Certain testing procedures will overcome some of the problems associated with the use of essay tests. Essay tests may not validly indicate achievement because the questions included in the test may not fairly represent instructional content. Reliability may be a problem because of variations in examinee response in different situations, in test markers from one time to another, and between different markers. To offset validity problems, only those content areas which have been taught should be assessed. Writing ability should be tested at different times, using different types of essays and as many questions as possible. Questions should be phrased clearly and pretested before use. Intra-marker reliability is increased when graders score all examinee responses on a question, before moving to the next question. Inter-marker reliability is enhanced when the rating scale has seven to fifteen points, and clear distinctions are made between each mark. In analytic marking, values are assigned for each of the evaluation criteria--such as organization, style, content, and mechanics--and then totalled. In impressionistic or wholistic scoring, papers are sorted into three piles representing varying quality; then each pile is further sorted. (Six steps for judging reliability of scoring procedures are appended). (GDC)