Exact Tests in SPSS

[Bridgette Portman]

Hi, I am new to this list.

I am trying to analyze a relationship between two nominal variables each with
more than two levels. I can't use Pearson's Chi-Square because more than 20%
of the cells have expected counts less than 5. However, I can't do a Fisher's
Exact Test because it's larger than a 2x2 table and SPSS will not allow that
unless I get the Exact Tests add-on Module, which I'm not sure I can afford,
and I don't even know if it's compatible with my version.

Does anyone have any advice about what I could do short of buying the Exact
Tests module? Like, would it be okay if I just broke it up into every
combination of 2x2, and ran Fisher's Exact Tests? That would be time-consuming
but would it be an acceptable thing to do?

Bridgette

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Anthony Babinec]

Post the table to the list.

Tony Babinec
[hidden email]


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Bridgette Portman
Sent: Friday, November 05, 2010 6:58 PM
To: [hidden email]
Subject: Exact Tests in SPSS

Hi, I am new to this list.

I am trying to analyze a relationship between two nominal variables each
with more than two levels. I can't use Pearson's Chi-Square because more
than 20% of the cells have expected counts less than 5. However, I can't do
a Fisher's Exact Test because it's larger than a 2x2 table and SPSS will not
allow that unless I get the Exact Tests add-on Module, which I'm not sure I
can afford, and I don't even know if it's compatible with my version.

Does anyone have any advice about what I could do short of buying the Exact
Tests module? Like, would it be okay if I just broke it up into every
combination of 2x2, and ran Fisher's Exact Tests? That would be
time-consuming but would it be an acceptable thing to do?

Bridgette

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command SIGNOFF SPSSX-L For a list of
commands to manage subscriptions, send the command INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Bridgette Portman]

I don't think attaching the table will do any good. This is going to be a
repeated problem with my data, since I have a lot of nominal and ordinal
variables that will need to be tested via Chi-Square, and a small sample size,
about N=200. I'm asking about what to do in general.

Bridgette

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Bruce Weaver]

You can download a 30-day trial version of StatXact from www.cytel.com.  That will buy you a little time, at least.  I also found one online calculator, although I see nothing that explains exactly which test is being implemented.

   http://www.physics.csbsju.edu/stats/exact_NROW_NCOLUMN_form.html 

Finally, there appears to be a procedure in R that you could use.

   http://www.mail-archive.com/[e-mail]/msg31808.html

HTH.

Bridgette Portman wrote

I don't think attaching the table will do any good. This is going to be a
repeated problem with my data, since I have a lot of nominal and ordinal
variables that will need to be tested via Chi-Square, and a small sample size,
about N=200. I'm asking about what to do in general.

Bridgette

=====================
To manage your subscription to SPSSX-L, send a message to
[e-mail] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD