Part I: Bilinear spline growth models (BLSGMs) w(w/o) time-invariant covariates (TICs) in the framework of individual measurement occasions
Manuscript Title:
Obtaining interpretable parameters from reparameterized longitudinal models: transformation matrices between growth factors in two parameter-spaces (accepted for publication in Journal of Educational and Behavioral Statistics)
Description:
In this part, we developed four models in unstructured time framework:
- BLSGMs for estimating fixed knots (not in the manuscript)
- BLSGMs for estimating random knots (not in the manuscript)
- BLSGMs-TICs for estimating fixed knots (in the manuscript)
- BLSGMs-TICs for estimating random knots (in the manuscript)
Example data:
Source Code:
R package: nlpsem
The models developed in this project are now part of R package nlpsem (dependency: OpenMx), where we provide functions capable of 'calculating' starting values from the input and generate the estimates described in the manuscript.
- R package: nlpsem (For OS, R version, and OpenMx version, see the demo)
MPlus 8
- BLSGMs for estimating fixed knots (not in the manuscript, provide for the cases that the TICs that are not the primary interest)
- BLSGMs for estimating random knots (not in the manuscript, provide for the cases that the TICs that are not the primary interest)
- BLSGMs-TICs for estimating fixed knots
- BLSGMs-TICs for estimating random knots
Part II: Bilinear spline growth mixture models (BLSGMMs) in the framework of individual measurement occasions
Manuscript Title:
Two-step growth mixture model to examine heterogeneity in nonlinear trajectories (accepted for publication in Journal of Behavioral Data Science)
Description:
In this part, we developed two models in unstructured time framework:
- Two-step BLSGMMs for estimating fixed knots (1) First step: multivariate Gaussian mixture models for clustering trajectories with considering uncertainty; (2) Second step: investigate predictors for clusters
- One-step BLSGMMs for estimating fixed knots (Mixture of experts models for clustering and estimating coefficients simultaneously)
Example data:
Demo:
- R package: OpenMx (For OS, R version, and OpenMx version, see the demo)
Source Code:
R package: OpenMx
The Two-step model developed in this project is now part of R package NonLinearCurve (dependency: OpenMx), where we provide functions capable of 'calculating' starting values from the input and generate the estimates described in the manuscript.
(https://github.com/Veronica0206/NonLinearCurve/blob/main/R/BLSGMM_2steps.R)