I was using the 3rd language (French/German...) to determine this year's top 10% (which can be closely correlated to the lower tier IP Schools COP) earlier. This year 3rd language was offered to 253+0 but not 251+0.

I am not sure exactly where the cut off is as I know one 252+0 wasn't offered but somewhere in the forum someone feedback that 252+2 was offered. Possibly the cut off is above 252 but lesser than 253

Looking at the current IP COP for CHS & TJC (http://www.kiasuparents.com/kiasu/forum ... 48&t=84817), they are both 253. So using the 3rd language to gauge the top 10% is quite accurate.

And the top 10% t-score is never the same every year due to the distribution shape changes, depending on the difficulties of the papers. T-score is merely trying to force the distribution to be a normal bell curve but in reality, the curve can be skewed, causing changes to the t-score. It can be bloated a bit at the tails or taper off thinner from year to year. This will cause the top whatever % t-score to change.

Eligibility to choose IP stream is not indicated, only Express/NA/NT stream is stated.


It is much easier if they indicate on the eligibility letter if one is qualified for IP. Save all the uncertainty & guess work as the score fluctuates from year to year.

The 3rd language option is used to gauge if the score falls within the top10%. Similar for those who are offered EESIS where it is used to gauge the top3% to enter the top tier schools.


It was indicated that IP is offered to the top 10% of the cohort. So that’s the correlation.

From what I gathered in the forum, 3rd language (French/German…) offered for 2013 = 251; 2012 = 249


2012 score didn’t tally with the lower tier IP schools COP (252)??? I am not sure if 249 is correct or there wasn’t sufficient space to accommodate all IP students in that year?

GreenA:

Hi! parents,


I have some questions regarding T-score:

If I understand correctly, T-score is statistic value. It is assumed that the distribution of pupils score is a Gaussian Normal Distribution.

The formula is:

Subject score = 50 + 10 * (Raw score - Mean) / Standard-Deviation

We can call the part (Raw score - Mean) / Standard-deviation Sigma, which is defined in statistic.

So that we can write as follow:

Subject score = 50 + 10 * Sigma

T-score = summation of Subject Score of all four subjects

Therefore, suppose a pupil's T score is 260. His average subject score is 260 / 4 = 65

Sigma value is (65 - 50)/10 = 1.5

Check the Standard Normal Distribution table, area below sigma 1.5 is
0.5 + 0.4332 = 0.9332

That means: a child with T-score of 260 stands at 93.32% of his cohort.

Each and every year, regardless of Mean and Standard Deviation, a child of 260 T-score stands at same place of 93.32%.

However, from this year's discussion, I come to understand that the EESIS was offered to top 3% of pupil. And each year, the cut off point for EESIS is different:

2013: 258
2012: 263
2011: 261
2010: 260

if the the cut off point of EESIS does points to top 3%, then it is contrary to my above calculation.

Does this means the PSLE result is actually not a normal distribution? Or is there anything wrong with my interpretation of the T-score.

Awaiting for comments. Numbers do not add up. It is really puzzling.

Due to the skewed nature of the T-score distribution being forced into a normal-distribution, the actual top 3% T-score < statistical 93.32% (= top 6.68%) already indicates that there are lesser students at this end of the distribution.