Hacker News Reader

Here is a story of something that happened to a friend of mine. This is from New Zealand.

He was taking 7th form Latin. Latin has enough students in the 3rd form but by the time you get to 7th form (The final year before university) there were only 20 or so students in the entire country taking the course. He was taking the course by correspondence because the school obviously wouldn't provide a teacher for a single student.

He scored 95% in the final exam. In NZ the end of year standardized tests are scaled to fit a curve across the entire country. He was scaled down to 43%. A failing grade.

Obviously in this case the only students that would bother taking Latin all the way up to 7th form are going to people who care enough about it to learn it well. The standard scaling method used doesn't make sense at all with this group of students.

He wrote a letter to the education minister of NZ complaining about this issue. Months later he got back a form letter explaining why test scores are scaled with no regard to the special circumstances.

Can you clarify what you mean? How does a curve cause a 95% to get scaled down to a 43% failing grade?

If you have 10 people who score:

    91%, 92%, 93%, 94%, 95%, 96%, 97%, 98% 99%, 100%

and scale the grades from 0-100, those students' curved grades are:

    0%, 11%, 22%, 33%, 44%, 55%, 66%, 77%, 88%, 100%

In other words, they turn the students' grades into their ranked percentile within their cohort. It's an absolutely brain-dead decision.

Oh, I see. Essentially curving from both ends, to fit a desired distribution.

That idea has never made sense to me, at least if the class is using a system where having a score less than some specific percent implies failure.

Stack ranking does the same to employees in elite teams in real-world companies.

No it means the exam was way too easy. 90% of the questions were giveaways. If its ok to fail 400 out of 1000 students doing English, then its ok to fail 4 out of 10 doing Latin. Once you adjust for the easiness of the exam, he actually did not do too well.

Are you trolling? With a subject like Latin, the actual situation is that all the students are really good by the time you have only 20 students left in the whole country. The exam is probably not easy at all, and someone who didn't prepare well would get a much worse score.

You are assuming that it makes sense to make grading to a curve with a sample that is highly selective. It does not.

(I did not study Latin. Just as an item of curiosity, my country's public radio service YLE broadcasts regular radio news in Latin. You can listen and also read some of the stuff here: http://ohjelmaopas.yle.fi/1-1931339 )

In related news, the fifth-best player on an NBA team is a failing basketball player, because 80% of the starting players are better than he is.

You try to fit the marks to a normal distribution. It assumes that all classes/courses have the same distribution, with the same midpoint and that the only variation is that the test was too easy.

So, (simplistically) if the expected midpoint is 75, and everyone is in the 95-100 range, a grade of 97.5 would be shifted down to 75.

It does penalize students for having really smart friends and getting together to take a difficult course.

http://en.wikipedia.org/wiki/Grading_on_a_curve

Curving only makes sense when there's a normal distribution. Sounds like, at that point, the students were no longer in a normal distribution and therefore curving failed.

People often underestimate the bitterness that these experiences create in smart people. It's not a petty thing, either - it's life changing in many situations.

We need smart people to take control of schools.

The road of life is bumpy. I would not seek to smooth out the few remaining bumps in school. We are already in danger of creating fragile people who shatter at the first bump.

All this, even if the teacher's idea were quite wrong - but it's not. By "clipping" the score, the teacher created a hybrid between curved and uncurved grades. If curved and uncurved are acceptable, why not the hybrid?

Neither curved nor uncurved grades remove points that a student earned from the student's score. Uncurved grades obviously just give you what you earned, and traditional curved grades give you more than what you earned. This "hybrid" gave me *fewer( points than what I earned.

> and traditional curved grades give you more than what you earned.

Traditional curves (mapping grades onto a bell-shaped curve based on the class average) do potentially give lower scores than standard 10% brackets (90% = A, etc.)

There is a common thing used in some places (particularly secondary -- don't know that I've ever seen it higher ed) often called a curve which just takes the highest grade in the class or the grade at a certain percentile point and calls it 100% and renorms scores based on that, and that system never gives people lower than the pre-adjusted score. That may be what you are thinking of.

Yes, traditional curves do that, and that is a laughably preposterous practice.

Smoothing out bumps or making it "easy" and worrying about fragile stuff is totally unrelated to acting intelligent.

In early grades, I had amazing teachers. Plus they had "enrichment" where they pulled us out of less-useful classes, gave us a lab and a dedicated teacher who just said "hey, what are you interested in?" I excelled there.

After moving a few times, I unofficially dropped out of school in 7th grade (12 years old), after grilling my teacher and realizing she didn't really have a clue. Other teachers were worse.[1]

I attended for a few more years (partially coerced by illegal police force[1]), then at 15 dropped out for good by leaving the country.

Things turned out OK for me. But I lost a LOT by not getting a real education, not having a chance at college. Just the lack of maths alone takes a lot of effort to rectify. Yes, it's still my fault, but the system shouldn't be encouraging 12-15 year olds to make such mistakes.

Kids will have enough bumps with bullying and school life in general. Adding idiocy bumps that make a mockery of education and intelligence just make kids cynical and tune out.

(It also makes such graduates more OK with the idea of corruption, stupid bureaucracy, mediocrity, etc. - a shrug and a "that's how things work". Such things should be vigorously fought, not beaten into children.)

1: One "science" teacher: "Carbon-14 dating is a lie told by scientists that hate god."

2: Somehow there was enough of a "problem" to haul me in front of a judge who pre-emptively revoked my drivers licenses for 3 years. I told him I'd just get a pilot's license instead and enrolled in flight school the next month. I also found it hilarious how people in court used the term "infinite wisdom" when referring to the judge. Illegal part was that I am not a U.S. person so they didn't have jurisdiction for enforcing mandatory schooling any more than they'd have on any other B-2 visa'd tourist.

> Smoothing out bumps or making it "easy" and worrying about fragile stuff is totally unrelated to acting intelligent.

I stared at this sentence for a while and didn't quite decode it. Sounds like you've had your share of bumps. I don't know how old you are, but you may eventually decide you wouldn't change a thing.

But quite a few kids raised in good environments seem to have missed our on bullies (real, not cyber), and stupid, cruel authoritarian teachers.

This results in a naive and fragile person who is not equipped for the real world.

Have you read The Diamond Age? Neal Stephenson makes a similar point when the headmistress (really YT from Snow Crash) answers a complaint about a teacher.

Sorry, poorly written. I'm just making the point that there's a far cry from producing "fragile" people/overprotecting versus creating and defending a stupid system.

In the former case, it can be good to let kids have to get over some problems or deal with unfairness.

In the latter, encouraging or defending stupidity on the part of teachers or the school system only encourages cynicism or acceptance of such brokenness. School systems should try to encourage the ideals, teach people to stand up to bad things, not accept mediocrity, etc.

I'm 33 and have kids of my own, if that's relevant. I liked Snow Crash but don't remember those specifics.

At the risk of sounding completely ignorant, what's wrong with what she did? In most classes I've been a part of, your total score was based on a weighted average of scores on particular tasks. The normalization here is a typical test curve -- a normalization that attempts to compensate for measurement error in one particular test, independent of the total class score. That's why she referred to "extra credit".

This was explicitly not the case in this class. A test curved to 85 points was worth less than another test curved to 90 points.

And honestly, even if each test was normalized to 100 points after the curve, I would still object (on ethical grounds, not mathematical) to not being given >100% when the test is curved to below my score.

Ah, I didn't catch that. That's... annoying.

Perhaps you might object mathematically too. I think of a curve as fitting the test scores to a specific distribution as a corrective to testing error. It's not the fitting that's the issue, it's how justified the target distribution is. If you truncate it, you have to justify doing that somehow (all prior test results had that distribution?). Of course, if the sample distribution squashes it enough, you might end up clustered at 100% anyway, but...