Do you think assignments graded relative to other students is fair?

[Leros]

So, bell curving? That is almost as old as universities.

Its not bell curving. Its linear interpolation.

 

[sdifox]

Its not bell curving. Its linear interpolation.

:shrug:, we called it bell curving, no matter what the final curve looks like.

 

[alkemyst]

Uhhh, no. In the context of your VLSI class, learning how to do things optimally is a very important lesson so I think there SHOULD be a metric based on the area consumed in the design. What I hate about the current grading structure is that the student has no idea when to stop obsessing over the design.

If they know that getting it in X um^2 will get them 30/30 and it'll scale down based on how much larger you are from it, then it gives the student a metric that they can work with and know when to stop. That's why I say use last year's data as a metric because then they have a number to work with. If they use this year's number, they won't know when to stop because that number won't be known until after everyone turns in their design.

IMHO the best design should always be sought, whether overkill or not. This is how things get smaller...not from doing cookie cutter or what the last guy did.

In my CPU class, everyone knew the code how to make a program to duplicate the CPU and registers in C++. I went to the teacher and asked if I could do it in VBSCript, one as a challenge and two because the program was already floating around.

 

[TuxDave]

IMHO the best design should always be sought, whether overkill or not. This is how things get smaller...not from doing cookie cutter or what the last guy did.

In my CPU class, everyone knew the code how to make a program to duplicate the CPU and registers in C++. I went to the teacher and asked if I could do it in VBSCript, one as a challenge and two because the program was already floating around.

"Best" design is a vague target and when the intent is to make people work hard, you don't need a floating target. So you're asking the student to continuously redesign until the turn-in date. That is insane.

In the industry, you'd be surprised how far the designs are from the "best' design known. Complexity and risk (and sometimes legal issues) keep engineers from doing the best thing conceivable. And that is assuming there is a single metric for "best" since you can trade off area with speed.

 

[JTsyo]

Here in Ontario anyway, college/university teachers aren't supposed to bell curve. I think some of them secretly do though because bad marks across the board make them look bad.

I agree that bell curving is absolutely not fair. You don't get bell curved in the real world. If you get a 90%, you should get that 90. If you get a 0, you probably deserved it. Other people's performance shouldn't reflect on you, unless of course your the leader on a group assignment or something.

Trouble is that counts on a completely fair test. There needs to be something to adjust for the testing. Of course you can have a class of equals where they don't fall on a bell curve, I would expect the professor to take that into account too.

 

[CountZero]

I agree with what you are all saying. Suggest a better method for grading.

This is an integrated circuit design class. The 30% about resources is concerning how much area we use in our design.

Suggestions?

This system makes perfect sense. When the project involves designing something where correctness is only part of the design making the performance part competitive will drive students to do their best possible work. In the real world if your IC design is 100x bigger than the average (and you are targetting area) you have essentially failed if anything 70% for just correctness is pretty friendly, at a minimum you should all be expected to get that. The sticky thing is what you do when someone has a design that isn't fully functional. It also works better when there are a few metrics which offer trade offs.

 

[TuxDave]

This system makes perfect sense. When the project involves designing something where correctness is only part of the design making the performance part competitive will drive students to do their best possible work. In the real world if your IC design is 100x bigger than the average (and you are targetting area) you have essentially failed if anything 70% for just correctness is pretty friendly, at a minimum you should all be expected to get that. The sticky thing is what you do when someone has a design that isn't fully functional. It also works better when there are a few metrics which offer trade offs.

And if we're talking on the scale that everyone got their design within 1% of each other, the poor guy who forgot to optimize that last line of code ends up with a C-? Tough luck?

 

[MrPickins]

And if we're talking on the scale that everyone got their design within 1% of each other, the poor guy who forgot to optimize that last line of code ends up with a C-? Tough luck?

Yeah, that's what seems really unfair to me, the grading scheme doesn't factor in variance.

 

[CountZero]

And if we're talking on the scale that everyone got their design within 1% of each other, the poor guy who forgot to optimize that last line of code ends up with a C-? Tough luck?

I had this second paragraph that talked about curves and professor discretion which I removed before posting. But the essence was that a curve is guidance and does well in 90%+ of cases but the prof. needs to look at whats going on to make sure that makes sense. In the highly unlikely scenario of everyone getting a working design where the spread over say 30 students is 1% then yes, the curve might be nonsensical. If the worst student is 30% worse than the best then I don't think its unfair to order them in their 1% intervals. And this is from someone who has been knocked down a few pegs grade wise for being less than 1% away from other folks. I had several classes that had competitive elements. It teaches you to look at your design beyond just meeting the bare minimum. By the time you are in an IC class meeting the bare minimum should be easy, making a good design is what you should be learning.

Also curves fall apart in small classes, 30 is probably the smallest class you could get away with it in. But 30+ I'd be absolutely shocked at everyone being in a 1% window of area without some type of cheating going on.

 

[nageov3t]

If cs profs wanted grading to reflect real world scenarios, wouldnt the grade be based on how fast you complete the project without regard to quality?

 

[alkemyst]

"Best" design is a vague target and when the intent is to make people work hard, you don't need a floating target. So you're asking the student to continuously redesign until the turn-in date. That is insane.

In the industry, you'd be surprised how far the designs are from the "best' design known. Complexity and risk (and sometimes legal issues) keep engineers from doing the best thing conceivable. And that is assuming there is a single metric for "best" since you can trade off area with speed.

Well, a student should know if they did their best and if not take what goes with it. If their best was subpar, then they need to deal with that level.

As a student, one should always push their limits. Sadly so freaking few do. In quite a few of my classes esp Zoology and Animal Anatomy and Physiology classes my grades were omitted from the curves. I usually scored above a 100 with bonus being offered. The next was usually 80 or below. Calculus kicked my ass a bit though.

 

[mfenn]

That would be well and good, depending on the system of distribution. What makes this system so bad is a linear distribution of grades. You could set the median resources used to a mark of 65, and then line people up and assign them marks based on a regular bell curve. That would mean you're not guaranteeing 50% of people to fail that section.

He says that they are linearly interpolating, NOT that they are using a linear distribution.

For example:
Best student uses 50mL of solution
Worst student uses 80mL of solution.

Student A used 70 mL, so he gets 10%
Student B used 60 mL, so he gets 20%
Student C used 50 mL, so he gets 30%
Student D used 51 mL, so he gets 29%
Student E used 51 mL, so he gets 29%
Student F used 80 mL, so he gets 0%

Seems perfectly fair to me.

EDIT:
I think that this is actually a really good way to grade the class. 70% of the grade is displaying competency. 30% of the grade is assigned based on merit. The 70% is key, because as long as you are competent, you will pass with a C. Do you want the A? Then you damn well better be excellent.

 

[mfenn]

Sounds crappy to me. The only way grades should be affected by other students is in a case where they are totally offset, then they will do a bell curve. Basically if everyone fails, they make the best person have close to 100% and rearrange the rest of the grades accordingly, so more people pass. I have never, ever seen or heard of this happening though, it's more or less a myth.

Happened all of time in my comp sci. classes (Clemson University). For example, in my advanced algorithms class, the highest grade I made on any test was a 56%. I still got an A in the class.

 

[HannibalX]

Grading on a curve is why stupid people pass. Each student should be graded on their own merits, period.

As for the corporate world working this way, no, it’s different. In school you have personal goals to complete, in a business you have departmental goals to complete. If your department doesn’t reach a goal, it is reflected on everyone, even if you did your part. You are being graded on your personal goals and performance but also on the performance of the department which you are part of, contribute to and were hired to work in.

 

[BrownTown]

Yes, it is completely fair, all course should be graded on a curve. If you are below average you get a below average grade, if you are above average then you get above average grade. Several of my professors said at the beginning of the semester, "I already know all the grades in the class, I just don't know who is going to get them". Guess what happened when I got to real world, the boss said, "I already know how many bonuses are going to be given out, I just don't know who gets them". Based on what I learned in my college experience I was one of the ones who got a bonus. I guess you can think its unfair to be graded against your peers, and then I guess you can think its unfair when you don't get a bonus, but the real world is all about COMPETITION. You are working hard for a limited number of management positions, raises, etc. If a boss retires and you have 4 good people in the group they don't jsut name 4 new bosses, they pick the BEST person and that is the new boss (OK, or maybe its the one who sucks the most dicks, depends on what sort of management your company has I guess ).

 

[TuxDave]

Well, a student should know if they did their best and if not take what goes with it. If their best was subpar, then they need to deal with that level.

As a student, one should always push their limits. Sadly so freaking few do. In quite a few of my classes esp Zoology and Animal Anatomy and Physiology classes my grades were omitted from the curves. I usually scored above a 100 with bonus being offered. The next was usually 80 or below. Calculus kicked my ass a bit though.

Make no mistake, I'm not advocating for slacking students. I just think the professor shouldn't be such a slacker and should come up with a design metric to hit. Want to make the students work hard? Make the target hard and if no student can hit it, make the highest scoring student the new target. Either way, give something to the student that'll give them an idea that they're doing things horribly wrong.

 

[TuxDave]

...but the real world is all about maximizing the amount of visible contributions...

Fixed for you! I completely agree with your last statement that the "best' person won't get the high bonuses if he was stuck fixing some crap that another person dumped on him and later took as credit.

Am I being a corporate humbug?

 

[BrownTown]

Fixed for you! I completely agree with your last statement that the "best' person won't get the high bonuses if he was stuck fixing some crap that another person dumped on him and later took as credit.

Am I being a corporate humbug?

School is probably even worse though when it comes to the number of people sucking up the the professors or especially cheating.

 

[TubeTote]

Only a real douche bag would use such a dumb ass grading system. Who's to say that you can't have a highly intelligent class of students?

We have enough challanges and competition in our daily lives, as I believe most students are aware of.

 

[jhu]

I'm glad I don't have study for grades anymore. OTOH, when you're in a class of 200+, 150 of them keep the other 50 in good standing!

 

[alkemyst]

Grading on a curve is why stupid people pass. Each student should be graded on their own merits, period.

As for the corporate world working this way, no, it’s different. In school you have personal goals to complete, in a business you have departmental goals to complete. If your department doesn’t reach a goal, it is reflected on everyone, even if you did your part. You are being graded on your personal goals and performance but also on the performance of the department which you are part of, contribute to and were hired to work in.

well in a real curve the stupid would fail. What most do is a 'half curve' and give out at least a C to everyone.

 

[ConstipatedVigilante]

Bell curves have only ever helped me. There are always quite a few idiots or slackers in any class. There is never a class where 80% of the class gets an 80, while the rest get 100. It just doesn't happen unless the test is waaay too easy.

 

[mugs]

Isn't that the same as using this year's data as the metric?

No, because last year's data is available before you turn your assignment in... you couldn't figure that out from when he said "if they knew ahead of time what it took to get 0/30 vs 30/30"?


What do you mean by linearly interpolated? Do you mean everyone's grades are evenly spaced in the span of 0 to 30 (so if there are 31 students the 2nd best gets 29, 3rd best gets 28, etc)? Or do you mean that if the best submission has an area of X and the class dumbass has an area of 5x (he's a dumbass) and your submission has an area of 2x you'd get 22.5/30 regardless of how many people are ahead of or behind you? (because there is a difference of 4 between the best and the worst, and a difference of 1 between the best and yours, therefore you lose 1/4 of the points)

The first method is unreasonable IMO, the second is reasonable IMO.

 

[ShawnD1]

What do you mean by linearly interpolated? Do you mean everyone's grades are evenly spaced in the span of 0 to 30 (so if there are 31 students the 2nd best gets 29, 3rd best gets 28, etc)? Or do you mean that if the best submission has an area of X and the class dumbass has an area of 5x (he's a dumbass) and your submission has an area of 2x you'd get 22.5/30 regardless of how many people are ahead of or behind you?

Your second example sounds more fair. If most of the class is very close to what the best design is, then most of the class should get a high mark for the overall design.

 

[Leros]

No, because last year's data is available before you turn your assignment in... you couldn't figure that out from when he said "if they knew ahead of time what it took to get 0/30 vs 30/30"?


What do you mean by linearly interpolated? Do you mean everyone's grades are evenly spaced in the span of 0 to 30 (so if there are 31 students the 2nd best gets 29, 3rd best gets 28, etc)? Or do you mean that if the best submission has an area of X and the class dumbass has an area of 5x (he's a dumbass) and your submission has an area of 2x you'd get 22.5/30 regardless of how many people are ahead of or behind you? (because there is a difference of 4 between the best and the worst, and a difference of 1 between the best and yours, therefore you lose 1/4 of the points)

The first method is unreasonable IMO, the second is reasonable IMO.

The second one is how linear interpolation works.

The resulting grades will probably be pretty reasonable. A few morons will get really high areas and most of the class will probably do pretty well getting mostly higher grades.