The Gibbs sampler can be used to obtain samples of arbitrary size from the posterior distribution over the parameters of a structural equation model (SEM) given covariance data and a prior distribution over the parameters. Point estimates, standard deviations and interval estimates for the parameters can be computed from these samples. If the prior distribution over the parameters is uninformative, the posterior is proportional to the likelihood, and asymptotically the inferences based on the Gibbs sample are the same as those based on the maximum likelihood solution, e.g., output from LISREL or EQS. In small samples, however, the likelihood surface is not Gaussian and in some cases contains local maxima. Nevertheless, the Gibbs sample comes from the correct posterior distribution over the parameters regardless of the sample size and the shape of the likelihood surface. With an informative prior distribution over the parameters, the posterior can be used to make inferences about the parameters of underidentified models, as we illustrate on a simple errors-in-variables model.

Abstract not found

Enterprise systems entail complex organizational interventions. Accurately gauging the impact of any complex information system requires understanding its multidimensionality, and the development of a correspondent, standardized, validated, and robust measurement instrument. Despite the popularity and potential of enterprise systems in modern organizations, no acceptably valid and reliable enterprise system success assessment scale has heretofore been developed. The present study tests the reliability and construct validity of the enterprise system success (ESS) measurement model and variants against new empirical data. Results from a confirmatory factor analysis utilizing structural equation modeling techniques confirm the existence of four distinct and individually important dimensions of ESS: individual impact, organizational impact, system quality, and information quality. Based on the analysis of results, the ESS instrument demonstrates strong reliability and validity.

According to Kenny and McCoach (2003), chi-square tests of structural equation models produce inflated Type I error rates when the degrees of freedom increase. So far, the amount of this bias in large models has not been quantified. In a Monte Carlo study of confirmatory factor models with a range of 48 to 960 degrees of freedom it was found that the traditional maximum likelihood ratio statistic, TML, overestimates nominal Type I error rates up to 70 % under conditions of multivariate normality. Some alternative statistics for the correction of model-size effects were also investigated: the scaled Satorra–Bentler statistic, TSC; the adjusted Satorra–

Somer obustness questions in str uctur al equation modeling (SEM) ar intr duced. Factor that a#ect the occuruv ce of nonconver gence and impr: er solutions arr/7 ewed in detail.

In this paper, in order to derive the utilization priority of functional information systems utilized in the process of supply chain integration and suggest a set of advisable strategies for IS utilization, the relationship analysis among three major functions (Creation functions, Connection functions, Support functions) of information systems utilized for supply chain management and supply chain management performance is carried out by means of LISREL. As a result of the analysis, this paper derives an IS utilization strategy for supply chain integration based on the priority of (support functions → creation functions → connection functions), and, through the further analysis, discloses that, in order for the derived strategy to be implemented successfully, the establishment of proper relationship with external utilization mechanism of the system and the proper role shift of information systems under the developmental stage of supply chain is required.

We are writing this editorial because it appears to us that some researchers in the Information Systems community view partial least squares modeling (PLS; also referred to as path analysis with composites or soft modeling) as some type of magical silver bullet. These researchers are less critical about the use of PLS than they should be. In spite of cautiously proposed rules of thumb available in the PLS literature, we are frustrated by sweeping claims made by some researchers that PLS modeling can or should be used (often instead of the covariance-based approach) because it makes no sample size assumptions or because somehow “Sample size is less important in the overall model ” (Falk and Miller 1992, p. 93). We are seeing an increasing number of such claims in papers submitted for review. It would be nice to think that such statements would be weeded out in the review process. However, more and more studies across a number of disciplines are creeping into the literature in which the samples are dwindling to ridiculously small sizes, despite the inferential intentions of the studies and the magnitude of parent populations. The use of small samples in these studies is frequently legitimized by references to the original developers of the PLS approach. Even MIS Quarterly has published at least one such study, which incorrectly states that “the PLS approach does not impose sample size restrictions for the underlying data. ” While many MIS Quarterly readers use PLS correctly, we are writing this editorial to combat the mistaken belief held by some in the IS community that PLS may be used in all cases when the sample size is small. Rather, we wish to stress the importance of adequate sample size, as well as the related issue of standard errors, in PLS research. The claim about the desirability of larger sample sizes when using PLS is not new. For example, Hui and Wold (1982) determined