Logistic Regression: A Self-learning Text, Third Edition (Statistics in the Health Sciences)

IV. The model and odds ratio for a nominal exposure
variable (no interaction case)(pages 82–84)
A. No interaction model involving a nominal
exposure variable withkcategories:
logit PðÞ¼X aþb 1 E 1 þb 2 E 2 þþbk 1 Ek 1

þ~

p 1

i¼ 1

giVi;

whereE 1 ,E 2 ,...,Ek 1 denotek1 dummy

variables that distinguish thekcategories of the

nominal exposure variable denoted asE, i.e.,

Ei¼1 if categoryior 0 if otherwise.

B. Example of model involvingk¼4 categories of

occupational status:

logit PðÞ¼X aþb 1 OCC 1 þb 2 OCC 2 þb 3 OCC 3

þ~

p 1

i¼ 1

giVi;

where OCC 1 , OCC 2 , and OCC 3 denotek 1 ¼ 3

dummy variables that distinguish the four

categories of occupation.

C. Odds ratio formula for no interaction model

involving a nominal exposure variable:

RORE*vs:E**¼exp

E* 1 E** 1



b 1 þ E* 2 E** 2



b 2

þþE*k 1 E**k 1



bk 1

"

;

whereE*¼(E 1 *,E 2 *,...,Ek* 1 ) andE**¼

(E 1 **,E 2 **,...,E**k 1 ) are two specifications of the

set of dummy variables forEto be compared.

D. Example of odds ratio involvingk¼4 categories

of occupational status:

ROROCC*vs:OCC**

¼exp

OCC* 1 OCC** 1



b 1 þ OCC* 2 OCC** 2



b 2

þ OCC* 3 OCC** 3



b 3

"

:

V. The model and odds ratio for several exposure

variables (no interaction case)(pages 85–87)

A. The model:

logit PðÞ¼X aþb 1 E 1 þb 2 E 2 þþbqEq

þ~

p 1

i¼ 1

giVi;

whereE 1 ,E 2 ,...,Eqdenoteqexposure variables

of interest.

Detailed Outline 93