ODDSRATIOS Subcommand (CSLOGISTIC command)
The ODDSRATIOS
subcommand estimates odds ratios for the specified factors or
covariates. Note that these odds ratios are model-based and are not directly computed by using the
observed data.
A separate set of odds ratios is computed for each category of the dependent variable (except the reference category).
If the FACTOR
keyword is specified, the odds ratios
compare the odds at each category j with the odds at category J, where
J is the reference category defined in parentheses following the variable
name of the factor. All other factors and covariates are fixed as
defined on the CONTROL
keyword.
If the COVARIATE
keyword is specified, the odds
ratios compare the odds at value x with the odds at value x + Δx,
where Δx is the change in x defined in parentheses following the variable
name of the covariate. To define the value x, specify the covariate
and the value on the CONTROL
keyword. All other factors
and covariates are fixed as defined on the CONTROL
keyword.
If a specified factor or covariate interacts with other predictors
in the model, the odds ratios depend not only on the change in the
specified variable but also on the values of the variables with which
it interacts. If a specified covariate interacts with itself in the
model (for example, X*X), the odds ratios depend on both the change
in the covariate and the value of the covariate. The values of interacting
factors and covariates can be customized by using the CONTROL
keyword.
The CSLOGISTIC
procedure sorts levels of each
factor in ascending order and defines the highest level as the last
level. (If the factor is a string variable, the value of the highest
level is locale-dependent.)
- Multiple
ODDSRATIOS
subcommands are allowed. Each subcommand is treated independently. - Either the
FACTOR
keyword and one or more factors, or theCOVARIATE
keyword and one or more covariates, but not both, are required. All other keywords are optional. - The
FACTOR
,COVARIATE
, andCONTROL
keywords must be followed by an equals sign and one or more elements enclosed in square brackets. - If a variable is specified on the
FACTOR
orCOVARIATE
keyword and is also specified on theCONTROL
keyword, theCONTROL
specification for that variable is ignored when the variable’s odds ratios are computed. Thus,FACTOR = [A B] CONTROL = [A(1) B(2)]
estimates odds ratios for factor A holding factor B at level 2 and for factor B holding factor A at level 1.
FACTOR = [option]. Valid options are one or more factors
appearing on the factor list. Optionally, each factor may be followed
by parentheses containing the level to use as the reference category
when computing odds ratios. Keyword LOW
or HIGH
,
or a value, may be specified. Put the value inside a pair of quotes
if the value is formatted (such as date or currency) or if the factor
is of string type. By default, the highest category is used as the
reference category.
- If a value is specified but the value does not exist in the data,
a warning is issued and the default
HIGH
is used. - Any factor may occur only once on the
FACTOR
keyword.
COVARIATE = [option]. Valid options are one or more covariates appearing on the covariate
list. Optionally, each covariate may be followed by parentheses containing one or more nonzero
numbers giving units of change to use for covariates when computing odds ratios. Odds ratios are
estimated for each distinct value. The default value is 1. Any covariate may occur only once on the
COVARIATE
keyword.
CONTROL= [option]. Specifies the factor and/or covariate
values to use when computing odds ratios. Factors must appear on
the factor list, and covariates must appear on the covariate list,
of the CSLOGISTIC
command.
- Factors must be followed by the keyword
LOW
orHIGH
, or a value, in parentheses. Put the value inside a pair of quotation marks if the value is formatted (such as date or currency) or if the factor is of string type. If keywordLOW
orHIGH
is used, each odds ratio is computed by holding the factor at its lowest or highest level, respectively. If a value is used, each odds ratio is computed by holding the specified factor at the supplied value. If a factor is not specified on theCONTROL
option, its highest category is used in odds ratio calculations. - If a factor value is specified but the value does not exist in
the data, a warning is issued and the default
HIGH
is used. - Covariates must be followed by the keyword
MEAN
or a number in parentheses. If the keywordMEAN
is used, each odds ratio is computed by holding the covariate at its overall mean. If a number is used, each odds ratio is computed by holding the specified covariate at the supplied value. If a covariate is not specified on theCONTROL
option, its overall mean is used in odds ratio calculations. - Any factor or covariate may occur only once on the
CONTROL
keyword.