Date: Wed, 11 Jun 1997 10:07:53 EDT
Reply-To: "Gil Klajman, Sociology" <klajmang@ALPHA.MONTCLAIR.EDU>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: "Gil Klajman, Sociology" <klajmang@ALPHA.MONTCLAIR.EDU>
Subject: Re: Force to show var/val labels
Michael Gregersen inquired:
> Is it possible to make SPSS show the variable/value labels in say a=20
> crosstab, when you dont have any cases matching the values. In the=20
> output you just have 'blanks' or zeros. My problem is, that I only get=
> an output of the variables and values with corresponding data.
A brute-force option which _might_ create problems is to compute for all
existing cases a variable (e.g.: COMPUTE CONSTANT = 1.) Then weight the
cases by CONSTANT. You could then create phony cases to represent combinations
for which there are no values in the real data. For these cases you could set
the value of CONSTANT at a very small number, say 0.0000000001
These cases would show up in your crosstabulation with apparent frequencies of
0. The effect on some statistics would be negligible. For others (e.g.,
the degrees of freedom and the probability for the chi-squared statistic)
which are influenced by the number of cells in the table, you would of course
have incorrect calculations.
But you can of course do those analyses with an without the phony cases, and
report the correct statistics along with an explanatory note.
Gil Klajman, Sociology, Montclair State University, Upper Montclair, NJ 07043