Controlling for a variable

In statistics, controlling for a variable is the attempt to reduce the effect of confounding variables in an observational study or experiment. It means that when looking at the effect of one variable, the effects of all other variable predictors are taken into account, either by making the other variables take on a fixed value (in an experiment) or by including them in a regression to separate their effects from those of the explanatory variable of interest (in an observational study). In statistics, controlling for a variable is the attempt to reduce the effect of confounding variables in an observational study or experiment. It means that when looking at the effect of one variable, the effects of all other variable predictors are taken into account, either by making the other variables take on a fixed value (in an experiment) or by including them in a regression to separate their effects from those of the explanatory variable of interest (in an observational study). Experiments attempt to assess the effect of manipulating one or more independent variables on one or more dependent variables. To ensure the measured effect is not influenced by external factors, other variables must be held constant. These variables that are made to remain constant during an experiment are referred to as the control variables.