ABSTRACT
With a small number of ordinal or nominal scores, two raters have more of a chance of agreeing compared to raters who are using continuous or quantitative scores. For example, if two raters are using say three scores to assign to subjects, they can disagree as follows: If rater A assigns a score of 1, then rater B can disagree by assigning a score of either 2 or 3. Thus for every assignment of a score by rater A, rater B can disagree in two ways. On the other hand, if raters are using a continuous score, in principal, they can disagree in an infi nity of ways. Of course, on a practical level, even using continuous scores, there is only a fi nite number of possible score assignments that are actually used, but if that number is “large”, say 100, there is a huge number of ways to disagree. In practice, since raters are judging the same subjects, their scores are correlated and the number of disagreements is limited, compared to randomly assigning scores. By quantitative or continuous scores is meant variables like weight, height, blood pressure, tumor size, SUV (standardized uptake values in radiology), blood glucose values. etc.
