عنوان

Exploring Predictors of Subjective Well-Being Using Machine Learning and Propensity Score Techniques

شماره

TL3cp6b0k3

زبان متن نوشتاري يا گفتاري و مانند آن

انگلیسی

عنوان اصلي

Exploring Predictors of Subjective Well-Being Using Machine Learning and Propensity Score Techniques

نام عام مواد

[Thesis]

نام نخستين پديدآور

Margolis, Seth Michael

نام ساير پديدآوران

Lyubomirsky, Sonja

نام ناشر، پخش کننده و غيره

UC Riverside

تاریخ نشرو بخش و غیره

2020

کسي که مدرک را اعطا کرده

UC Riverside

امتياز متن

2020

متن يادداشت

The majority of people around the world report wanting "the good life." But how do they achieve it? Most research in well-being science operationalizes "the good life" as subjective well-being, which is comprised of positive affect, negative affect, and life satisfaction. This project uses a nationally representative publicly available dataset from the Midlife in the United States (MIDUS) project (N = 4,378) to investigate predictors of subjective well-being. Importantly, this dataset contains measures of most of the previously identified predictors of subjective well-being. In addition to determining which predictors are stronger than others, this project also explores the utility of machine learning models and propensity score methods. Machine learning models are used in this project to determine the extent to which non-linear and interaction effects predict subjective well-being. I also evaluate the value of a propensity score method for identifying causal effects on subjective well-being. Linear effects accounted for the vast majority of variance in subjective well-being. Machine learning models that could model non-linear and interaction effects predicted subjective well-being approximately as accurately as linear multiple regression models that only allowed for linear effects. Furthermore, linear multiple regression models appeared well-suited to model non-linear and interaction effects via variable transformations. Indeed, these models predicted subjective well-being as accurately or more accurately than machine learning models. Unfortunately, a propensity score method provided little value in identifying causal effects because it failed to eliminate relationships between a predictor of interest and other predictors. The role (or lack thereof) that machine learning and propensity score techniques could play in subjective well-being research is discussed. Replicating previous research, sociability, physical health, disengagement from goals, sex life quality, wealth, and religious activity were among the strongest predictors of subjective well-being. Consistent with previous research in the U.S., demographic factors appeared to be relatively weak predictors of subjective well-being. Finally, control over one's life-and financial and work matters in particular-was a strong predictor of subjective well-being, an effect that previous research may have downplayed.

مستند نام اشخاص تاييد نشده

Margolis, Seth Michael

مستند نام اشخاص تاييد نشده

Lyubomirsky, Sonja

مستند نام تنالگان تاييد نشده

UC Riverside

نام الکترونيکي

فرمت انتشار

p

نوع ماده

[Thesis]

کد کاربرگه

276903

سطح دسترسي

a

تكميل شده

Y