Use the GLOW_RAND data from Section 9.4 and fit a cluster-specific model using adaptive quadrature for estimation. Use FRACTURE as the response and covariates: calculated fracture risk score...

Use the GLOW_RAND data from Section 9.4 and fit a cluster-specific model using adaptive quadrature for estimation. Use FRACTURE as the response and covariates: calculated fracture risk score (FRAC_SCORE), age (AGE), and self-perception of risk (RATERISK). Include a random intercept for physician id (PHY_ID).

(a) Does the number of quadrature points appear to impact the model estimates?

(b) What happens if you test each covariate for a random slope?