Intraclass correlation coefficient and its CONFIDENCE LIMITS

[Luba]

Using this macro <> and the instruction provided i was able to obtain the different ICC for my reliability data.

How do I obtain corresponding 95% confidence limits????

Thanks.

[Luba]

CORRECTION:

this macro: http://support.sas.com/kb/25/031.html

Using this macro

http://support.sas.com/kb/25/031.html

and the instruction provided i was able to obtain the different ICC for my reliability data.

How do I obtain corresponding 95% confidence limits????

Thanks.

[lvm]

There are many versions of the ICC, and McGraw and Wong (1999; Psychological Methods, vol. 1, no. 1, pp. 30-46) discuss many of them. They show how to get confidence intervals. This is not built into the macro. A different macro is given in:
http://www.nesug.org/proceedings/nesug07/sa/sa13.pdf
This one calculates a much more limited set of ICCs, but it does give confidence intervals (using one of the possible methods).

[Chapi]

Hello, The link that was pasted above doesn't work. Is it got updated or removed any idea?

[FreelanceReinh]

Hello @Chapi and welcome to the SAS Support Communities!

Whenever you need a paper from one of the major US or international SAS conferences, https://www.lexjansen.com is probably the most comprehensive source you can find. The link to the paper in question on that website is https://www.lexjansen.com/nesug/nesug07/sa/sa13.pdf.

[Chapi]

Thank you so much!! I will look up there.

[Luba]

thanks...

yes, i used the macro that you sent as well... but i switched to the other one cause it offered more ICCs....

but good to know that i was on the right track....

[statsplank]

In any case, whether the analytical version (i.e. formula) for CI exists or not, you can obtain the CI by bootstrap.

[Ksharp]

Here is my code for ICC 95% CI based on this macro: http://support.sas.com/kb/25/031.html