A New Perspective on Method Variance: A Measure-Centric Approach
Spector, Paul E.
A New Perspective on Method Variance: A Measure-Centric Approach - 855-880 p.
A widespread methodological concern in the organizational literature is the possibility that observed results are due to the influence of common-method variance or mono-method bias. This concern is based on a conception of method variance as being produced by the nature of the method itself, and therefore, variables assessed with the same method would share common-method variance that inflates observed correlations. In this paper, we argue for a more complex view of method variance that consists of multiple sources that affect each measured variable in a potentially unique way. Shared sources among measures (common-method variance) act to inflate correlations, whereas unshared sources (uncommon-method variance) act to attenuate correlations. Two empirical examples, one from a simulation study and the other from a single-source survey, are presented to illustrate the complex action of multiple sources of method variance. A five-step approach is suggested whereby a theory of the measure is generated for each measured variable that would inform strategies to control for method variance by assessing and modeling the actions of identified method variance sources.
Common-Method Variance Uncommon-Method Variance Method Bias Psychometrics Construct Validity
A New Perspective on Method Variance: A Measure-Centric Approach - 855-880 p.
A widespread methodological concern in the organizational literature is the possibility that observed results are due to the influence of common-method variance or mono-method bias. This concern is based on a conception of method variance as being produced by the nature of the method itself, and therefore, variables assessed with the same method would share common-method variance that inflates observed correlations. In this paper, we argue for a more complex view of method variance that consists of multiple sources that affect each measured variable in a potentially unique way. Shared sources among measures (common-method variance) act to inflate correlations, whereas unshared sources (uncommon-method variance) act to attenuate correlations. Two empirical examples, one from a simulation study and the other from a single-source survey, are presented to illustrate the complex action of multiple sources of method variance. A five-step approach is suggested whereby a theory of the measure is generated for each measured variable that would inform strategies to control for method variance by assessing and modeling the actions of identified method variance sources.
Common-Method Variance Uncommon-Method Variance Method Bias Psychometrics Construct Validity