Reference > Commands > T (tabulate)

For statistics tables, the form of the T (tabulate) command is as follows:

[+] T table [(F= ...)] [(ID='text')] = @variable * [banner] [+weight],

The rules are as for distribution tables, except that row variable must be an svar, an ivar or a wvar and must be preceded by '@'. The '@' following the = sign defines this table as a statistics table. The value of the row variable is the score used to calculate the statistics.

A four-row table is incremented.

The rows and the formats that print them are:

PTR Print Total Row total number of records in table

BST Base for Statistics total number of records excluding those where the score value has the value U (undefined)

SUM Sum of Scores addition of all the scores

SSQ Sum of squares addition of the squares of the scores

From these rows of a statistics table, or from the same values incremented at print time from the <V> label controls of a distribution table or from the values found in a value distribution table, the following statistics are computed, depending on the relevant formats:

AVG Average

SDV Standard Deviation

SER Standard Error

EVR Error Variance

TTF F-tests/t-tests

TTT t-test matrix

Note that other statistical functions, chi-square, Kolmogorov-Smirnov, Mann-Whitney, medians, quartiles, min and max, are produced from distribution tables.

The rules for redefining tables require that table type is preserved between definitions; a statistics table may not be redefined as a distribution table or vice versa.

Statistics tables may be "converted" into distribution tables, see %STA in the chapter on functions.

Examples of T statistics commands:

t #16 (f=evr) = @$iunit * $break,

+t #16a (f=evr) = @$iunita * $age,