Tiffin round 2 today

[Snowdaddy]

Thanks, I am more interested in :-

(1) Why do they need to defer the decision based on raw scores? They should be decisive and stick to a rule ( this is quite contrary to the TGS brand !)

(2) This ambiguity (standardised score vs age standardised score) will likely lead to a greater number of appeals.

(3) The stage 2 exam scores are comprised 40% Maths and 60% English. It may make sense to age standardised one of the components and not the other - what to do then?

(4) What will sway the decision to age standardised or not? What particular thing are they trying to address? We can only guess. It would be good to hear what happened last year.

This decision could have a bigger impact on the October and August born. TGS will need to have a ready made answer by allocation day.

I am not a statistics expert but from my understanding

1) if you standardise, what split do you choose, 12 months or 6 sets of 2 months etc. if you ran both scenarios and the results were different then you would have a problem justifying your rules. Easier to choose afterwards
2) see (1)
3) shouldn't really mix and match, it suggests result are fixed
4) I think in past only round 1 was standardised

The problem is you have a truncated set, only have the top 500 sitting (or whatever exact number was), unlike first set (of about 1700 candidates), now only have about 30%, if you plot a normal distribution then take 30% at one end you have a strange shape distribution (so any techniques based on normal distributions aren't valid).
Further you get into small sets, lets say 500 candidates gives 12 monthly sets of about 42 (and as births aren't even through the year some sets might be nearer only 30). Then to standardise need to get rid of freak results so lets say you drop top 5 and bottom 5 from each set, and use average of remainder. Almost certainly now have some sets under 30 and that means cannot apply certain regression techniques without risk of unreliable results. So what happens next, fewer bigger sets, but who chooses the size, or one set of all. Hope you start to see the problem.

[stateschoolman]

Thanks.

I guess TGS knows best and have wisely kept their options open considering the issues you have outlined above.

[MCLC]

Thanks, I am more interested in :-

(1) Why do they need to defer the decision based on raw scores? They should be decisive and stick to a rule ( this is quite contrary to the TGS brand !)

(2) This ambiguity (standardised score vs age standardised score) will likely lead to a greater number of appeals.

(3) The stage 2 exam scores are comprised 40% Maths and 60% English. It may make sense to age standardised one of the components and not the other - what to do then?

(4) What will sway the decision to age standardised or not? What particular thing are they trying to address? We can only guess. It would be good to hear what happened last year.

This decision could have a bigger impact on the October and August born. TGS will need to have a ready made answer by allocation day.

I am not a statistics expert but from my understanding

1) if you standardise, what split do you choose, 12 months or 6 sets of 2 months etc. if you ran both scenarios and the results were different then you would have a problem justifying your rules. Easier to choose afterwards
2) see (1)
3) shouldn't really mix and match, it suggests result are fixed
4) I think in past only round 1 was standardised

The problem is you have a truncated set, only have the top 500 sitting (or whatever exact number was), unlike first set (of about 1700 candidates), now only have about 30%, if you plot a normal distribution then take 30% at one end you have a strange shape distribution (so any techniques based on normal distributions aren't valid).
Further you get into small sets, lets say 500 candidates gives 12 monthly sets of about 42 (and as births aren't even through the year some sets might be nearer only 30). Then to standardise need to get rid of freak results so lets say you drop top 5 and bottom 5 from each set, and use average of remainder. Almost certainly now have some sets under 30 and that means cannot apply certain regression techniques without risk of unreliable results. So what happens next, fewer bigger sets, but who chooses the size, or one set of all. Hope you start to see the problem.

Is the Maths and English weighted that way? roughly 1700 sat the first round?

[sharvari.bandal]

My DD said Maths was easy for her still she left 2 questions at the end but I am quite sure whatever she has done will have doo accuracy.
I am more worried about her English as comprehension and creative writing is not her strong point and liking.
Will come to know on 1st March.

[tiffinboys]

Aren't you underestimating your DD? She hasn't disappointed you in any of the several tests that she sat for. Now relax until 1st March. Good Luck.