[loughnane]

I agree. Finding a balance is tough though.

On one end of the spectrum you grade work throughout the semester, which has the failings you mention above.

On the other end a “final” of some sort is worth 100% of the grade, which feels too high stakes to be a reliable representation of what students truly understand.

I hope to find an explicit recommendation for a good balance.

[professoretc]

The solution I use is to allow later grades to replace earlier grades. This works best if you can assign each graded item to a specific "topic", so that you can say "your grade on each topic is the highest grade you receive throughout the quarter" or whatever. Then at the end, you calculate their final course grade from some combination of the topic grades. (The easiest way is just to set a pass/fail threshold for each topic and then base their final grade on the number of topics passed.) You can adjust the ratio of replacement to averaging so that you're not totally throwing out earlier results, if you want to do that.

[loughnane]

That’s a neat approach. I like it.

[professoretc]

I got the idea for this system from Nilson and Stanny's Specifications Grading:

https://styluspub.presswarehouse.com/browse/book/97816203624...

[cosmojg]

I can imagine a system where students are graded relative to the median performance on each exam and assignment, and if what the article is saying is true, we can expect this median to rise on its own as the semester progresses. Naturally, this would place more weight on later performance and less weight on earlier performance.

Alternatively, one could keep their existing grading system and simply use some sort of quadratic weighting scheme which continuously increases the effect of an exam or assignment on a student's final grade as the end of the semester approaches. This would be a lot easier to write down on a syllabus but might disincentivize effort spent earlier in the class.