Youran Lin, University of Alberta

In applied psycholinguistics, researchers often deal with complex, interrelated factors that contribute to large variations in outcome measures – Such factors may influence the outcomes directly or indirectly. Bilingual development is an issue of this nature, as it is shaped by multiple interacting dimensions, such as past language experience, current language exposure, cognitive abilities, and cultural identity.

Furthermore, these constructs are often not directly measured, but estimated with multiple sub-measures, which requires careful synthesis and integration.

Using simulated and realistic bilingual datasets, this course introduces statistical approaches for addressing such complexities. Participants will learn to:

1. Use directed acyclic graphs (DAG) to conceptualize and visualize complex relationships (e.g., moderation and mediation),

2. Use structural equation modeling (SEM) to model such relationships statistically, and

3. Incorporate factor analysis (FA) in structural equation modeling to derive latent variables from multiple sub-measures.

This course is designed as a guided workshop rather than a theoretical or mathematically intensive treatment of statistical methods. Participants will have ample time to engage in hands-on coding and practice applying the methods. Although bilingualism is used as an example of a complex psycholinguistic phenomenon, the methods introduced in this course are broadly applicable to research involving multidimensional constructs and questionnaire/assessment data.

Prior knowledge: Proficiency with R and experience with regression modeling in R

Participants should bring their own laptop. They are welcome to bring their own datasets. If you wish to practice with your own datasets, they should include continuous dependent variable(s) and multiple independent variables with potential mediating relationships.