Research staff at the Human Flourishing Program seek to employ some of the most rigorous approaches to quantitative empirical analysis while also integrating this work with scholarship from the humanities. The Program’s work has in fact helped pioneer new methodological approaches and some of these have arisen precisely from engagement with trying to study flourishing more comprehensively (such as outcome-wide studies) or from engagement with humanities scholarship (such as new approaches to measurement). A summary of some of methodological principles and innovations arising out of the Program’s work is described below.
- Causal Inference -
In our empirical work on the determinants of well-being we try to employ, whenever possible, the most rigorous principles of causal inference including employing longitudinal designs, rich confounder adjusting, controlling for baseline outcome and prior levels of exposure. We have published papers on study design considerations and confounding control principles. We also routinely use sensitivity analysis to assess how robust or sensitive our conclusions are to potential unmeasured confounding factors so as to better assess the strength of evidence. Some of our also work employs more sophisticated causal models such as marginal structural models and causal mediation analysis models to examine the effects of time-varying exposures or mechanism. We have employed these approaches, for example, in our work on the effects of religious service attendance on mortality risk and depression and the potential mechanisms governing these relationships.
- VanderWeele, T.J. (2021). Can sophisticated study designs with regression analyses of observational data provide causal inferences? JAMA Psychiatry, 78:244-246.
- VanderWeele, T.J. (2019). Principles of confounder selection. European Journal of Epidemiology, 34:211-219.
- VanderWeele, T.J. and Ding, P. (2017). Sensitivity analysis in observational research: introducing the E-value. Annals of Internal Medicine, 167:268-274.
- VanderWeele, T.J., Jackson, J.W., and Li, S. (2016). Causal inference and longitudinal data: a case study of religion and mental health. Social Psychiatry and Psychiatric Epidemiology, 51:1457-1466.
- VanderWeele, T.J. (2015). Explanation in Causal Inference: Methods for Mediation and Interaction. New York: Oxford University Press.
- Outcome-Wide Studies -
Most quantitative empirical studies attempting to assess causality examine only a single exposure and a single outcome. However, this does not allow one to easily see how the exposure or phenomenon under study may affect other outcomes, or what the relative effect magnitudes are, or whether there may be harmful effects on certain outcomes and beneficial effects on others. A more comprehensive study of flourishing requires examining multiple outcomes simultaneously. The outcome-wide longitudinal design developed at the Program extends classical approaches for causal inference to examine multiple outcomes simultaneously. We have published both brief and more comprehensive introductions to this analytic approach and have used this approach to examine the effects on a wide range of outcomes of numerous psychosocial phenomena and exposures including parental warmth, parenting practices, religious service attendance, religious upbringing, forgiveness, social cohesion, volunteering, hope, purpose in life, life satisfaction, character strengths, and financial conditions.
- VanderWeele, T.J., Mathur, M.B., and Chen, Y. (2020). Outcome-wide longitudinal designs for causal inference: a new template for empirical studies. Statistical Science, 35:437-466.
- VanderWeele, T.J. (2017). Outcome-wide epidemiology. Epidemiology, 28:399-402.
- Meta-Analysis -
The strongest evidence often comes from meta-analyses that combine evidence over multiple studies. We have developed new metrics for meta-analyses that better characterize evidence when effects may be heterogeneous across settings and potentially have beneficial effects in some contexts and detrimental effects in others. We have also developed methods to help assess whether meta-analyses are robust to potential unmeasured confounding and to publication bias (wherein some studies end up not being published in the research literature and are thereby excluded from such meta-analyses). We have used these approach to help try to resolve controversies in meta-analysis around the effects of violent video games, and media exposure to suicide, smoking, and sexual behaviors, as well as to gain additional insight into supportive employment interventions and job-crafting practices at work.
- Mathur, M. and VanderWeele, T.J. (2020). Sensitivity analysis for unmeasured confounding in meta-analyses. Journal of the American Statistical Association, 115:163-172.
- Mathur, M. and VanderWeele, T.J. (2019). New metrics for meta-analyses of heterogeneous effects. Statistics in Medicine, 3:1336-1342.
- Mathur, M. and VanderWeele, T.J. (2020). Sensitivity analyses for publication bias in meta-analyses. Journal of the Royal Statistical Society, Series C, 69:1091-1119.
- Measurement -
The assessment of psychosocial constructs is a perennial challenge in attempts to study well-being. A large psychometric literature has developed along with a host of empirical methodological tools. Unfortunately, most of the empirical approaches are based purely on correlations and ignore potential causal relations between the potential factors under study. Most of the literature on psychometric assessment also tends to ignore the rich insights and important distinctions that have arisen within philosophy and theology concerning the relevant constructs. We have been attempting to develop a more integrated theory of measurement, taking into account causal relationships and incorporating insights and analytic frameworks and definitions from the philosophical and theological literatures. Although this approach is still under development, a number of important critiques of existing practices have already emerged including issues concerning causal relationships between factors under study, differential causal relationships between different indicators of the same construct, and causal interpretation of composite measures using scales or indices. We have also been working towards incorporating insights from the philosophical and theological literature to produce more conceptually satisfactory measures of meaning, suffering, and spiritual well-being, along with current ongoing projects on measurement related to hope, optimism, and love.
- VanderWeele, T.J. (2022). Constructed measures and causal inference: towards a new model of measurement for psychosocial constructs. Epidemiology, 33:141-151.
- VanderWeele, T.J. and Vansteelandt, S. (2022). A statistical test to reject the structural interpretation of a latent factor model. Technical Report.
- VanderWeele, T.J. and Batty, C.J.K. (2022). On the dimensional indeterminacy of one-wave factor analysis under causal effects. Technical Report.
- Humanities Scholarship -
The study of flourishing is inherently interdisciplinary, with various disciplines providing important and distinct insights. Although much of the research at the Human Flourishing Program is empirical, we also contribute humanities scholarship on well-being including work in philosophy, theology, and history. Moreover, we are working towards trying to better integrate or synthesize knowledge across disciplines so as to allow insights from one discipline to contribute the pursuits of another. Such work has included developing empirical hypotheses based on philosophical and theological traditions; allowing claims in the humanities to potentially be challenged by empirical research; incorporating philosophical and theological insights and distinctions into measure development; using philosophy and theology to enrich the interpretation of empirical results; employing empirical social science research methodologies to evaluate interventions based on philosophical or theological insights; and bringing insights together from different disciplines and synthesizing them. We have written a short working paper on some of these approaches. While much work remains to be done in developing more systematic approaches to the integration of knowledge across disciplines, we remain committed to an interdisciplinary approach to the study of human flourishing and to continuing to explore how various academic disciplines might better engage with one another.
- Case, B.W. and VanderWeele, T.J. Integrating the humanities and the social sciences: six approaches and case studies. Harvard University Technical Report.
We very much hope that the use of these various rigorous methodologies, and the integrating of insights from the humanities, will help us better understand, and thereby also better promote, human flourishing.