You can specify one or more summary functions
for variables in any one dimension. For category variables, summaries
can be specified only for the variables at the lowest nesting level.
Thus, in the table expression
/TABLE SEX > (HAPPY + HAPMAR) BY AGECAT
you can assign summaries to HAPPY and HAPMAR or
to AGECAT, but not to both and
not to SEX.
If a scale variable appears in a dimension, that
dimension becomes the statistics dimension, and all statistics must
be specified for that dimension. A scale variable need not be at the
lowest level of nesting. Thus, the following is a valid specification:
CTABLES /TABLE AGE [MINIMUM, MAXIMUM, MEAN] > SEX > HAPPY.
A multiple response variable also need not be
at the lowest level of nesting. The following specification is a valid
specification:
CTABLES /TABLE $MLTCARS [COUNT, RESPONSES] > SEX.
However, if two multiple response variables are
nested, as in $MULTCARS > $MULTNEWS, summaries can be requested only for the variable at the innermost
nesting level (in this case, $MULTNEWS).
The general form for a summary specification is
[summary 'label' format, ..., summary 'label' format]
The specification follows the variable name in the
table expression. You can apply a summary specification to multiple
variables by enclosing the variables in parentheses. The following
specifications are equivalent:
The brackets are required even if only one summary
is specified.
Commas are optional.
Label and format are both optional; defaults are used if label and format are not specified.
You can include )CILEVEL in the quoted label to display the confidence level
for confidence intervals.
If totals or subtotals are defined for a variable
(on the CATEGORIES subcommand),
by default, the same functions that are specified for the variable
are used for the totals. You can use the keyword TOTALS within the summary specification to specify different
summary functions for the totals and subtotals. The specification
then has the form [summary ‘label’
format ... TOTALS [summary ‘label’ format...]]. You must still specify TOTAL=YES on the CATEGORIES subcommand
to see the totals.
Summaries that are available for category variables
are also available for scale variables and multiple response sets.
Functions that are specific to scale variables and to multiple response
sets are also available.
If case weighting is in effect, summaries are calculated
taking into account the current WEIGHT value. To obtain unweighted summaries, prefix a U to the function name, as in UCOUNT. Unweighted functions are not available
where weighting would not apply, as in the MINIMUM and MAXIMUM functions.
Figure 2. Table with custom percentiles and custom column labels
Each summary function for the row variable appears
by default in a column.
Labels for standard deviation and the 90th percentile
override the defaults.
Because TVHOURS is recorded in whole hours and has an integer print format, the
default general print formats for mean and standard deviation would
also be integer, so overrides are specified.
Table 1. Summary functions:
all variables
Function
Description
Default Label*
Default Format
COUNT
Number of cases in each
category. This is the default for categorical and multiple response
variables.
Count
Count
ECOUNT
The adjusted count used in effective base weight
calculations if an effective base weight variable is specified on the WEIGHT
subcommand. Otherwise, the value is the same as COUNT.
Adjusted Count
Count
ROWPCT.COUNT
Row percentage based on
cell counts. Computed within subtable.
Row %
Percent
COLPCT.COUNT
Column percentage based
on cell counts. Computed within subtable.
Column %
Percent
TABLEPCT.COUNT
Table percentage based on
cell counts.
Table %
Percent
SUBTABLEPCT.COUNT
Subtable percentage based
on cell counts.
Subtable %
Percent
LAYERPCT.COUNT
Layer percentage based on
cell counts. Same as table percentage if no layers are defined.
Layer %
Percent
LAYERROWPCT.COUNT
Row percentage based on
cell counts. Percentages sum to 100% across the entire row (that is,
across subtables).
Layer Row %
Percent
LAYERCOLPCT.COUNT
Column percentage based
on cell counts. Percentages sum to 100% across the entire column (that
is, across subtables).
Layer Column %
Percent
ROWPCT.VALIDN
Row percentage based on
valid count.
Row Valid N %
Percent
COLPCT.VALIDN
Column percentage based
on valid count.
Column Valid N %
Percent
TABLEPCT.VALIDN
Table percentage based on
valid count.
Table Valid N %
Percent
SUBTABLEPCT.VALIDN
Subtable percentage based
on valid count.
Subtable Valid N %
Percent
LAYERPCT.VALIDN
Layer percentage based on
valid count.
Layer Valid N %
Percent
LAYERROWPCT.VALIDN
Row percentage based on
valid count. Percentages sum to 100% across the entire row.
Layer Row Valid N %
Percent
LAYERCOLPCT.VALIDN
Column percentage based
on valid count. Percentages sum to 100% across the entire column.
Layer Column Valid N %
Percent
ROWPCT.TOTALN
Row percentage based on
total count, including user-missing and system-missing values.
Row Total N %
Percent
COLPCT.TOTALN
Column percentage based
on total count, including user-missing and system-missing values.
Column Total N %
Percent
TABLEPCT.TOTALN
Table percentage based on
total count, including user-missing and system-missing values.
Table Total N %
Percent
SUBTABLEPCT.TOTALN
Subtable percentage based
on total count, including user-missing and system-missing values.
Subtable Total N %
Percent
LAYERPCT.TOTALN
Layer percentage based on
total count, including user-missing and system-missing values.
Layer Total N %
Percent
LAYERROWPCT.TOTALN
Row percentage based on
total count, including user-missing and system-missing values. Percentages
sum to 100% across the entire row.
Layer Row Total N %
Percent
LAYERCOLPCT.TOTALN
Column percentage based on total count,
including user-missing and system-missing values. Percentages sum
to 100% across the entire column.
Layer Column Total N %
Percent
* This is the default on a U.S.-English system.
The .COUNT suffix can be omitted from percentages that are based on cell counts.
Thus, ROWPCT is equivalent to ROWPCT.COUNT.
Table 2. Summary functions:
scale variables, totals, and subtotals
Function
Description
Default Label
Default Format
MAXIMUM
Largest value.
Maximum
General
MEAN
Arithmetic mean. The default
for scale variables.
Mean
General
MEDIAN
50th percentile.
Median
General
MINIMUM
Smallest value.
Minimum
General
MISSING
Count of missing values
(both user-missing and system-missing).
Missing
General
MODE
Most frequent value. If
there is a tie, the smallest value is shown.
Mode
General
PTILE
Percentile. Takes a numeric
value between 0 and 100 as a required parameter. PTILE is computed the same way as APTILE in the TABLES command. Note that in the TABLES command, the default percentile method was HPTILE.
Percentile ####.##
General
RANGE
Difference between maximum
and minimum values.
Range
General
SEMEAN
Standard error of the mean.
Std Error of Mean
General
STDDEV
Standard deviation.
Std Deviation
General
SUM
Sum of values.
Sum
General
TOTALN
Count of nonmissing, user-missing,
and system-missing values. The count excludes valid values hidden
via the CATEGORIES subcommand.
Total N
Count
ETOTALN
Adjusted total N used in effective base weight
calculations if an effective base weight variable is specified on the WEIGHT
subcommand. Otherwise, the value is the same as TOTALN. Not available for multiple
response sets.
Adjusted Total N
Count
VALIDN
Count of nonmissing values.
Valid N
Count
EVALIDN
Adjusted valid N used in effective base weight
calculations if an effective base weight variable is specified on the WEIGHT
subcommand. Otherwise, the value is the same as VALIDN. Not available for multiple
response sets.
Adjusted Valid N
Count
VARIANCE
Variance.
Variance
General
ROWPCT.SUM
Row percentage based on
sums.
Row Sum %
Percent
COLPCT.SUM
Column percentage based
on sums.
Column Sum %
Percent
TABLEPCT.SUM
Table percentage based on
sums.
Table Sum %
Percent
SUBTABLEPCT.SUM
Subtable percentage based
on sums.
Subtable Sum %
Percent
LAYERPCT.SUM
Layer percentage based on
sums.
Layer Sum %
Percent
LAYERROWPCT.SUM
Row percentage based on
sums. Percentages sum to 100% across the entire row.
Layer Row Sum %
Percent
LAYERCOLPCT.SUM
Column percentage based on sums. Percentages
sum to 100% across the entire column.
Row percentage based on
responses. Total number of responses is the denominator.
Row Responses %
Percent
COLPCT.RESPONSES
Column percentage based
on responses. Total number of responses is the denominator.
Column Responses %
Percent
TABLEPCT.RESPONSES
Table percentage based on
responses. Total number of responses is the denominator.
Table Responses %
Percent
SUBTABLEPCT.RESPONSES
Subtable percentage based
on responses. Total number of responses is the denominator.
Subtable Responses %
Percent
LAYERPCT.RESPONSES
Layer percentage based on
responses. Total number of responses is the denominator.
Layer Responses %
Percent
LAYERROWPCT.RESPONSES
Row percentage based on
responses. Total number of responses is the denominator. Percentages
sum to 100% across the entire row (that is, across subtables).
Layer Row Responses %
Percent
LAYERCOLPCT.RESPONSES
Column percentage based
on responses. Total number of responses is the denominator. Percentages
sum to 100% across the entire column (that is, across subtables).
Layer Column Responses %
Percent
ROWPCT.RESPONSES.COUNT
Row percentage: Responses
are the numerator, and total count is the denominator.
Row Responses % (Base: Count)
Percent
COLPCT.RESPONSES.COUNT
Column percentage: Responses
are the numerator, and total count is the denominator.
Column Responses % (Base:
Count)
Percent
TABLEPCT.RESPONSES.COUNT
Table percentage: Responses
are the numerator, and total count is the denominator.
Table Responses % (Base:
Count)
Percent
SUBTABLEPCT.RESPONSES.COUNT
Subtable percentage: Responses
are the numerator, and total count is the denominator.
Subtable Responses % (Base:
Count)
Percent
LAYERPCT.RESPONSES.COUNT
Layer percentage: Responses
are the numerator, and total count is the denominator.
Layer Responses % (Base:
Count)
Percent
LAYERROWPCT.RESPONSES.COUNT
Row percentage: Responses
are the numerator, and total count is the denominator. Percentages
sum to 100% across the entire row (that is, across subtables).
Layer Row Responses % (Base:
Count)
Percent
LAYERCOLPCT.RESPONSES.COUNT
Column percentage: Responses
are the numerator, and total count is the denominator. Percentages
sum to 100% across the entire column (that is, across subtables).
Layer Column Responses %
(Base: Count)
Percent
ROWPCT.COUNT.RESPONSES
Row percentage: Count is
the numerator, and total responses are the denominator.
Row Count % (Base: Responses)
Percent
COLPCT.COUNT.RESPONSES
Column percentage: Count
is the numerator, and total responses are the denominator.
Column Count % (Base: Responses)
Percent
TABLEPCT.COUNT.RESPONSES
Table percentage: Count
is the numerator, and total responses are the denominator.
Table Count % (Base: Responses)
Percent
SUBTABLEPCT.COUNT. RESPONSES
Subtable percentage: Count
is the numerator, and total responses are the denominator.
Subtable Count % (Base:
Responses)
Percent
LAYERPCT.COUNT. RESPONSES
Layer percentage: Count
is the numerator, and total responses are the denominator.
Layer Count % (Base: Responses)
Percent
LAYERROWPCT.COUNT.RESPONSES
Row percentage: Count is
the numerator, and total responses are the denominator. Percentages
sum to 100% across the entire row (that is, across subtables).
Layer Row Count % (Base:
Responses)
Percent
LAYERCOLPCT.COUNT.RESPONSES
Row percentage: Count is the numerator,
and total responses are the denominator. Percentages sum to 100% across
the entire column (that is, across subtables).