Some researcher might oppose our approach by stating that our study does not control for idiosyncratic firm competences but rather for best practices. And this is partly true, however, if we wanted to control for idiosyncratic firm competences, then we would have delivered a set of case studies that show how a competitive advantage (sustainable or not) might be established and managed. Case studies, which can be an appropriate methodological mean to discuss such a complex construct as the resource-based view, have to face difficulties in terms of comparability and do not necessarily provide general guidelines for how performance differences can be transferred to other companies. Nevertheless, they are of course very valuable to explain superior management for example in a very detailed way. We would like to refer to our recent book on industrial excellence, which is focusing on case studies as a mean to explain the superior management skills and development of company resources (Loch et al. 2003). The book describes best plant operating practices, excellent either in the execution of certain or whole sets of processes.
KeywordsConfirmatory Factor Analysis Measurement Model Discriminant Validity Convergent Validity Team Leader
Unable to display preview. Download preview PDF.
- 53.Referring to Homburg and Baumgartner’s study, we have to mention that the sample for their study has been 112 applications of causal analysis from international marketing literature and only 17 applications of German marketing literature. This should be kept in mind by authors who cite this paper as a justification for the use of the ULS approach. 54 However, it is controversial how far they can testify convergent validity (Fritz 1995, pp. 135-136).Google Scholar
- 55.It has to be noted that the x2-test and in turn the GFI, AGFI and the quotient of x2/df are calculated under the assumption of multivariate normal distribution. Given however that the potential bias decreases the model fit, they can also be shown in studies using the unweighted least square (ULS) method (Browne (1984)).Google Scholar
- 56.We can generally distinguish between respondents from the same and from different organizations. Within the organization we can distinguish between respondents according to their function (or department) and according to their hierarchical level.Google Scholar
- 57.According to Widaman (1985, p. 3), the correlation matrix of observed variables can be expressed as: ∑=ΛΦΛ+Ψ. where Λ = the MT x R factor loading matrix Φ = the R x R matrix of correlations among factors Ψ= the MT x MT diagonal matrix of unique factor loadings for mt measures that represent T traits and M methods for R common factors.Google Scholar
- 58.Three basic situations of moderators are distinguished: a discontinuous two-valued moderator variable, a discontinuous multi-valued moderator and a continuous moderator. Given that the scales are considered metric, only the third case needs to be discussed in the following.Google Scholar