Pre-Conference Workshop

New Features in Mplus

Lecturers:  Bengt Muthén, Tihomir Asparouhov and Ellen Hamaker

The pre-conference workshop runs from 8:30 am to 6:00 pm, and has two parts:

(1) Analyses you probably didn’t know that you could do in Mplus

  • with highlights from the forthcoming book

Regression and Mediation Analysis Using Mplus (Muthen, B., Muthen, L., & Asparouhov, T.)

(2) Preview of the forthcoming Mplus Version 8

Morning Session

Part (1) will sample examples from a range of Mplus topics such as heteroscedasticity modeling for regression and mediation analysis using the CONSTRAINT option and random coefficient modeling; Heckman modeling; Bayesian approach to missing data on binary covariates; Monte Carlo studies of moderated mediation;sensitivity analysis of mediator-outcome confounding; counterfactually-defined causal effect mediation modeling with binary outcomes and binary and nominal mediators; and two-part mediation analysis with floor and ceiling effects.

Afternoon Session

Part (2) provides a preview of the forthcoming Mplus Version 8 which features new methods for analyzing intensive longitudinal data such as that obtained from ecological momentary assessments. Long time series for both n=1 and n >1 are discussed. The focus is on time-series analyses with auto-regressive and moving-average components both for observed-variable models such as regression and mediation analysis and for latent-variable models such as factor analysis and structural equation modeling. Applications to be discussed include:

  • multilevel AR(1) model with random mean and random AR parameter
  • latent multilevel AR(1) model with multiple indicators and random innovation error
  • latent multilevel VAR(1) model and dynamical networks
  • dynamic SEM
  • dynamic latent class analysis using hidden Markov and Markov-switching AR(1) models

The pre-requisites for this workshop are an understanding of Linear and Logistic regression; computer is not needed.

Note:  The discussion of methods for intensive longitudinal data continues on Tuesday morning with the keynote presentation by Bengt Muthen (Which methods do we need for intensive longitudinal data?) and on Tuesday afternoon with the state-of-the-art presentation by Tihomir Asparouhov (Latent class analysis for intensive longitudinal data, Hidden Markov processes, Regime switching models and Dynamic Structural Equations in Mplus 8).

Note from Bengt Muthén

Dear participant in the M3 pre-conference workshop on Mplus,

 

As the current M3 pre-conference workshop description shows, we begin at 8:30 on Monday morning the 23rd and might go as long as 6pm.  The location is the Business Building (next to the building the meeting usually takes place in), room 106.  Lunch (not provided) will be from 12 – 1:30.

 

The Part 1 morning session that features book highlights will involve logistic and probit regression so if you haven’t looked at that in a while you want to refresh your memory.   The morning as well as the afternoon will also draw on Bayesian analysis so you might want to look at the handout and video from our Topic 9 short course and/or the paper on our website under Papers, Bayesian Analysis:

 

Muthén, B. (2010). Bayesian analysis in Mplus: A brief introduction. Technical Report. Version 3. Clickhere to view Mplus inputs, data, and outputs used in this paper.
download paper contact author show abstract

 

The time-series topic of the Part 2 afternoon session will be presented by Ellen Hamaker with contributions from Tihomir Asparouhov about the Mplus implementation.  Ellen has considerable experience with these kinds of analyses.  My contribution to this topic will be presented on Tuesday in my keynote address in the morning (Which methods do we need for intensive longitudinal data?). 

Note 1: The afternoon will be long.  After Ellen Hamaker has led two sessions, Tihomir Asparouhov will lead a session 5-6:30 (or a bit longer).  So make late dinner plans. I don’t think you will regret trying to stay focused that long.

There will be no computing so you don’t have to bring a computer with Mplus on it.

Mplus Version 8 is currently in its development stage.  It will not be released until this Fall and the work we discuss might be released in two steps as 8.0 and 8.1.  The workshop afternoon provides a preview to give you an opportunity to learn more about the topic and to give us feedback on what we have done and what we have in mind to do.

Note 2: Also, Tihomir’s state-of-the art talk Tuesday afternoon (Latent class analysis for intensive longitudinal data, Hidden Markov processes, Regime Switching Models, and Dynamic Structural Equations in Mplus 8).  So if you are coming to the workshop to hear about the future of Mplus Version 8, you might want to consider staying an extra day.

 

I look forward to seeing you soon,

 

Bengt