Structural Equation Modeling (Cancelled)

Due to unforeseen circumstances, this class has been cancelled. We plan to hold this class in the Fall. Dates will be announced later and posted to our Events Calendar.

This course introduces Structural equation modeling (SEM) with Stata software. This statistical method tests theoretically derived models with observed and unobserved variables. This methodology is different from other regression models in that this method does not assume that the variables have been measured without error. The relationship between theoretical constructs is tested using latent variables, which are variables with at least two observable measures that mathematically can represent unobserved abstract constructs. The relationship between variables is analyzed using direct, indirect, and total effects. In this manner, researchers can test mediation effects to identify underlying mechanisms that influence the relationship between a key independent variable and outcome. Moreover, this type of model allows researchers to test theoretical models that have more than one dependent variable. The class will focus on models for continuous variables and discuss options to analyze models with categorical variables.

The course will cover how to perform the following steps:

1. Measure confirmatory factor analysis (CFA) to evaluate the validity of the construct
2. Model identification
3. Model fit
4. Test hypotheses
5. Model the missing data of continuous independent variables
6. Interpret results

Requirements: 1) Students must know how to model and interpret correlations and ordinary least square (OLS) regressions. 2) Know how to use the basic Stata functions, i.e., enter and save data, create variables, use command window, and do file. 3) Knowledge of logistic regression models is not required but is helpful.

Instructor: Eugenia Conde

Dr. Conde is a Statistical Consultant at The H. W. Odum Institute for Research in Social Science at the University of North Carolina at Chapel Hill. She provides consultations to students and faculty on research methods and statistics. Her Ph.D. is in sociology with a concentration in demography and medical sociology. In addition, she holds an MSPH in epidemiology. Before working at the Odum Institute, she worked at Rutgers University and at Duke University as a statistical consultant for graduate students and as a statistician for researchers from different disciplines, including political science, economics, psychology, sociology, and public health.

She is co-authoring a statistics book with Dr. Dudley L. Poston and Dr. Layton Field, Applied Regression Models in the Social Sciences (Forthcoming, Cambridge University Press). Her research interests include social inequalities, research methods, and statistics with a focus on missing data and methodologies to study people of color.

Registration fees:

• $0, with a $20 deposit to hold your spot (deposit is refundable upon your attendance for at least 66% of the course)

Additional course information:

• This course has been cancelled. Those who are registered will have their registration fees refunded automatically.

For questions regarding the status of this class, please contact Jill Stevens at jill_stevens@unc.edu