Unit of analysis: States (39)
Time: 3 years
Total observations: 117
Dependent variable: % children graduating within 4 years from high school
Independent variable: 8 composite scores ranging from 1-30. These scores did NOT change throughout the period of examination (i.e. 8 measures are time-invariant).
I am trying to determine whether to proceed as a fixed effect or mixed effect model. Does the fact that I have only 39 states (and not the population of all 50) mean I have to go mixed effect?
A followup: if it is mixed, I assume 1) my fixed variable would be the IVs, and 2) my random would be state?