Chi Square and freqency counts

[Nichols, David]

Q. I have count data and want to run CROSSTABS or perhaps some
loglinear models using GENLOG, HILOGLINEAR or LOGLINEAR. How do
I enter this data into SPSS without making a data record for
every single case?

A. Entering count data in SPSS is quite simple. We refer to
this as aggregated data. You need enter only one data record
for each cell of your design, regardless of the number of
cases in that cell. Empty cells can be omitted.

For example, if you have a 2x3 table of data with one empty
cell (a 0 count) in the (2,3) location, it can be entered as
follows:

DATA LIST FREE / A B COUNT.
BEGIN DATA
1 1 20
1 2 50
1 3 80
2 1 10
2 2 5
END DATA.
WEIGHT BY COUNT.
CROSSTABS A BY B.

The trick here is to use the WEIGHT command to inform SPSS that
the COUNT variable represents the number of times the combination
of values of A and B occurs. Here we have left off the (2,3) cell
to simulate a 0 count. You can also include the 0 counts; they do
elicit warnings, and they can have important statistical
implications, but they are handled just fine in general SPSS
processing.

It is not uncommon for users to enter data this way but forget to
apply the WEIGHT command. If you do this, you will of course get
a table with a count of 1 in each cell. If you are using a version
of SPSS with dialog boxes, you can enter the three columns of data
directly into the data editor, then apply the weights by selecting
Data..Weight Cases from the menus and moving the COUNT variable
into the Frequency Variable box in the dialog box.

David Nichols
Principal Support Statistician and
Manager of Statistical Support
SPSS Inc.
nic...@spss.com

>----------
>From: Randy Smith[SMTP:smi...@ARES.OBU.EDU]
>Sent: Wednesday, March 04, 1998 6:12 PM
>To: SPS...@UGA.CC.UGA.EDU
>Subject: Chi Square and freqency counts
>
>I know there are 2 ways to get a chi square--through crosstabs and the chi
>square function itself. Is there a way in either procedure to enter
>frequency
>counts if you already have them? In other words, does SPSS always have to do
>the counting for you?
>Thanks--Randy
>
>
>
>=========================================================
>| Randolph A. Smith smi...@alpha.obu.edu |
>| Department of Psychology smi...@ares.obu.edu |
>| Ouachita Baptist U. (870) 245-5108 |
>| Arkadelphia, AR 71998-0001 (870) 245-5086 fax |
>=========================================================
>