A. DAG describing a causal system where an individual-level phenotypic trait X [ a ] , like age, affects their overall tendency to give interactions, , and their general tendency to receive interactions, . The * and symbols mark our estimands: the causal effects we wish to estimate. The rest of the DAG is similar to Fig 2A . B. Mapping between structural parameters (SCM 2) and statistical parameters. As for Fig 2B , the transparent arrows indicate that these effects are possible (e.g., simulation 3), but do not exist in SCM 2. C. Fixed-effect estimates, for two statistical models fitted to the data generated with SCM 2. Left : fixed effects of the social relations model adjusted by X [ a ] (statistical model 2). Right : fixed effects of the non-adjusted social relations model (statistical model 1). The target values of the well-adjusted model are shown in grey. Deviations from them allow us to understand how X [ a ] affects the varying effects of the model, when X [ a ] is not adjusted for.