Structural Equation Modelling for Ecologists and Evolutionary Biologists (SEMR02)
This course will be delivered by Dr. Jonathan Lefcheck (author of the piecewise package) from the 19th - 23rd November 2018 in Glasgow City Centre.
The course is a primer on structural equation modelling (SEM) and confirmatory path analysis, with an emphasis on practical skills and applications to real-world data.
Structural equation modelling is a rapidly growing technique in ecology and evolution that unites multiple hypotheses in a single causal network. It provides an intuitive graphical representation of relationships among variables, underpinned by well-described mathematical estimation procedures. Several advances in SEM over the past few years have expanded its utility for typical ecological datasets, which include count data, missing observations, and nested or hierarchical designs.
We will cover the basic philosophy behind SEM, provide approachable mathematical explanations of the techniques, and cover recent extensions to mixed effects models and non-normal distributions. Along the way, we will work through many examples from the primary literature using the open-source statistical software R (www.r-project.org). We will draw on two popular R packages for conducting SEM, including lavaan and piecewiseSEM.
Participants are encouraged to bring their own data, as there will be opportunities throughout the course to plan, analyze, and receive feedback on structural equation models.
Introduction to SEM
Module 1: What is Structural Equation Modeling? Why would I use it?
Module 2: Creating multivariate causal models
Module 3: Fitting piecewise models
Readings: Grace 2010 (overview), Whalen et al. 2013 (example)
SEM Using Likelihood
Module 4: Fitting Observed Variable models with covariance structures Module 5: What does it mean to evaluate a multivariate hypothesis?
Module 6: Latent Variable models Module 7: ANCOVA revisited & Nonlinearities
Readings: Grace & Bollen 2005, Shipley 2004
Optional Reading: Pearl 2012, Pearl 2009 (causality)
Module 8: Introduction to piecewise approach
Module 9: Incorporation of random effects models
Model 10: Autocorrelation Reading: Shipley 2009; Lefcheck 2016
Advanced Topics with Likelihood and Piecewise SEM
Module 11: Multigroup models and non-linearities
Module 12: Composite Variables
Module 13: Phylogenetically-correlated data
Module 14: Prediction using SEM
Module 15: How To Reject A Paper That Uses SEM
Readings: Grace & Julia 1999, von Hardenberg & Gonzalez‐Voyer 2013
Open Lab and Final Presentations
Email [log in to unmask]
Check out our sister sites,
www.PRstatistics.com (Ecology and Life Sciences)
www.PRinformatics.com (Bioinformatics and data science)
www.PSstatistics.com (Behaviour and cognition)
You may leave the list at any time by sending the command
to [log in to unmask], leaving the subject line blank.