ATAR Notes: Forum
VCE Stuff => Victorian Technical Score Discussion => Topic started by: kingmar on November 11, 2007, 12:03:24 pm
-
I want to propose an idea, I'm sure we've all had it...
No-one has managed to come up with a Study Score calculator yet. But how would this work?
Consider this. Study Scores are normally distributed. Correct?
Thus, if everyone from this year gets statements of study scores, we can possibly plot points and get an estimation of what is needed for a certain study score. Not to mention, mid-year VCAA release their mean and standard deviation statistics.
If we get a sample of like 5 study scores or ranges of marks for a certain study score, does anyone think it possible that we can normalise these marks ourselves and predict study scores?
Just a thought.
This wouldn't work for Chemistry or English by the way, as the study design is changing too much for old stats to be relevant.
-
difficult because it dpends on difficulty of the exams as well as strength of the competition. Also some scores mark sacs harder than others ( But i dont know if we rank according to schools in victoria )
-
Yeah i'd like someone to create one, but it would be hard too many variables.
-
The scores reported in the SoS are moderated already.
Thus, it's easier to normally distribute. But, I'm not the maths whiz here. All we'd need are like 4 reference points - we could estimate the rest reasonably accurately. After all, all you need are two sets of points to determine mean and SD.
VCAA provide these statistics in due course. When they are released, we can make an accurate guess, I presume. But I need data.
-
Yes thats true, we could create an accurate calculator for last years data...but this years scaling, difficulty, competiion is all going to change. Eg the exams are different difficulty, hence subject scaling AND score skew is different... as well as the people themselves being different.
But youre right, we could probably get a really rough indication./
-
Cohorts don't change much between years.
What we derive in one subject we can relate to similarly scaled subjects, such is the nature of scaling.
Which is what makes this great. BM can be likened to H&HD, Eco can be likened to Accounting.
Yes, the ranges all differ. But proportionally, we can thus make an accurate estimation, as opposed to a random guess out of the blue.
-
Just calculate your STUDY using normal distr, by percentage of doing well.
mu = 30
sd = 7
Pr(score <= X) = 0.85, top 15%
X = ~37.25.
Gonna have to use grade distributions to find percentages of people who got A's etc ..
-
Just calculate your STUDY using normal distr, by percentage of doing well.
mu = 30
sd = 7
Pr(score <= X) = 0.85, top 15%
X = ~37.25.
Gonna have to use grade distributions to find percentages of people who got A's etc ..
Yes, that was what I was referring to.
But specifically, that doesn't tell you what kind of EXAM SCORES you would need to achieve, or scaled SAC scores.
Which is my aim after I'm done with this pain in the arse here named VCE.
-
or...you could just study really hard and let VCAA do all that.
-
That would probably be unreliable, as the sample size would be too small.
-
"textual abuse"
You don't like me much, do you? :cry:
Have a look at all the "predict my study score" threads out there.
If we got some semblance of a table going, it would help them greatly.
I'm only trying to help. Genuinely help, as opposed to your put-downs and unconstructive comments.
Don't tell me, tell all those people out there that go on about what study score they want. It's a service I'm thinking about. You don't have to use it if you hate it or me so much.
That would probably be unreliable, as the sample size would be too small.
It would be more reliable than plain guessing at study scores as per all the thread predictions we have going on.
It's a thought. If there's too many doombringers or negative opposition to it (can't see why, but still), then I won't bother.
-
uh... theres no dislike, OR put downs in any post on this thread...
-
Tell me when rustic_metal has said something useful in response to me, and I'll eat my chocolate hat.
If you can't, I'll eat it anyway, because it's chocolate.
EDIT: Ask him. I reserve my judgement, and I will not talk about it further.
Now, back on topic...
Anyone care to get Statements of Study Scores? Mine won't suffice on its own. At least two are needed to predict means and SD's, not to mention making generalisations. Five as a minimum is preferable.
-
lol, not being useful isnt the same as put downs and dislike.
-
or...you could just study really hard and let VCAA do all that.
good idea :)
-
Alright. Bad idea, I resign.
Seems like my idea of helpful differs from most people.
Keep guessing those study scores then! :P
-
Aww dont worry kingmar, rustic_metal is like that to everyone lol ...
I guess it's because his first impression of you wasn't a good one (bus.man forum). You'll get used to him dw :)
-
Huh, I think it's a great idea. Farkin difficult yes but I don't understand why ppl are bagging it.
I'll be happy to help with the subjects I have when I get my SoMs :)
(I'm planning on getting SoMs for all my subjects this year just to torture myself even more :))
-
I like looking at other people's results. I'm sorta nosy. Hahaha.
-
I'll probably get SoMs for my two 3/4s this year - Business Management and VET IT: Network Administration (ICA99 training package, not the new ICA05 though).
-
I think it's a good idea too, I can't say that I haven't thought about it before, but it does seem really difficult.
I have SoM for English in 2006, and my mid-year Psych one. I'm not planning to buy any others though, but when the examiner's report comes out I may be able to give you fairly accurate scores for Psych and Methods.
-
Huh, I think it's a great idea. Farkin difficult yes but I don't understand why ppl are bagging it.
I'll be happy to help with the subjects I have when I get my SoMs :)
(I'm planning on getting SoMs for all my subjects this year just to torture myself even more :))
nOONES FKING BAGGING IT OMG lol
:@:@:@ ANd no, noones being helpful either. were just pointing out the god damn problems.
ARGH
-
Huh, I think it's a great idea. Farkin difficult yes but I don't understand why ppl are bagging it.
I'll be happy to help with the subjects I have when I get my SoMs :)
(I'm planning on getting SoMs for all my subjects this year just to torture myself even more :))
nOONES FKING BAGGING IT OMG lol
:@:@:@ ANd no, noones being helpful either. were just pointing out the god damn problems.
ARGH
Calm down mate lol ...
But it's true we arent bagging the idea...
-
AARAGHHH@!! YOU WONT LIKE ME WHEN IM ANGRY
*VEINS BULGE **SKIN GOES GREEN* *MUSCLES GROW (SEEMINGLY FROM NOWHERE)*
ZOMG1
-
Wow :shock: ooook .... sorry geez
-
Wow :shock: ooook .... sorry geez
:wink:
-
When you approximate an approximation you're gonna get errors. That's the way it is.
Maybe I should explain the train of thought.
This is not about "Let's let VCAA do all the work and just work hard".
This is about the countless "predict my SS plozzzz" threads going around.
I am not saying this is guaranteed to work. I am simply launching a discussion for a PROPOSED SS calculator/predictor.
It's all well and good to say what grades will get what, but wouldn't it be better if we had actual scores? If you knew you were at a certain level in SACs, by varying the exam score you could get a prediction occurring.
Now, the way I see it is this:
We will have VCAA's standard deviations and means. This forms the basis 30 study score.
Using proportions, we can deduce required SAC scores.
Say Methods. The exams are worth 66% of your score. Thus, your "out of 100" SACs will be moderated so that they match up with your exams.
A note: this is based on the assumption that the mean in everything gives you a 30.
By using normal probability, we can predict study scores and the apparent percentage needed to get that score.
With real data, it will become all the more meaningful.
I can already do it without real data, but it will be extremely biased towards my results, and will seem a bit weird (again, assuming the results are distributed equally)
Thus, the main problem with this is the assumptions. Does it completely render the calculator invalid? That's for you to decide. After exams, I'll design a prototype table. I'm crap at programming (only VB for me), so maybe Daniel15 or someone could do that.
At the moment, it's at a philosophical idea stage.
-
AARAGHHH@!! YOU WONT LIKE ME WHEN IM ANGRY
*VEINS BULGE **SKIN GOES GREEN* *MUSCLES GROW (SEEMINGLY FROM NOWHERE)*
ZOMG1
Whoa! Get off the steroids... i mean the spinach, boy.
-
A note: this is based on the assumption that the mean in everything gives you a 30.
I'm concerned about this. Based on a lot of anecdotes I hear, getting mean scaled SAC and mean exam score doesn't necessarily guarantee you a 30. Either a lot of people lying, or there is further moderation done on scores ...
-
All this moderation stuff leaves me with headaches.Its what the VCAA duded so behind closed doors that somehow counts ...
-
I'm 99% sure VCAA are dirty scammers.
I've been writing this calculator, and we can see valid results for a 30 score:
(http://img222.imageshack.us/img222/3167/wtf1qo8.png)
But trying for a 45 score:
(http://img235.imageshack.us/img235/7436/wtf2li7.png)
More than 100% for GA1 and GA2? What?
Assumption 1: I was under the impression that, for example, a median in each GA would contribute equally EDIT: proportionally, not equally, equitably to a median study score (where GA1 = 50, GA2 = 50, GA3 = 400 marks) .......
Assumption 2: Pr(STUDY > 45) = 1 - 0.983938. Hence, taking the inverse of each GA for 0.983938 should give the required score for each GA?
Or have I got it wrong? ... this is all based off exact data provided by VCAA about 2006 ..
It's not a coding error either, my calculator agrees with me.
F U VCAA.
-
Funny. We both get the same results.
I couldn't write jack for DOS programs.
If you look at the proportions, the distributions are NOT normally distributed. That's why it must all be manually inputted.
I'm sure you'll find the French Exam scores are a little too high as well. As long as all your code is correct, I'm sure that's the issue. That's why assumptions are no good.
To be honest, I suspect VCAA do their own twiddling in the meantime. I'm certain that even with crap exam scores, I can still do well because I suspect they take your OTHER studies into account when moderating your scores.
i.e. I get average 45 in all my subs except a 30 in methods.
VCAA inexplicably lift my methods score to 33. Why? Because of the moderating process.
They don't tell you everything, though... So yes, it is difficult. That's what real data can help you with.
EDIT: I might do some more research into the grading process undertaken by VCAA. This looks like an interesting side project.
-
Meh, as coblin said, the lack of transperancy in the system is a bad thing...
Here's my code anyway. Written in C++, for gnu gcc. Using boost::regex
study_calc.cpp - http://pastebin.com/f4755d394
study_calc.h - http://pastebin.com/f4f938292
Binary + sources: http://download.yousendit.com/0D05BBE02DCF3328
Only one set of grade data included (for French), but it can be generated by getting a PDF scaling report from VCAA, saving it as .txt with Foxit PDF Reader, and my program will parse the rest.
My thoughts: this is bloody stupid. vcaa says study's are normally distributed mu = 30 sd = 7, no?
-
That is pretty hardcore
-
edited
just wait for your SCORE! either way you get the same score wether you spent 50 years making the study calculator or you didnt and enjoyed life!
lol
-
I think you're talking about an ENTER calculator. They're working on a study score calculator. It's different :)
-
This side project is on hold while I have exams, so I don't have anything at the moment.
This is not for me. I don't think anyone understands. I can already guess reasonably accurately my scores, so I don't care.
This is intended to be a solution to all the "predict my study score! plz plz plz i'll give you candy" kind of threads that spam the forums. A lot of people want to know, so if something could be devised that addresses it, wouldn't that be good?
I'm not stupid. Don't I know that I'll get the same scores whatever I do?
I repeat, this is NOT ABOUT ME, it's about THOSE THAT WANT TO KNOW SCORES.
Read the WHOLE topic before you make premature judgements. If you have nothing constructive to say, don't say it. I'll do my work when I'm through with exams and I'll post what I end up discovering.
Until then, wait on. Only 6 days until I'm done with chinky-chi-cho.
-
I'm 99% sure VCAA are dirty scammers.
I've been writing this calculator, and we can see valid results for a 30 score:
(http://img222.imageshack.us/img222/3167/wtf1qo8.png)
[other stuff...]
Why'd you take a screenshot of text instead of copy-and-pasting it? :P
Edit: Ewww, spaces for indentation in the .cpp file :shock:. Hard tabs FTW ;)
saving it as .txt with Foxit PDF Reader,
Foxit can save PDFs as .txt files?
I've been copying and pasting stuff from PDFs to get the text from them! :P
-
I spent two hours on it tonight.
At the moment, based on rough guesses and my Accounting scores (and some guy's BM marks), I still need to adjust some values...
I will post an excel spreadsheet when I work out the true values. I'm getting there...
Otherwise, just sit tight with this.
-
Funny. We both get the same results.
I couldn't write jack for DOS programs.
If you look at the proportions, the distributions are NOT normally distributed. That's why it must all be manually inputted.
I'm sure you'll find the French Exam scores are a little too high as well. As long as all your code is correct, I'm sure that's the issue. That's why assumptions are no good.
To be honest, I suspect VCAA do their own twiddling in the meantime. I'm certain that even with crap exam scores, I can still do well because I suspect they take your OTHER studies into account when moderating your scores.
i.e. I get average 45 in all my subs except a 30 in methods.
VCAA inexplicably lift my methods score to 33. Why? Because of the moderating process.
They don't tell you everything, though... So yes, it is difficult. That's what real data can help you with.
EDIT: I might do some more research into the grading process undertaken by VCAA. This looks like an interesting side project.
does that mean, if you do badly in 4 of your subjects but the other 2 are good e.g: 45, 45, 25, 25, 20, 20
then your 45's will go down to like 42 or something?
So much for moderation....
-
To be honest, I suspect VCAA do their own twiddling in the meantime. I'm certain that even with crap exam scores, I can still do well because I suspect they take your OTHER studies into account when moderating your scores.
i.e. I get average 45 in all my subs except a 30 in methods.
VCAA inexplicably lift my methods score to 33. Why? Because of the moderating process.
They don't tell you everything, though... So yes, it is difficult. That's what real data can help you with.
EDIT: I might do some more research into the grading process undertaken by VCAA. This looks like an interesting side project.
Was that example merely an illustration of a suspicion?
I don't think it's true. They have a similar idea where they look at your other raw scores in order to determine the scaling of the subject, but it is aggregated statewide. They find the average of "average raw scores" that every student taking a particular subject got, and they set that score as the score that 30 will scale to. The way this works is because if you have an especially smart cohort taking Specialist Maths, the competition will be high. In order to standardise this against other subjects, they look at how every student in Specialist Maths went in their other subjects, and take the average of the "average raw score of each student" (that's a mouthful!) to determine the scaling.