In Canada, 80% is an A-, in the States, it's a B-. Does that mean....

[pasmith]

This is a poor way of assigning grades. A much better way is for the instructor to write an exam with some idea of what he/she should represent what the average student in the class should know and set that as an average score. After grading the papers, make a histogram of the numerical grades, possibly adjust expectations a little and then look for natural divisions between clusters of scores that can be used to differentiate, say, an A- from a B+, where the numerical score itself is not relevant other than to place the scores into bins. The scores WILL fall into clusters (which may seem surprising, but try it and see).

If you decide that the average score should be 50% and that represents a B or a C or whatever, then simply placing cutoffs between clusters will do the rest. This also allows you to use the entire numerical range from 0-100% meaningfully instead of being restricted to artificially predetermined percentages, grading on a curve or other rather arbitrary ways of assigning grades.

And then you apply a piece-wise linear transformation so that the grade boundaries follow those prescribed by the institution.

 

[stevendaryl]

In my opinion, there is no objective meaning to a cutoff of 70%, 80%, 90% for an A, B or C. Depending on how questions are designed, getting less than 100% might indicate you don’t understand the material. Or maybe getting anything right at all might indicate you have learned the material.

For an example from elementary school, showing that you can add fractions with different denominators. If you understand it, you will get close to 100%. Anything significantly less than 100% means you don’t understand the concepts.