Quantile regression and random forest analysis on influencing factors of the subjective well-being of the elderly
10.3760/cma.j.cn371468-20240329-00142
- VernacularTitle:老年人主观幸福感影响因素的分位数回归与随机森林分析
- Author:
Hongqiang SHI
1
;
Yuxi LIU
;
Xiangren YI
;
Haifeng DING
;
Xianglei ZHU
;
Chonghua WAN
Author Information
1. 广东医科大学附属东莞第一医院，东莞 523808
- Publication Type:Journal Article
- Keywords:
Subjective well-being;
Influencing factors;
Quantile regression;
Random forests;
Elderly
- From:
Chinese Journal of Behavioral Medicine and Brain Science
2025;34(6):545-551
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To explore the influencing factors of the subjective well-being of elderly individuals and improve the accuracy of well-being predictions.Methods:A multi-stage stratified sampling method was conducted in Guangdong province and Guangxi Zhuang autonomous region from August 2019 to October 2020. The health status and subjective well-being of older adults were assessed using a self-administered multidimensional health questionnaire, while overall well-being was evaluated with a single-item 5-point scale. The statistical analysis was performed using R software (version 4.3.2), with analytical methods including χ2 tests, univariate analysis, quantile regression and random forest model. Results:A total of 2 620 participants were recruited, and 70.9% of whom reported feeling happy.The depression symptom score of the elderly was 3.00(1.00, 6.00), emotional status score was (28.26±4.14), activities of daily living (ADL) score was 69.00(52.00, 81.00), and the body mass index(BMI) was (22.68±3.81)kg/m 2.The quantile regression results indicated that marital status (married)(Q 10: β=0.049; Q 50: β=0.014; Q 90: β=0.005) and depressive symptoms (no depression, depressive tendency)(Q 10: β=1.000; Q 50: β=1.000; Q 90: β=0.025) had varying effects on older adults' subjective well-being across three quantiles (Q 10, Q 50, Q 90)(all P<0.05). In contrast, social activity (no social activity)(Q 10: β=-0.072; Q 50: β=-0.011; Q 90: β=-0.006), ADL (severe disability, moderate disability, mild disability)(Q 10: β=-0.103, Q 50: β=-0.006, Q 90: β=-0.048; Q 10: β=-0.063, Q 50: β=-0.012, Q 90: β=-0.005; Q 10: β=-0.078, Q 50: β=-0.016, Q 90: β=-0.009), and emotional status (poor emotion, fair emotion)(Q 10: β=-0.261, Q 50: β=-0.009, Q 90: β=-0.025; Q 10: β=-0.140, Q 50: β=-0.017, Q 90: β=-0.029) had a negative impact on elderly subjective well-being at the three quantile points(all P<0.05). Random forest results indicated that depressive symptoms, ADL, emotional status, household monthly income per capita, BMI, social activity, multimorbidity status, and marital status were the most important determinants of subjective well-being. Conclusion:Depressive symptoms and emotional status have a significant impact on the subjective well-being of older adults. Additionally, ADL, household income per capita, BMI, and social engagement also play important roles in influencing well-being, while the emotional support provided by marriage should not be overlooked.
