02.11.2022Talk by Dr. Charles Driver

Multivariate longitudinal models such as the cross-lagged panel model and similar varieties offer hopes of deep insight into the relations between developmental constructs. Many complexities in such modelling exist, and all too often such models are misunderstood and misapplied, with sometimes fantastical 'causal' conclusions resulting. I will discuss the use of such models as representations of theory, some traps and partial solutions for causal interpretation, as well as what we can learn from such modelling approaches even when we accept the models are imperfect. Empirical applications include considerations of the relation between mathematics and language development, based on data from the Mindsteps online learning software in use in Switzerland.
Flyer (PDF, 661 KB)