Date of Award

Spring 1991

Project Type


Program or Major

Reading and Writing Instruction

Degree Name

Doctor of Philosophy

First Advisor

Grant L Cioffi


Schools, districts, states, and testing organizations routinely assess writing ability with timed, prompted writing samples. Based on these results, those in control swing the gate open or slam it shut for thousands of students, teachers, administrators and schools, in decisions about promotion, admission, retention or funding.

This study demonstrates that timed writing samples poorly predict actual classroom writing performance, underestimating the weakest and poorest while overrating the strongest. Self-selected portfolios from 263 randomly selected students in grades five, eight, and eleven across eleven school unions also provide a clearer picture of the development of writing abilities from elementary through high school than prompted writing samples.

High school students in this study are less able than their middle and elementary school peers to predict adult ratings of their work, and less frequently agree with adults about what makes their writing strong. High schools also provide students with the shortest times to develop and revise their writing. Conversely, students seem to progress from grades five to eight in their ability to keep their work open to revision longer and to maintain a sense of adult standards. Finally, students who develop adult-like judgment on the emotional rather than the conceptual or language qualities of papers tend to produce more highly rated papers.