I have a set of symptoms that may belong to an A condition, B condition, C (both A+B) or D (neither of the options) and I asked two groups of clinicians to rate the belongingness of each symptom to one of the four conditions. I used Gwet´s AC1 to assess interrater agreement of the overall 16 symptoms over each group and ran a paired t-test to evaluate group differences in their ratings (two groups assessing the same set of symptoms). I am unsure about three issues:

  • to what extent AC1 fits better than other agreement statistics to analyze these results (raters classifying symptoms -a dependent variable with nominal categories).
  • how convenient is to evaluate agreement over each symptom instead of the set of symptoms (I wasn´t able to perform this analysis using both irrCAC and pairedCAC R packages)
  • whether t-test is the right statistic to perform the comparison. I am pretty sure I am wrong b/c there is not really a mean comparison --dependent variable is nominal. thanks in advance!
  • More Fernán Arana's questions See All
    Similar questions and discussions