Controlling for a variable

In <a href="/facts/Causal_model/TC6UJglD">causal models</a>, controlling for a variable means <a href="/facts/Data_binning/542EIHcu">binning data</a> according to measured values of the variable. This is typically done so that the variable can no longer act as a <a href="/facts/Confounder/G4TVx1dt">confounder</a> in, for example, an <a href="/facts/Observational_study/zEhxI4eh">observational study</a> or <a href="/facts/Experiment/esltM54f">experiment</a>.
When estimating the effect of explanatory variables on an outcome by <a href="/facts/Regression_analysis/n6z5Tf7K">regression</a>, controlled-for variables are included as inputs in order to separate their effects from the explanatory variables.
A limitation of controlling for variables is that a causal model is needed to identify important confounders (backdoor criterion is used for the identification). Without having one, a possible confounder might remain unnoticed. Another associated problem is that if a variable which is not a real confounder is controlled for, it may in fact make other variables (possibly not taken into account) become confounders while they were not confounders before. In other cases, controlling for a non-confounding variable may cause underestimation of the true causal effect of the explanatory variables on an outcome (e.g. when controlling for a <a href="/facts/Mediator_variable/UrpjgGVc">mediator</a> or its <a href="/facts/Reachability/D8cQpSFQ">descendant</a>). <a href="/facts/Counterfactual/GdfYSv1k">Counterfactual</a> reasoning mitigates the influence of confounders without this drawback.

Controlling for a variable open-in-new

Controlling for a variable