I have some questions about how to translate my data structure and research Qs to syntax for the lmer function (lme4 package).
I am looking to predict teenagers scores on a mental health questionnaire (0-10) from 3 variables: age (continuous), sex (M/F) and my variable of interest, let's call it X for simplicity (continuous). I demeaned both my continuous predicted variables. Both the teenager and their parent filled out the questionnaire about the teenager, adding a third (within-subject) variable rater. I restructured the dataset into long format such that each subject has 2 rows for each outcome value: parent-report & self-report.
Number of subjects ~85
Number of outcome observations ~170
I am interested in the following effects:
- The fixed effect of X on outcome score (main interest)
- The interaction between X and sex on the outcome ("does X affect the outcome in one sex but not the other?")
- The fixed effect of sex on scores
- The fixed effect of age on scores
But I would also like to know whether the effects above are dependant on who is the rater? In this sense, rater is not a nuisance grouping variable whose effect on the outcome I want to account for. I would like to perform a test similar to a MANOVA but given that some subjects are missing some observations, I would prefer to use mixed models. As I understand it, linear mixed models can be used for multiple outcome data but I do not know how to phrase the syntax such that:
- I declare non-independant observations within subjects (rater falls within subject)
- I do not have a random slope for every subject (I have a relatively small sample size)
Using some specific examples, I'd like to know which (if any!) of the following capture my needs...
m1 <- lmer(score ~ X*sex+ age + (1+rater), data = mydata )
m2 <- lmer(score ~ X*sex + age + (1|rater), data = mydata )
# same as m1?
m3 <- lmer(score ~ X*sex + age + (1|ID/rater), data = mydata )
# Error: number of levels of each grouping factor must be < number of observations
# An issue related to missing data??
Any help (for any part of the above) is appreciated!
ID
is the subject ID. How many raters are there, and can you explain how rater is nested in subject ? $\endgroup$X*sex + age
as fixed effects but also in how these fixed effects vary by rater ? Is this an accurate description of your setup and research questions ? $\endgroup$