* Chapter 6: Model Diagnostics using Stata .

* Fit Model 6.3 (final model) using REML estimation .
xi: xtmixed vsae i.sicdegp*age_2 age_2sq, || childid: age_2 age_2sq, noconst covariance(unstruct) variance

**************************************************************************************.
* Plot standardized conditional residuals vs. fitted values for each level of SICDEGP .
**************************************************************************************.

predict stresid, rstandard

predict fitted, fitted

twoway (scatter stresid fitted), by(sicdegp)

**************************************************************************.
* Assess whether variance of conditional residuals is constant across ages.
**************************************************************************.

predict resid, resid

scatter resid age_2

******************************************************.
* Assess normality of residuals within each SICD group.
******************************************************.

qnorm resid if sicdegp == 1

qnorm resid if sicdegp == 2

qnorm resid if sicdegp == 3

******************************************************.
* Generate Q-Q plots of EBLUPs of random child effects.
******************************************************.

predict raneffs*, reffects

qnorm raneffs1

qnorm raneffs2

********************************************************.
* Generate scatterplots of EBLUPs within each SICD group.
********************************************************.

twoway (scatter raneffs2 raneffs1), by(sicdegp)

***********************************************************.
* Plot conditional predicted values vs. actual VSAE values,
* within each SICD group.
***********************************************************.

twoway (scatter vsae fitted), by(sicdegp)