Algebra Seminar
Title: Linear Structural Equation Models
Abstract: Linear structural equation models (L-SEMs) are a class of multivariate statistical models which study possible causal dependencies among variables. These models are associated with a path diagram, a graph with a directed acyclic part and a bidirected part. When a model has been specified, it is of interest to determine whether the model parameters can be recovered from the covariance matrix which they define. In this talk we will introduce the topic of causal inference using L-SEMs and present recent progress on generic identifiability using algebraic methods.