Desktop productivity for business analysts and programmers

Residuals Chi-square test

Reply
N/A
Posts: 1

Residuals Chi-square test

Hi,

First of all, statistics is not my best part Smiley Happy, but now I have to do quite a lot of table analysis with chi-square tests.

Therefore, I have to calculate the residuals for each cell.

My question:

Is there any possibility to calculate the standardized residuals by a chi-square test in SAS Enterprise Guide 7.1?

In "Cell statistics", there is choice to indicate "Cell contribution to Pearson Chi-square" e.g.

Or can I better calculate the residuals by myself (excel).

Thanks in advance!

Grand Advisor
Posts: 17,360

Re: Residuals Chi-square test

I don't know how to do it via point and click.  Via code you can request the deviation amount and expected amount.  Perhaps those are options in the output statistics section?

I'm not familiar with standardized residuals for chi square tests.  Given the statistical nature of this question you may want to post/move it to the SAS Statistical Procedure section instead of EG.

ods trace on;

proc freq data=sashelp.class;

table sex*age/deviation expected;

ods output crosstabfreqs=want;

run;

ods trace off;

proc print data=want;

run;

Grand Advisor
Posts: 9,582

Re: Residuals Chi-square test

Maybe you need this :

proc freq data=sashelp.class;

table sex*age/deviation expected chisq cellchi2 ;

ods output crosstabfreqs=want;

run;

Trusted Advisor
Posts: 1,056

Re: Residuals Chi-square test

Ah! Now that I see the SAS code, I can suggest the "point and click" method.

1. With your dataset open, select Describe | Table Analysis.

2. On the "Data" tab, drag your variables into the Table Variables selections.

3. On the "Tables" tab, drag one variable to the top, and the other to the left.

4. On "Cell Statistics", select Cell Frequency ... and Expected Cell Frequency

5. On "Table Statistics" / "Association", select Chi-square tests

Run the task.

How close is this to what you want?

  Tom

Ask a Question
Discussion stats
  • 3 replies
  • 718 views
  • 1 like
  • 4 in conversation