crosstabs/chisquares/correlations

[Deepa Lakshmi Shanadi]

Dear all,

Apologies in advance if this is more of a statistics questions rather than a SPSS question.

I know for 2by2 tables, I use chisquares to see if 2 groups are statistically different.

For anything more than a 2by2 table, i.e. 3 groups by 3 responses, can I still use the chi-square?

If not, what else would I use in SPSS? What is the menu option?

Thanks,
Deepa

<input onclick="jsCall();" id="jsProxy" type="hidden">

[Tesiny, Ed]

Short answer, yes you can still use chi square assuming you have a reasonable N.

 

Ed Tesiny

[hidden email]

 

---

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Deepa Lakshmi Shanadi
Sent: Friday, September 17, 2010 9:22 AM
To: [hidden email]
Subject: crosstabs/chisquares/correlations

Dear all,

Apologies in advance if this is more of a statistics questions rather than a SPSS question.

I know for 2by2 tables, I use chisquares to see if 2 groups are statistically different.

For anything more than a 2by2 table, i.e. 3 groups by 3 responses, can I still use the chi-square?

If not, what else would I use in SPSS? What is the menu option?

Thanks,
Deepa

<INPUT id=jsProxy onclick=jsCall(); type=hidden>

[John F Hall]

AFAIK chi-square can be used with anything, but 2 x 2 tables may be a special case where a different (chi-square based) statistic might be appropriate.  Have a  look at Jim Ring's specially written notes:

http://surveyresearch.weebly.com/statistics-notes-to-accompany-course.html

 

Chi-square is sensitive to sample size: I used to have  rule of thumb that it will be significant once you got 200 or more cases (and Pearson's r has a maximum of 0.707 with 3 x 3 tables) but I could be wrong.  No doubt one of the regulars will give a better answer.

 

John Hall
[hidden email]
http://surveyresearch.weebly.com

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Friday, September 17, 2010 3:21 PM

Subject: crosstabs/chisquares/correlations

Dear all,

Apologies in advance if this is more of a statistics questions rather than a SPSS question.

I know for 2by2 tables, I use chisquares to see if 2 groups are statistically different.

For anything more than a 2by2 table, i.e. 3 groups by 3 responses, can I still use the chi-square?

If not, what else would I use in SPSS? What is the menu option?

Thanks,
Deepa

<INPUT id=jsProxy onclick=jsCall(); type=hidden>

[Bruce Weaver]

John F Hall wrote

AFAIK chi-square can be used with anything, but 2 x 2 tables may be a special case where a different (chi-square based) statistic might be appropriate.  Have a  look at Jim Ring's specially written notes:
http://surveyresearch.weebly.com/statistics-notes-to-accompany-course.html

Chi-square is sensitive to sample size: I used to have  rule of thumb that it will be significant once you got 200 or more cases (and Pearson's r has a maximum of 0.707 with 3 x 3 tables) but I could be wrong.  No doubt one of the regulars will give a better answer.

John Hall
[e-mail]
http://surveyresearch.weebly.com 

John is right on both counts.  The chi-square  test of association can be used for tables larger than 2x2 (when certain conditions are met); and the 2x2 table is a special case where there are better options than the ordinary test.  I have some notes you might find useful.

   https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/notes/chisqr_assumptions

Cheers,
Bruce