Note:
For the first question the key tools are described in the help file
that was created for Exercise #2 -- that is, we can use "cc" with the
"by" option to obtain stratified odds ratios, or the "mhodds"
procedure.
Also, for some descriptives you may want to use "table" and "tabulate".
See the DO file for the second question of Exercise #2 -- there I use
"table" to compute means for each clinic.
Also:
If we want to stratify on more than one variable, then we can use
"mhodds" to do this. We'd specify:
mhodds Dvar Evar Svar1 Svar2
And this would return an odds ratio adjusted for both Svar1 and Svar2.
Since I can't make "cc" take 2 variables with the "by" option, I could
as an alternative to mhodds, simply create a new combined variable --
AgeAlc = 1000*age + alc, and use this as the stratifying variable. By
creating this new variable we create a unique value for every (Age,Alc)
combination. I could then use this as the "by" argument with "cc".
****************
****************
** logistic ** logistic regression
****************
****************
Overview: The "logistic" command estimates a logistic regression for
a binary outcome variable. The default summary is in terms
of exp(beta_j), which is an odds ratio when beta_j is the
coefficient of a main effect in the model.
Usage: "logistic Dvar X1 X2 X3"
Where - Dvar is the disease variable (1=disease, 0=control)
Xj are predictor variables
Summaries: The "logistic" command returns estimates of adjusted odds
ratos, exp( beta_j ).
After fitting the model using logistic, the command "logit" will
display the regression coefficients, beta_j.
Options: (1) "logistic y x1 x2 [freq=count]" -- this is used when the
data are in a "grouped" format with the number of
cases/controls that have a certain covariate combination
given by the variable "freq".
(2) "xi: logistic y i.x1 x2" -- this is used to create
dummy variables for variable x1.