VIII. The hierarchical backward elimination
approach(pages 184–185)
A. Flow diagram representation.
B. Flow description: evaluateEViVjterms first,
thenEViterms, thenViterms last.
C. Use statistical testing for interaction terms, but
decisions aboutViterms should not involve
testing.
IX. The hierarchy principle for retaining variables
(pages 185–187)
A. Definition: If a variable is to be retained in the
model, then all lower order components of that
variable are to be retained in the model forever;
examples.
C. Rationale: Tests about lower order components
can give different conclusions depending on the
coding of variables tested; such tests should be
independent of coding to be valid; therefore, no
such tests are appropriate; example.
X. An example(pages 188–192)
A. Evans County CHD data description.
B. Variable specification stage.
C. Final results.
XI. Summary(pages 192–193)196 6. Modeling Strategy Guidelines