Help using Base SAS procedures

different types of correlations using proc corr

Accepted Solution Solved
Reply
Frequent Contributor
Posts: 143
Accepted Solution

different types of correlations using proc corr

Using this:

 

ods exclude all;
proc corr data=SomeInputData;
ods output PearsonCorr=SomeOutputData;
	var SomeMeasure;
	by SomeCategory;
run;
ods exclude none;

Gives me the Pearson correlation including p values. Can I also obtain the Kendall rank, Spearman and Point-Biserial correlation using proc corr (or other procs)?


Accepted Solutions
Solution
3 weeks ago
SAS Super FREQ
Posts: 3,903

Re: different types of correlations using proc corr

Posted in reply to csetzkorn

The answer is the same as last time you asked, but replace the word 'Pearson' by 'Spearman.'.

See https://communities.sas.com/t5/SAS-Procedures/add-p-values-to-output-dataset-of-proc-corr/m-p/430081

 

If I may suggest, it will be easier to ask and answer questions if you post code that uses one of the built-in SAS data sets in the SASHELP libref.  See also the SAS/STAT data sets.

 

My favorites in Sashelp are Class,Cars, Heart, and Iris.  You can use PROC CONTENTS to see the variables and PROC PRINT (OBS=5) to see a few records.For example, here is your solution using Cars:

 

proc sort data=sashelp.cars out=cars;
by origin;
run;

ods exclude all;
proc corr data=cars Spearman;
ods output SpearmanCorr=P;
	var mpg_city weight;
	by Origin;
run;
ods exclude none;

proc print data=P; run;

 

View solution in original post


All Replies
Super User
Posts: 21,478

Re: different types of correlations using proc corr

Posted in reply to csetzkorn

Somewhat, see the options on the PROC CORR statement in the documentation.

I don't know if Point-Biserial correlation is available. 

 

 

Highlighted
Super User
Posts: 12,148

Re: different types of correlations using proc corr

Posted in reply to csetzkorn

Did you SAS install not come with ANY online help or documentation?

 

You can request Fisher Hoeffding Kendal Pearson Polychoric Polyserial and Spearman correlations with proc corr.

SAS Super FREQ
Posts: 3,903

Re: different types of correlations using proc corr

To follow-up on Ballardw's comment, all documentation is posted online. Here are some URLs that I have bookmarked for easy reference:

 

Most Base SAS Procedures: CONTENTS, COMPARE, MEANS, RANK,...

Base SAS Statistical Procedures: CORR, FREQ, and UNIVARIATE  <== relevant for current question

DATA Step Functions and Call Routines

SAS/STAT User's Guide

SAS/IML User's Guide

 

Frequent Contributor
Posts: 143

Re: different types of correlations using proc corr

Hi all,

 

Thanks and apologies for not doing enough research. I wrote this on the mobile last night and did not have reliable Internet access.

 

I am currently using this code to obtain the Spearman correlation:

 

ods exclude all;
PROC CORR DATA=SomeInputData SPEARMAN out=SomeOutputData;
   var SomeMeasure;
	by SomeCategory;
RUN;
ods exclude none;

Is there a way to get p values similarly to the Pearson correlation code in my original question?

Solution
3 weeks ago
SAS Super FREQ
Posts: 3,903

Re: different types of correlations using proc corr

Posted in reply to csetzkorn

The answer is the same as last time you asked, but replace the word 'Pearson' by 'Spearman.'.

See https://communities.sas.com/t5/SAS-Procedures/add-p-values-to-output-dataset-of-proc-corr/m-p/430081

 

If I may suggest, it will be easier to ask and answer questions if you post code that uses one of the built-in SAS data sets in the SASHELP libref.  See also the SAS/STAT data sets.

 

My favorites in Sashelp are Class,Cars, Heart, and Iris.  You can use PROC CONTENTS to see the variables and PROC PRINT (OBS=5) to see a few records.For example, here is your solution using Cars:

 

proc sort data=sashelp.cars out=cars;
by origin;
run;

ods exclude all;
proc corr data=cars Spearman;
ods output SpearmanCorr=P;
	var mpg_city weight;
	by Origin;
run;
ods exclude none;

proc print data=P; run;

 

☑ This topic is solved.

Need further help from the community? Please ask a new question.

Discussion stats
  • 5 replies
  • 191 views
  • 6 likes
  • 4 in conversation