Strenght of association between Ordinal and Dichotomous Variables

[E. Bernardo]

Dear all,

 

I am doing chisquare analysis for the ordinal (with three categories) and dichotomous variables.  The sample size meets the chisquare requirement; specifically, no expected value less than 5.  I need your opinion on the following:

 

1. Is chi-square appropriate to test the hypothesis that the two variables are not associated? 

2. Can I use contingency coefficient to measure the strenght of association between variables? Please suggest.

 

Thank you.

 

Eins

[Bruce Weaver]

Eins Bernardo wrote

Dear all,
 
I am doing chisquare analysis for the ordinal (with three categories) and dichotomous variables.  The sample size meets the chisquare requirement; specifically, no expected value less than 5.  I need your opinion on the following:
 
1. Is chi-square appropriate to test the hypothesis that the two variables are not associated? 
2. Can I use contingency coefficient to measure the strenght of association between variables? Please suggest.
 
Thank you.
 
Eins

Re point 1, take a look at this (disregard the title of the document, which is incorrect):

   http://www.uvm.edu/~dhowell/methods7/Supplements/OrdinalChiSq.html

As Howell explains, the test of "linear-by-linear" association is related to Pearson r computed on the data in the contingency table.  So one option, I suppose, would be to report that r.  But if you want something more conventional for a cross-tabulation, you might consider Goodman & Kruskal's gamma.

   http://en.wikipedia.org/wiki/Gamma_test_%28statistics%29

HTH.

[E. Bernardo]

Problem Solved!  Thank you Bruce.

 

Eins

--- On Sat, 7/17/10, Bruce Weaver <[hidden email]> wrote:

From: Bruce Weaver <[hidden email]>
Subject: Re: Strenght of association between Ordinal and Dichotomous Variables
To: [hidden email]
Date: Saturday, 17 July, 2010, 1:37 AM

Eins Bernardo wrote:

>
> Dear all,
> Â
> I am doing chisquare analysis for the ordinal (with three categories)Â and
> dichotomous variables.  The sample size meets the chisquare requirement;
> specifically, no expected value less than 5.  I need your opinion on the
> following:
> Â
> 1. Is chi-square appropriate to test the hypothesis that the two variables
> are not associated?Â
> 2. Can I use contingency coefficient to measure the strenght of
> association between variables? Please suggest.
> Â
> Thank you.
> Â
> Eins
>
>
>

Re point 1, take a look at this (disregard the title of the document, which
is incorrect):

   http://www.uvm.edu/~dhowell/methods7/Supplements/OrdinalChiSq.html

As Howell explains, the test of "linear-by-linear" association is related to
Pearson r computed on the data in the contingency table.  So one option, I
suppose, would be to report that r.  But if you want something more
conventional for a cross-tabulation, you might consider Goodman & Kruskal's
gamma.

   http://en.wikipedia.org/wiki/Gamma_test_%28statistics%29

HTH.



-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Strenght-of-association-between-Ordinal-and-Dichotomous-Variables-tp1266395p1274724.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (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