This note demonstrates that in applied regression analysis, the variance of a coefficient of interest may decrease from the inclusion of a control variable, contrasting with Clarke’s assertion (2005, 2009) that the variance can only increase or stay the same. Practitioners may thus be well advised to include a relevant control variable on this basis alone, particularly when it is weakly correlated with the variable of interest.
Dieser Beitrag ist mit Zustimmung des Rechteinhabers aufgrund einer (DFG-geförderten) Allianz- bzw. Nationallizenz frei zugänglich. - This publication is with permission of the rights owner freely accessible due to an Alliance licence and a national licence (funded by the DFG, German Research Foundation) respectively.