Application of Dynamic Multivariate Panel Models in Social Work Research

Background

Intensive longitudinal data collected using experience sampling and diary methods, a.k.a., ecological momentary assessment (EMA) data, are growingly important in social work research, especially that on behavioral health. EMA data generally comprise of multiple subjects who provide frequent reports on events of their daily lives in the format of electronic diaries. Reports could be time-based, following a fixed-schedule, randomly triggered, or event-triggered.

Contingent on the research questions, data characteristics, and the assumed distribution of the response variables, EMA data can be analyzed in various models such as dynamic panel models, fixed effects models, dynamic structural equation models, cross-lagged panel models, and their various extensions. However, these modeling approaches typically impose some restrictive assumptions on the data-generating process, such as Gaussian errors, effects that are constant in time, or univariate responses.

Based on a Bayesian approach, the dynamic multivariate panel model (DMPM) supports both time-varying and time-invariant effects, multiple simultaneous responses across a wide variety of distributions, and arbitrary dependency structures of lagged responses of any order. The DMPM is especially relevant to social work research, since it can easily be applied in intervention studies in which the treatment effects increase or decrease over time, enabling more rigorous causal inference and richer interpretation. Given the lack of studies based on the DMPM in social work, this workshop will discuss the steps in building a DMPM, strategies for model appraisal, and reporting guidelines based on real data.

Objectives

Upon completion of the workshop, participants will be able to:

1. recognize how the dynamic multivariate panel model is promising in modeling intensive longitudinal data,

2. demonstrate understanding of the steps of model building for dynamic multivariate panel models, and

3. use the dynamite package in R to fit dynamic multivariate panel models.

Content

Based on the pedagogical techniques of learning-science-by-doing-science and problem-based learning, this workshop aims to address the following content:

1. summary of the characteristics and advantages of dynamic multivariate panel models,

2. a concise and systematic set of procedures to establish dynamic multivariate panel models,

3. application of the dynamic multivariate panel model in social work research using a problem example based on real data,

4. a step-by-step demonstration of model construction, estimation, prediction, posterior inference, visualization, and diagnostics using the dynamite package, and

5. how to draw causal inferences and conclusions.

Implications

Overall, this focus on the dynamic multivariate panel model can be invaluable to social work researchers, faculty, students, and professionals who wish to leverage state-of-the-art panel models to derive more rigorous causal inferences and conclusions. Featuring real data analyzed with a cutting-edge statistical package using the Bayesian approach, this workshop simplifies esoteric panel models making them more applicable to social work research. By becoming familiar with the best practices in applying the dynamic multivariate panel model, attendees can gain the expertise necessary to use a powerful statistical tool that can help enhance the examination of causal effects and advance intervention research in social work.