Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Intraclass correlation

16 views
Skip to first unread message

frank.d...@gmail.com

unread,
May 18, 2013, 4:22:21 AM5/18/13
to
Hi there,

I am calculating intraclass correlation coefficients to assess the reliability of visual scoring of brain scans in Parkinson's.

However, I did this three times:
*in SPSS,
*in software available from the university of Hong Kong at http://department.obg.cuhk.edu.hk/researchsupport/IntraClass_correlation.asp
*and then in Excel using the formulas given on http://luna.cas.usf.edu/~mbrannic/files/pmet/shrout1.htm

The results I got from the calculation 'by hand' were exactly equal to those delivered by the Hong Kong website. To my amazement, however, these differ from the SPSS results.
Could anyone explain what is happening here?

Thanks,

Frank

Art Kendall

unread,
May 18, 2013, 7:29:37 AM5/18/13
to
There are many places in SPSS where one can get some of the many kinds
of ICCs.
The algorithms are available in SPSS under <help>

I suggest that you
-- Google "nabble SPSSX"
--join that discussion group which has many very experienced statisticians
--post your query with the syntax you are using and a description of you
data. What are the variables? What is their level of measurement? What
is the response scale? How many scorers? How many times did a scorer
see the same scan? How many scans are there?

How different are the results?

If the data is not extensive and is not confidential you might also
attach your data.

As a courtesy to this list, post what you find out back on this list.

Art Kendall
Social Research Consultants

Bruce Weaver

unread,
May 18, 2013, 12:32:56 PM5/18/13
to
On 18/05/2013 7:29 AM, Art Kendall wrote:
> There are many places in SPSS where one can get some of the many kinds
> of ICCs.
> The algorithms are available in SPSS under <help>

--- snip ---

Robert Yaffee used to have some nice notes on how to compute the various
types of ICCs described in the well-known Shrout & Fleiss article with
SPSS. Unfortunately, it is no longer up. But you can see a version of
it via the "internet way back" archive.

http://web.archive.org/web/20120102084447/http://www.nyu.edu/its/statistics/Docs/intracls.html

HTH.


--
Bruce Weaver
bwe...@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/Home
"When all else fails, RTFM."
0 new messages