Re: Bimodal grades

[Leigh Palmer <palmer@SFU.CA>]

> My experience with bimodal distributions in intro physics is that it
> represents the overlap of two distinct levels of preparation for the
> course. Simplistically, one group could have tested out of the course, and
> the other is in bad need of remedial help.
>
> When my colleagues talk about "teaching to the top 10% of their class", I
> remind them that they are "constructing a winning team out of all American
> level players", which is *not* to be confused with "teaching". Karl

If a class is sufficiently large that a bimodal distribution can
be meaningfully inferred then it is too large to teach. I had a
mildly bimodal distribution* on my first exam this semester and
I do believe (from talking to the students) that the better ones
*learned* more than the lower group. The exam was done entirely
without numerical data (or calculators). I tell them that they
will not be able to use high school methods to solve problems in
this university course. Some protest having to learn anything
new, but many learn!

Leigh

*Average grade = 50.2 %

Range of    Number
 scores    in range

  6-10         2
 11-15         1
 16-20        13
 21-25        10
 26-30        32
 31-35        21
 36-40        32
 41-45        22
 46-50        27
 51-55        23
 56-60        15
 61-65        21
 66-70        21
 71-75        25
 76-80        17
 81-85        11
 86-90         7