[R] Cohen's Kappa for beginners

Matthew Schmidt matthew.schmidt at gmail.com
Thu Feb 10 06:23:34 CET 2011


Refresh list was used, but a restart of the client was necessary for
updates to become active.

Please advise as to appropriate list. Many thanks!

On Wednesday, February 9, 2011, David Winsemius <dwinsemius at comcast.net> wrote:
>
> On Feb 9, 2011, at 3:29 PM, matthew.schmidt wrote:
>
>
>
> I'm a newb with R and am trying to move from proprietary stats tools to open
> ones. Hopefully this post might help others doing the same.
>
> I downloaded the R package for Mac
>
>
> Posted to the wrong list.
>
>
> and hacked together a couple scripts, but
> couldn't seem to get the right output. After installing the psych package, I
> tried the test script included in this thread on the console, but I got
> similar results to a prior poster:
>
>
> x
>
> [1] "red"    "yellow" "blue"   "red"
>
> y
>
> [1] "red"  "blue" "blue" "red"
>
> ck <- cohen.kappa(cbind(x,y))
>
> Error in counts[i, j] <- sum(scores[i, ] == score.levels[j], na.rm = TRUE) :
>  subscript out of bounds
> In addition: Warning messages:
> 1: NAs introduced by coercion
> 2: NAs introduced by coercion
>
> Since I got the impression that there may be a version problem, I updated
> the psych package to the most recent version, 1.0-94 (from source). After
> this, I tried over and over to run the cohen.kappa command, and still kept
> getting the same error.
>
> So I quit R and restarted it, then went to the package manager and selected
> the checkbox next to "psych" to load it, and tried to re-run the command.
>
> This time I got the expected output:
>
> Call: cohen.kappa1(x = x, w = w, n.obs = n.obs, alpha = alpha)
>
> Cohen Kappa and Weighted Kappa correlation coefficients and confidence
> boundaries
>                  lower estimate upper
> unweighted kappa  0.098      0.6  1.10
> weighted kappa   -0.693      0.0  0.69
>
> Number of subjects = 4
>
> Lesson learned: close and restart the Mac client if you're using the GUI
> front-end in order for changes to the package manager to take effect.
>
>
> I have a small screenshot for you to look at:
>
>



More information about the R-help mailing list