Atar Calculator Scaling (1 Viewer)

[saksham21]

How do atar calculators convert HSC marks to scaled marks? I've seen that HSCNinja's atar calculator provides a table with all the scaled marks for your particular HSC mark from recent years (I've attached a picture to show you if you don't know what I'm talking about). Does anybody know where they get this data from and how accurate it is?

 

[Nikolerak]

How do atar calculators convert HSC marks to scaled marks? I've seen that HSCNinja's atar calculator provides a table with all the scaled marks for your particular HSC mark from recent years (I've attached a picture to show you if you don't know what I'm talking about). Does anybody know where they get this data from and how accurate it is?

View attachment 29808

I am unable to answer your question, however I can confirm that HSC Ninja appears to be quite accurate. It overestimated my ATAR by 0.60. Matrix calculator was even more accurate underestimating my ATAR by only 0.05.

 

[Trebla]

They are most likely sourced from the UAC scaling reports published each year. In particular, Table A3 of that report which gives particular percentiles of scaled marks and most ATAR calculators attempt to interpolate the gaps in between.

 

[Jojofelyx]

They are most likely sourced from the UAC scaling reports published each year. In particular, Table A3 of that report which gives particular percentiles of scaled marks and most ATAR calculators attempt to interpolate the gaps in between.

so does that mean the closer you get to 99.95 the more accurate the prediction becomes?

 

[idkkdi]

They are most likely sourced from the UAC scaling reports published each year. In particular, Table A3 of that report which gives particular percentiles of scaled marks and most ATAR calculators attempt to interpolate the gaps in between.

just to confirm, UAC uses raws when calculating ATAR right? So even if both 80,81 raw align to the same score, UAC will be using scaled scores which are different, e.g. 41.2, 41.9 right?

 

[Trebla]

just to confirm, UAC uses raws when calculating ATAR right? So even if both 80,81 raw align to the same score, UAC will be using scaled scores which are different, e.g. 41.2, 41.9 right?

Correct. Scaled marks are derived from raw marks, not aligned marks. Table A3 should not be read as a “conversion” between aligned marks and scaled marks.

Also, a technicality to point out is that a raw mark of say 80 vs 81 will actually give different exact aligned marks, but these can round to the same aligned mark.

 

[Jojofelyx]

Correct. Scaled marks are derived from raw marks, not aligned marks. Table A3 should not be read as a “conversion” between aligned marks and scaled marks.

Also, a technicality to point out is that a raw mark of say 80 vs 81 will actually give different exact aligned marks, but these can round to the same aligned mark.

https://www.uac.edu.au/assets/documents/scaling-reports/preliminary-report-2020-nsw-hsc.pdf

So when you look at the A3 conversion table, how can you approximate the mark for the centiles provided?
Say we look at chemistry, P90 = 45.5 HSC mark, is there a way of determining what centile (other than the ones seen, 99, 90 75...etc) correlates to the appropriate mark? I hope that makes sense lol

 

[Trebla]

https://www.uac.edu.au/assets/documents/scaling-reports/preliminary-report-2020-nsw-hsc.pdf

So when you look at the A3 conversion table, how can you approximate the mark for the centiles provided?
Say we look at chemistry, P90 = 45.5 HSC mark, is there a way of determining what centile (other than the ones seen, 99, 90 75...etc) correlates to the appropriate mark? I hope that makes sense lol

If you are talking about to finding percentiles in between say 90th and 99th then you can estimate by interpolation (either fitting a straight line or some curve). You most likely won’t get an exact number though (there is no way to verify accuracy without the actual data anyway) but the ballpark approximation should be reasonable.

 

[Jojofelyx]

If you are talking about to finding percentiles in between say 90th and 99th then you can estimate by interpolation (either fitting a straight line or some curve). You most likely won’t get an exact number though (there is no way to verify accuracy without the actual data anyway) but the ballpark approximation should be reasonable.

I see, is there like an approximate for every subject percentile to know what atar a person will get. Like for example, if I'm in P90 for all my subjects, is that a 90 ATAR?

 

[Trebla]

I see, is there like an approximate for every subject percentile to know what atar a person will get. Like for example, if I'm in P90 for all my subjects, is that a 90 ATAR?

A 90th percentile for everything does not equal 90 ATAR. It’s more complicated than that.

For that, you’ll need to know:
- The scaled marks for that percentile for all subjects
- The sum of those scaled marks (out of 500)
- What ATAR that sum corresponds to

The latter involves another interpolation using even more limited data mapping the aggregated scaled mark with the corresponding ATAR.

You can see there is a noticeable room for error but this is effectively how ATAR calculators work.

 

[Jojofelyx]

For that, you’ll need to know:
- The scaled marks for that percentile for all subjects
- The sum of those scaled marks (out of 500)
- What ATAR that sum corresponds to

The latter involves another interpolation using even more limited data mapping the aggregated scaled mark with the corresponding ATAR.

You can see there is a noticeable room for error but this is effectively how ATAR calculators work.

ahhhh so there is a LOT of guessing that the calculators have to do I see, but if I'm aiming for like 99.85 then would the calculators on average spit out more reliable required marks, rather than say if I was going for 94.3 or something?

 

[Trebla]

ahhhh so there is a LOT of guessing that the calculators have to do I see, but if I'm aiming for like 99.85 then would the calculators on average spit out more reliable required marks, rather than say if I was going for 94.3 or something?

That seems to be the case based on ancedotal evidence, but I haven’t seen this hypothesis being tested.

 

[Jojofelyx]

That seems to be the case based on ancedotal evidence, but I haven’t seen this hypothesis being tested.

ah well, doesn't really matter too much, just gotta give all my subjects my best shot and see how it goes

 

[idkkdi]

ahhhh so there is a LOT of guessing that the calculators have to do I see, but if I'm aiming for like 99.85 then would the calculators on average spit out more reliable required marks, rather than say if I was going for 94.3 or something?

99.85 is the most useless aim. you miss out on literally every major thing you would need that high of an atar for lol.

 

[Jojofelyx]

99.85 is the most useless aim. you miss out on literally every major thing you would need that high of an atar for lol.

Heyyyyy ahaha, that was just a hypothetical, I'm actually aiming for 99.96 :\