TAB <data>,L
makes cross-tabulations of <data> according to selected variables and
saves the results in the output file (see OUTPUT) and from line L
onwards in the current edit field, if L is given.
Also means and standard deviations of a cell variable can be computed
(see CELL specification below).
The table can also be saved as a Survo data file (see TABFILE?).
The IND and CASES specifications may be used to restrict the set of
observations.
Tables produced by TAB have a header and letters as line labels.
The header and the labels are omitted by giving a specification LABELS=0 .
The variables controlling classification can be numeric or string
variables and they are to be listed by the specification
VARIABLES=<column var>,<row var 1>,<row var 2>,...,<row var n>
or
VARIABLES=<column var>:<row var 1>,<row var 2>,...,<row var n>
The first alternative of the VARIABLES specification implies computing
of one n-dimensional table.
The second one (with a colon ':' after <column var>) implies computing
of n two-dimensional tables with a common <column var>.
There is still a more general alternative
VARIABLES=<cvar 1>:<cvar 2>:...:<cvar m>:<rvar 1>,<rvar 2>,...
which allows m nested column variables for each row variable.
The classification of a numeric variable has to be given in the form
<name of var>=L1,U1,U2,U3,...
where L1=lower limit of the 1st class
U1=upper limit of the 1st class
U2=upper limit of the 2nd class etc.
For example, Age=0,6,12,21 specifies age groups 0-6,7-12 and 13-21.
Any class can be given a name in parentheses after the upper limit.
For example, Color=1,1(white),2(red),3(blue).
If the class widths are equal, classification may also be given in the
form <name of var>=<up.limit_of_1.st_class>(<class_width>)<last_class>
For example, Age=10(5)25 is same as Age=5,10,15,20,25. In this
abbreviated notation no class names can be given.
By default, cases with missing values in classifiers are omitted.
However, for numeric classifiers a special class MISSING can be
given as the last class. Example; Age=5,10,15,20,25,MISSING
The classification of a string variable is described by examples:
1) Assume that 'Weekday' is a string variable with values 'Su','Mo' etc.
A classification for 'Weekday' could be
Weekday=/Mo/Tu/We/Th/Fr(Workdays),/Sa/Su(Weekend)
thus forming 2 classes Workdays=(Mo,Tu,We,Th,Fr) and Weekend=(Sa,Su).
2) Assume that 'Letter' is a string variable with values 'A','B','C' etc.
To classify letters to wowels and consonants the following
specification is valid:
Letter=/A/E/I/O/U(wowel),/-(consonant)
where /- denotes all the remaining alternatives.
The printout of the (frequency) table may be include column and row
sums, too. In addition to frequencies various tables of percentages
can obtained. These options are selected by a RESULTS specification
of the form
RESULTS=CSUMS,RSUMS,C%,R%,T%
with following consequences:
CSUMS sums of columns
RSUMS sums of rows
C% table of percentages in columns
R% table of percentages in rows
T% table of percentages from the grand total
Any of the key words in RESULTS may be omitted.
An example of a complete TAB operation for a 3-dimensional contingency
table with column and row sums:
TAB COMMUNES,27
VARIABLES=Industry,Province,People RESULTS=CSUMS,RSUMS
Industry=0,2,5,9 (proportion of people working in industry 0-9)
Province=/UUS/KYM,/TUR/AHV,/- (3 classes)
People=0,5000(small),20000(medium),500000(large)
Tables of means, standard deviations and sums:
To compute means and standard deviations for a selected variable,
a CELL specification may be included. It has the form
CELL=<cell variable>,<format for the results, ###.# for example>
Sums of the cell variable are computed (instead of means and standard
deviations) by giving the CELL specification in the form
CELL=<cell variable>,<format>,<FSUM or SUM>
For FSUM, both frequencies and sums are printed. For SUM, sums only are
printed.
T = More information on TAB operations
S = General information on statistical operations
Help System (web edition)