Analysis of a Stratified Cluster Sample

[Alice P Chiu]

Dear all,
I would like to seek some advice regarding the analysis of a Stratified Cluster Sample.
I would like to find the association between two variables. Within the 2x2 table, there was an expected count of  less than 5.
I was told that Fisher's exact test should not be used in such a scenaario.
Can you please advice which test should be used instead?
Thanks in advance!

Alice.


 
____________________________________________________________________________________
Get your own web address.  
Have a HUGE year through Yahoo! Small Business.
http://smallbusiness.yahoo.com/domains/?p=BESTDEAL

[Dominic Lusinchi]

Alice,

I am assuming you are using the Complex Sample module in SPSS, and have
specified what sampling design was used to collect the data.

To your question of which test to use chi-square v. Fisher's exact: the
latter should be used because in a 2x2 table if only one cell has an
expected frequency of less than 5, one of the assumptions for the chi-square
test is violated. SPSS gives you automatically the results of Fisher's test
in Crosstab when the analysis is on a 2x2 table.

Good luck.

Dominic Lusinchi
Statistician
Far West Research
Statistical Consulting
San Francisco, California
www.farwestresearch.com


-----Original Message-----
From: Alice P Chiu [mailto:[hidden email]]
Sent: Monday, March 05, 2007 2:29 AM
Subject: Analysis of a Stratified Cluster Sample

Dear all,
I would like to seek some advice regarding the analysis of a Stratified
Cluster Sample.
I would like to find the association between two variables. Within the 2x2
table, there was an expected count of  less than 5.
I was told that Fisher's exact test should not be used in such a scenaario.
Can you please advice which test should be used instead?
Thanks in advance!

Alice.



____________________________________________________________________________
________
Get your own web address.
Have a HUGE year through Yahoo! Small Business.
http://smallbusiness.yahoo.com/domains/?p=BESTDEAL

[Guan-Yu Lin]

Dear All,



I have several outcome measures.

For each DV, I conducted two separate repeated measure ANOVAs.

One had A & B as between group variables as well as C as within group
variable on each DV.

The other had A,B, & D as between group variables as well as C as within
group variable on each DV.



I found that the main effect of C on DV1 got significant as D was not
included into the analysis.

Moreover, the main effect of A and the interaction effect between A and C on
DV2 got significant as D was not included into the analysis.



What do these results mean? What results should I report?



Thank you in advance for any help you can offer.



Guan