American Time-Styles: A Finite-Mixture Allocation Model for Time-Use Analysis

Abstract

Time-use has already been the subject of numerous studies across multiple disciplines such as economics, marketing, sociology, transportation and urban planning. However, most of this research has focused on comparing demographic groups on a few broadly defined activities (e.g., work for pay, leisure, housework, etc.). In this study we take a holistic perspective, identifying a typology of latent “time-styles,” that defines the different ways people allocate the 24 hr in a day across multiple competing daily activities. We propose a finite-mixture time-allocation model that accounts for differences in life priorities across individuals, taking into consideration the fact that we all have the same “budget” of 24 hr to spend every day and that this allocation leads to highly sparse, truncated data. This model is then applied to time-use data from the American Time Use Survey collected by the U.S. Bureau of Labor Statistics in 2006.

Notes

^a Average time spent only among those engaged in the activity.

¹To avoid the well-known label-switching problem with latent-class models, the classes were ranked by size so that parameter estimates were comparable across solutions.