Skip to contents

Dynamic structural equation models

Codecov test coverage Documentation

Package dsem fits dynamic structural equation models, which includes as nested submodels:

  1. structural equation models
  2. vector autoregressive models
  3. dynamic factor analysis
  4. state-space autoregressive integrated moving average (ARIMA) models

The model has several advantages:

  • It estimates direct, indirect, and total effects among system variables, including simultaneous and lagged effects and recursive (cyclic) dependencies
  • It can estimate the cumulative outcome from press or pulse experiments or initial conditions that differ from the stationary distribution of system dynamics
  • It estimates structural linkages as regression slopes while jointly imputing missing values and/or measurement errors
  • It is rapidly fitted as a Gaussian Markov random field (GMRF) in a Generalized Linear Mixed Model (GLMM), with speed and asymptotics associated with each
  • It allows granular control over the number of parameters (and restrictions on parameters) used to structure the covariance among variables and over time,

dsem is specifically intended as a minimal implementation, and uses standard packages to simplify input/output formatting:

  • Input: time-series defined using class ts, with NA for missing values
  • Input: structural trade-offs specified using syntax defined by package sem
  • Output: visualizing estimated trade-offs using igraph
  • Output: access model output using standard S3-generic functions including summary, predict, residuals, simulate, and AIC

Please see package vignettes for more details regarding syntax and features.

Citation

Thorson, J. T., Andrews, A. G., Essington, T., & Large, S. (2024). Dynamic structural equation models synthesize ecosystem dynamics constrained by ecological mechanisms. Methods in Ecology and Evolution 15(4): 744-755. https://doi.org/10.1111/2041-210X.14289