I have genetic data. I have two tests that pick up changes in the genome (amplifications and deletions) at several different points in the DNA. There are three basic outputs- high, normal, low. I want to check the agreement between the two tests for these three states. I'm not sure but I think this is a kappa statistic. If so how is that any different to just assessing the percentage of points that agree and disagree?
You can use a chance-adjusted index of ordinal reliability such as a weighted kappa. This will be different from percent agreement in two ways. First, it will adjust for the amount of estimated chance agreement. Second, it will award partial credit for being one category off.