But now that I've read and thought about the issue in more detail I can't continue to employ TGPs. Thankfully, Feldman has some suggestions :-) He groups his first suggested practices into those involved in making grading easy to understand, mathematically sound, and accurately indicate a student's academic performance. One practice is to avoid the zero. It's rarely the case when a student knows absolutely nothing about a topic so the zero is not accurate in that way. Also, when coupled with a 0-100 point scale, the zero disproportionately punishes students. Zeroes for a few late or missing assignments can significantly affect a students final grade under TGPs, but they do not indicate final understandings and so are inaccurate in that way as well.
The 0-100% grading scale itself is problematic because it allows for many more ways to fail than to succeed. Often under TGPs, more than half of the scale counts as failing with an F letter grade. It's also too granular. I certainly can't tell a student or parent what the difference is between an 85% and an 86% in terms of a student's understanding of a topic. There's no valid academic reason to have that degree of gradation unless you're simply looking to rank students against each other.
So Feldman recommends three things: 1) Avoid zeroes, 2) Change the scale to something like 0-4 (five levels instead of 100), and 3) minimum grading. I'll have more to say about the 0-4 scale later, but if the system in which you are working requires the use of a 0-100, then at the very least you can set a minimum failing grade a student can earn on that scale. Depending on the particulars of your school or district grading scale, that minimum might be between 50-69%. An example: let's say a student has five assessments on the same topic and gets an 85% (B) on four of them, but is missing the fifth. Averaging a zero in for the missing one gives the student a 68% (D). Averaging in a 59% for the missing assignment gives the student an 80% (low B)--which probably better reflects what the student understands about that topic.
Speaking of averaging...why do we do that? Let's say on a particular topic a student earns these five scores: 82, 84, 40, 94, 94. The average is 78.8, the middle score is 84, and the score that is most frequent is 94. In such a scenario, what score more accurately reflects what the student now understands? It seems like the student was understanding at a proficient level, then had a really bad day, and ended up performing at a really high level. A 78.8 average for this student doesn't accurately reflect what they now know.
Feldman writes "Averaging a student's performance over time doesn't accurately reflect what the student came to understand over time and lowers grades for students who took longer to learn and demonstrate proficiency. Instead, a student's most recent performance can more accurately describe their achievement." Although Feldman doesn't specifically call out ways to calculate a student's grade and give more weight to the more recent ones, there are many in the Standard Based Grading (SBG) community that have. Calculation methods like a Decaying Average put more weight on more recent scores and of course teachers can just use their professional judgement as well. (This of course makes more sense in situations in which the student is being assessed multiple times on the same topic or learning standard.) I'll also note that widely used computerized grading systems such as those in PowerSchool and Infinite Campus have the capability of calculating grades with methods other than a mean.
Lastly in this section, Feldman points out that group grades reward students for "...working together to create a product regardless of whether each student learned from that group's work. Awarding grades based on student learning subsequent to the group work refocuses students on its purpose and provide us with opportunities for formative feedback." I've often graded lab work as group work even though I've required lab papers from every student. Feldman's comments here (and later) have made me rethink how I will assess lab work going forward.
So, for more accurate grading:
- Avoid zeroes
- Change the scale (0-4)
- Apply Minimum grading
- Weight more recent performance
- Grade individual performance, not group performance
Next post: Bias-Resistant Grading Practices