Friday, July 21, 2006

Topic for Discussion: "Pacing" of a class

In comments to a previous post, Patrick Dugan mentions a "progressive curve of increasing [grades] as the semester progresses".

This reminded me of something I've been ruminating on for awhile. If you have a class with regular assignments throughout, what's the best way to weight the assignments with respect to time?

Most of the classes I've taken as a student follow a roughly constant curve; each homework is about as heavily weighted as any other, with occasional (seemingly random) spikes for large assignments, plus of course a heavier weighting on the mid-term and final.

On the other hand, most games I've played (especially boardgames) follow an increasing curve; early victories or setbacks are small, later ones are more important. This is great for games, because it keeps players interested to the very end. Also, players will know that even if they start out poorly, they can come from behind to win if they do particularly well in the endgame, so no one loses hope. Would this progression be worthwhile in a (game-like) classroom?

  • Students given incentive to try harder in the last half of the course rather than drop.
  • Early assignments are safe experiences, where students can learn my grading style without huge risk.


  • If every class did this, students would have too little work to do early in the semester, and be overloaded as finals approach.
  • Puts too much emphasis on the last half of the course, so may only be suitable for classes that build on earlier topics, rather than "survey" classes that cover a variety of unrelated material.

So, what's your opinion? Constant weighting, or progressive increasing? Or is there some other method that might work even better that I haven't considered?


Duncan said...

I see your biggest issue being one of "User Model" expectations. Students will be expecting the default system of weighting. You will be providing a different Grading Model. I see this causing frustration and confusion, especially as the class gets harder.

In the game landscape the easier sections are easier because the players are learning a totally new system. They are learning the controls. They are getting the hang of the strategy. They are simply weak characters, or don't have access to the good units. Students, on the other hand, know what they have to do for a good mark (at least by second semester). They know how to jump from assignment to assignment. They know how to navigate the common class because they all work on the same Grading Model and other common factors. They don't need an extra learning curve.

I also think that if you don't challenge them from the start then you'll lose some of them. They will get complacent, or think that the course has nothing to offer - nothing to challenge them with. Not to mention that every class already gets busy near end of semester (whether planned or not), shifting yours to be even more so would be harder on students.

Adam Yulish said...

Some classes use cumulative learning, where each assignment and project builds on a previous assignment. Other classes teach certain skills, emphasizing practice and refinement (and therefore emphasizing the last and presumably best effort). In my experience, graduate classes tend to follow the latter, while undergraduate classes follow the former. Many classes mix both types.

Make the grading system and pacing of work conform to the class type. If your "players" need to build up points throughout the class, they need to know early on what they're in for, and should have the opportunity to fail or succeed early on. Students should know by the halfway point what their chances of success are. If instead your "players" learn and refine skills, then the big projects and grades can wait until later in the class. For skill refinement classes, grading should emphasize the endgame.

By the way, I think this question would make a great game design class assignment. Tell the class to design a grading system that is fair, encourages students to learn, and simplifies grading for the professor, for the following classes. 1) A business class with a couple of student team projects, 2) A music class where you learn to play an instrument, 3) A computer programming class where you will learn certain basic programming concepts. This could lead to a great discussion about how to encourage and discourage certain activities by design. What's a good way to keep a team of students accountable to each other for group projects? If some instruments are more difficult than others, how do you balance that according to grading? How does that affect class enrollment at a university level? Is there a way to create computer programming assignments that make it impossible for students to simply give each other copies of their code to turn in, but not turn into a grading nightmare for the professor? Grading design could lead to an eye-opening discussion about macro- and micro-level design issues, and flesh out differences between design and implementation.

Patrick Dugan said...

Being a Game Design class its accetpable for the first assignment to be either designing the grading system, or (more likely) learning the game-like grade system you've designed. The discourse of the class, learning game design theory and applying it to board and card designs, or maybe digital prototypes, should reflect the form of the class.

Here's my design - students have four attributes, instead of the usual one:

- their standing grade (hearts)

- their cap grade (heart capacity)

- their failure resilience (defense power)

- their actual grade (standing grade / heart capacity)

You allow students to daily lose and gain hearts through class participation and performance (on quizzes or something more creative and to the point), the dominant strategy, if there is one, will involve being proactive to the point of getting a decent dose of hearts daily, thereby covering their ass. The incentive is that enough proacitivity, despite mistakes, should earn you an A, teaching the most imporant lessons in game development rather than game design in particular. Then theres a dualistic strategy in earning, at rarer intervals throughout, either armor upgrades (so that standing grade penalties, which increase over the course, are lessened) or heart containers (raising the grade cap); based on whether the projects display conservative refinements over iterations, or wild spread into creative zones, such as designing card game or RPG about social dynamics, which for me is pretty new wave. Coming up with an objective criteria for that is tough, but there you go.

You'd probably want to write a spread sheet program for this.

Evan said...

I find that the topic of the class largely defines the structure. Language classes, for example, are always cumulative, whereas in many math classes I've had we learn a formula or whatever at the beginning of the class then don't touch it again until the final. Seems to me like this should directly affect the sort of grading system used. A math class, where you learn a series of information which may or may not be particularly related or interreferential, should probably be graded on the same scale throughout the class. With language, however, it could be interesting to scale it to be harder at the beginning and easier at the end, since one always has to remember every word and grammatical form taught during the class by the end.

A political science professor of mine from last semester here at Zoomass did something with his midterm and final I had never seen before: the class average on the midterm was 68, so he declared that if one does very well on the final, I think even just by comparison to their grade on the midterm, he would scale the final's grade so that it counts for more. It's a good way of giving pre-midterm slackers incentive to catch up (which I did).

Lewis said...

Hi Ian,

I weight grading toward the end of any class, so as not to hurt unduly folks who know nothing of the topic to begin with, or who have not learned to cope with how I do things.

And maybe it provides hope to those who don't do so well to begin with.

Lew Pulsipher

Gilmoy said...

1. At Washington State U., my Human-Computer Interfaces (HCI) class started out with minimal loading, and became almost a death march at the end. This was compounded by giving students too many "any 1 of N" assignment choices up-front. Human nature being what it was, most students maximally deferred all of these assignments, so they all came due right around the time the main-line death march was starting.

2. My current advisor (for whom I am the TA) has an intriguing policy: 90%+ on the final is an automatic A for the class, regardless of anything previous. We've had at least 1 student for whom 89 meant F.

The course otherwise follows your "constant curve" model. I think this policy entices some marginal students to not drop right after midterm shock.

Ian Schreiber said...

Interesting, Gilmoy! Thinking in terms of game mechanics, I assume that the rationale behind "90+ on final = A in the course" is that if you can demonstrate proficiency at the end, then struggling before doesn't really matter. By that logic, you could consider adjusting the policy: your final grade is the higher of (your cumulative grade in the course) OR (your grade on the final). 89 on final = 89 in class.