Criterion-referenced grades are determined by an individual’s performance against specified learning outcomes, whereas proportionate grades are determined by individual performance compared to the performance of all others in a class.
With criterion-referenced grading, student grades are based on how well they perform against specified learning outcomes and standards.
Effective assessments give instructors data with which to accomplish the following:
- Determine if students have achieved the course learning outcomes
- Identify content areas where students are performing either strongly or weakly
- Improve learning activities and assessment practices within the course
With this data in hand, instructors also calculate course grades. Depending on the individual preferences of the instructor and course policy, scores are usually based on either performance standards (criterion-referenced) defined in the course curriculum and syllabus, or by comparing scores among students (norm-referenced or curving).
- Instructors assess and score based upon the individual student’s performance against the specified learning outcomes and performance standards.
Skills and knowledge associated with a particular course are usually defined and documented within the course syllabus.
- Skills are introduced, modeled, practiced, improved on, and mastered within the instructional activities of the course, both inside and outside of the formal class meeting.
- It is possible for all students to excel or for all to perform poorly.
- Since scores are not standardized, it is possible for the distribution to take on virtually any shape.
- Students who perform exceptionally well may not be as distinguishable from those who barely exceed the established performance standard.
Here is a list of tips to effectively use criterion-reference assessments:
- Define the content for the assessment in terms of knowledge and skill that the student must demonstrate.
- Define a scoring rubric for judging student performance. If the assessment is a test, the rubric will be a simple “correct/incorrect” judgment. If the assessment is a presentation, project, essay, term paper, etc., then the rubric will be more detailed and include the major knowledge and skill categories along with their respective scoring standards.
Make sure that the assessment employs the same skills that the students used within the learning activities leading up to the assessment.
Proportionate Grading or Curving
Most of the current literature on assessing student learning discourages grading on a curve, although many authors acknowledge that the practice of curving grades is sometimes a mandated policy within individual academic departments at many universities.
Sometimes faculty may feel that curving actually helps students because it minimizes the effect of lower than expected scores. However, if student scores are lower than expected, the instructor should examine the assessment and verify that it is aligned with the corresponding learning outcomes and activities.
Additionally, some may argue that curving grades prevents grade inflation. Although curving does limit the number of “A” grades in a course, it may also interfere with the accuracy of grades as indicators of the actual work and learning accomplished by students.
The main concern regarding norm-referenced assessments is that instructors take on significant risk in implementing such assessments without sufficient statistical data to adequately prove the validity and reliability of their students’ scores.
If curving grades is a requirement in a department, instructors should take into account some critical questions:
- What is your purpose for curving the scores on this exam? Is it a department requirement? Are there other reasons?
- What data to you have to back up your decision to curve the scores on this exam? Have the test questions been through validity checks?
- Do you have a sufficient number of students who have taken the exam so that you can be confident with the reliability statistics?
- Are you confident with the comparisons of students within the scores distribution?
Curving test scores is usually accomplished by computing the mean of the scores and the standard deviation and then plotting the scores along the adjusted curve and assigning grades within that revised distribution. At the course level, these assessments are not usually put through the rigors of validity or reliability studies.
Read the full version of this article from the Fall 2005 edition of Focus on Faculty.