glmdiagnostics

glmdiagnostics - BASIC DIAGNOSTICS Fitted values and...

Info iconThis preview shows pages 1–6. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: BASIC DIAGNOSTICS Fitted values and residuals for GLMs are defined as we expect: Y ij = X ij e ij = Y ij- Y ij and are easily obtained from PROC MIXED : MODEL y = / OUTPM = diag; The data set diag will contain both the fitted values and the residuals. They can be used for plotting within SAS or exported to another package for graphing. Many of the diagnostics for the usual ( iid ) regression models can be used for GLMs as well. 147 We need to check: normality assumptions independence assumptions variance assumptions presence of outliers or influential points presence of collinearity among covariates linearity assumptions for covariates. 148 NORMALITY Consider box plots or frequency histograms of the e ij . Look for symmetry (no skew) and unimodality. Consider normal probability plots of the e ij . Look for a straight line. Plot residuals vs. fitted values and look for an approximately equal number of positive and negative residuals across most fitted values (this is another check for skew). 149 INDEPENDENCE Here we refer to the assumed independence of the clusters from each other. This is usually assumed to hold unless an obvious violation is known. EXAMPLE Cluster=person, but study participants belonged to family groups, which was not recorded. 150 VARIANCES NOTE: The e ij do not necessarily have constant variance!! Compare your model-estimated correlation matrix ( RCORR from PROC MIXED output) to your EDA residual correlation matrix. They should look approximately similar in both the type of correlation structure and the size of the esti- mated correlations. Plot Y ij vs. Y ij and look for tightly and evenly clustered points around the diagonal line Y = Y ....
View Full Document

Page1 / 17

glmdiagnostics - BASIC DIAGNOSTICS Fitted values and...

This preview shows document pages 1 - 6. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online