Jay, Thanks for the feedback. You are right that I'm not going to generalize to other judges. The question has been answered. : )

--Steve

At 08:24 PM 7/3/2004 -0400, Goodwin, Gerald F wrote: >Assuming that your two judges are the only two you care about (i.e., you >are not going to argue that these two judges ratings are generalizable to >a broader set of judges), yes your approach is correct. > >HTH > >Jay > > >---------- >From: Stephen Salbod >Sent: Sat 7/3/2004 7:56 PM >To: SPSSX-L@LISTSERV.UGA.EDU >Subject: Intraclass Correlation Coefficient > > >I'm looking for confirmation: > >I selected 2 judges (fixed). >Each judge counts the number of fear words used by a participant in a >paragraph. >The average of the 2 measures will be the unit of measure in the study. >I have 59 participants. >Thus, data editor, has 59 participants (targets) with 2 measures each. > >I plan to use ICC (3, k); > >Details: > ICC (3, 2): Two-way mixed model; unit of analysis is average > rating; and >consistency definition. > >I was unable to read the original 1979 Shrout & Fleiss article; however, I >was able to obtain the 1996 McGraw & Wong article, and, also I was able to >download from the web, Yaffee's article for SPSS V8. I think I understand >the material, but I want to get a second opinion. > >Thus, I would appreciate any comments, suggestions, or additional readings. >Thank you and have happy 4th. > >Stephen Salbod, Pace University NYC