Multivariate and univariate odd ratios (logistic regression) -------------------------------------------------------------
^mulogit^ depvar indepvar1 indepvar2 ... indepvarp [^if^ exp]
Generates multivariate and univariate odds ratios (ORs) and 95% confidence intervals (CIs) for variables in ^varlist^. The resulting univariate and multivariate ORs are appended to the rightside of the data editor, so that they can be cut and pasted into other graphics packages or publication tables.
The purpose of ^mulogit^ is to write the ORs and 95% CIs limits as new variable values in the data editor, and then generate plots of the ORs and CIs using the new variables.
The twoway rcap graph of Version 10 is first used to generate a separate plot for multivariate ORs and CIs, then a separate plot for univariate ORs and CIs. The variable names (value labels) in the plots are appended with one of the three labels designating the significance level for each variable:
* - p<0.05 ** - p<0.01 *** - p<0.001
The following plots are saved to disk:
^multuniv.gph^ - a twoway rcap plot of both the multivariate and univariate ORS and CIs
^mult.gph^ - a twoway rcap plot of multivariate ORs and CIs
^univ.gph^ - a twoway rcap plot of univariate ORs and CIs
Finally, the ^combine^ graphics command is used to combine the separate multivariate and univariate plots into a single graph. The ^ycommon^ option for ^combine^ is also employed to display the same scale (ranges) of multivariate and univariate ORs and CIs. ^Examples:^ ^---------^
. mulogit low age lwt smoke ptl ht ui ftv . mulogit low age lwt smoke ptl ht ui ftv if race==1
^Author:^ ^-------^ Leif E. Peterson Center for Biostatistics The Methodist Hospital Research Institute (TMHRI) Houston, Texas 77030 USA Tel: +001 (713) 441-6121 e-mail: lepeterson@tmhs.org