SELECT Subcommand (REGRESSION command)
By default, all cases in the active dataset are
considered for inclusion on REGRESSION
. Use SELECT
to include a subset
of cases in the correlation matrix and resulting regression statistics.
- The required specification on
SELECT
is a logical expression. - The syntax for the
SELECT
subcommand is as follows:/SELECT=varname relation value
- The variable named on
SELECT
should not be specified on theVARIABLES
subcommand. - The relation can be
EQ
,NE
,LT
,LE
,GT
, orGE
. - Only cases for which the logical expression on
SELECT
is true are included in the calculation of the correlation matrix and regression statistics. - All other cases, including those with missing values
for the variable named on
SELECT
, are not included in the computations. - If
SELECT
is specified, residuals and predicted values are calculated and reported separately for both selected and unselected cases by default. See the topic RESIDUALS Subcommand (REGRESSION command) for more information. - Cases deleted from the active dataset with
SELECT IF
, a temporarySELECT IF
, orSAMPLE
are not passed toREGRESSION
and are not included among either the selected or unselected cases.
Example
REGRESSION SELECT SEX EQ 'M'
/VARIABLES=AGE,STARTPAY,YRS_JOB,SALARY
/DEPENDENT=SALARY
/METHOD=STEP
/RESIDUALS=NORMPROB.
- Only cases with the value M for SEX are included in the correlation matrix calculated
by
REGRESSION
. - Separate normal P_P plots are displayed for cases with SEX equal to M and for other cases. See the topic RESIDUALS Subcommand (REGRESSION command) for more information.