Book/Relative grading

Grading by comparison with the performance on the group (relative grading – norm-referenced) is based on the performance of the student in the context of the group. It derives the grade from the student's ranking within a certain group. We rank the students according to their performance on the test and then assign them grades according to pre-agreed limits. Relative grading assumes that the performance of different study groups (across space and time) is basically the same. From the student's point of view, this method of classification contains an obvious injustice, because the evaluation depends not only on their own performance, but also on the performance of others. It is therefore possible that if the student had been in a different group (for example, if he or she had taken the test in a different school year), they would have received a better grade with the same level of knowledge.

If we want to use an evaluation based on relative standardization, two decisions need to be made. First of all, it is necessary to determine what grade we will assign to the average performance. In the frequently used four-level grading system A, B, C, D, we can intuitively choose the border between B and C as the grade corresponding to average performance; however, that is not the only option.

Furthermore, it is necessary to decide in advance on the boundaries separating the individual grade levels. For example, z-scores or percentile rankings are used to determine specific grades based on performance, in a similar way to that described in the chapter on relative test-passing cut-offs.

For a four-level grading scale then, the boundaries between the individual grade levels correspond to, for example, z-scores –2, 0, 2, as indicated in Figure 5.7.1:

Fig. 5.7.1 Example of grading a student's result on a test using z-scores (relative grading). The group is divided according to the test scores in such a way that the resulting subgroups are separated from the average by an agreed number of standard deviations. We grade the subgroups created in this way according to the relevant classification scale – in this case four-level. Note that the highest grade “excellent” is given in this case to only 2.2% of the test takers, the average grades “very good” and “good” are given to a large majority of students (each of these groups comprises 47.7%) and again only a completely negligible number of students (2.2%) gets the lowest grade of “fail”.

Splitting into subgroups based on standard deviations from the mean (z-scores) is not the only option. An alternative is to divide the group according to the achievement scores into subgroups of equal size, and give these subgroups the same grade. For example, the distribution could be such that the top 25% get an “excellent” grade, another 25% get a “very good” grade, etc.

Advantages and Disadvantages

• Grading systems based on student comparisons are simple and easy to use.
• They work well in situations where students need to be lined up, for example as part of entrance and admission tests to a department of study in which there is a limited number of places.
• They are appropriate for large courses that do not encourage collaboration among students, but generally emphasize individual achievement.
• The obvious disadvantage is that individual are graded not only on results, but the results of other students also determine the student’s results.
• The assessment threshold can only be established after the test has been given. Therefore, it is not possible to comment in advance on how difficult the test will be (although it is known in advance how the threshold will be determined).
• Relative assessment will be more applicable in large non-selective groups that will be representative of the entire student population. In small classes (under 40), this group may not be a representative sample. One student may get an excellent grade because he is in a low-achieving group, while their classmate with the same result in a better group gets a lower grade.
• A second objection to grading in relation to others is that it encourages competition rather than cooperation. This method of assessment sets up a relationship of direct competition between students. When students are pitted against each other for a few top grades to be handed out, they are less likely to cooperate with each other in their studies.

A compromise solution for small groups is to use so-called “anchoring” in the relative evaluation. This means grading will be adjusted according to the overall (average) level of the students in the group. If a teacher has used a similar test repeatedly in different years, he or she can use the accumulated test results as an anchor. The teacher then compares the current group with this collected large group. Similarly, a well-constructed pretest can be used as an anchor, in which we estimate the ability of the entire group using absolute criteria. Modifying the relative grading system with anchoring helps reduce feelings of competition between students because then they are no longer competing only with each other.