Question about the ATAR (1 Viewer)

[xGhanem]

Okay, so I understand that when a person achieves top rank internally for a certain subject, and that if a different person beats that guy in the exams, the person that was 1st takes their exam mark, which seems unfair to me.
(If this is wrong, then please correct me)
My question is, does that mean that the internal ranks are more important than the external ranks in terms who gets the higher final mark?

 

[Laurennnnnn]

My understanding is that the #1 ranked will get the top exam mark (assessment mark) and their own true exam mark (external mark); the average of the two marks will be their final mark. So, your rank is super important in ensuring a good final mark, but if your rank isn't flash you can still pull yourself up with your own external mark. This is my understanding of it anyway!

 

[rumbleroar]

My understanding is that the #1 ranked will get the top exam mark (assessment mark) and their own true exam mark (external mark); the average of the two marks will be their final mark. So, your rank is super important in ensuring a good final mark, but if your rank isn't flash you can still pull yourself up with your own external mark. This is my understanding of it anyway!

^

this is quite true

a lot of emphasis is placed upon who is first and who is last (last receives the lowest exam mark). You generally want to be clustered around the top and have a strong internal rank (i.e. the relative gaps in between each rank are very small, so the distribution of the marks work in your favour more - if person A was ranked first and person B was second and the gap between them was 1%, they will only have 1% difference in between when they distribute the internal assessment marks [n.b. correct if wrong]) and have strong external HSC mark, so your averaged HSC mark is good.

 

[obliviousninja]

SB would have gotten a state rank if his internals didn't drag him.

 

[cem]

Okay, so I understand that when a person achieves top rank internally for a certain subject, and that if a different person beats that guy in the exams, the person that was 1st takes their exam mark, which seems unfair to me.
(If this is wrong, then please correct me)
My question is, does that mean that the internal ranks are more important than the external ranks in terms who gets the higher final mark?

It is not unfair as the student with the top exam mark keeps that exam mark as their own exam mark but...

it also sets the top assessment mark for the cohort and that mark becomes the assessment mark of the 1st ranked student - who had to earn that rank anyway. If they don't get the top exam mark they will be brought down by their own exam mark, which again they keep.

So - top exam mark 98 becomes the top assessment mark but say that top ranked student only got 78 in the exam they end up with (98+78)/2 = 88.

If the top exam mark is 98 and that student was 2nd in the ranks and say three marks behind first they would end up with an assessment mark no more than 3 marks off the top assessment mark and more likely two (due to aligning) thus (98+96)/2 = 97

How is that unfair - the student who earnt the top rank is rewarded for that effort but penalized for their poor exam performance while the consistent student who came first externally and second internally is rewarded for consistency.

 

[Queenroot]

If you're not the top student make sure the top student is doing really well and make sure the last student is passing

 

[braintic]

It is not unfair as the student with the top exam mark keeps that exam mark as their own exam mark but...

it also sets the top assessment mark for the cohort and that mark becomes the assessment mark of the 1st ranked student - who had to earn that rank anyway. If they don't get the top exam mark they will be brought down by their own exam mark, which again they keep.

So - top exam mark 98 becomes the top assessment mark but say that top ranked student only got 78 in the exam they end up with (98+78)/2 = 88.

If the top exam mark is 98 and that student was 2nd in the ranks and say three marks behind first they would end up with an assessment mark no more than 3 marks off the top assessment mark and more likely two (due to aligning) thus (98+96)/2 = 97

How is that unfair - the student who earnt the top rank is rewarded for that effort but penalized for their poor exam performance while the consistent student who came first externally and second internally is rewarded for consistency.

There IS an unfairness in the system when seen in a different light. To see it, take an extreme (ie. unlikely) scenario:

Assume that there are a large number of students in the cohort, with person A and B being the standout top two.
Person A ranks 1st all year, beating person B by 10% in every assessment.
Person A has an off day in the HSC exam ... person B gets exactly as expected, but A only beats B by 1%.
Thus the maximum mark for the assessment component is much lower than expected.
So person B gets a much lower assessment mark than expected, due entirely to the off performance by person A.

Although this is an extreme case, it does highlight the unfairness in determining the 'mark window' based on the performance of just two people. Moderation should be based on the performance of ALL students in the cohort (using mean and standard deviation), not just two people.

 

[cem]

Moderation IS determined by the performance of the entire cohort including mean and standard deviation.

The top and bottom exam marks set the top and bottom internal marks. The total number of marks earned by the cohort will equal, give or take a mark or two, the total marks awarded as internal marks. The mean is the same as the exam and the standard deviation is also the same as the exam.

The exam is the only thing that can be used as that is the common assessment task done by ALL students across the state.

If a student performs unexpectedly poorly or well they can be deemed to be an outlier and taken out of the process - as would probably happen in your scenario.

 

[braintic]

Moderation IS determined by the performance of the entire cohort including mean and standard deviation.

The top and bottom exam marks set the top and bottom internal marks. The total number of marks earned by the cohort will equal, give or take a mark or two, the total marks awarded as internal marks. The mean is the same as the exam and the standard deviation is also the same as the exam.

The exam is the only thing that can be used as that is the common assessment task done by ALL students across the state.

If a student performs unexpectedly poorly or well they can be deemed to be an outlier and taken out of the process - as would probably happen in your scenario.

As I said, my scenario is an extreme case. If the top person underperforms by a couple of marks they would not be removed from the moderation process. Yet this WOULD affect the marks of the people below.

Reallocating the total number of marks DOES NOT preserve the standard deviation. It takes only a simple example to show this is wrong.

Lets say that for a class of 4 people, the following marks are submitted to the board:
90 70 60 30

Let say that in the exam they score:
90 80 60 30

Try reallocating the total exam mark while preserving BOTH relative gaps AND standard deviation AND setting the top and bottom marks to 90 and 30.

Again it is an extreme example, but the problem is still there in typical cases.

In any case, here is a quote from the Board of Studies site (http://www.boardofstudies.nsw.edu.au/hsc-results/moderation.html):
"Please also note that the standard deviation is never used to moderate school assessment marks."

From the same page:
"some students are initially excluded from the moderation of the assessments of their group. These include:students whose performance in the examination ... is markedly below what was expected on the basis of their performance relative to the group in the assessment"
In other words, they DON'T exclude people whose results are markedly HIGHER than expected.

 

[cem]

The BOS doesn't need to use standard deviation for the assessment marks as they have used it on the exam marks and the exam marks set the assessment marks and the assessment marks will match the range, total, mean and standard deviation of the exam. That is why they say they don't use standard deviation on the moderated marks - because they don't - they use them on the exam marks.

You are also assuming that the real gaps and relative gaps have to remain the same - they don't. In your example the 90 and 30 would stay the same but the middle two marks would be adjusted to reflect the relative gaps so student two would end up with an assessment mark in the high 70s while student 3 would end up in the mid-60s - keeping 'relative' gaps not 'real' gaps the same across the cohort.

 

[braintic]

The BOS doesn't need to use standard deviation for the assessment marks as they have used it on the exam marks and the exam marks set the assessment marks and the assessment marks will match the range, total, mean and standard deviation of the exam. That is why they say they don't use standard deviation on the moderated marks - because they don't - they use them on the exam marks.

You are also assuming that the real gaps and relative gaps have to remain the same - they don't. In your example the 90 and 30 would stay the same but the middle two marks would be adjusted to reflect the relative gaps so student two would end up with an assessment mark in the high 70s while student 3 would end up in the mid-60s - keeping 'relative' gaps not 'real' gaps the same across the cohort.

No they have NOT used standard deviation on the exam marks, because the moderation process is performed on the RAW exam marks.
And even if they had, that would not remove the problem I have described.

I have most certainly not confused real gaps with relative gaps.
And your calculation has most definitely NOT preserved relative gaps.
Before moderation, the gaps were 20, 10 and 30.
That is in the ratio 2 : 1 : 3.
After your moderation, assuming your marks were 90, 77.5, 65, 30, the gaps are now 12.5, 17.5 & 35.
Simplifying this ratio by setting the middle number to 1 as in the previous ratio, you get 0.71 : 1 : 2.
So the relative gaps are nowhere near preserved.

 

[D94]

No they have NOT used standard deviation on the exam marks, because the moderation process is performed on the RAW exam marks.
And even if they had, that would not remove the problem I have described.

I have most certainly not confused real gaps with relative gaps.
And your calculation has most definitely NOT preserved relative gaps.
Before moderation, the gaps were 20, 10 and 30.
That is in the ratio 2 : 1 : 3.
After your moderation, assuming your marks were 90, 77.5, 65, 30, the gaps are now 12.5, 17.5 & 35.
Simplifying this ratio by setting the middle number to 1 as in the previous ratio, you get 0.71 : 1 : 2.
So the relative gaps are nowhere near preserved.

The moderated marks would be 90, 71.8, 62.7 and 35.5 (1 d.p.). BOS would raise the lower bound to ensure the mean is the same as the mean of the HSC exam marks and relative gaps are the same as the school marks. The lower bound does not have to be fixed - it can be raised, but not lowered.

(NB: I am not claiming standard dev. is used)

In your first response, that is not a straightforward case of unfairness. Student A does not have the responsibility to carry the cohort. The HSC exam is first and foremost an individual assessment task. Student B cannot simply rely on student A to get them better marks. Maybe student A had an off day, but student B and the rest of the cohort are still capable of achieving better marks. A mark cannot be 'expected' without comparing it to other marks. In your case, one could interpret student B as achieving above 'expectation' as opposed to student A achieving below 'expectation'.

Undoubtedly, there are other instances where I found the nature of assessment tasks made the HSC and moderation process 'unfair', but unfairness is subjective and this is probably the most fair system to date.

 

[braintic]

The lower bound does not have to be fixed - it can be raised, but not lowered.

Quote from the Board of Studies page linked to below:

"... In such cases the bottom moderated assessment mark may need to be moved up or down."


The fact that a person's mark can be affected by the performance of a single person IS unfair.
Do you think the bottom person deserves to have his mark raised to 35.5 based solely on the distribution of the rest of the marks, when an identical person at an "anti-school" might get their mark lowered to 24.5?

 

[D94]

Quote from the Board of Studies page linked to below:

"... In such cases the bottom moderated assessment mark may need to be moved up or down."


The fact that a person's mark can be affected by the performance of a single person IS unfair.
Do you think the bottom person deserves to have his mark raised to 35.5 based solely on the distribution of the rest of the marks, when an identical person at an "anti-school" might get their mark lowered to 24.5?

Ok, I stand corrected, but that's beside the point.

The person's mark was not affected by a single person, it was affected by the whole cohort. The mean of the exam marks is factored in, and as you said above,

Moderation should be based on the performance of ALL students in the cohort (using mean and standard deviation), not just two people.

It was not raised only due to the distribution; it was raised because the mean of the exam marks was higher than the mean of the school marks. Moderation is based on the performance of all students, and a person's mark was affected by the performance of all students. You can't have it both ways, whether it's 2 students at the top, or a student at the bottom.

But you are looking at this in a black and white case. You are claiming a person did not perform to expectation but what measure is there for expectation? Why must it be that student A performed below expectation, and not that student B performed above expectation? It's expected that student A does the best, and you described student A as beating student B by 1%, so that expectation is met. So how can you objectively claim that student B achieved to 'expectation' and student A didn't?

The scenario you just mentioned is ridiculous. You are assuming the exact same assessment was conducted at both schools, where the quality of teaching is exactly the same, that the influences of others in both schools affect that student in the exact same way, and that they were the worst student at that school as well. Seriously, does your question still warrant a response?

 

[braintic]

Ok, I stand corrected, but that's beside the point.

The person's mark was not affected by a single person, it was affected by the whole cohort. The mean of the exam marks is factored in, and as you said above,



It was not raised only due to the distribution; it was raised because the mean of the exam marks was higher than the mean of the school marks. Moderation is based on the performance of all students, and a person's mark was affected by the performance of all students. You can't have it both ways, whether it's 2 students at the top, or a student at the bottom.

But you are looking at this in a black and white case. You are claiming a person did not perform to expectation but what measure is there for expectation? Why must it be that student A performed below expectation, and not that student B performed above expectation? It's expected that student A does the best, and you described student A as beating student B by 1%, so that expectation is met. So how can you objectively claim that student B achieved to 'expectation' and student A didn't?

The scenario you just mentioned is ridiculous. You are assuming the exact same assessment was conducted at both schools, where the quality of teaching is exactly the same, that the influences of others in both schools affect that student in the exact same way, and that they were the worst student at that school as well. Seriously, does your question still warrant a response?

Well it is clear you are wedded to this system, so there is not much point in me continuing, is there.
Except to say that any moderation system that attempts to force a compromise between the distribution of the assessment marks and the distribution of the exam marks is flawed.