Estimation of Optimum Sample Size in Structural Equation Modeling Assessing Its Adequacy for Social Researchers

Structural equation modeling (SEM) as a multivariate method for data analysis is spreading out gradually. The main aim of the article is to introduce the most popular and acceptable methods to sample size estimation (acceptable power of the null hypothesis test and the accuracy in estimation of the free parameters) that can be conf ident to the results of the free parameters estimation in a specified model.Moreover, introducing criteria and methods, which can be used by social researches to sample size determination and its adequacy assessment, is an added endeavor of the paper. Using general rules, in other words, rules of thumbs to estimate sample size are popular now. What these rules are is one of the questions that I try to answer in this article. Although, there are some advantages of the rules such as simplicity, but there are some deficiencies that ignoring them can lead to inaccurate or incorrect results. One of the most important of objections in using rules of thumbs is that the researchers could not be conf ident in a specific situation if there are enough sample size to parameter estimation from one hand, and the test of the analyzed model, on the other. In such situation, specifically when a given model is a complicated one, because of the number of observed var iables, const ruct s and parameter s and the pre-assumption of multivariate normality can be violated. As mentioned earlier, this disadvantage is more clearly observable. Satorra-Saris method which is founded on the test power and chi-square difference test (comparing a nested model with a mother one) and Monte Carlo method which is founded on the accuracy of free parameters, are some methods that can be used to assess sample size adequacy.