Mark wrote...
Is there a mechanized way to take a series of grades and bump them u or
down to fit a forced distribution, eg. no more than 10% get A's, 30
B's etc...
I've never understood the logic behind grading on a curve. If the tes
is fair, then if, say, 1/4 of students taking it score > 90%, shouldn'
all of them get As? Likewise, if more than 1/2 score below 50%
shouldn't those fail?
Now if there were some methaphysical fairness involved, to wit, th
teacher would be fired if s/he needed to skew the scores consistentl
up or down, then curves would be OK.
As a pure thought experiment, if the top 10% of scores should get As
then with the named range Scores containing the scores and x one o
those scores, the grade for x would be given by
=LOOKUP(PERCENTRANK(Scores,x),{0;0.3;0.6;0.9},{"D";"C";"B";"A"})
This won't give precise cutoffs if there are any ties for the scores a
the 90th, 60th, etc. percentiles. Scores at such thresholds would ge
the lower grade. If you want them to receive the higher grade, use th
formula
=LOOKUP(COUNTIF(Scores,"<="&x)/COUNT(Scores),{0;0.3;0.6;0.9},
{"D";"C";"B";"A"}