Common Method Variance Analysis of Structural Models

A frequently cited limitation of survey-based research in the organizational sciences refers to possible method biases in investigations of the relationships between study variables which may lead to inaccurate conclusions being drawn from the data. This is often associated with method biases, which may account for a significant amount of variance among the research constructs at hand (Brannick et al., 2010; Pace, 2010; and Podsakoff & Organ, 1986). Method biases are often referred to as “common method variance effects and biases”, and they can take many forms. While some observed relationships might be affected by the method chosen, others could be distorted by response biases held by the individual who completes the questionnaire, thereby increasing the chance of a measurement error in the observed relationships. Such methodical and/or response-based biases are of great concern for organizational scientists as they threaten the validity of conclusions drawn about study variables and their inter-relationships (Podsakoff et al., 2003). Even though there is likely to be a random component to such biases, a systematic impact on variance produced by common method biases is more problematic because it potentially provides an alternative, non-hypothesized explanation for the observed relationships. In other words, the effects of common method biases may pose a rival explanation for the observed relationships between the study variables and therefore have the potential to lead to invalid or misleading conclusions.