A recent comment (and some answers) was about a test of being
"uneven". As a statement, this is OK, but I really don't know if this
is aasumed to be bad or good. Let's develop a little "theory of tests".
Take, for example, a course in calculus (with which I have more
experience). Among possible test questions there are many so easy
that a student not knowing them should really flunk, while there are
also questions which you expect only an "A" student to know how to
solve. You want to be "fair", so you ask say 20 "C-questions". Hence
every "C-student" should get them all correct. How do you distinguish
the "A-students". Whom do you give a "D" and whom do you flunk?
Let's go to extremes: give a calculus test asking to differentiate
100 polynomials. Very few students will get them all correct. Those are the
"A"s. Most students will do irrelevant mistakes, and let's decide
that 80 correct is a "C". Who didn't get 60 correct flunks. This would
be a very "even" test. And if you want to be "tough" and lower the
grades, just dim the lights in the test room and put the thermostat
at 90 degrees F.
Regards Emilio