They take more time to create and use. There are more possibilities for raters to disagree. It is more difficult to achieve intra- and inter-rater reliability on all of the dimensions in an analytic rubric than on a single score yielded by a holistic rubric. There is some evidence that raters tend to evaluate grammar-related categories more harshly than they do other categories (McNamara, 1996), thereby overemphasizing the role of accuracy in providing a profile of learners' proficiency. There is some evidence that when raters are asked to make multiple judgement, they really make one. Care must be taken to avoid a "halo effect" and focus on the individual criteria to assure that diverse information about the learner's performance is not lost.