Reading focus
Syllabus: How Can We Study Change Within People?
Compare the traditional and random-intercept models in the parent–adolescent illustration, keeping their assumptions visible.
Why you’re reading it. Evaluate what a temporal association describes before interpreting it as a mechanism.
A question to consider. If a lagged association disappears after separating stable differences, what conclusion is warranted, and why would “there can be no causal effect” go beyond the result?
Why this reading
A model can include repeated observations yet mix stable differences between people with fluctuations within them. Hamaker and colleagues show how that mixture changes the interpretation of cross-lagged associations and introduce an alternative decomposition.
Before you read
A cross-lagged panel model relates one variable at a later wave to both variables at an earlier wave. A random intercept represents stable differences in a person’s level. Separating that component changes what the remaining lagged coefficient describes.
Read for
- Compare the model diagrams, simulations and parent–adolescent illustration. Identify what each coefficient refers to.
- Examine the authors’ qualifications about measurement error, observation intervals, changing processes and alternative models.
Consider afterwards
- A model links parental responsiveness to later adolescent symptoms. If that association changes after separating stable differences, which interpretation has changed—and what still prevents a causal conclusion?
Editorial context
The paper establishes a problem for particular within-person interpretations of the traditional model. It does not make the random-intercept model universally preferable, nor does model fit establish a causal mechanism. Its alternative still requires assumptions about stability, measurement and time; other models may better match a different question.