|
Theories in social, behavioral and economic science are becoming more complex as modern research produces data that requires sophisticated analysis methods. As a result, social scientists use increasingly complex statistical modeling to test their theories. Modern modeling software avoids difficult mathematical equations by relying on a high-level command language or a graphical model specification. The emergence of these powerful and easy-to-use modeling tools makes it important for students to understand both the possibilities and limitations of such techniques. This course, then, is aimed at students in social, behavioral and economic sciences who plan to go into research.
Structural Equation Modeling (SEM) can be viewed as the successful marriage of two techniques treated in this course: multiple regression analysis and factor analysis. Multiple regression analysis examines the effect of multiple independent variables on a single dependent variable. In SEM, this is generalized to path analysis, a technique that allows not only multiple dependent variables, but also intermediate variables that mediate the relations between the independent and dependent variables. Factor analysis explores the existence of unobserved factors from the relationships among observed variables. In SEM, this is generalized to confirmative factor analysis, a technique that allows us to test the validity of our factor model statistically. Finally, SEM allows us to combine observed variables and latent factor in one path model. The result is a powerful and flexible modeling tool.
Format
In the first part of the course, students receive a general introduction to structural equation modeling, the problems and solutions involved, and special issues such as assumptions and methodological and technical choices to be made. Class time consists of lectures, computer demonstrations, presentations, and discussions. Students also work in computer labs to analyze small data sets using the program AMOS. After the break, students analyze a real data set, present the results in class, and write a report following professional publication standards (a social science publication style guide will be provided). Additionally, guest lecturers speak to the class about using SEM in their own research.
|