Tags:
create new tag
, view all tags, tagging instructions

  • logisticHtml1.sas: For printing data sets created by multiple calls of the Logistic macro.

MacroHeaderForm edit

Macro Name

Logistic Regression

Description This is a macro for performing logistic regression with a binary outcome, checking model assumptions, and producing output that is arranged in a way suitable for use in publications. Some features that add convenience include automatically creating quartile categories for specified predictors, rescaling numeric predictors to make odds ratios more interpretable, and showing counts for categorical predictors.

Relevant links:

Logistic.sas: The main macro.
LogisticPrint.sas: For printing data sets created by multiple calls of the Logistic macro.
logisticHtml.sas: Creates hyperlinked and highlighted output file.
SampleLogisticRun.sas: Example sas program using the Logistic and LogisticHtml macros.
LogisticHtmlExample.html: The file produced by the above example program.
Check list: Suggestions of what to look for in the output and how to deal with some issues that you may find.

Macro Parameters Parameters for Logistic Macro:
DSName = name of SAS data set
Outcome = name of outcome variable. This variable is assumed to be coded 0, 1 and the model predicts the
probability of a 1 outcome.
Predictors = list of predictor variables. This list should include class variables and variables to be 'quartiled.'
Interaction variables may be included in which case the predictor list should be quoted, e.g.
Predictors = %quote(Age Gender Age*Gender). Note: all variables appearing in interaction terms
should also appear as main effects.
Classvar = classification variables with reference level. Use %quote function. For example,
Classvar=%quote(Gender(ref='Male')Race(ref='Asian')) Gender and Race could be numeric variables with
formatted values of Male or Asian or they could be alphanumeric variables with string values of 'Male'
and 'Asian'. Do not include variables for which quartiles are to be generated.
Quartiles = list of numeric variables which should be entered into the model as class variables with four quartile
levels. Variables in this list will not be included in the model as linear predictors.
Quad = Y/N for creating and testing quadratic terms (default = Y)
Out = name of output SAS data set containing logistic estimates, p-values, etc.
Print = Y/N (default = N)
Include = F,P,FP,or N
F = frequencies of outcome variable are displayed for each class variable value, e.g., Sex Female 31/100
P = percentages are displayed with class variable values, e.g., Sex Female 31.0%
FP = frequencies and percentages are displayed with class variable values, e.g., Sex Female 31/100 (31.0%).
N = no frequencies or percentages are displayed with class variable values (default = FP)
Obs = Y/N If Y, then the number of observations in each outcome group and the total are displayed. (default = Y)
LL = Y/N If Y, then -2log(likelihood) is displayed. (default = N)
HL = Y/N If Y, then the Hosmer-Lemeshow p-value is displayed along with the number of groups of estimates
(default = N).
Programmer UCSF - Peter Bacchetti
Revision

Disclaimer

Example(s)

Special Features Special Features of Logistic macro:
The macro will automatically test a quadratic term for numeric (non-class) predictors with more than two values.
The macro will automatically output odds ratios that have been rescaled until the units > 1/10 of the IQR.
Called Macros Macros called by Logistic Macro:
%words, %checkvar, %univariate, and %quartiles
Notes Pending Work
1. Investigate and fix warning messages in SampleLogisticRun.
2. Add standard headings to all macros.
3. Investigate the best way to have hyperlinks in LogisticHtml output open in a new window.
4. A macro similar to LogisticHtml that produces an output file suitable for pasting selected results tables into manuscripts.
See Also Research Topic: Logistic Regression
Topic attachments
I Attachment Action Size Date Who Comment
elsesas SampleLogisticRun.sas manage 3.1 K 18 Aug 2009 - 17:47 ChengshiJin Example sas program using the Logistic and LogisticHtml macros.
elsesas SampleLogisticRunCall.sas manage 0.2 K 18 Aug 2009 - 17:45 ChengshiJin Call for internal testing of completeness of postings
elsesas logisticHtml.sas manage 8.4 K 18 Aug 2009 - 17:46 ChengshiJin Creates hyperlinked and highlighted output file.
elsesas logisticHtml1.sas manage 8.4 K 18 Aug 2009 - 12:45 ChengshiJin For printing data sets created by multiple calls of the Logistic macro.
Topic revision: r27 - 31 Mar 2010 - 12:41:52 - MaryBanach
CTSpedia.LogisticRegressionMacro moved from CTSpedia.LogisticRegression on 12 Jun 2009 - 17:04 by MaryB? - put it back
 

Copyright & by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding CTSPedia? Send feedback