In a message dated 98-07-09 Sheila King advised a new AP teacher who asked how hard to make tests:
>Make your tests hard. OK, I should qualify that. Put some >reasonable/medium/easy questions on the test, but also put one or two of >the difficulty level of AP questions. > >I also try to never include a question that doesn't require some amount >of thinking. Anything that can be done automatically, by rote, is not a >very good question.
I agree completely - we must ask the challenging questions! Here's what I do for quizzes and tests (yes, I use alterante accessment techniques too, but I'm not addressing that here): All classes at our school are supposed to use a 90-80-70-65 scoring scale on all quizzes and tests but I can't ask the types of questions we're talking about with this scale. So, I petitioned for, and won, approval to significantly widen the scale for AP Calc. tests - much closer to what the May exam will use. Why must we confine ourselves to the top 10 - 20 points of the percentage scale? A wider scale allows me to challenge the students with much more interesting (i.e. difficult) questions without "sacrificing their grade" and stressing them out. It doesn't change the letter grades - this is not "inflation" as long as the test content is appropriate to the scale. It gives them more room for improvement, the test becomes an additional learning tool (and they're always cumulative from day 1), and it helps students acclimate to a higher level of difficulty. I don't advise making all questions challenging however. My tests have a varitey of material - from easy to tough, objective (adjusted for guessing) and free response. This is NOT to be confused with a "curve" - I do not curve. The scale is announced BEFORE the test. I use the same one all year long - seems less arbitrary this way but to each his own.
In terms of quizzes during each chapter, I usually give two kinds: take-home (mostly hard questions but I do use the 90-80-etc scale here since _everything_ except another person is at their disposal) and the 10-20 minute easy in-class kind (also 90-80-etc) to make sure they have command of the necessary rote mechanics, formulas, etc. There's very little creative thinking or need for justification on the short in-class quizzes. All of that "cool meaty stuff" is saved for the take-homes which students say take them from 1.5 to 3 hours to complete (I always think it should be less). If they have a time conflict that night, they just need to explain this to me and I usually grant more days - I'm more interested in quality here, not quickness). Students tell me this is more effective than studying before a quiz and as much as they groan about them during the year, at the end they always recommend I keep using them. It frees up class time, allows assessment of more than just a sampling of the topics covered, gives an occasional venue for students who have "time anxiety", and is a great learning tool - instead of just leaving a problem blank because they don't see their way through it in time, they keep working at it. Also, since they've put more time into it and written each problem at least twice - once until they're happy with it and then the final "clean" copy to turn in - they seem to have better retention. The major drawback for the teacher however, is that take-homes are extremely time-consuming to correct. (By the way, my tests are never take-homes.)