dropout probability follows a seasonal pattern, peaking within the first 3 months before declining and stabilizing over time. Short-term memberships and low attendance frequency significantly increase dropout risk, confirming that membership conditions and time-related variables are key predictors of retention. Additionally, macroeconomic factors particularly unemployment play a crucial role in increasing dropout probability.

These results could help both recreational facility managers and policymakers to design strategies to reduce dropout among physical activity practitioners, in particular, in the current situation generated by the COVID-19 crisis.