ESRA 2019 Programme at a Glance
Causality in Health Inequalities Research: New Solutions to Old Challenges 1?
|Session Organisers|| Mrs Katharina Loter (Martin Luther University Halle-Wittenberg)
Professor Oliver Arránz Becker (Martin Luther University Halle-Wittenberg)
|Time||Wednesday 17th July, 09:00 - 10:30|
Questions that motivate most studies on the emergence and reproduction of health inequalities over the life course are longitudinal and causal in nature. Fortunately, after several decades of socio-epidemiological research that produced numerous cross-sectional (and thus associational) findings, the need for studies aiming at disentangling causal effects and mechanisms from non-causal associations is becoming increasingly recognized. Especially, attempts to estimate causal effects of social background variables (e.g., marital and family status or occupational status) and/ or life changing events (e.g., pregnancy, separation and divorce, death of a family member, retirement) on health over time – including selection and reverse causality – have to be undertaken using advanced data structures and analytical techniques suited for causal inference. Following recent developments in causal theory including directed acyclic graphs (DAG), new opportunities for innovative health research are being generated by the increasing availability of rich panel data with multiple times of observation and different types of health measures (biomarkers, fitness tracker data, results of objective medical tests, and a variety of subjective indicators). Nonetheless, causal analyses on health inequalities over time remain challenging for many reasons, and one of them is health being a strong predictor of attrition in panel studies (due to nonresponse or death). Several analytical approaches have been proposed as a remedy for the manifold challenges regarding causal inference in health research. The first one deals with health-related self-selection as a missing data problem and uses techniques for correcting causal effect estimates such as weighting or dropout models for nonignorable nonresponse (e.g., Heckman selection models). The second one focuses on reducing unobserved heterogeneity (e.g., via instrumental variables, propensity score matching, fixed-effects regression). The third one places emphasis on testing potential intervening mechanisms (mediator and suppressor effects). The fourth one corresponds with the use of agent-based modeling (e.g., with SIENA) enabling simultaneous estimation of selection and causation effects in a social network framework. Finally, the consideration of measurement error while dealing with latent variables may serve to adjust causal effect estimates for (lack of) reliability. In the light of the above challenges, we invite submissions applying any of the mentioned approaches (or other innovative analysis techniques) with the goal to improve understanding of causality and selection in (preferably longitudinal) research on health inequalities. Welcome are contributions that aim at unraveling the causal complexity between social background variables and/ or life course transitions and events, and any kind of objective or subjective health outcome.
Keywords: health inequalities, life course, causal inference, selection, panel data analysis
Trajectories of Mental Well-Being around Critical Life Events: Longitudinal Analyses Based on the SF12-Questionnaire
Dr Marco Giesselmann (DIW Berlin) - Presenting Author
Since the beginning of the 2000s, a refined measurement of mental well-being has been integrated into several large scale, representative longitudinal surveys: the mental components of the Short Form 12/36 health questionnaire (SF12/36). In this study, we use measures derived from this questionnaire to analyze trajectories of mental well-being around critical life-events. Thus, our study relates to a strand of life course research that investigates the impact of critical live events on (mental) well-being—research that mostly uses simple measures of general life satisfaction as dependent variable. Our aim is to test, in how far results from this research scan be reproduced, if a SF12/36-based measure is used as indicator of mental well-being. Our study starts with discussing the survey-methodological implications and substantive differences between the standard measures of well-being (general life-satisfaction) and a SF12/36-based indicator, the Mental Component Summary score (MCS). To explore life course related differences between these indicators, we conduct transition centered empirical designs using data from the German Socio-Economic Panel (SOEP) study. In order to model trajectories of well-being, we use simple mean-difference based designs as well as parametric approaches, controlling for unit-specific heterogeneity through fixed effects (FE) estimation. We explain empirical differences in the trajectories on the basis of conceptual disparities in the indicators. We also discuss the implications of these disparities for life course related research on mental well-being.
Protective Effect of Marriage on Health: Instant or Cumulative, Short- or Long-Term?
Dr Malgorzata Mikucka (Mannheim University) - Presenting Author
Professor Oliver Arranz-Becker (Martin Luther University Halle-Wittenberg)
Professor Christof Wolf (Gesis, Mannheim University)
The idea that marriage has a protective effect on health has extensive theoretical fundaments, but the empirical evidence is less conclusive. One of the reasons is the difficulty to empirically disentangle the self-selection into marriage from the causal effects of marriage. Although past studies often addressed the “selection or protection” question, they relied predominantly on FE regression which controls for selection on premarital health but not for selection on premarital health slopes. This could produce biased estimates, potentially overestimating the protective effect of marriage on health.
This paper is the first one to estimate the effect of marriage on health using fixed effects models with individual slopes (FEIS) in order to eliminate selection on health slopes from the causal marriage effect estimates. We use SOEP data (2002-2016) on marital histories and estimate the effect of marriage on physical and mental health as captured by the SF-12 scale.
The conventional fixed-effects regression with group-specific slopes (FEGS) showed that men who eventually married exhibited more sustainable mental health (e.g., smaller declines over time) already before marriage. When taking into account the self-selection on premarital health slopes (FEIS models), we observed surprisingly few effects of marriage on health. Regarding mental health, we found only temporary positive shifts around the transition into marriage among women. Hence, the beneficial mental health effects reported in the literature may in part capture short-term rather than sustainable protection, and they may partly reflect the self-selection on health slopes. The analysis for physical health showed a temporary decline directly after the transition into marriage, but this was countered by positive cumulative effects in the long run. These results suggest that marriage has a protective effect on physical health, but these long-term effects may even go undetected depending on time span of observation and additional controls employed in the model.
Health Dynamics around Marital Separation: Anticipation and Adaptation Effects on Parental Health, Conditional on Children’s Age
Mrs Katharina Loter (Martin Luther University Halle-Wittenberg) - Presenting Author
Professor Oliver Arránz Becker (Martin Luther University Halle-Wittenberg)
Dr Małgorzata Mikucka (The Mannheim Centre for European Social Research (MZES))
Professor Christof Wolf (GESIS Leibniz-Institute for the Social Sciences)
Our study aims at estimating the intra-individual causal effect of marital separation on parental health conditional on the age of the youngest biological child. Considering potential buffering effects, child’s age (distinct categories: childless, parents of pre-school, primary school, adolescent and adult children) is supposed to play a moderating role in the separation-health nexus. We restrict our sample to women and men who were first married when entering the SOEP panel and count for them respectively 816 and 552 transitions to separation between 2002 and 2016. Our dependent variable is mental health-related quality of life (SF-12). We estimate distributed fixed-effects (dummy impact functions), covering the time span of up to three years before to four (and more) years after marital separation. This kind of modelling enables us to carefully examine patterns of temporal dynamics prior to the event (anticipation), in the year of the event and shortly afterwards (immediate effect) and following the event (adaptation), while taking into account person-related time-constant unobserved heterogeneity. Additionally, we model effect heterogeneity on the group level using interaction terms multiplying impact function by children’s age group. Our results show that childless men are mentally affected by marital separation, at least within the first year after the event, and childless women are not. Further, we observe both significant anticipation and adaptation effects for mothers of pre-school children and also adult children, however the nature of the patterns is different. Whereas mental health of mothers with pre-school children starts to deteriorate already one to two years before marital separation and continues to decline afterwards, mothers of adult children seem to experience rather a short-term decline around marital separation. In contrast, parents of adolescent children seem to be negatively affected by marital separation immediately after the event but recover mentally within the first two years.
Poverty Dynamics and Subjective Health. New Evidence from Fixed Effects Models with Individual Health Trajectories?
Mr Nico Seifert (Institute of Medical Sociology (IMS), Martin Luther University Halle-Wittenberg) - Presenting Author
Dr Anja Knöchelmann (Institute of Medical Sociology (IMS), Martin Luther University Halle-Wittenberg)
Professor Matthias Richter (Institute of Medical Sociology (IMS), Martin Luther University Halle-Wittenberg)
Although the association between poverty and subjective health is well established in social epidemiology, the relative contribution of social causation and health selection remains an unsolved puzzle. Several studies relied on panel data with fixed effects models to account for selection of individuals with poor health into poverty. A systematic review of these studies concluded that there is still evidence for a weak but significant association between income poverty and subjective health, indicating that some part of the association cannot be attributed to selection. In this study, we argue that standard fixed effects models may fail to fully account for selection because they erroneously assume that the health trajectories of exposed and unexposed individuals follow parallel trends. Substantially parallel trends imply that, had the exposed individuals not slipped into poverty, their health would have developed parallel to the health of unexposed individuals. We argue that this assumption is violated whenever there is selection of individuals with less favourable health trajectories into poverty. Using data from the German Socio-Economic Panel (GSOEP) study 1994-2016, we analysed whether the association between income poverty and subjective health can still be found in a fixed effects model with individual slopes (FEIS). This model allows the slope of the effect of aging on health (i.e. the health trajectory) to vary between individuals as a result of stable unobservables interacting with age. The results confirm that standard fixed effects models yield an upwardly biased estimate of the poverty effect, especially among men. When accounting for the correlation between individual health trajectories and risk of poverty, no significant association between poverty and subjective health remained. These findings let us conclude that while fixed effects models are a strong toolkit for disentangling social causation and health selection, model assumptions may need to be adjusted when the aim is to identify social determinants of health.