How accurate are the ATAR calculators? (1 Viewer)

Salehhh

Member
Joined
Jan 28, 2013
Messages
183
Gender
Male
HSC
2013
People say 5-10 marks off and people say 0.5 - 1 marks off...
 

SuperMike96

Member
Joined
Jan 30, 2013
Messages
253
Gender
Male
HSC
2013
I'd take it as a general guide. Expect to be within a few points of the ATAR calculators estimate, but don't think it will be 100% accurate.
 
Joined
Jun 7, 2013
Messages
340
Gender
Female
HSC
2013
Use a few of them and they will give you a general guide of where you are headed but it definitely shouldn't be as big of a gap as 5-10.
 

D94

New Member
Joined
Oct 5, 2011
Messages
4,426
Gender
Male
HSC
N/A
Well you need to understand how ATAR calculators work first before you can ask how accurate they are.

The calculators use the previous year's data from UAC, i.e. the UAC tables. So Table A3 indicates a HSC mark and a scaled mark, as well as a percentile. What ATAR calculators do is they take in your HSC mark, convert it to the scaled mark, then add up the scaled marks to form your aggregate. They then refer to table A9 and most likely interpolate to determine your ATAR.

So since scaling is carried out afresh each year, last year's data is obviously not completely valid for an estimate for this year. Even if it were accurate, there's not enough data points to determine the exact scaled mark - it is not a linear relationship, nor is aggregate to ATAR. In the actual calculations, UAC takes in your raw HSC marks and scales them, but they publish the HSC mark that is presumably aligned - this means there can be a set of raw HSC marks that correspond to an aligned HSC mark. This does increase the error because there isn't enough data.

Since there is less data available for the lower percentiles in Table A3, the error is increased. For the popular subjects, especially the ones the, I guess, "bright" students take, the scaling change would be minimal, so the data 'could' be averaged over a period of time to determine what might be the scaled marks for this year. But again, it is an estimate, and scaling is carried out afresh each year so one cannot assume there will be no change at all.

Essentially, you cannot even guess the error in estimates. But it's reasonably safe to assume the higher your estimate, the more likely it is within 0.5-1. The lower your estimate, it is possible the range is 5 and in even lower estimates, it could be within 10.
 

Users Who Are Viewing This Thread (Users: 0, Guests: 1)

Top