Structural equation modeling provided the means of simultaneously assessing the quality of measurement in our model and the relationships among constructs. Although our measures were simple and rather explorative, we were able to test the complex path model in a rigorous way because it was specified a priori. However, sample size only allowed us to use a limited number of manifest variables and the dependent variable customer- relationship performance, a four-item construct, was treated as a mean-score of the items in the estimation of the full model. The three-item construct, IT knowledge,was treatedthe same way.