Dear Statalisters,
I have a 65 observation, we want to look at the variable only walking >=6 minutes correctly assign in the outcome variable Frail or No Frail. Below is cross tabulation of variables. I did, two analyses, slightly change the data. I got two separate answers. Please advise.
Code:
| frail walk | No Yes | Total -----------+----------------------+---------- >=6 | 5 60 | 65 ----------+----------------------+---------- Total | 5 60 | 65 kap frail walk Expected Agreement Agreement Kappa Std. Err. Z Prob>Z ----------------------------------------------------------------- 92.31% 92.31% 0.0000 0.0000 0.00 0.5000 . kappaetc frail walk Interrater agreement Number of subjects = 65 Ratings per subject = 2 Number of rating categories = 2 ------------------------------------------------------------------------------ | Coef. Std. Err. t P>|t| [95% Conf. Interval] ---------------------+-------------------------------------------------------- Percent Agreement | 0.9231 0.0333 27.71 0.000 0.8565 0.9896 Brennan and Prediger | 0.8462 0.0666 12.70 0.000 0.7131 0.9792 Cohen/Conger's Kappa | 0.0000 0.0000 . . 0.0000 0.0000 Scott/Fleiss' Pi | -0.0400 0.0180 -2.22 0.030 -0.0760 -0.0040 Gwet's AC | 0.9169 0.0387 23.68 0.000 0.8396 0.9943 Krippendorff's Alpha | -0.0320 0.0180 -1.78 0.080 -0.0680 0.0040 ------------------------------------------------------------------------------ . kappaetc , benchmark showscale Interrater agreement Number of subjects = 65 Ratings per subject = 2 Number of rating categories = 2 ------------------------------------------------------------------------------ | P cum. Probabilistic | Coef. Std. Err. P in. >95% [Benchmark Interval] ---------------------+-------------------------------------------------------- Percent Agreement | 0.9231 0.0333 0.99 0.988 0.8000 1.0000 Brennan and Prediger | 0.8462 0.0666 0.25 0.988 0.6000 0.8000 Cohen/Conger's Kappa | 0.0000 0.0000 . . . . Scott/Fleiss' Pi | -0.0400 0.0180 . . . . Gwet's AC | 0.9169 0.0387 0.98 0.980 0.8000 1.0000 Krippendorff's Alpha | -0.0320 0.0180 . . . . ------------------------------------------------------------------------------ *create frail and walk into one variable gen frail_walk=1 if walk==1 & frail==1 replace frail_walk=0 if frail_walk~=1 gen nofrail_walk=1 if walk==1 & frail==0 replace nofrail_walk=0 if nofrail_walk~=1 kap frail_walk nofrail_walk Expected Agreement Agreement Kappa Std. Err. Z Prob>Z ----------------------------------------------------------------- 0.00% 14.20% -0.1655 0.0205 -8.06 1.0000 . kappaetc frail_walk nofrail_walk Interrater agreement Number of subjects = 65 Ratings per subject = 2 Number of rating categories = 2 ------------------------------------------------------------------------------ | Coef. Std. Err. t P>|t| [95% Conf. Interval] ---------------------+-------------------------------------------------------- Percent Agreement | 0.0000 0.0000 . . 0.0000 0.0000 Brennan and Prediger | -1.0000 0.0000 . . -1.0000 -1.0000 Cohen/Conger's Kappa | -0.1655 0.0766 -2.16 0.034 -0.3185 -0.0125 Scott/Fleiss' Pi | -1.0000 0.0000 . . -1.0000 -1.0000 Gwet's AC | -1.0000 0.0000 . . -1.0000 -1.0000 Krippendorff's Alpha | -0.9846 0.0000 . . -0.9846 -0.9846 ------------------------------------------------------------------------------ Confidence intervals are clipped at the lower limit. . kappaetc , benchmark showscale Interrater agreement Number of subjects = 65 Ratings per subject = 2 Number of rating categories = 2 ------------------------------------------------------------------------------ | P cum. Probabilistic | Coef. Std. Err. P in. >95% [Benchmark Interval] ---------------------+-------------------------------------------------------- Percent Agreement | 0.0000 0.0000 . . . . Brennan and Prediger | -1.0000 0.0000 . . . . Cohen/Conger's Kappa | -0.1655 0.0766 . . . . Scott/Fleiss' Pi | -1.0000 0.0000 . . . . Gwet's AC | -1.0000 0.0000 . . . . Krippendorff's Alpha | -0.9846 0.0000 . . . . ------------------------------------------------------------------------------
Sugi
0 Response to Agreement (kappa)
Post a Comment