Detailed Outline
I. Overview(page 45)
A. Focus:
Simple analysis
Multiplicative interaction
Controlling several confounders and effect
modifiers
B. Logistic model formula whenX¼(X 1 ,X 2 ,...,Xk):
PðÞ¼X
1
1 þe
aþ~
k
i¼ 1
biXi
:
C. Logit form of logistic model:
logit PðÞ¼X aþ~
k
i¼ 1
biXi:
D. General odds ratio formula:
RORX 1 ,X 0 ¼e
~
k
i¼ 1
biðÞX 1 iX 0 i
¼
Yk
i¼ 1
ebiðÞX^1 iX^0 i:
II. Special case – Simple analysis(pages 46–49)
A. The model:
PðÞ¼X
1
1 þeðÞaþb^1 E
B. Logit form of the model:
logit P(X)¼aþb 1 E
C. Odds ratio for the model: ROR¼exp(b 1 )
D. Null hypothesis of noE, Deffect: H 0 :b 1 ¼0.
E. The estimated odds ratio exp(^b) is computationally
equal toad / bcwherea, b, c,anddare the cell
frequencies within the four-fold table for simple
analysis.
III. Assessing multiplicative interaction(pages 49–55)
A. Definition of no interaction on a multiplicative
scale: OR 11 ¼OR 10 OR 01 ,
where ORABdenotes the odds ratio that compares
a person in categoryAof one factor and categoryB
of a second factor with a person in referent
categories 0 of both factors, whereAtakes on the
values 0 or 1 andBtakes on the values 0 or 1.
B. Conceptual interpretation of no interaction
formula: The effect of both variablesAandBacting
together is the same as the combined effect of each
variable acting separately.
Detailed Outline 65
