2017 Stata Summer School:

Longitudinal Data Analysis Using Structural Equation Modeling

Taught by Paul Allison, Ph.D.
August 17-18, Hotel Birger Jarl Conference
Stockholm, Sweden 

Read reviews of this seminar 

To see a sample of the course materials, click here.


For the past ten years, Professor Paul Allison has been teaching his acclaimed two-day seminars on Longitudinal Data Analysis Using SAS and Longitudinal Data Analysis Using Stata. In this new seminar he takes up where those courses leave off, with methods for analyzing panel data using software for structural equation modeling (SEM).

Panel data have two big attractions for making causal inferences with non-experimental data:

  • The ability to control for unobserved, time-invariant confounders.
  • The ability to determine the direction of causal relationships.

Fixed effects methods that control for unobserved confounders are now well known and widely used. Cross-lagged panel models have long been used to study the direction of causality. But trying to combine these two approaches poses serious problems.

To deal with these difficulties, econometricians have developed the Arellano-Bond methods for estimating dynamic panel data models. However, Arellano-Bond is known to be statistically inefficient and can also be severely biased in important situations.

Professor Allison has recently shown that dynamic panel models can easily be estimated by maximum likelihood with SEM software (ML-SEM). His recent work demonstrates that the performance of ML-SEM is superior to the econometric methods. Plus, ML-SEM offers better capabilities for handling missing data, evaluating model fit, relaxing constraints, and allowing for non-normal data.

This seminar takes a deep dive into the ML-SEM method for estimating dynamic panel models, exploring the ins and outs of assumptions, model specification, software programming, model evaluation and interpretation of results. We’ll work through several real data sets in great detail, testing out alternative methods and working toward an optimal solution.

This is an applied course with a minimal number of formulas and a maximal number of examples. Although the methodology is cutting edge, the emphasis is on how to actually do the analysis in order accomplish your objectives. 

It’s also a hands-on course with lots of opportunities to practice the methods, including at least two hours of carefully-designed and supervised exercises using real data. You are strongly encouraged to bring your own data and try out the new methods. 

At the end of this seminar, you should be able to confidently apply the ML-SEM method for dynamic panel data to your own research projects. You will also have a thorough understanding of the rationale, assumptions, and interpretation of these methods.

CAUTION: This is not a survey course but rather an intensive study of a specific methodology—one that has enormous potential for improving causal inference from panel data. This seminar does NOT cover several popular methods that use SEM to analyze panel data, including:

  • Latent growth curve models
  • Latent trajectory models
  • Common factor models 

The methods covered in this course are NOT typically useful in analyzing data from randomized experiments. That’s because dynamic panel models are designed for situations in which the predictor variables of central interest vary over time. These methods are also not designed for prediction—rather, the goal is to maximize the ability to make valid causal inference. To use these methods, you need panel data with at least three time points, and the number of individuals should be substantially larger than the number of time points.


WHO SHOULD ATTEND?

This seminar is designed for those who want to analyze longitudinal data with three or more time points, and whose primary interest is in the effect of predictors that vary over time. You should have a solid understanding of basic principles of statistical inference, including such concepts as bias, sampling distributions, standard errors, confidence intervals, and hypothesis testing. You should also have a good working knowledge of the principles and practice of linear regression. 

It is desirable, but not essential, to have previous training in either longitudinal data analysis, structural equation modeling, or both. If you’ve taken Professor Allison’s courses on either of these topics, you should be well prepared. You should also be an experienced user of Stata.


FORMAT, AND MATERIALS

Participants receive a bound manual containing detailed lecture notes (with equations and graphics), examples of computer printout, and many other useful features. This book frees participants from the distracting task of note taking.


COMPUTING

This seminar will use Stata 14 for the many empirical examples and exercises. To participate in the hands-on exercises, you are strongly encouraged to bring a laptop computer.  If you do not already have Stata installed, a temporary license will be provided free of change. Lecture notes using SAS, Mplus and R are available on request to registered participants. A power outlet and wireless access will be available at each seat.

If you plan to bring your own data, it’s best to bring it in both the wide form (one record per individual) and the long form (multiple records per individual). SEM estimation of dynamic panel models requires that the data be in the wide form. 


REGISTRATION AND LODGING

 Please go to the Metrika website for information on registration, and discounted hotel accommodations.


SEMINAR OUTLINE

  • Characteristics of panel data
  • Objectives of panel modeling methods
  • Review
    • Random effects models
    • Fixed effects models
    • Cross-lagged panel models
    • SEM modeling
  • Problems
    • Dependence among repeated measures
    • Lagged dependent variables
    • Problems with reciprocal causation
  • Difference scores for T=3
  • Econometric solutions
    • Arellano-Bond
    • System estimator
    • ML methods
  • SEM for random effects
  • SEM for fixed effects
  • SEM with lagged endogenous and predetermined variables.
  • SEM compared with Arellano-Bond
  • SEM details
    • Evaluating model fit.
    • Relaxing constraints
    • Higher-order lags
    • Likelihood ratio tests
    • Allowing coefficients to vary with time
  • Missing data by FIML
  • Random coefficients
  • Higher-level clustering
  • Alternative estimators
  • Other latent variables
  • Categorical outcomes

COMMENTS BY RECENT PARTICIPANTS  

“Dr. Allison is a spectacular teacher! He obviously has years of experience with teaching because he organizes his slides well, knows the material inside and out, is very patient and responds to questions very well, and can judge the optimal pace of the class. I found the preparatory materials (recommended articles and slides) extremely helpful to review before the class began. I would highly recommend this class to other colleagues, especially those with a background in SEM.”
  Mary M. Mitchel, Johns Hopkins University

“This is a very informative, intense course on a novel approach that combines fixed effects model with cross-lagged model using the SEM approach. I truly appreciate that organized way that Dr. Alison provides information. It helps me to understand difficult concepts more easily.”
  Hye Won Kwon, University of Iowa

“As a graduate student, this course was extremely helpful. I learned more in two days than two semesters of the same stat courses. Dr. Allison is very knowledgeable and presents complex material beautifully and clearly.”
  Daniel Paulus, University of Houston

“I really appreciate the state-of-the-art insight shown by Paul Allison. Even as an only moderately experienced SEM user, I could follow well. The methods taught are very powerful for strategy/organization behavior studies.”
  Sebastian Fourné, WHU – Otto Beisheim School of Management

“This course provided a better explanation of fixed versus random effects and showed how they are hidden within procedures like PROC GLM. It gives many examples of reciprocal causation for different number of time points, lending to a better appreciation of the limitation of having fewer waves of data. This is a good segue into growth curve modeling.”
  Andy Kin On Wong, McMaster University & University Health Network

“This course provided an excellent discussion of how to apply different SEM models to panel data. It pulled from the latest publication on these topics and provided a framework to analyze my personal data. I was greatly appreciative of learning not just how to code with SEM, by the advantages and tools each software package offered.”
  Elizabeth Hecht, North Carolina State University