User talk:Practical123

Test message

Origen

I certainly didn't find the reference in Butterworth itself, but in some secondary source. I am not sure which Wikipedia article you are referring to, or the exact statement added; if I knew these, I might be able to be more exact. This reference work might have been upon which one I drew at the time. Hgilbert 04:23, 19 August 2006 (UTC)[reply]

It may have been from this source, which I perhaps should have checked in the original. Hgilbert 10:19, 19 August 2006 (UTC)[reply]

Do go ahead and delete the passage, as there is real uncertainty as to its validity. I can certainly not vouch for it. Hgilbert 17:43, 19 August 2006 (UTC)[reply]

Fleiss' kappa

Hi, can you tell me what needs cleaning up about the article? - Francis Tyers · 07:29, 3 April 2007 (UTC)[reply]

Hello Francis.

It would take me less time to rewrite the article myself than to list the various problems. For starters:

1. Why not begin the article with a short (1 or 2-word introduction), then list the contents?

2. "The scoring range is between 0 and 1." Don't kappa values range from -1 to 1?

3. "Significance": this word invites confusion with the issue of statistical significance (which the article doesn't address). "Intrepretation" is a better term in connection with the Landis & Koch table.

4. How does one determine statistical significance of kappa (p-values)?

5. This sentence is just goobledygook!:

"Agreement can be thought of as follows, if a fixed number of people assign numerical ratings to a number of items then the kappa will give a measure for how consistent the ratings are."

a. 'Agreement' - actually, reliability or consistency would be better words here

b. 'can be thought of' - awkward expression here

c. 'if a fixed number' - With Fleiss' kappa, are you sure the number of raters needs to be 'fixed'?

d. 'of people' - would 'raters' or 'judges' be better terms here? Do raters need to be people? Could they be algorithms, biochemical assay results, etc.?

e. 'assign numerical ratings' - do the ratings need to be numerical? Can Fleiss' kappa be used with non-numerical ratings (e.g., medical diagnoses)?

f. 'then the kappa' - odd phrase; better just to say 'kappa'

g. 'will give a measure' - better: will express

Practical123 13:56, 3 April 2007 (UTC) (J. Uebersax)[reply]