The purpose of this study was to examine the stability of measures of student writing performance across types of discourse (genres) and across response modes (selected response: multiple choice; constructed response: single paragraph, and full length essay). The study addressed the following: (1) the relationship/stability of writing scores within and between genres when students are tested on different occasions (on writing tasks in the same discourse mode, in different discourse modes); and (2) the relationship/stability of writing scores across selected response items (multiple choice items), brief constructed response items (single paragraph task), and longer constructed response item (full essay 1-2 pages, 3-5 paragraphs). Results indicated that the relationship between a student's scores on essays written in the same mode of discourse is stronger than the correlation between two essays in different discourse modes. These findings suggest that generalizations about student writing competency should take care to reference the discourse domain rather than the more general domain of writing, as students' command of different discourse structures vary. (Author/GK)