Posts
In my previous post, I listed down some of the demerits of the CGPA system. While most agree to it that the CGPA system has its drawbacks, they feel that this is the best system available.
Here are some of their arguments
- You cannot distinguish between someone who has secured 91 and someone who has secured 93. So you give them the same grade
- Suppose you miss a grade in a 4 credit course, your CGPA drops by (1x4)/200 = 0.02 only.
- The likelihood of missing a grade in a subject is 5/100 ( You are among the 5 unfortunate people who missed the higher grade in a class size size of 100), i.e. 1 in 20.
- Rash commented: The Indian Education System is so fake that it is best to ignore it. The marks awarded are so shallow that you don't feel pushing for changes is really worth all the trouble.
- If you cannot distinguish between 91 and 93, why segregate 89 and 91?
- You get 0.02 when you average the grade-loss over 8 semesters. Suppose, someone misses a grade in the first semester, the grade loss would be 0.02x8=0.16 in the first semester which is more than sufficient for someone to miss his department change.
- Your probability holds only when the event is unbiased.
- @Rash: All I can say is - If it isn't worth all the trouble, why do we have examinations at all? Couldn't we just have projects/actual work?
However, as Rash suggested, I am presenting my suggestions. You are free to brainstorm/critically analyze the suggestions.
Centralise
Centralise the grading system. Give two people getting the same marks the same grade. This way, students in 'high performing sections' will not lose out. The grade will be purely based on merit.
Uniformity
There has to be a parameter which should decide the cutoff marks for the subject. Do not leave it to the Professor. Further, apply the criterion uniformly to all subjects.
Feedback
The students should be given a proper feedback on their performance. And the students should know why/where they lost their marks. I am sure IITs can provide feedback sheets having critical data such as class average, class highest and the weak areas of the students - say analytical abilities, calculation errors or weak concepts.
Time Bonus
A student who completes his exam well before time definitely has clearer concepts than someone who takes the complete 3 hours to complete the exam.
One could award 10% of the marks secured by the student as bonus per hour. So, if I get 90 and I finish half an hour early, my final marks should be 90 + (10% of 90)/2, i.e. 94.5
This way, plagiarism would reduce. A student finishing half an hour early would realise that returning the answer scripts is a better option than trading answers in the surplus time.
Cluster Grading
It doesn't make sense in having multiples of 5 and 10 as cut-offs.
Give grades in clusters. Say, there are 9 to 10 students in and around 90. Give the same grade to all of them (the cluster). Further, the cluster size may not be greater than 3 to 4 marks. If there are 10 clusters, give 10 grades on a scale of 5 to 10 - 5.0, 5.5 and so on till 10.
11 comments
First, the exam system exists because it's the easiest way to roughly evaluate a person. I would prefer a Personal Interview over any other method, but I do accept that it's not feasible to conduct PI for so many ppl (And maintain the same Environment/Mood/Question Bank etc.)
It's a good thing to make certain guidelines on how to give marks, but sadly not every professor follows (wants to follow?) it. At the end of the day, unless the same prof evaluates everything, there will be a lot of bias/non uniformity.
Feedback is something we should encourage more. And if possible, it should be both ways (Though I don't know how effective it'll be).
And I strongly disagree on the Time Bonus point. Completing your paper in 2 hours, when you're given 3 hours, doesn't prove that you're smart. It proves that you're too smart and over confident (And stupid, if I may add). If you're given a time frame, you must learn to utilize the whole time frame in the most efficient way. Done your paper in 2 hours? Revise. Elaborate more in some points. Highlight important keywords. Heard about Usability & UX in Web Designing? It holds true in real world too! :P
Dunno about Cluster Grading/Centralising. Here grades are allotted using a Histogram (Marks vs No. of students) and they try and divide it into distinct groups (whenever possible). Or atleast that's what they tell us! :D
Anyway it's good to see someone so bothered about cgpa. Frankly, I just got the result of one of my subjects, and I scored much more than those who studied for the paper. All I concluded was that just you're grades DO NOT PROVE ANYTHING. But we have to use it till humans can invent some device to measure aptitude, intelligence, talent, creativity etc. :P
I believe usability/highlighting keywords are all taken into account while evaluating. All I am saying is - here is a candidate who has attempted 90 marks. The rest of the 10 marks is out of his reach/he hasn't studied the topic. So, instead of sitting in the exam hall trading answers and indulging in plagiarism, let the fellow get out of the examination hall.
There has to be a mechanism that reduces plagiarism in the examinations - and invigilators are proving no good.
I am sure no candidate would be willing to sacrifice his revision time for the time bonus. The time bonus awards would award a very small %, say 10% on the marks obtained (not on the marks attempted!)
Any other mechanism to reduce plagiarism would surely be very welcome
It is sorry state of the perspective that the indian society has built about "how a successful student
be described?" n wat the unanimous answer to it is "Marks". The objective of education is to gain
mastery in performing a job in future. Wen one becomes an engineer hes not a 90% or a 70%
engineer. The successful application of wat one reads is SUCCESS. However as it can be readily
seen it is wrong to compare such a success as people have their own ways to apply things. Marks is
just an incentive for the students to evaluate themselves... but research shows it makes no
difference between a stud who gets 91 or 93. Usually most of the successful people in the world
dont have great grades. The best placement that one gets depends to what extent one can apply
them in real world n not how mch that stud gets in the exam. I really feel bad about the way
students nowadays value marks... value rote knowledge and call it precision at the end of the day.
Maybe the answer lies in the Darwin's concept of "survival of the fittest" where the different
schools and colleges compare students based on marks rather than a full profile. In a nutshell..." It
is waste of time to think about a metrics to evaluate marks which is actually unpragmatic ".
-Shantanu Biswas
MBA-IB @ SIIB,Pune
@shantanu: If you think that way, it is unpragmatic to bring about any changes in any system whatsoever. The successful student will always find a way to surmount barriers.
I think you do not very well understand the system in IITs & in many other educational institutions. Marks are not just an incentive - they are the gateway to many kinds of incentives - scholarships, jobs & department changes.
When we think of marks, let us not limit ourselves in our thinking - Marks are not just a parameter to evaluate the student, they are a parameter that are taken into account in many other areas.
Do you still think bringing about changes in the cgpa system is a complete waste of time?
If we are living in a well... it makes sense to devise methods that would take us out the well... rather than device changes in there so that we can adjust to the well eco-system.
The present system is at fault, because its not productive. It fails to reciprocate the message that students should focus on its practicality and how wat they study can be implemented. Thousands of engineers in India don't call them enggs wen they sit for an interview in future, primarily because they all ran after marks and rote mantra... which might give you good marks bt wanes of in course of time.
I have been in a CGPA system for a total of 5 yrs now(4 in engg + 1 in mba). Evaluating a person's capabilities is no joke. HR executives n researchers keep on devising new methods to recruit people. In India, where education is in bulks, ... resort to a simple number turns out to be the choice of lackadaisal authorities. I support the interview system which might be backed by a technical objective/subjective written test(if necessary).
This is a vast topic and as far as branch changes are considered... thats a topic in itself... why ask very high marks 4m one candidate of branch A to shift to branch B wen most of the studs of branch B hav secured less than dat? Statistically most of the engineers in India dont pursue their career in the branches they opt for. Career counselling is pathetic and product like N-CAT by reliance is a good initiative.For Branch changes, an above average perfromance, with aptitude in requested branch and a strong statement of purpose should do.
I dont think I would value being a scholar if I have high marks on my CV but am low on applicability of my skill. But if u want some
discounts in ur study expenditure, I dont think changing the system makes any sense in that.
You will get my point of view when u think from a completely independent angle. What runs the world are businesses. If an education doesnt help a student to contribute to the business its of no use.
Hope you have understood wat I want to say!!
@Shantanu Biswas:
I get your point.
The present system is at fault - no doubt about that. A mere number is no indicator of one's talent - no doubt about that. CGPA system cannot accurately somebody's potential - agreed. (In fact, no system can - not even interviews). One should focus on the overall system than focussing on the CGPA system. Right?
But what I wanted to point out are some glaring discrepancies (you can read my post on inconsistencies in the cgpa system) & suggestions to rectify them. The discrepancies are akin to - 2+2=5. Mathematical faults. A student with the same mark gets different grades! (Think of how you feel when the shopkeeper refuses to give you back the change. You feel cheated. Right?)
CGPA system was supposed to be better than the absolute system! But it isn't. (Atleast the absolute system was mathematically consistent).
PS: I do not agree that someone in Naval Architecture would end up doing Mining later on in life. That probably merits another post. :)
Lets get messi... :P
The CGPA system that is followed by KIIT University(Bhubaneswar) is quite similar to the one followed in IITKgp with a few differences.
Centralised grading system is strictly followed. All the students who wrote the same paper are pooled together and are then graded. The CGPA there for scoring subjects being absolute(spread of 10 points per grade) but less scoring ones have relative grading but the spread of each grade is less than 10marks/points. So the difference between two performances are scaled such that the student who performs the best is consiered as Grade O - Outstanding(like EX in iitkgp). This makes up for the loss of percentage that a "low scoring subject" topper would have in case of absolute percentage grading. Consistency in a batch is a MUST if the question paper was the same for all
the students for a certain paper/subject.
The award of numerical marks which suffer from a number of imperfections like inter and intra-examiner variability, imperfection of tools,
sampling of contents, sampling of objectives
within the content, arbitrary time limits for answering individual questions in testing situations, etc. The CGPA turns out to be a better system than that.
The distinction among individuals on the basis of single-number right score warrants absolute
precision of judgement on the part of teachers/evaluaters, which is not humanly possible. The answer lies in grading which places students of similar potential in one grade. Two similar graded students can score more than each other in subsequent retakes of the same paper.
Need to distinguish between 89 and 91 bcoz a normal distribution ought to be distributed in spreads satisfying the degree of confidence
of the person interpreting the marks.(this would make sense to u if hav knowledge of normal probability distribution).
Consistent display of the highest grade by an individual over a period of time in a particular area of study may be used as a criterion for the identification of talent. The student getting all 71s,81s,91s or 60s,70s,80s cannot be consistent in getting that. So such candidates can be detected from a series of past perfromances in their CV.
CGPA awarded through relative grading method provides for not only inter-subject comparability but also comparability over the
years.(assuming each batch is equally talented). It also does not get affected by the difficulty of test. This method invariably covers the entire range of distribution regardless of the nature of subject - a far precise index of students’ performance than what is obtained through marks. It also eases avoidable pressure on students and parents and literally save them from passing through traumatic experiences.
The grades denote the relative performance of students within a specified group, provided they are awarded using relative grading method.Thats real world man... as an engineer you will compete in your own team and company rather than acroos the industry. Even in
IT companies grading has more tough rules to avoid any bias. Only the top 20% in a team get A grades - nxt 20% B grade - and so on.
Based on which the employees are remunerated and promoted.
The criterion used for deciding the number of cases against each letter grade is determined on the basis of the spread of scores divided
into nine equal parts. These eventually correspond to 4%,7%, 12%, 17%, 20%, 17%, 12%, 7% and 4% in a normal distribution(the %
model can vary 4m inst 2 inst), which is expected in an examination involving a large number of students.This is pure stats... :P
Introduction of grades dampens the zeal of bright students is absolutely a myth...under this method the award of grade is determined by
the relative performance of the group. Even if 90% of questions have been confidently attempted... the stud can't be guarateed an EX
grade... so need to push further. At the same time the evaluation should also limit wat 2 answer and the answers should be to the point.
Some faculties in Symbi do penalise studs for answers which have the whole chapter in them rather than just the answer. CBSE Board
also promotes this implicitly.
As far as uniformity is considered... the cgpa system in Symbiosis calculates the cutoff as the 70% of the average performance of the
batch. So if average is 80 marks then cut-off is 56marks. The spreads of the grades are distributed normally betw cutoff and highest
marks scored.
Quantitative feedback is nt possible as the student can calculate the respective percentage marks and add to meaningless disputes...
and kill the very benefits 4 which they were introduced. Qualiative feedback is possible within the same subject as ex, very G, G, Ok etc. to be given to diferent sections of the same paper. (Would be great to have them... dont think they are present newhr in the world..)
Time bonus cannot be evaluated as the end result matters in the real world. Nadal cannot be given more prize money for defeating
Federer in straight sets than for a five set victory. The level/degree of precision in distinction that you might be searching for from the cgpa system cannot be satisfied for the same reasons metnioned before which makes it better from the marks system. Such distinction should be achieved/proved by doing something outside the curriculum like workin on a paper, prizes from reknown olympiads r contests r n indigenious invention. And belive me it would be far more influential on the CV than a time-valued cgpa which gives ya slight edge over others.
Plagiarism is different topic altogether... hav seen studs help out others even wen they might risk their own job.
Cluster grading gets cared of when being followed by normal distibution based calculation of grade spreads.
There's another problem though... a more statistically correct method could be more justified but similar efforts need also be made in order to make normal public(the analyser) understand the way to analyse them... the way they should be.
As of branch selection is considered... chk these stats out 4m newhr... as to how many enggs go for MBA... the branch distibution of IT employees of different IT companies... the available job pool in different branches and the no of grads passing out from those branches
in India. This wud giv ya an approx picture of wat i said earlier.
Thinking of opening a blog of my own nxt week!!! c ya thr!!!
Finally, you agreed to look into my points rather than arguing whether changes are essential or not.
The trouble with the mechanism of CGPA gradings you mentioned is - they have fixed number of grades. So even if I have 7 clusters (which I would expect if there are around 7 questions in the question paper), the professor/software has to allot 5 different grades (not 7!). That is what you are missing.
I agree that I should have dealt with plagiarism separately. However, I do not think it would be fair to compare it with tennis that way. Think of the challenging system that was introduced because the chair umpires weren't as good. Similarly, we need a mechanism that controls plagiarism. Time bonus is an idea that I mooted. You are free to come up with yours.
I do not think that if there is an inferior mechanism followed outside, IITKgp needs to ape that. Kgp needs to set an example.
And lastly, by giving a good grade to someone who has scored poorly in English is something I will never agree in principle. The one who is good at it suffers because of this.
Good luck with your blog! Would be looking forward to more ideas! :)
The fact that there are fixed number of clusters depends on the level of confidence that the interpreter wants from that marks. If he is very specific so the number of grades need to go up and vice-versa. To standardise this throughout the education system they hav e arrived at that based on some raw data and analysis. But the number of ques in a paper has nothin to do with the number of grades.
Metrics differ with what we need to measure. As needs are different for different poeple so might be the metrics... A metrics can only be compared when the needs of both the systems match or are similar.
What we hav in symbi is a bit different though... (even such a mahabharat I wrote earlier missed this out :P)... The batch average gets calculated and then certain multiples get grades e.g
1.7 times average > A+
1.5 times average > A and so on.
I dont know the stat behind this so am not able to comment on this. In this system, if all the studs get close to average then no one gets an A grade. While the passing marks is 70% of average. This is to foster competition and is less dependent on capabilities. Statistically, if a stud who got good marks but somehow has < 70% of average then he wud fail. The way I see it, hr the need of the metrics is not similar to the one followed in IIT. This is because we have many papers where thr r case studies and there is no right answer or wrong answer but a better or a poor answer. Even in stat papers a large weightage is given to its interpretation than its calculation.
The english eg: The spread of each grade is less comapred to a math paper bcoz the studs score less in it. So a diff of 5 marks in eng is more than a diff of 5 marks in maths. Mathematically, 100(highest),90,80,70,60,50.... in maths can be equivalent to... 70(highest),63,56,49... whr the spreads r 10 n 7 respectively. So thrs nothing wrong if a stud A gets 90 in maths(B grade), 70 in eng(A grade) while stud B gets 100 in maths(A) and 63 in eng(B).Both studs r even this way. The only way the eng topper wud lose if cgpa grades were given based on percentile system and when the stud A gets "A" in Eng for 70marks while the Stud B gets "B" grade in eng for 55marks.
True, very true. The metrics need to differ with what we need to measure.
That is why we need flexibility in the system. Flexibility in the number of grades that can be given.
Consider the example you took - what if the highest in English was 77? (The guy is really good at English). Do you now give the 63 guy a B grade or a C grade? Trouble. Isn't it?
The trouble here at IIT is - half the people don't know how the grades are given. In fact, the system isn't clearly known.
I just hope some transparency comes into the system. It really seems as if the grades are given on the whims of the Professor. And I'll continue to feel that way till they tell us more about how they give the grades. (Nobody knows whether it is based on statistical data or not)
In between, really insightful discussion! Learned a lot about grading criterion at different places. Thanks a lot!
Post a Comment