A study investigated possible reasons for the low performance in 2000 on the writing portion of the Delaware Student Testing Program (DSTP) by students, especially in grades 3 and 5. The study also investigated ways to improve classroom instruction in writing. A panel of teachers reviewed the anchor papers and the process of testing. Panel members re-scored the anchor papers. A second panel of teachers participated in a re-scoring of a sample of 100 text-based student writings per grade (without using anchor papers). Panel members also discussed related issues in test administration, test development, scoring, and classroom instruction. Results include: (1) two text-based writing prompts should not be given on the same day; (2) instructions should be written to draw students' attention; (3) the new scores and the original scores on anchor papers were highly consistent in grades 3 and 5, and moderately high in grade 8 and grade 10; (4) passages should be engaging and the difficulty level should be consistent from year to year; (5) wording in the prompt should always direct students back to the text; (6) tenth grade teachers pointed out that most of the writing done by high school students is text-based, and that text-based writing is not a separate type of writing; (7) fifth grade teachers suggested that committees or teams develop questioning activities for teachers to use to improve students' performance on text-based writing; and (8) grade patterns for correlation coefficients were identified, with grade 3 being the lowest and grade 10 the highest. Changes planned for the Spring 2001 writing assessment include: the range for the total writing scores will be 1 to 15; two text-based writing tasks will be administered on different days; a prewriting sheet and scratch paper will be provided; and the text-based writing prompts will be formatted closer to the stand-alone prompts. Contains 6 tables of data. Appendixes contain the writing scoring rubric and a correlation matrix between reading and writing. (RS)