Structural equation modelling and causal inference: an online course
This online course consists of ten lectures in the form of videos plus the RStudio files used in the videos. These videos vary from 1 to almost 3 hours in length, with a total length of 18 hours and 5 minutes. The videos are based on the book “Cause and correlation in biology: A user’s guide to path analysis, structural equations modelling, and causal inference” and approximately follow the order of presentation in that book, although there are some newer details that are not found in the book. I recommend that you view these videos in the order below. Although the videos are self-contained, you will get the best learning experience if you first view a video and then read the associated section(s) of the book.
The first two videos are freely available on YouTube; the links to these first two videos are given below and you are welcome to distribute these two links to anyone who might be interested. The links to the remaining eight videos will be sent to you in a series of emails to your @gmail.com address. In order to use these links, you must have a @gmail.com email address. If you don’t yet have a gmail.com email address, then go to the Google site and set one up. These links will remain available to you for two months. The only way to view these videos is via these links, so don’t loose them!
Most of these videos include practical sessions in which I explain how to carry out analyses in the R statistical environment. Therefore, you will need a recent version of R as well as RStudio on your computer if you wish to do the associated analyses yourself.
Finally, purchasing this online course as an individual also allows you to contact me (within reason…) via email, TEAMS, SKYPE or ZOOM if you have any questions. Online courses for groups automatically include group sessions via TEAMS, SKYPE or ZOOM.
Lectures
Introduction https://youtu.be/TgBq4uOA4ck (42 minutes): Axioms of causality; Randomized vs. controlled experiment; limitations; Physical vs “observational” control. Chapter 1 of book.
DAGs, d separation and data https://youtu.be/TgBq4uOA4ck (2 hours, 5 minutes): Directed Acyclic Graphs (DAGs) as a universal translator; conditioning in a DAG; d-separation; ggm library, Markov condition; consequences of d-separation; logic of causal inference. Chapter 2 of book.
Dsep tests Link provided upon purchase (1 hour, 22 minutes): History of SEM + Sewall Wright; D-sep tests + extensions (glm, mixed models, phylogenetic corrections); estimating the path coefficients; decomposition of causal effects. Chapter 3 of book.
PiecewiseSEM Link provided upon purchase (2 hours, 20 minutes): Using the piecewiseSEM package in R to conduct dsep tests of path models. Not in book.
Equivalent models and AIC Link provided upon purchase (1 hour, 49 minutes): Defining and finding d-separation equivalent models; using AIC statistics to choose between non-equivalent models. Not in book.
Covariance-based path analysis Link provided upon purchase (1 hour, 40 minutes): D-seps tests vs classical SEM; path analysis and SEM; Covariance algebra; population vs sample covariance matrices; structural equations; fixed/free parameters and estimation via maximum likelihood; Maximum likelihood X2 statistic; degrees of freedom; logic of ML SEM testing​. Chapter 4 of book.
Covariance based path analysis with lavaan Link provided upon purchase (1 hour, 14 minutes): Using the lavaan package in R to perform covariance-based path analysis. Chapter 4 of book.
Latent variables and measurement models Link provided upon purchase (2 hours, 45 minutes): The notion of a latent variable; the notion of a measurement model; types of latent variables; fitting measurement models​ in lavaan. Chapter 5 of book.
Structural equation models Link provided upon purchase (2 hours, 45 minutes): The full structural equation model​; robust methods for non-normality; small sample sizes; bootstrap; modification indices.​ Chapter 6 of book.
Multigroup models Link provided upon purchase (1 hour, 47 minutes): Fitting, using and understanding SEM in a multigroup context. Chapter 7 of book.