1

I am struggling with how to describe level 2 data in my Multilevel Model in R. I am using the nlme package.

I have longitudinal data with repeated measures. I have repeated observations for every subject across many days.

The Goal:

  • Level 1 would be the individual observations within the subject ID
  • Level 2 would be the differences between overall means between subject IDs (Cluster).

I am trying to determine if Test scores are significantly affected by study time, and to see if it's significantly different within subjects and between subjects.

How would I write the script if I want to do "Between Subjects" ?

Here is my script for Level 1 Model

model1 <- lme(fixed = TestScore~Studytime, random =~1|SubjectID, data=dataframe, na.action=na.omit)

Below is my example dataframe

      `Subject ID` Observations TestScore Studytime
 1            1            1        50       600
 2            1            2        72       900
 3            1            3        82       627
 4            1            4        90      1000
 5            1            5        81       300
 6            1            6        37       333
 7            2            1        93       900
 8            2            2        97      1000
 9            2            3        99      1200
10            2            4        85       600
11            3            1        92       800
12            3            2        73       900
13            3            3        81      1000
14            3            4        96       980
15            3            5        99      1300
16            4            1        47       600
17            4            2        77       900
18            4            3        85       950

I appreciate the help!

Vanessa Lin
  • 21
  • 2
  • 1
    There's something fundamentally wrong with using a random effect, if you are interested in seeing if there is a significant difference between the groupings of your random effects. The random effect is not a measure of `mean` but of `variance`. So you can measure the difference in mean based on study time, but not across subjects. You can see if there is a significant variability between subjects by testing if the random effect is 0 (or approximately zero). I am also not sure of your difference in "between subjects" and just "subjectID". – Oliver Mar 17 '21 at 20:34

0 Answers0