Multilevel Modeling Growth Curve in Python

Dataset description:Permalink

This project uses the “drvisits” dataset. In 1997 Germany reformed their healthcare system to try to reduce costs. Ninety percent of the population is enrolled in this healthcare system and it is publically funded. These data track a sample of the German population before and after the reform. The outcome of interest is numvisit, the number of doctor visits in the last 3 months before the interview. Additional variables collected are: * age and education (in years) ; * married (0/1); * self-reported health status (badh==1 if bad health); * loginc is log of household income; * reform (0==pre; 1==post); * id – subject identifier. Note, there are at most two records for any one id, and in some cases there is just one.

Step 1: Evaulate the need for transformation of variablesPermalink

Step 2: Examine the distributions of outcome variablePermalink

Step 3: Clean some variables and generate some interactions for subsequent modelsPermalink

Step 4: Perform a one sample (Paired) t-test on the difference between pre- and post- reform outcomesPermalink

Step 5: Fit an initial MLM modelPermalink

Step 6: Evaluate modelsPermalink

Step 7: Make predictionsPermalink

Step 8: Critique resultsPermalink