Using Mplus to fit and test measurement and structural equation models

The videos are mp4 files which can be downloaded for viewing.  The slides are PDF files.

 If you have trouble viewing mp4 files, you could try using VCL or GOM (for example).

Questions or feedback? Email Alan Taylor.

Video 1 Introduction to observed and latent variable models (30 minutes).  Powerpoint slides

Video 2A brief introduction to Mplus (22 minutes).  Powerpoint slides

 Video 3Getting data into Mplus.  Powerpoint slides

     Part 1: (28 minutes)  Writing a tabbed data file with SPSS, creating an Mplus command file and running it.

     Part 2: (20 minutes)  Dealing with missing data when writing and reading the data, and using Stat Transfer.

Outtake 1: (3 minutes)  How not to create an Mplus data from Excel.

Outtake 2: (2 minutes)  A digression on the length of variable names in Mplus.

Video 4A multiple regression analysis with Mplus.

      Part 1: (2 minutes)   Outline of what is covered in video 4.  Powerpoint slides

      Part 2: (7 minutes)   Carrying out a multiple regression, and obtaining standardised results with the output command.

      Part 3: (12 minutes)  Obtaining a path diagram with Diagrammer and using @ to set a parameter to a fixed value.

      Part 4: (22 minutes)  Obtaining graphs with the plot command, saving data produced by Mplus with savedata, and merging the saved
                                       data with an SPSS data file.  The auxiliary and idvariable options are also used.
      Part 5:  (10 minutes)  A comparison of how Mplus and AMOS set up models (a follow-up of Part 3).   A techncal report
(tech1) is produced.  Powerpoint slides
      Part 6:  (6 minutes)   Using the define command to create an interaction term, and to centre the variables. 

Video 5A path analysis with observed variables.  Powerpoint slides (including a brief description of estimators used by Mplus)

       Part 1:  (6 minutes)   Outline of what is covered in video 5.

      Part 2:  (16 minutes)  Setting up and testing the basic model, using model indirect to test indirect effects, and adding a covariance.

      Part 3:  (8 minutes)  Using bootstrapping and requesting confidence intervals.

      Part 4:  (12 minutes)  Including a categorical dependent variable, seeing how Mplus automatically chooses estimators, and specifying the
                                       estimator to be used.  A postscript on mediation.

Video 6A confirmatory factor analysis - measurement model.  Powerpoint slides

       Part 1:  (4 minutes)   Outline of what is covered in video 6.

      Part 2:  (6 minutes)  Why are we testing a measurement model?  (A rambling and repetitive disquisition for those with spare time.)

      Part 3:  (15 minutes)  Setting up and running the model, assessing GOF, and requesting modification indices (with a discussion of MIs).

      Part 4:  (4 minutes)  Adding a covariance between indicator residuals.  Mention of an article on the importance of taking shared method
                                     variance into account in SEm

Video 7A path model with latent variables.  Powerpoint slides

       Part 1:  (2 minutes)   Outline of what is covered in video 7.

      Part 2:  (9 minutes)  Fitting the model and comparing the results with those for the same model with observed variables.

      Part 3:  (7 minutes)  Testing a direct effect and following up an unexpected result.

      Part 4:  (8 minutes)  Examining the reason for the unexpected direct effect using the @ operator.

       Part 5:  (14 minutes)  Obtaining MIs and using estimators which are robust to non-normality.

 Video 8How Mplus handles missing data.  Powerpoint slides

       Part 1:  (12 minutes)   Outline of what is covered in video 8, a review of missing data mechanisms, and the Mplus approach to
                                        missing data.

      Part 2:  (6 minutes)  How the missing data were generated, and the use of type=basic to obtain information about missing data.

      Part 3:  (4 minutes)  Investigation of the pattern of missing data for the variables involved in the analysis, and fitting an
                                     observed-variable regression model with missing data.

      Part 4:  (7 minutes)  Bringing all variables into the model by requesting a covariance, and a description of Full Information
                                     Maximum Likelihood.

       Part 5 :  (4 minutes)  Bringing all variables into the model by requesting variances and means.

       Part 6:  (5 minutes)  A latent variable path model with missing data.

       Part 7:  (4 minutes)  Multiple imputation (MI) and the outline of what is to be covered on MI.

       Part 8:  (12 minutes)  Producing MI data in SPSS, writing the data out for Mplus and analysing the data in Mplus.

       Part 9:  (12 minutes)  Producing MI data in Mplus using the H1 (unrestricted) and H0 (model-based) methods.

       Part 10:  (19 minutes)  Analysing the Mplus-produced MI data in Mplus, and a description of the Fraction of Missing Information index.

       Part 11 :  (8 minutes)  Using auxiliary variables to improve FIML estimates in the presence of missing data, and setting up a simulation.

       Part 12:  (10 minutes)  Using auxiliary variables in Mplus.

Video 9:  Latent growth models - LGMsPowerpoint slides

       Part 1:  (14 minutes)   Outline of what is covered in video 9, and an introduction to latent growth models.

       Part 2:  (19 minutes)  Using Mplus to fit a LGM to simulated data.

       Part 3:  (10 minutes)  Using Mplus to fit a LGM to real data with three time points, and applying constraints to allow a
                                       quadratic component.  Using the plot command.

       Part 4:  (8 minutes)   Saving factor scores

       Part 5:  (17 minutes)  Ftting a LGM to data for eight time points and with missing data.  Allowing estimation of the paths to some
                                       time points.