Author Information
TzeHoung Lee, Singapore University of Social Sciences, SingaporePeter Tay, Singapore Institute of Technology, Singapore
Abstract
Why do some people actively seek time alone while others prefer constant social contact? We examined this question using behavioral data from 203 older adults in Singapore, measuring 104 aspects of their lives—from daily activity patterns to personality traits to health status. We compared three analytical approaches: (1) traditional factor analysis with ordinary least squares regression, (2) confirmatory factor analysis for construct validation, and (3) a two-stage ensemble method inspired by dual-process theories of cognition. The two-stage approach explained 28% of variance in solitude preference, substantially outperforming traditional methods (13%) and single-model machine learning (21%). Critically, proper hyperparameter tuning revealed that the analytical refinement stage (gradient boosting) added meaningful value beyond initial pattern recognition (random forest), increasing explained variance by 7 percentage points. The strongest predictors were behavioral patterns (hours spent alone, solitary activities) rather than personality traits—extraversion ranked only 15th among 104 predictors. Confirmatory factor analysis validated solitude preference as a distinct construct, separating from loneliness (r = .42), social anxiety (r = .18), and extraversion (r = −.36). These findings advance understanding of solitude as a meaningful individual difference with practical implications for well-being in later life and demonstrate methodological advantages of ensemble methods for complex behavioral data with mixed variable types.
Paper Information
Conference: AGen2026Stream: Loneliness
This paper is part of the AGen2026 Conference Proceedings (View)
Full Paper
View / Download the full paper in a new tab/window
To cite this article:
Lee T., & Tay P. (2026) Understanding Preference for Solitude: A Data-Driven Approach Based on a Dual-Process Architecture ISSN: 2432-4183 The Asian Conference on Aging & Gerontology 2026: Official Conference Proceedings (pp. 241-256) https://doi.org/10.22492/issn.2432-4183.2026.20
To link to this article: https://doi.org/10.22492/issn.2432-4183.2026.20
Comments
Powered by WP LinkPress