Modeling Evaluation - Essay Example

A Review of a Structural Equation Modeling Process Structural equation modeling (SEM) is a statistical modeling technique that provides a comprehensive approach to research questions (Burnette & Williams, 2010). SEM is becoming increasingly more popular in researches that involve the behavioral sciences because of its ability to specify and test linear relationships among variables. Oftentimes, a simple bivariate experiment is not a feasible option when researchers investigate human behavior in its natural setting (Haar & Brougham, 2010).

Furthermore, researchers in the areas of organizational behavior, management, business, and applied psychology are often interested in multivariate relationships among some or all of the variables in a specified model. Incidentally, SEM provides a viable statistical tool for exploring all of these relationships. It should be noted, however, that SEM is largely a confirmatory tool rather than an exploratory procedure (Diomantopoulos, Riefler, & Roth, 2008). This means that SEM will most likely be used by researchers in cases when the validity of a certain model is to be established, rather than it being used to find the appropriate model.

Typically, the models being investigated depict processes presumed to underlie values obtained with sample data, and these processes are assumed to result in measures of association, like correlation, among the variables in the model (Ringle, Gotz, Wetzels, & Wilson, 2009). By and large, SEM is largely a “glorified” regression procedure which, unlike ordinary regression, doesn’t assume that measurement error is zero and can simultaneously estimate parameters representing the whole model rather than just pieces of the model (Ringle, Gotz, Wetzels, & Wilson, 2009).

In the model proposed by Williams, Vandenberg, & Edwards (2009), it is greatly emphasized that researchers must give due consideration into the validity of a theory embedded or implied in the proposed measurement since there are many cases when indicators could be viewed as causing rather than being caused by the latent variable measured by the indicators. A latent variable is a variable that cannot be directly observed and must be inferred from measured variables (Burnette & Williams, 2010).In many cases, researchers will not be able to detect all possible causes of error because there may be some which have neither been discussed in prior literature nor revealed by exploratory research (Grace & Bollen).

Furthermore, Williams, Vandenberg and Edwards argue that some goodness-of-fit indexes, such as chi-square, are meaningless while others such as GFI have become less useful. Thus, their paper suggests three goodness-of-fit indexes which are the comparative fit index, the root-mean-square error of approximation, and the standardized root mean residual (Williams, Vandenberg, & Edwards, 2009). According to Williams et al, the construct level term represents these missing causes. This indicates, according to the authors, that the more comprehensive the set of formative indicators specified for the construct, the smaller the influence of the error term.

The authors further not that in this context, as the variance of the residual increases, the meaning of the construct becomes progressively ambiguous. Thus, they strongly suggest that the variance of the error term be used as indication of construct validity, since the error term actually captures the aspects of the construct’s domain that the set of indicators neglect (Williams, Vandenberg, & Edwards, 2009).ReferencesBurnette, J., & Williams, L. (2010). Structural equation modelling: An introduction to basic techniques and advanced issues. In R. Swanson, & E. I. Holton, Research in organizations.

Beurett-Koehler Publishers.Diomantopoulos, A., Riefler, P., & Roth, K. (2008). Advancing formative measurement models. Journal of Business Research .Grace, J., & Bollen, K. (n.d.). Representing general theoretical concepts in structural equation models: The role of composite variables. Journal of Environmental and Ecological Statistics , 191-213.Haar, J., & Brougham, D. (2010). Cultural and organizational perceptions of support towards mental health outcomes: A study of Maori employees. Unpublishe White Paper .Ringle, C., Gotz, O.

, Wetzels, M., & Wilson, B. (2009). On the use of formative measurement specifications in structural equation modelling: A Monte Carlo simulation study to compare covariance-based and partial least squares model estimation methodologies. Munich Personal RePEc Archive .Williams, L., Vandenberg, R., & Edwards, J. (2009). Structural equation modelling in management research: A guide for improved analysis. The Academy of Management Annals , 543-604.