Dynamic Structural Equation Modeling - Online Course
A 3-Day Livestream Seminar Taught by
Dan McNeishWednesday, November 11 —
Friday, November 13, 2026
10:00am-12:30pm (convert to your local time)
1:30pm-3:30pm
Dynamic structural equation modeling (DSEM) is a recently developed analytic framework that combines aspects of multilevel modeling, structural equation modeling, and time-series analysis. Although DSEM has many applications, it is particularly useful for intensive longitudinal data.
Roughly speaking, longitudinal data is considered intensive if you have 15 or more repeated measures on the same individuals over a relatively short time span, such as a few days or weeks. This type of data has become increasingly common as technological advances like smartphones and wearables continue to transform how data are collected, how studies are designed, and what research questions can be asked.
While traditional longitudinal models focus on growth over longer durations, intensive longitudinal models focus on momentary changes over short durations. For instance, a growth model may be interested in how anxiety changes over 12 months, but an intensive longitudinal model may be interested in why anxiety was low at 12pm, spiked at 4pm, and receded at 8pm.
In this seminar, you’ll learn about both foundational and intermediate topics in DSEM. The course will emphasize the capabilities of DSEM that distinguish it from more traditional methods for intensive longitudinal analysis, such as mixed models and univariate time-series models. These capabilities include:
-
- Modeling outcomes that are latent or based on measurement scales composed of multiple item responses.
- Estimating multivariate models for mediation (e.g., how chains of effects unfold over time).
- Dyadic data (e.g., how related individuals like romantic couples or managers/employees affect each other over time).
Starting November 11, this seminar will be presented as a 3-day synchronous, livestream workshop via Zoom. Each day will feature two lecture sessions with hands-on exercises, separated by a 1-hour break. Live attendance is recommended for the best experience. If you can’t join in real time, recordings will be available within 24 hours and accessible for four weeks after the seminar.
Closed captioning is available for all live and recorded sessions. Captions can be translated to a variety of languages including Spanish, Korean, and Italian. For more information, click here.
ECTS Equivalent Points: 1
More details about the course content
After completing this seminar, you’ll have a solid foundation in:
-
- The differences between intensive longitudinal data and traditional longitudinal data.
- The opportunities and unique research questions that can be asked and answered with DSEM and intensive longitudinal data.
- The conceptual underpinnings of DSEM and how it can handle the special features of intensive longitudinal data.
- How to leverage concepts from multilevel and structural equation modeling to build models that capture idiosyncratic features of intensive longitudinal data and designs.
After completing this seminar, you’ll have a solid foundation in:
-
- The differences between intensive longitudinal data and traditional longitudinal data.
- The opportunities and unique research questions that can be asked and answered with DSEM and intensive longitudinal data.
- The conceptual underpinnings of DSEM and how it can handle the special features of intensive longitudinal data.
- How to leverage concepts from multilevel and structural equation modeling to build models that capture idiosyncratic features of intensive longitudinal data and designs.
Computing
Examples and software code will be provided in Mplus (version 8 or higher is required for DSEM).
DSEM is a relatively new method, so commercial software like Mplus currently has the most functionality and greatest ease of implementation.
R and Stata can accommodate some types of DSEM models, but Mplus currently has the most advanced DSEM capabilities. Course examples and instruction will exclusively be in Mplus, but you may work in the software package of your choice. Note that some of the advanced models covered in the course may not currently be supported or may be difficult to fit outside of Mplus.
If you’d like to familiarize yourself with Mplus basics before the seminar begins, we recommend reading through UCLA’s short guide here.
Examples and software code will be provided in Mplus (version 8 or higher is required for DSEM).
DSEM is a relatively new method, so commercial software like Mplus currently has the most functionality and greatest ease of implementation.
R and Stata can accommodate some types of DSEM models, but Mplus currently has the most advanced DSEM capabilities. Course examples and instruction will exclusively be in Mplus, but you may work in the software package of your choice. Note that some of the advanced models covered in the course may not currently be supported or may be difficult to fit outside of Mplus.
If you’d like to familiarize yourself with Mplus basics before the seminar begins, we recommend reading through UCLA’s short guide here.
Who should register?
This seminar will benefit graduate students, postdocs, early career researchers, and experienced researchers in academia, government, or industry looking to gain a better understanding of intensive longitudinal data and how DSEM can be applied to understand moment-to-moment dynamics in these data.
The course does not assume any background in DSEM, multilevel modeling, or structural equation modeling. However, you should have knowledge of the principles and practice of linear regression, which will not be reviewed. Any experience or familiarity with multilevel modeling or structural equation modeling will be helpful, but no knowledge of these topics is assumed, and relevant topics from these areas will be covered to the extent needed to engage with DSEM.
Some equations will be provided to foster a deeper understanding of the modeling methods covered, but the focus will be on application and interpretation of the methods. By the conclusion of the course, you should have the knowledge and tools you need to fit DSEM models to your own data.
This seminar will benefit graduate students, postdocs, early career researchers, and experienced researchers in academia, government, or industry looking to gain a better understanding of intensive longitudinal data and how DSEM can be applied to understand moment-to-moment dynamics in these data.
The course does not assume any background in DSEM, multilevel modeling, or structural equation modeling. However, you should have knowledge of the principles and practice of linear regression, which will not be reviewed. Any experience or familiarity with multilevel modeling or structural equation modeling will be helpful, but no knowledge of these topics is assumed, and relevant topics from these areas will be covered to the extent needed to engage with DSEM.
Some equations will be provided to foster a deeper understanding of the modeling methods covered, but the focus will be on application and interpretation of the methods. By the conclusion of the course, you should have the knowledge and tools you need to fit DSEM models to your own data.
Seminar outline
Day 1
-
- Overview of intensive longitudinal data
- Recent increase in behavioral science for mood, affect, and health research
- Broad overview of dynamic structural equation models
- Combination of multilevel, SEM, and time-series analysis to accommodate features of intensive longitudinal data and psychological/behavioral data
- Within/Between person decomposition to isolate momentary and habitual effects
- Latent mean centering to account for unreliability and missing data
- Multilevel autoregressive models for N > 1 data
- Establishing between-person variance
- Time-varying covariates, time-invariant covariates, and moderation
- Overview of Bayesian estimation
- Kalman filter for unequally spaced intervals
Day 2
-
- Heterogeneous variance DSEM
- Variance as the main outcome
- Predictors associated with more or less stability in time series
- Accommodating non-stationarity
- Residual DSEM for systematic trends
- Vector autoregressive DSEM
- Dynamics of multiple outcomes simultaneously
- Dynamics of multiple related people simultaneously (e.g., dyadic DSEM)
Day 3
-
- Categorical DSEM
- Dynamics of binary or ordinal variables
- Probit models
- Missing data
- Missing at random methods
- Selection models and shared parameter models for data that are not missing at random
- Measurement Error models
- Within-person reliability
- Estimating effect while theoretically removing measurement error
- Within-person mediation in DSEM
Time permitting
-
- Cross-classified DSEM
- Parameters vary randomly across multiple clustering units
- Person-specific and time-specific mediation
- Latent variable DSEM
- Modeling dynamics in latent constructs
- Building measurement models
- Assessing person-invariance and time-invariance
Day 1
-
- Overview of intensive longitudinal data
- Recent increase in behavioral science for mood, affect, and health research
- Broad overview of dynamic structural equation models
- Combination of multilevel, SEM, and time-series analysis to accommodate features of intensive longitudinal data and psychological/behavioral data
- Within/Between person decomposition to isolate momentary and habitual effects
- Latent mean centering to account for unreliability and missing data
- Multilevel autoregressive models for N > 1 data
- Establishing between-person variance
- Time-varying covariates, time-invariant covariates, and moderation
- Overview of Bayesian estimation
- Kalman filter for unequally spaced intervals
- Overview of intensive longitudinal data
Day 2
-
- Heterogeneous variance DSEM
- Variance as the main outcome
- Predictors associated with more or less stability in time series
- Accommodating non-stationarity
- Residual DSEM for systematic trends
- Vector autoregressive DSEM
- Dynamics of multiple outcomes simultaneously
- Dynamics of multiple related people simultaneously (e.g., dyadic DSEM)
- Heterogeneous variance DSEM
Day 3
-
- Categorical DSEM
- Dynamics of binary or ordinal variables
- Probit models
- Missing data
- Missing at random methods
- Selection models and shared parameter models for data that are not missing at random
- Measurement Error models
- Within-person reliability
- Estimating effect while theoretically removing measurement error
- Within-person mediation in DSEM
- Categorical DSEM
Time permitting
-
- Cross-classified DSEM
- Parameters vary randomly across multiple clustering units
- Person-specific and time-specific mediation
- Latent variable DSEM
- Modeling dynamics in latent constructs
- Building measurement models
- Assessing person-invariance and time-invariance
- Cross-classified DSEM
Payment information
The fee of $995 USD includes all course materials.
PayPal and all major credit cards are accepted.
Our Tax ID number is 26-4576270.
The fee of $995 USD includes all course materials.
PayPal and all major credit cards are accepted.
Our Tax ID number is 26-4576270.