The Effect of Selected Composition Errors on Grades Assigned to Essay Examinations

Abstract

For many years measurement specialists have pointed out the deficiencies of the essay examination as an instrument for measuring achievement in content subjects. Writers have noted that the grades assigned to a set of essay exams may reflect, in addition to the quality of the answers, factors that are unrelated or indirectly related to the academic goals of the subject covered by the examination. The quality of the students' written composition is one factor that would logically appear to be a serious cause of low inter-reader reliability. Flaws in writing are readily apparent to most teachers and certainly distract teachers from concentrating on the content of the answer. However, teachers are differentially sensitive to such flaws and, thus, grades on most essay exams reflect to an unknown extent the composition skills represented in the papers being graded. Most teachers and measurement specialists are unwilling to abandon essay examinations. However, if measures of achievement are to be useful, they must reach a certain minimal level of reliability, and a problem of continuing interest is that of finding ways to improve the reliability of marks assigned to essay exams. It has been argued that reliability would be improved if grades did not reflect the quality of written composition (Adams, 1964; Bean, 1953). However, relatively little research has been directed to the identification of the composition skills that influence grades or to the development of grading procedures that minimize this effect.