Exploring Outliers and Problematic Data Distributions
Numeric variable outliers (suspiciously extreme minimum
or maximum values) may be apparent in the analysis above. But you
should explore them further as demonstrated below, while also consulting
with your data source and the literature to verify what are reasonable
minimum and maximum values. Some variables will not meet the assumptions
of statistical tests (for example, will not be normally distributed).
The PROC UNIVARIATE code below will get us started with useful output
for the variable FG, which is fasting blood glucose value.
ODS listing close; ODS rtf file="c:/temp/TIADout.rtf"; proc univariate data=TIAD; var FG; qqplot FG/normal(mu=est sigma=est color=red l=4) square; run; ODS rtf ...
Get Data Analysis Plans: A Blueprint for Success Using SAS now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
