Posts

Showing posts from March 23, 2010

Structural Equation Model

Image
By reading several articles on web site, I am trying to explain the SEM here: ALARM Researchers should not design a complex experiment with too many variables when the knowledge of the subject matter is not well-understood. SEM is built upon a measurement model and a structural model. MEASUREMENT MODEL 1. Error reduction : Using reliability techniques like Cronbach's alpha 2. Determining latent trait : Using factor analysis 3. Predicting : Using regression techniques Source: http://www.creative-wisdom.com/teaching/WBI/SEM.shtml In the above figure, latent trait (the ellipse) has been extracted through the factor analysis of observed variables (rectangles). Observed variable is affected by errors (circles). STRUCTURAL MODEL It can be understood through input-throughput and output relationship following definite path. In other words, input structure predicted throughput and throughput predicted output. Figure source: http://www.creative-wisdom.com/teaching/WBI/SEM.shtml Source: htt