In article <[EMAIL PROTECTED]>, Rich Ulrich <[EMAIL PROTECTED]> wrote: >On 6 Feb 2003 06:30:59 -0800, [EMAIL PROTECTED] (Robert J. MacG. >Dawson) wrote:
>> Herman Rubin wrote: >> > In article <[EMAIL PROTECTED]>, >> > EAKIN MARK E <[EMAIL PROTECTED]> wrote: >> > >The students did understand that I said that 79 was a C. Some told me >> > >later that since many faculty round 79 up to a B, they feel that I >> > >should also round it up regardless of what I said on the syllabus. >> > I do not give that type of exams. I would have 70% to be an A. >If they moved the cutoff, I guess Harvard could use the same >tests that they used 35 years ago. The 'scoring inflation' would >be illustrated by the change in the labels. I was not discussing moving the cutoff. In fact, I consider lowering the level of a course while keeping the same title and recognition to be a major crime. However, it seems to be what is done everywhere; an American undergraduate degree is now a joke in general. >However, that's not the reason I'm joining this discussion. >What Robert says next seems so *unthinking* - >> I'm afraid I don't see the rationale for this. An examination has >> little value *except* as a way of setting grades; and setting an exam so >> hard that a student performing at an excellent level will only get 70% would >> seem to reduce the sensitivity of the test as a measuring instrument in >I imagine several extra values, beyond setting grades; >I don't see that 70% passing is "low sensitivity" for the test -- >This is scaling, and weighting. What should count for what? >Over the term, do you want to tend-to-pass the individual >who once aced the work, or do you want to tend-to-fail >the person who blew off one exam? Fortunately for my evaluation of students, I had a student not doing too well show real understanding on the final. It is a common mistake to assume that grades from the typical type of routine examinations mean anything about what the student can do later, which is the only thing which should count. >Other purposes for a test? -- >In my own experience in high school, I almost *never* saw >any test or any homework that I missed any problems on, >except by carelessness. As a result, I was *not* inspired >to read anything extra. To some extent (I think), my folks >had chosen our town and its schools, because of great >reputation. Is that supposed to be a good education? This is part of the current miseducational strategy. If we had really good education, an "A" would usually be an indication that the student was placed in too low or too slow a course. The aim is to learn for the future, not for an easily graded test, and so we should design our tests with that in mind, and also make it expensive for a teacher or other official to delay the academic progress of a student because of age. Most of our courses are very bad, concealing the ideas behind "mastery" of routine. ............... >So you test on next month's curriculum, a few quick questions. >Is the SAT, etc., a test with rotten sensitivity because the >average number correct is set to be near 50% The best information I have seen on prospective graduate students is for the Indian students. Their grading scale is 60% for first class, 50% for second class, and 33% for pass class. Much more than 60% is unusual. >Almost all the tests that I had were formally scored with >"100 points total" with the 10-point intervals for A, B, C, D. > - As a statistician, I it is fine to set a point total if you want; >but I think that transforming percents into letter-grades has >to be *ignorant* behavior, almost to the point of >being outrageous, if it weren't so 'traditional'. >I know that I have heard teachers/ professors say such >anti-educational things as, "I'm going to have to make the >next test easier in order to bring up the averages." >Well, if they weren't wedded to the superstition about >making the percents equivalent to the letters, they would >never be tempted to screw up their tests that way; they >would go ahead and ask the questions that they THOUGHT >they should ask, instead of trying to predict the eventual >grade point score. -- This address is for information only. I do not claim that these views are those of the Statistics Department or of Purdue University. Herman Rubin, Deptartment of Statistics, Purdue University [EMAIL PROTECTED] Phone: (765)494-6054 FAX: (765)494-0558 . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
