How to explain the interaction term in the repeated measurement?

CCJ — Wed, 20 Apr 2022 02:20:56 GMT

Hello~

I want to explore the change in cognitive decline associated with air pollution exposure. I have two measurements of cognitive score (continuous variable) and multiple measurements of PM (continuous variable). My SAS code is as following:

proc glimmix empirical ;
class id;
model cognitive_score=PM year_air PM*year_air / dist=normal link=identity solution;
random intercept / subject=id;
run;

The individual effect of PM and year_air on cognitive change is significantly negative, but the effect of interaction term on cognitive change is significantly negative.

How I explain the interaction term? Should I plot the related plot to explain?

In addition, I found the intercept and the estimate of PM are too strange and too large when I put the variable of time (year_air) in the model. Why is it?

Did anyone can help me? Thank you.

Re: How to explain the interaction term in the repeated measurement?

PaigeMiller — Wed, 20 Apr 2022 11:43:17 GMT

How I explain the interaction term? Should I plot the related plot to explain?

Yes, plot it.

I found the intercept and the estimate of PM are too strange and too large when I put the variable of time (year_air) in the model.

This can happen when you add terms into the model (or remove terms from the model) if the term added (or removed) is highly correlated with other terms in the model.

topic How to explain the interaction term in the repeated measurement? in Statistical Procedures

How to explain the interaction term in the repeated measurement?

Re: How to explain the interaction term in the repeated measurement?