Endogeneity (applied statistics)

Endogeneity (of "endogenous ") means in the regression analysis that a relationship between the explanatory (independent) variables and the disturbance is. An endogenous explanatory variable is correlated with the error term. This means that the covariance of the explanatory variables and the disturbance is non-zero:

It is important that as there is no endogeneity, otherwise the estimator is inconsistent, the estimate is biased. In order to test endogeneity, there is Endogenitätstests as the Hausman test.

Is it because that relevant variables have been neglected, we speak of under fitting, too many explanatory variables of overfitting. A possible remedy is to use a fixed-effects estimator in conjunction with panel data. Other techniques are often used instrumental variables and regression discontinuity analysis.

• Regression analysis