Department of Psychology, Cornell University, Ithaca, NY 14853.
Department of Government, Cornell University, Ithaca, NY 14853.
Proc Natl Acad Sci U S A. 2023 Mar 21;120(12):e2212867120. doi: 10.1073/pnas.2212867120. Epub 2023 Mar 13.
In 2004 through 2016, three studies in the national Midlife in the United States (MIDUS) project asked participants the open-ended question "What do you do to make life go well?". We use verbatim responses to this question to evaluate the relative importance of psychological traits and circumstances for predicting self-reported, subjective well-being. The use of an open-ended question allows us to test the hypothesis that psychological traits are more strongly associated with self-reported well-being than objective circumstances because psychological traits and well-being are similarly self-rated-meaning that they both ask respondents to decide how to place themselves on provided and unfamiliar survey scales. For this, we use automated zero-shot classification to score statements about well-being without training on existing survey measures, and we evaluate this scoring through subsequent hand-labeling. We then assess associations of this measure and closed-ended measures for health behaviors, socioeconomic circumstances, biomarkers for inflammation and glycemic control, and mortality risk over follow-up. Although the closed-ended measures were far more strongly associated with other multiple-choice self-ratings, including Big 5 personality traits, the closed- and open-ended measures were similarly associated with relatively objective indicators of health, wealth, and social connectedness. The findings suggest that psychological traits, when collected through self-ratings, predict subjective reports of well-being so strongly because of a measurement advantage-and that circumstance matters just as much when assessed using a fairer comparison.
在 2004 年至 2016 年期间，美国国家中年生活研究（MIDUS）项目中的三项研究向参与者提出了一个开放式问题：“你做了什么来让生活顺利？”。我们使用这个问题的逐字回答来评估心理特征和环境对预测自我报告的主观幸福感的相对重要性。使用开放式问题可以检验这样一种假设，即心理特征与自我报告的幸福感的关联比客观环境更为密切，因为心理特征和幸福感都是自我评估的，这意味着它们都要求受访者根据提供的和不熟悉的调查量表来决定如何定位自己。为此，我们使用自动化的零镜头分类来对关于幸福感的陈述进行评分，而无需对现有调查措施进行培训，并且我们通过后续的手动标记来评估这种评分。然后，我们评估了这种衡量标准以及封闭式衡量标准在健康行为、社会经济环境、炎症和血糖控制的生物标志物以及随访期间的死亡率风险方面的关联。尽管封闭式措施与其他多项自我评估，包括大五人格特质，具有更强的关联，但封闭式和开放式措施与相对客观的健康、财富和社会联系指标具有相似的关联。研究结果表明，当通过自我评估收集心理特征时，它们之所以能够如此强烈地预测主观幸福感报告，是因为存在一种测量优势，并且在使用更公平的比较进行评估时，环境也同样重要。
