In coming up with grading schemes a lot of people put individual standards (or targets) instead of assignments. This can be a little overwhelming because you can cover dozens of targets and they change every grading period. From a grade book/scheme, I like to think in terms of complexity/difficulty. Lets take a learning target created by a math teacher:
I can calculate the average of a set of numbers and find the missing data in a set of data.
For this particular target she can develop different items that measure a students mastery of that target. For example the three problems below (think back to the bouyancy problems).
- Calculate the mean, median, AND mode of the following data set. 54, 62, 75, 76, 77, 77, 78, 82, 85, 95, 99
- Lucy wants to get a 93 in her math class. Her first four test scores were 96, 92, 94, and 93. What will she need to score on her fifth test to reach her goal?
- There are five numbers in a data set. The lowest number in the set is 54, and the highest number in the set is 82. If 92 is added to the data set and the average is 85, what are the missing data values?
Wether you want to go with PISA levels, Blooms Taxonomy or Depth of Knowledge these are not equivalent items related to complexity/difficulty. She is going to have the same thing happen with other learning targets. So in thinking about her class we asked, "What does a student getting an A in your class tell oyu about that student?" From this conversation we came up with the idea of C -> Conceptual/Computation B -> Basic Application A -> Advanced Application. When constructing an assessment she can make sure that she not only covers targets, but that she addresses the different levels of complexity. Of course you won't cover all targets at all levels, but you do want to make sure you cover enough topics and all the complexity levels.