help separate dialog: separate
-------------------------------------------------------------------------------
Title
[D] separate -- Create separate variables
Syntax
separate varname [if] [in] , by(byvar|exp) [options]
options description
-------------------------------------------------------------------------
Main
* by(byvar) categorize observations into groups defined by
byvar
* by(exp) categorize observations into two groups defined
by exp
Options
generate(stubname) name new variables by suffixing values to
stubname; default is to use varname as prefix
sequential use as name suffix categories numbered
sequentially from 1
missing create variables for the missing values
shortlabel create shorter variable labels
-------------------------------------------------------------------------
* Either by(byvar) or by(exp) must be specified.
Menu
Data > Create or change data > Other variable-transformation commands >
Create separate variables
Description
separate creates new variables containing values from varname.
Options
+------+
----+ Main +-------------------------------------------------------------
by(byvar|exp) specifies one variable defining the categories or a logical
expression that categorizes the observations into two groups.
If by(byvar) is specified, byvar may be a numeric or string variable
taking on any values.
If by(exp) is specified, the expression must evaluate to true (1),
false (0), or missing.
by() is required.
+---------+
----+ Options +----------------------------------------------------------
generate(stubname) specifies how the new variables are to be named. If
generate() is not specified, separate uses the name of the original
variable, shortening it if necessary. If generate() is specified,
separate uses stubname. If any of the resulting names is too long
when the values are suffixed, it is not shortened and an error
message is issued.
sequential specifies that categories be numbered sequentially from 1. By
default, separate uses the actual values recorded in the original
variable, if possible, and sequential numbers otherwise. separate
can use the original values if they are all nonnegative integers
smaller than 10,000.
missing also creates a variable for the category missing if missing
occurs (byvar takes on the value missing or exp evaluates to
missing). The resulting variable is named in the usual manner but
with an appended underscore; e.g., bp_. By default, separate creates
no such variable. The contents of other variables are unaffected by
whether missing is specified.
shortlabel creates a variable label that is shorter than the default. By
default, when separate generates the new variable labels, it includes
the name of the variable being separated. shortlabel specifies that
the variable name be omitted from the new variable labels.
Examples
---------------------------------------------------------------------------
Setup
. sysuse auto
Create variable mpg0 containing values of mpg when foreign is 0
(Domestic) and create variable mpg1 containing values of mpg when foreign
is 1 (Foreign)
. separate mpg, by(foreign)
Plot quantiles of mpg0 against the quantiles of mpg1
. qqplot mpg0 mpg1
---------------------------------------------------------------------------
Setup
. sysuse auto, clear
Create variable mpg0 containing values of mpg when price is not greater
than 6000 and create variable mpg1 containing values of mpg when price is
greater than 6000
. separate mpg, by(price>6000)
---------------------------------------------------------------------------
Setup
. sysuse auto, clear
Same as above, but call variables mpgpr0 and mpgpr1
. separate mpg, by(price>6000) gen(mpgpr)
Plot quantiles of mpgpr0 against the quantiles of mpgpr1
. qqplot mpgpr0 mpgpr1
---------------------------------------------------------------------------
Saved results
separate saves the following in r():
Macros
r(varlist) names of the newly created variables
Also see
Manual: [D] separate
Help: [R] tabulate oneway, [R] tabulate twoway, [R] tabulate,
summarize()
