SPSSX Discussion

Chi Square question

Classic

List

Threaded

3 messages Options

lts1

Chi Square question

Hi SPSSers,

I am having a problem trying to run a Chi-square. I have data that includes a set of 6 of dummy variables called "situation" and a 3-level variable called "status". The problem is the situation variables involve multiple responses, and I don't have the multiple response set option, that's why they're dummy coded as six different variables. I need to run a Chi-sq on the full set of six situation variables across the 3 status levels. I can run frequencies and get them for each level, but how do I run a Chi-sq on this? Can anybody help? Thanks in advance.

Best,
Lisa

Lisa T. Stickney
Ph.D. Candidate
Fox School of Business
Temple University
[hidden email]

====================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

Kylie Lange

Re: Chi Square question

Hi Lisa,

I don't see any way of analysing the set of situation variables by
status data with a chi-square test, as the data is not independent.
Perhaps consider conducting 6 separate chi-square tests - one for each
situation category against status (ie, a 2x3 crosstabs table). Would
that be useful?

Cheers,
Kylie.

On 11/01/2008 10:00 AM, Lisa Stickney wrote:

> Hi SPSSers,
>
> I am having a problem trying to run a Chi-square. I have data that includes a set of 6 of dummy variables called "situation" and a 3-level variable called "status". The problem is the situation variables involve multiple responses, and I don't have the multiple response set option, that's why they're dummy coded as six different variables. I need to run a Chi-sq on the full set of six situation variables across the 3 status levels. I can run frequencies and get them for each level, but how do I run a Chi-sq on this? Can anybody help? Thanks in advance.
>
> Best,
> Lisa
>
> Lisa T. Stickney
> Ph.D. Candidate
> Fox School of Business
> Temple University
> [hidden email]
>
> ===================
> 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
>

ViAnn Beadle

Re: Chi Square question

You CAN get a chi-square statistic from such a table in CTABLES but it
adjusts the statistic to handle the overlap issues coming from multiple
responses. I replicated the CTABLES table's counts by restructuring the file
from wide to long using VARSTOCASES and then trim out all the 0 values to
get a single vector for situation variables. The chi-square reported by
CROSSTABS was really different than the chi-square reported by CTABLES.
Although I don't exactly how the adjustment is done, I suspect that the
CROSSTABS version will always be larger, especially when the number of
responses greatly exceeds the number of respondents.

Having said that, I still wonder what hypothesis is really being tested for
this kind of table--independence or respondents or response. To make the
long story short--CTABLES is required to calculate an adjusted chi-square
test on a multiple dichotomy variable.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Kylie Lange
Sent: Thursday, January 10, 2008 5:10 PM
To: [hidden email]
Subject: Re: Chi Square question

Hi Lisa,

I don't see any way of analysing the set of situation variables by
status data with a chi-square test, as the data is not independent.
Perhaps consider conducting 6 separate chi-square tests - one for each
situation category against status (ie, a 2x3 crosstabs table). Would
that be useful?

Cheers,
Kylie.

On 11/01/2008 10:00 AM, Lisa Stickney wrote:
> Hi SPSSers,
>
> I am having a problem trying to run a Chi-square. I have data that
includes a set of 6 of dummy variables called "situation" and a 3-level
variable called "status". The problem is the situation variables involve
multiple responses, and I don't have the multiple response set option,
that's why they're dummy coded as six different variables. I need to run a
Chi-sq on the full set of six situation variables across the 3 status
levels. I can run frequencies and get them for each level, but how do I run
a Chi-sq on this? Can anybody help? Thanks in advance.

>
> Best,
> Lisa
>
> Lisa T. Stickney
> Ph.D. Candidate
> Fox School of Business
> Temple University
> [hidden email]
>
> ===================
> 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

=====================
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