I'm not aware of any means of forcing the SPSS
Frequencies procedure to display values for
which there are no valid cases.
If you have access to the SPSS Tables module, however,
you might have an option. The Tables procedure doesn't
automatically include empty values in Tables of
Frequencies either, but there is a fairly simple
"trick" you can do in this module to get the desired
Here's what you might do:
1. Compute a constant in the data file, such as
2. Add 5 bogus cases to the file, each with one of the
five values of the variable you wish to display, such
that all five are included (the order doesn't matter).
3. Enter a 2 (or any value different from the constant
above) for the "layer" variable for these 5 bogus
cases (the cases can have missing values on all other
4. Select Statistics...Custom Tables...Tables of
Frequencies. Put your variable of interest in the
"Frequencies For" box, and the "layer" variable in
the "Separate Tables" box. Click on OK, or select
any additional formatting options you desire.
This will generate a layered table, with the table
you want in the first layer, and a bogus, negligible
one in the second. The key point is that the first
table should include all values, whether there are
any valid cases for them or not. SPSS Tables builds
layered tables identically across layers, with as
many categories displayed as are necessary to accommodate
the most populated layer. This is just a little trick to
harness this feature of Tables to get empty row values
Here's a quick sample job that should illustrate the
DATA LIST / x 1 layer 3.
* Table of Frequencies.
/FORMAT BLANK MISSING('.') /TABLES
/STATISTICS COUNT ((F5.0) 'Count' ) .
I hope Douglas College has the Tables module so you
can try this.
Best of luck,
SPSS Technical Support
>From: Melissa Mack[SMTP:Melissa_Mack@DOUGLAS.BC.CA]
>Sent: Friday, September 05, 1997 12:39 PM
>Subject: Display values
>I am analysing a course evaluation in which students responded (using a
>likert scale, from 1-5) to several statements. From a preliminary
>glance at the results, I noticed that for some statements the range
>of response was smaller than the possible 1-5. I want to produce a frequency
>table that would include all five options even though some options were not
>chosen. In other words, if no one responded with a "1" to a particular
>statement, I still want a frequency table that includes the value of "1" with
>a "0" frequency for that value. Is there a means of accomplishing this?
>Although I do have access to SPSS 6.1.3 and SPSS 7.5.1, I would prefer the
>solution to be written as syntax.
>Thank you for your help.
>New Westminster, British Columbia, Canada