Info

SAS Code: proc mixed covtest; class phys; model y = pt_age/s;

random int pt_age/type = un subject = phys s;

SAS Code: proc mixed covtest; class phys; model y = pt_age/s;

random int pt_age/type = un subject = phys s;

run;

errors, degrees of freedom, t-statistics and two-sided significance levels are also provided. SAS Proc Mixed's estimates of the random effects correspond to the 60/s of Section 4.1.3. These estimates of random effects vary from — 7-67 (physician 003) to 10-57 (physician 013, not shown).

Estimates of Random Coefficients. The estimates of the random effects can be used to generate shrinkage estimates of the random coefficients or estimates of the individual physician adjusted mean satisfaction scores using = 68-0081 + 0-1533* (X. j — X..) + u0j-, where X. j is the mean age of patients in physician;"s practice and X.. is the mean age of patients in the sample (n = 1492). For physicians 003 and 013, the estimates of their mean satisfaction scores adjusted for patient age are ft* 003 = 60-4 and /?*13 = 78-6 . The OLS estimates of the adjusted mean satisfaction scores (intercepts) for these physicians (see Table IV) were 54-0 and 87-2, respectively. The discrepancies between the OLS estimates and the BLUPs can be attributed, in part, to shrinkage.

Table IX. One-way analysis of covariance (ANCOVA) model with random effects:

SAS Proc Mixed

Was this article helpful?

0 0

Post a comment