1
$\begingroup$

I am trying to fit a multilevel longitudinal model and I am not sure how to conceptualize the structure and how to specify it with lme4 notation.

My outcome of interest is the child's internalized symptoms scores, measured 12 times across childhood, by different informants at each wave of data collection. Not every informant was questioned at each wave. I see 2 different ways to conceptualize the nesting of the data :

Option 1 : Waves at level 1; Informants at level 2 and child at level 3.

enter image description here

Option 2 : Informants at level 1; waves at level 2 and child at level 3.

enter image description here

My question is how to specify a model with lme4 to account for the facts that measures by the same informant are correlated over time and that measures by different informants at a specific time point are correlated. My grouping variables are "id" for the child level; "wave" for the time level and "info" for the informant level. My covariates "X" are all at the child level. I would like to include random intercepts and slopes for each level. My guesses are :

Option 1 : Y ~ X + (wave|id) + (wave|id:info) Option 2 : Y ~ X + (info|id) + (info|id:wave)

Thanks,

Myriam

Improve this question
$\endgroup$

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.