Date: Mon, 10 Jan 2011 14:19:25 -0800
Reply-To: J McClure <mc006@pacbell.net>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: J McClure <mc006@pacbell.net>
Subject: Re: statistical test for 2x8 table with small expected values
In-Reply-To: <4D2B74FC.4080000@DrKendall.org>
Content-Type: text/html; charset=ISO-8859-1
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#ffffff">
Yes, thanks Art. I guess I had a brain fart (again). <br>
Thanks to you and the others on the listserv for providing such a
valuable service.<br>
Jan<br>
<u><br>
If I understand correctly what you have</u>, the cells and margins
of the 4 way crosstab should have all that you need.<br>
Here is the output from the simulation. Does it have what you need?<br>
<br>
On 1/10/2011 1:07 PM, Art Kendall wrote:
<blockquote cite="mid:4D2B74FC.4080000@DrKendall.org" type="cite">oops
it would be 4 way like this.
<br>
<br>
new file.
<br>
set seed 20110110.
<br>
input program.
<br>
loop id = 1 to 482.
<br>
compute depression = RV.BERNOULLI(.10).
<br>
compute PTSD = RV.BERNOULLI(.04).
<br>
compute alcohol = RV.BERNOULLI(.10).
<br>
compute suicidal = RV.BERNOULLI(.10).
<br>
end case.
<br>
end loop.
<br>
end file.
<br>
end input program.
<br>
formats id (f3) depression to alcohol (f1).
<br>
value labels depression to suicidal 0 'no' 1 'yes' .
<br>
CROSSTABS tables = depression by suicidal by PTSD by alcohol
<br>
/cells = all
<br>
/method =exact.
<br>
<br>
<br>
Art
<br>
<br>
On 1/10/2011 3:59 PM, Art Kendall wrote:
<br>
<blockquote type="cite">It appears that the METHOD=EXACT
specification in CROSSTABS might be what you need.
<br>
The simulation in the syntax below ran in a few seconds.
<br>
It assumes depression occurs in 10% of cases, PTSD in 4 %,
depression in 10%. You can change the RV arguments to be more
like your actual situation.
<br>
<br>
Open a new instance of SPSS. Paste the syntax in a syntax
window. Run it.
<br>
Is this what you need?
<br>
<br>
Art Kendall
<br>
Social Research Consultants
<br>
<br>
new file.
<br>
set seed 20110110.
<br>
input program.
<br>
loop id = 1 to 482.
<br>
compute depression = RV.BERNOULLI(.10).
<br>
compute PTSD = RV.BERNOULLI(.04).
<br>
compute alcohol = RV.BERNOULLI(.10).
<br>
end case.
<br>
end loop.
<br>
end file.
<br>
end input program.
<br>
formats id (f3) depression to alcohol (f1).
<br>
value labels depression to alcohol 0 'no' 1 'yes' .
<br>
CROSSTABS tables = depression by PTSD by alcohol
<br>
/cells = all
<br>
/method =exact.
<br>
<br>
<br>
On 1/10/2011 2:06 PM, J McClure wrote:
<br>
<blockquote type="cite">Hi,
<br>
From a survey I have three clinical characteristics
(depression, PTSD,
<br>
and alcohol misuse) which are either present (yes) or not
present (no).
<br>
I also have the answers to multiple questions about suicidal
thoughts
<br>
and behaviors which I have collapsed into a dichotomous
variable of yes
<br>
or no.
<br>
I want to see if the proportion of respondents answering
positively to
<br>
each of the questions differs by whether they have one or more
of the
<br>
clinical characteristics.
<br>
So, I created a variable of clinical characteristics
(Clin_Char) with 8
<br>
categories which represent all possible combinations including
none and
<br>
all three characteristics.
<br>
The problem is that with an n of 482, the numbers are small
for three of
<br>
the categories and when I do a chi-squared test there are
cells with an
<br>
inadequate expected value.
<br>
Collapsing categories or dropping some categories would loose
<br>
information but it seems since the contingency table is bigger
than 2x2
<br>
I can't use Yates correction Fisher's exact test.
<br>
Suggestions?
<br>
Thanks,
<br>
Jan
<br>
<br>
=====================
<br>
To manage your subscription to SPSSX-L, send a message to
<br>
<a class="moz-txt-link-abbreviated" href="mailto:LISTSERV@LISTSERV.UGA.EDU">LISTSERV@LISTSERV.UGA.EDU</a> (not to SPSSX-L), with no body text
except the
<br>
command. To leave the list, send the command
<br>
SIGNOFF SPSSX-L
<br>
For a list of commands to manage subscriptions, send the
command
<br>
INFO REFCARD
<br>
<br>
</blockquote>
</blockquote>
<br>
</blockquote>
</body>
</html>
=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@LISTSERV.UGA.EDU (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
| 
