something about GRE physics

genie05UID: 30358

发信人: beach (天空的颜色), 信区: AdvancedEdu
标  题: something about GRE physics
发信站: 瀚海星云 (2004年09月10日10:30:51 星期五), 站内信件 WWWPOST

發信人: ccchang@Palmarama (Powered by Peugeot 106) 看板: AdvancedEdu
標  題: 一份關於 GRE Subject 的資料 -- Physics
發信站: 台大計中椰林風情站 (Wed Dec 13 18:08:30 2000)

底下是一篇曾經在 ETS 擔任過 GRE 物理學科測驗委員的教授所寫的經驗談。
當中提到了準備 GRE 物理學科測驗的一些大致方向以及對 GRE 測驗該有的一
些觀念。有趣的是，裡面還提到了對於大陸學生用類似考古題的方式準備測驗
而獲得高分這回事提出了一些看法。雖然這些看法不代表所有的教授的意見，
但是至少反映出有老師注意到這回事。雖然談的主要是物理學科測驗，但我想
不管是那個領域這篇文章相當值得一讀。

原文是從 U. of Florida 物理系所轉載，若要做其他用途請告知原作者。

---------------------------------------------------------------------------

From six years of past service on the Physics GRE examination committee,
I can offer the following (slightly dated) information in response to the
recent questions and discussion:

1) Five or six different examinations were in the test bank at any one time,
   used in some random order over 10 or 12 years. Tests were evaluated and
   revised after their first use and at about "mid-life". Recently
   decommissioned tests have been released by ETS. Tests still in the test
bank
   are not available.

2) For statistical purposes (cross-normalization of scores, validation of a
   test, etc.) a new examination must have about 30 of its 100 questions
taken
   from a previous examination.

3) In the years of my service, the median score for the students from
mainland
   China taking the examination was about two standard deviations above the
   median score for US students taking the examination. Some other foreign
   student populations had similarly high scores, others did not. From this
   one can conclude that the scores in the 75th percentile rating and above
   are almost entirely earned by foreign students and that the median score
   for a US student is more nearly about 35th percentile

4) The median number of right answers on the exam hovered a little below 30.

5) The exam proporations are carefully described in the information booklet,
   and it is clear that roughly 60 questions deal with first and second year
   topics (classical mechanics, electricity and magnetism, modern physics) at
   a first- or second- year difficulty level.

6) Exams and exam writing committees vary in their priorities for such things
   as: expecting knowledge of relations and formulas, selecting of wrong
   answers(distractors) based on computational errors, dimensional errors,
   order of magnitude, asymptotic errors, ...
   Ex: some exams have answers which differ numerically by small pure numbers
   (2, pi, ...), and some answers which must be computed (from a variety of
   constants). In the latter cases some exams have answer differences of 2-3
   and others, emphasizing order of magnitude estimates, have answer
   differences of 10 or more. I think one must be prepared for both.

7) Questions are evaluated, and usually revised or discarded, if there is a
   significant difference in the percentage of right answers for men and
women.
   Questions are evaluated, and usually revised or discarded, if the
   performance on the question differs from the performance on the test as a
   whole (for example, if the students with high total scores found it more
   difficult than the student with low total scores).

As to insights about this information, it would, of course, be useful to know
how much of this is still an accurate reflection of the situation.

a) One piece of advice I draw from this information, is that haste is not
quite
   as important as 100 questions and 180 minutes makes it seem (1.8 minutes
per
   problem). If your goal is to hit an above-median score for the domestic
   population, get 30 right answers and you are doing really well! (that's 5
   minutes per problem!). One important strategy then is to scan through the
   exam quickly and first do those problems that you find easy. Don't let a
   systematic (sequential) strategy compell you to leave unexamined problems
   (some of which could be easy) at the end of the test.

b) One can ask whether the criteria for revision are valid. There is a wise
   (I think) effort to avoid language or topics which would give an unfair
   advantage to certain cultural, geographic, or gender groups. What degree
of
   homogeneity of the population justifies the assumption that each question
   should differentiate students in the same way that the total score
   differentiates them? If this is valid, it serves as a distinct reminder of
   the many other crucial credentials of graduate school bound students which
   are not measured by the GREs (aptitude for advanced work, persistence, >
   thoroughness, experimental talent and intuition, research > procedures,
   independence, creativity, ...). Test takers as well as graduate schools
   should accept the admonition that GRE scores do not measure the whole
   person, even by the standards of preparation for graduate school!
   Statistical evidence suggests that GRE scores give only a weak prediction
   of success in first year graduate courses, and that there is no clear
   predictive success for work beyond the first year in graduate school.

c) From more than a dozen independent sources (Chinese physics graduate
   students in the US and in China and Chinese physics faculty members) I
have
   heard that in most Chinese physics departments there are books in which
   students who have taken the tests report their recollections of the
   questions. Accumulated over time, these books have about an 80-85鴋ance
   of having all the questions on the next exam to be given. There is only
   about a 15-20鴋ance that a student who prepares with these books will
   encounter a new exam, but since that exam has 30 questions from prior
exams,
   this will lead only to a minor embarassment. As a consequence, most
   Chinese students are preparing from a large reservoir of GRE-style
questions
   and most are preparing from the actual examinations themselves. One would
   expect that good students would be very good, after this preparation,
   at both multiple choice exams and at getting good GRE scores. How this
   relates to their relative preparation for graduate work in physics cannot
   be inferred from the score differentials

-------------------------
Powered by Peugeot 106
--
☆ [Origin:椰林風情] [From: 140.115.5.20] [Login: **] [Post: 35]


--
我的梦想：
1流浪。
2开家公司，惨淡经营。
3买个大农场，一家人安安静静地生活。
※ 来源:·瀚海星云 bbs.ustc.edu.cn

apolloxpUID: 156355

For statistical purposes (cross-normalization of scores, validation of a
test, etc.) a new examination must have about 30 of its 100 questions
taken
from a previous examination.
每次考试有30道题目上次考过？
复现率那么高:rolleyes: