Main Section 3: Summary

This section provides an summary for the steps of SEM in AMOS.

1. Utilizing the summary of the steps when conducting SEM in AMOS.

AMOS (Analysis of Moment Structures) is one of the most frequently used and practical software in Structural Equation Modeling. Therefore, researchers need to know how to conduct Structural Equation Modeling analysis through AMOS. The first step of Structural Equation Modeling in AMOS is to open the data file to be analyzed. After opening the data file, the first thing to do is decide the normality assumption. Normality is assumed using Mardia's coefficient and the formula p*(p+2). In this formula, p is the number of observed variables. To meet the normality assumption, the value derived from the formula should be higher than Mardia’s coefficient. After the normality assumption is met, the structural model is created according to the research hypotheses. Observed and latent variables in the model are drawn on the main window with the help of the icons on the AMOS toolbar. After all the observed and latent variables are drawn on the main window, we should proceed testing the measurement model. In testing the measurement model, covariance between latent variables should be drawn. Once the measurement model is drawn, the analysis should proceed. First, factor loadings should be examined. Standardized factor loadings must be higher than .50 and ideally above .70 with statistical significance. After examining the factor loadings, the goodness of fit indexes of the model should be evaluated according to the criteria of Schermelleh et al. (2003). The goodness of fit indexes of the measurement model are expected to be within the range of good fit or acceptable fit criteria. Testing of the structural model requires the following testing of the measurement model. The same paths are followed for testing the measurement model. An essential point in the structural model is the addition of error terms to the dependent variables. Another critical point is that not all paths have to be meaningful in the structural model. It should be kept in mind that the insignificance of the path between the independent and dependent variables may be due to the strength of the mediator variable in the relationship. It is desired that the path between the independent variable and the dependent variable becomes insignificant when the mediator variable is included in the model. After that, the goodness of fit indexes of the structural model are examined and modifications are made if necessary. The important thing here is that all modifications should have a theoretical basis. If the modification indices are higher than 4, it is necessary to draw covariance between the two variables or error terms shown in the table. The last step of the structural equation model is determining the model validity. The bootstrapping method is used to determine the model validity. The important point here is to make the correct markings in the “Analysis Properties” menu. In the output that opens, standardized indirect effect of the independent variable on the dependent variable is examined. If there is no 0 between the lower bound and upper bound values of the standardized indirect effect, it is concluded that the mediation effect is significant. Lastly, proper reporting of all analyses is required.