SMART ORTAL

Online Resource and Technical Assistance Library | Inclusionary and Integrated Mental Health Education and Supports

Universal SEBMH Screening
Resource Links
Classroom Practices
Artifacts
Modules

This session introduces foundational principles of statistical modeling with an emphasis on real-world application across school mental health, injury prevention, and related fields. Designed for researchers with beginner to intermediate statistical experience, the session unpacks core concepts such as the goals of modeling (description, inference, prediction), common assumptions, and the difference between statistical and causal interpretations. Through a hands-on demonstration using R/STATA and a publicly available dataset, participants can explore how to fit and interpret a basic regression model, assess model assumptions, and think critically about what models can—and cannot—tell us. No prior experience with R/STATA is required, and all code is available through RStudio for optional exploration.

In this workshop, participants will learn what a mixed model consists of, review key concepts and notations (e.g., intercepts, slopes, etc.), and have the opportunity to apply this methodology using sample sleep study data in RStudio.

In this workshop, participants will learn how to use Exploratory Factor Analysis (EFA) to uncover the latent structure underlying a set of survey or assessment items—i.e., when multiple observed measures reflect a smaller number of unobserved (i.e., latent) constructs. We’ll build an intuitive workflow: when EFA is appropriate (vs. PCA/CFA), how to choose an extraction method and rotation, how to determine the number of factors (e.g., scree/parallel analysis), and how to interpret loadings, cross-loadings, and communalities to refine a measure. The session will include a live R walkthrough, with practical guidance on common pitfalls (over factoring, Heywood cases, and “pretty” solutions that don’t replicate).

In this workshop, participants will learn how to use Confirmatory Factor Analysis (CFA) to test whether a hypothesized factor structure adequately fits observed data—i.e., when you have a theoretical or empirically derived model (e.g., from EFA) specifying which observed indicators load onto which latent constructs, and you need to evaluate that model’s fit. We’ll build on the EFA workflow covered in the February session: when CFA is appropriate (vs. EFA/PCA), how to specify and identify a measurement model, how to evaluate model fit (e.g., chi-square, CFI, TLI, RMSEA, SRMR), and how to interpret standardized loadings, modification indices, and residuals to diagnose misfit. The session will include a live R walkthrough (using lavaan), with practical guidance on common pitfalls (model non-convergence, correlated residuals that mask misspecification, over-reliance on modification indices, and the difference between a model that “fits” and a model that’s theoretically defensible).

Find Slides & Recording here following the March 26th workshop.

Reports and Publications