Grading advice

The teacher has considerable flexibility in terms of how they convert marks into grades. While strategies like normalizing grades on a Gaussian distribution are widely used internationally, under the EPFL system, grades should be related to attainment of particular learning goals and not to comparisons with other students.

Here is a procedure that might be useful for helping with that process (it is based on a procedure outlined by our colleagues in ETHZ. Their guidelines document is accessible here):

First, identify what level of performance you would associate with two specific points on your scale:

• ‘Excellent’ (grade 6) should be reserved for students who show real mastery of the learning goals. The assessment should be designed so it is normal for some students to get a 6 (although this may not happen every year).
• ‘Minimum pass’ (grade 4) should reflect those who have shown that they have more or less achieved the learning outcomes outlined for the course without showing more than a basic competence.

Once you have defined the “excellent” and “minimum pass” grades, the marks corresponding to the intermediate levels can be read by linear interpolation between grades 6 and 4, and between grades 4 and 1.

Since teachers design a new assessment each year, it is normal that there will be some variability in difficulty of exams, and thus it is not required that the same relationship between marks are grades is applied each year. In altering the mark to grade conversion, however, the focus remains on attainment of learning goals rather than on comparison between students in a cohort.

If you grade a hundred exams over three days, can you be sure that the grade you will give a script on day one is the same as the grade you would give the same script on day three? If you have more than one person involved in grading, can you be sure that they would each give the same grade to the same script? Here are a few simple ideas that can help to ensure that your grading rewards able students rather than lucky ones:

• Have a clear grid which reminds you what marks are assigned for. For example, do you give some marks for applying the correct concepts or using the correct formula even if the calculation is inaccurate? If so, how many marks? If you write down clearly what marks are assigned for and refer to it continuously when grading, then you have a better chance of treating the last script the same way as the first one.
• If there is more than one person involved in the grading, then this ‘grading grid’ becomes even more important. It is a good idea too to start by having each member of the grading team grade the same few scripts and to compare their grades. This will allow you to calibrate your standards and to ensure greater uniformity across graders.
• Give the grading grid to students in advance of the exam. If students do not know what the criteria are, then some of them – lucky students – will meet your criteria by accident, while others will not meet them even if they actually understand the content.  If, on the other hand, you make the criteria clear, you give students who understand the material the information they need to be able to succeed in the exam.
• Give students a chance to practice. Even if you feel that your instructions are incredibly clear, some students will best understand what you want from them when they have had a chance to see it in practice. If you can provide them with a mock exam or a mid-term which uses the same format at the final exam then they will get a feel for what it required of them. In the case of assessment by projects or reports you might give them some (unmarked and anonymised) examples of previous work along with the assessment grid and invite them to try grading them for themselves. Again, this will give them a feel for how the grading criteria will be applied in practice.

A grading grid or grading rubric is simply a clear statement, as to what it is that you reward with marks and how many marks you give under each heading.

• Grids can be very simple (listing a few criteria and the maximum mark awarded for each criterion), or more complex (listing criteria and sub-criteria and describing in more what a satisfactory, good and excellent performance looks like under each criterion).
• Grids can be generic (describing criteria and marks for all assessments of a given type) or specific (describing what the teacher is looking for in each question).

In general, it could be argued that simple and generic grids are more suitable for exams which are based on questions which demand a similar set of skills (e.g. exams where the questions are based on problem sets or where all the questions are essay style), while specific and more detailed grids may be appropriate for coursework which requires multiple skills to be displayed (e.g. project reports that require design, conceptualisation, analysis, writing and presentation skills).

Why use grading grids?

• Grading grids can help increase consistency in grading.
• They can clarify for students what the teacher expects instead of either guessing “what the teacher wants”, or assuming that the criteria applied in other courses also apply here (ideally the teacher provides the students with the grading grid in advance of the exam).
• Some teachers find grading rubrics streamline the assessment process and so reduce the amount of time spent on grading.

What do grading grids look like?

There is no single right way of doing a grading grid.

A simple, generic grading grid for exam questions for an introductory Physics course, for example, might look like this:

But remember that grids can also be more complex and more specific. Here are some examples of other grading grids. Some are quite detailed, others more general:

Computer Programming (from California State University Long Beach site)

Essay style paper (from Penn State site)

Oral Presentation (from University of Florida site)

Some things to remember…

• A big part of the value of a rubric is making explicit to yourself and your students what exactly it is that you are (often implicitly) looking for. Don’t just pick up someone else’s rubric, take the time to tease out what exactly it is you want students to do.
• Ideally, when you make your assessment criteria explicit, you should find that they match your desired learning outcomes and your teaching approaches. If they don’t, then you should look again at the alignment of your goals, teaching and assessment.
• If there are multiple criteria for an assessment and marks awarded for each criterion, then there is a danger that all students will do well on some criteria but not on others and so everyone’s marks regress towards a mean value. This would mean the test no longer discriminates between those who have learned more and those who have learned less. There is, therefore a trade-off to be considered between assessing all the criteria that you would like and being able to satisfactorily distinguish between students. Setting minimum standards for key criteria (like in the example above where correct calculations and clear presentation scored zero if the conceptual understanding was inadequate) can help to address this problem.