Thinking Clearly About Correlations and Causation: Graphical Causal Models for Observational Data

Abstract

Psychologists in many fields face a dilemma. Whereas most researchers are aware that randomized experiments are considered the "gold standard" for causal inference, manipulation of the independent variable of interest will often be unfeasible, unethical, or simply impossible. One can hardly assign couples to stay married or get a divorce; nonetheless, one might be interested in the causal effect of divorce on well-being. One cannot randomly resettle individuals into different strata of society, but one might be concerned about the causal effects of social class on behavior. One cannot randomize children to different levels of adversity, yet one might care about the potential negative consequences of childhood adversity on health in adulthood. This article provides very general guidelines for researchers who are interested in any of the many research questions that require causal inferences to be made on the basis of observational data. Researchers from different areas of psychology have chosen different strategies to cope with the weaknesses of observational data. To circumvent the issue altogether, some researchers have implemented "surrogate interventions": If the real-life cause of interest cannot be manipulated, there might be a proxy that can be randomized in the lab. For example, an influential study on the effects of social class on prosocial behavior included an experimental manipulation of perceived social class. Participants were asked to compare themselves with either the top or the bottom of the "social ladder," so as to temporarily change their subjective 745629A MPXXX10.

Obsessive–compulsive symptoms in eating disorders: A network investigation

“…To ensure clinically relevant psychopathology, we excluded 160 individuals (11%) who scored below a clinical cutoff on the eating disorder examination questionnaire (EDE‐Q) version 6.0 (Fairburn & Beglin, ; threshold of two) and 150 (10%) who scored below a clinical cutoff on the Yale–Brown obsessive compulsive scale—self‐report (Y‐BOCS‐SR; Steketee, Frost, & Bogart, ; threshold of 16). Because using a symptom severity threshold as an inclusion criterion can alter network structure via Berkson's Bias (De Ron, Fried, & Epskamp, ; Rohrer, ), we also analyzed the data without applying an EDE‐Q threshold (n = 370; see Supplement) as a sensitivity analysis.…”

“…Because using a symptom severity threshold as an inclusion criterion can alter network structure via Berkson's Bias (De Ron, Fried, & Epskamp, 2019;Rohrer, 2018), we also analyzed the data without applying an EDE-Q threshold (n = 370; see Supplement) as a sensitivity analysis.…”

Behaviour Genetic Frameworks of Causal Reasoning for Personality Psychology

“…To the extent that each statement is true and conscientiousness is itself possible to manipulate, conscientiousness would represent an extremely important factor for public policy intervention. However, several causal models may generate data whereby conscientiousness and life outcomes are correlated (Rohrer, 2018). Perhaps conscientiousness does not cause student achievement, but rather students that achieve academically increase in conscientiousness.…”

Statistical Modelling for Sports Scientists: Practical Introduction Using R (Part 1)

“…Causal inference thus aims to achieve ceteris paribus conditions needed to make causal interpretations by careful considerations of the known and unknown biases involved (Shrier & Platt, 2008;Angrist & Pischke, 2015;Hernán, 2016Hernán, , 2017Rohrer, 2018;Hernán, Hsu & Healy, 2019;Lederer et al, 2019;.…”

“…On the other hand, it might be some hidden confounder that motivates to-be-tall athletes to choose basketball. Known and measured confounders from the observational studies can be taken into account to create ceteris paribus conditions when estimating causal effects (Shrier & Platt, 2008;Angrist & Pischke, 2015;Hernán, 2017;Rohrer, 2018;Lederer et al, 2019;.…”