I often read blogs from professors of statistics talking about how they prefer to teach models from a "distribution-centric" perspective rather than otherwise. However, I've never really been able to understand this point and it's differences from other methods of communicating models.
Could anyone provide an explanation for what is meant by this and provide a simple example (say a general linear model) and a more complex one (say a mixed-effects model)? Any literature which discusses the differences in pedagogy would also be immensely helpful.
I am a postgrad biologist who does some undergrad statistics tutoring, and I am very keen to learn new ways to approach teaching these topics.