search

ld-archer / E_FEM

This is the repository for the English version of the Future Elderly Model, originally developed at the Leonard D. Schaeffer Center for Health Policy and Microsimulation.
MIT License
3 stars 1 forks source link

Check T-test comparison populations #49

Closed ld-archer closed 3 years ago

ld-archer commented 3 years ago

Part of #45

Make sure the T-test comparison populations (from ELSA) are the exact same populations. I.e. there should be the same people in each group in CV2, and CV1 needs comparative. Look at age, sex, BMI, education, check some chronic diseases.

ld-archer commented 3 years ago

Need to also check the replenishing populations. Could be including/not including a replenishing population in any of the model runs that cause differences over time.

The changes to the minimal model T-test population have made this clear. The p-value in wave 3 is 0, which shows the means are not equal. This then improves to reach 0.993 in wave 5, and drops right back down to 6e-04 by wave 8. I think this shows either a problem with the replenishing population, or potentially a problem with the mortality rate? ELSA doesn't report any deaths in wave 7 or 8 so we can't check the mortality rate past this point.

ld-archer commented 3 years ago

CV2 needs a replenishing population, will create a new issue to go through that process.