chisq issue

[tmm]

so now I am thinking I should just export the table and then manually do stuff in excel to make it look like the example I was provided and import to SAS but that I am sure that is not what past programmers in my company have done. I am sure they figured out a way to do it all in SAS and my boss has even said I should do this all in SAS but I am left to figure it out on my own and as we can see by my number of posts I am not able too. Have read tons of books and just spinning my wheels and wasting time. I have my meeting Friday to show my data but I am no where near able to do that unless I figure this out.

[tmm]

I think I got it. I played and played and this is the code I used:

proc sql;

create table chi4 as

select * from

(select distinct who as source, who, svcd, svdesc, admission, count from chi)

union all

(select distinct "national", a.who, b.svcd, b.svdesc, b.admission, b.count from

chi as a inner join chi3 as b

on a.svcd = b.svcd and

a.svdesc = b.svdesc and

a.admission = b.admission)

order by who, svcd, source, admission);

run;

The output chi4 table looks like this:

who        who          svcd           svdesc          admission             count

1111       1111          1             minor              no                   1

1111       1111          1             minor              yes                  0

national   1111          1             minor              no                  31645

national   1111          1             minor              yes                  52

2222       2222          2             moderate           no                   1

2222       2222          2             moderate           yes                  0

national   2222          2             moderate           no                  41864

national   2222          2             moderate           yes                  154

etc

Now I have to do a step to drop the 2nd who column or rename the label or something so when I do my chi it looks at the who with the numbers and national and then the remaining columns. Because if I use the 2nd column then it will add all the counts together. If I use first column it will chi it correctly from what I can tell.

[tmm]

I added a data step because even though the labels in the chi4 table both show who, the actual column name is source and who so I did this step:

data chi5 (drop = who);

set chi4;

run;

It drops the 2nd who column and leaves the other who column which is actually source and looks the way it should. So let me see how my chi works now.

[mlamias]

Why on earth would you want to perform a Chi-Squared Test on a table with this many cells?

[SteveDenham]

And with so many cells with virtually no representation (counts<2)?  A sensible aggregation seems in order here.

Steve Denham