Margaret,
When I compute by hand (actually by spreadsheet), I get -0.61314.
What formula are you using? You'll want to do this in a spreadsheet as the Kappa stat is sensitive to rounding.
David Futrell
Senior Workforce Research Consultant
Eli Lilly and Company
----- Original Message ----
From: Margaret MacDougall <
[hidden email]>
To:
[hidden email]
Sent: Tuesday, January 16, 2007 8:35:09 AM
Subject: Cohens' Kappa Statistic with Binary Data Involving a Zero Frequency
Dear SPSS Users
Apologies for cross-posting
I would be most grateful for some confirmation re the appropriate use of Cohen's Kappa Statistic when considering binary data for which there is a zero frequency on cross-tabulation.
By way of illustration, suppose that there are two tests for diagnosing presence or absence of neuropathic pain and one wishes to assess agreement between the two. The observed frequencies are as listed below.
Absent with both methods: 0
Absent with Method 1 but present with Method 2: 7
Present with Method 1 but absent with Method 2: 6
Present with Method 1 and present with Method 2: 4
If I attempt to calculate Cohen's Kappa by hand, using the standard formula, I get a value of -0.625. However, if I use SPSS, Minitab or either of a few online calculators I have accessed, in each case the value generated is -0.613.
I would be most grateful to receive advice on which if any of these values is the correct one and an explanation for the discrepancy.
Many thanks
Best wishes
Margaret
---------------------------------
All new Yahoo! Mail "The new Interface is stunning in its simplicity and ease of use." - PC Magazine
____________________________________________________________________________________
Sucker-punch spam with award-winning protection.
Try the free Yahoo! Mail Beta.
http://advision.webevents.yahoo.com/mailbeta/features_spam.html
Locked
