A) read correlation matrix into sas; B) use that matrix as input to proc corr to compute reliability

[emaguin]

Exactly what the subject line says. Given, for example, a correlation matrix of y1-y6 in text format, read it into sas and use that matrix as input to proc corr to compute the reliability of the scale defined by y1-y6.

[Tom]

Reading a correlation matrix should not be any different than reading in any other type of data.

What is your formula for calculating reliability?

How would you calculate it from the original data?

Please share an example input dataset and the expected result for that data.

[emaguin]

In her reply, Paige says that proc corr won't take matix input and that is what I thought was true when I looked at the proc corr documentation but I wanted to ask because my sas knowledge would only fill a thimble but one with a hole in it. I know that sas has IML and it can be used to do amazing things, including, I'm sure, computing a reliability coefficient but I don't know how use IML.
I can do this in spss with two commands: matrix data and reliability (which accepts a matrix data file as well as regular data file as input).
Thank you for replying.

[PaigeMiller]

A correlation matrix cannot be input to PROC CORR.

How do you plan to compute reliability?

[emaguin]

I thought was true when I looked at the proc corr documentation but I wanted to ask because my sas knowledge would only fill a thimble albeit one with a hole in it. I know that sas has IML and it can be used to do amazing things, including, I'm sure, computing a reliability coefficient but I don't know how use IML.
I can do this in spss with two commands: matrix data and reliability (which accepts a matrix data file as well as regular data file as input).
Thank you for replying.

[ballardw]

Please show the text of your "correlation" matrix.

Type=Corr data sets have to be of very specific structures we need to see if you even have all of the required elements. You may want to search the online help for "Type=corr datasets" to see the basic details.

It is likely better to provide the raw data used to make your correlation matrix if available and let Proc Corr make the set or do the analysis on the raw data.

[emaguin]

Thanks for pointing out that sas has a corr dataset and many more types as well based on the search you suggested. I understand your point but the format and structure of a sas type=corr data set is that for a matrix dataset in spss and I know to make them, I've made them and am making them. Example A.2+Figure A.3 in that search shows how to do that and the result.
In her reply, Paige says that proc corr won't take matix input and that is what I thought was true when I looked at the proc corr documentation but I wanted to ask because my sas knowledge would only fill a thimble but one with a hole in it. I know that sas has IML and it can be used to do amazing things, including, I'm sure, computing a reliability coefficient but I don't know how use IML.
I can do this in spss with two commands: matrix data and reliability (which accepts a matrix data file as well as regular data file as input).
Thank you for replying.

[Ksharp]

https://support.sas.com/kb/30/333.html#errinvar

proc corr data=sashelp.class alpha;
var weight height age;
run;