Hi, I have survey data collected at over 350 colleges. My dependent variable is continuous and all independent variables/covariates are binary - group membership indicators. The dependent variable is the mean of several 0-1 variables then standardized to have a mean of 50 and a standard deviation of 25.
All variables in my model are observed and the model is fully saturated.
I have set up a model with the following parameters (skipping Names and usevariables).
VARIABLE: cluster is col_id ; missing are all . ; within = q16 enrl female dev ;
ANALYSIS: type = twolevel ;
MODEL: %WITHIN% ACC ON q16 fulltime female dev ;
OUTPUT sampstat stdyx ;
The independent variables are all coded as 0/1 where 1 represents the characteristic described by the variable (dev indicates a student is not college-ready - needs developmental courses).
The between level means estimate for ACC is 41.602 with a S.E. of 0.274. The between level variance estimate for ACC is 12.942 with a S.E. of 1.584. The mean estimate for ACC is quite a bit lower than I expected. It has been a long time since I have done this type of analysis so I'm wondering if I have missed something.
I'm planning to expand this model to a intercept and slopes as outcome model, but I want to make sure I understand what I am doing at a less complex level before I move on.