Structure and Logistics

• Are more appropriately public; involve an audience or panel.
• Do not rely on unrealistic and arbitrary time constraints.
• Offer known, not secret, questions or tasks.
• Are more like portfolios or a season of games, not one- shot.
• Require some collaboration with others.
• Recur-and are worth practicing for and retaking.
• Make assessment and feedback to students so central that school schedules, structures, and policies are modified to support them.

Intellectual Design Features

• Are “essential”-not needlessly intrusive, arbitrary, or designed to “shake out” a grade.
• Are “enabling”-constructed to point the student towards more sophisticated use of the skills or knowledge.
• Are contextualized, complex intellectual challenges, not “atomized” tasks corresponding to isolated “outcomes.”
• Involve the student’s own research or use of knowledge, for which “content” is a means.
• Assess student habits and repertoires, not mere recall or plug-in skills.
• Are representative challenges-designed to emphasize depth more than breadth.
• Are engaging and educational.
• Involve somewhat ambiguous tasks or problems.

Grading and Scoring Standards

• Involve criteria that assess essentials, not easily counted but relatively unimportant errors.
• Are graded not on a curve but in reference to performance standards (criterion-referenced, not norm-referenced).
• Involve demystified criteria of success that appear to students as inherent in successful activity.
• Make self-assessment a part of the assessment.
• Use a multifaceted scoring system instead of one aggregate grade.
• Exhibit harmony with shared schoolwide aims-a standard.

Fairness and Equity

• Ferret out and identify (perhaps hidden) strengths.
• Strike a constantly examined balance between honoring achievement and native skill or fortunate prior training.
• Minimize needless, unfair, and demoralizing comparisons.
• Allow appropriate room for student learning styles, aptitudes, and interests.
• Are attempted by all students, with the test “scaffolded up,” not “dumbed down,” as necessary.
• Reverse typical test-design procedures. A model task is first specified; then, a fair and reliable plan for scoring is devised.

(These are provided by Grant Wiggins, former director of research at CES; he gives credit to Ted Sizer, Art Powell, Fred Newmann, and Doug Archbald and to the work of Peter Elbow and Robert Glaser for some of these criteria.)