This paper discusses correlational and directional item context effects as two method biases that can threaten the validity of survey data. Two empirical studies are used to establish their presence in IS research. In addition, item separation with partial randomization is shown to be a viable way for researchers to control for correlational item context effects associated with inflated Cronbach’s alphas. This paper also presents a procedure to correct the inter-construct correlations and R2 values to account for directional item context effects in comparative experimental studies.
|Journal||Pacific Asia Journal of the Association for Information Systems|
|Publication status||Published - 2013|