All About T-Score

[CrazyDad]

NYCWTC:

CrazyDad:

Hi NYCWTC,

Are you from MOE?
The reason to start this topic is to lure the real MOE dude to put in the REAL Mean and Std Dev. Because whatever we put out here are just some guided guesstimation. Wonder why MOE cannot release these information.

I have been searching for sometime but get nothing... :?

No, I am not from MOE. But it really just seems impossible for the average for Chinese to be 78...and the others too.

This number actually make sense.
See there are 81% of students scored 75 and above, and 2% of student failed. If you fit this into Gaussian, then the mean will be 70. But this also means 50% of student will be more than 70 and this contradict the actual statistic (81% scored more than 75 marks). So now you have to use Skewed Gaussian, but by how much. So take one point at a time and perform the \"H test\". Now can apply Poisson Distribution with lambda to the power of one over two.
Still we need more real data.

[CrazyDad]

Now that you have estimated T score, it is about time to look for a school that fits this. Please go to this site http://app.sis.moe.gov.sg/schinfo/SIS_AdvSearch.asp, just key in the score in the last box, a list of suitable school will be shown

[justsay]

[quote]So now you have to use Skewed Gaussian, but by how much. [/quote]

Hi crazydad , can you help this suaku parent to put it in more layman's term. :oops:

Do you mean first T score formula shall apply using their raw score, then Skewed Gaussian will apply to moderate the T score further if the whole cohort don't do well, the Skewed Gaussian is used to push up the marks of the whole cohort ?

Thank you.

[P6boy-dad]

justsay:

[quote]So now you have to use Skewed Gaussian, but by how much.

Hi crazydad , can you help this suaku parent to put it in more layman's term. :oops:

Do you mean first T score formula shall apply using their raw score, then Skewed Gaussian will apply to moderate the T score further if the whole cohort don't do well, the Skewed Gaussian is used to push up the marks of the whole cohort ?

Thank you.[/quote]Statistically, the mean score and the standard deviation are already a moderating tools. What i can say about the T-Score for Chinese (with 80% scoring A and A*) is that, the mean score is VERY high at least 82 and the SD is about 7-8 marks. It means that your child has to score above 82 before he can obtain a T-score of 50 and above. Any marks below the mean score would result in score below 50 for Chinese. I am sure alot of parents dont realise that, thinking that as long as my child score 75 marks would be sufficient. That was why a student who scored 3A* and 1A (chinese) would get a score of 244. The A for Chinese would have given him a T-score of 41 if he scores only 75 marks.

[CrazyDad]

[quote]Hi All,

Base on the feedback from all parents and using the result from 2007. 2007 PSLE is pretty similar to 2009; very difficult math, easy Chinese and some tricky Science questions.

I have estimated the parameter to calculate the T score as followed;
CL: Mean = 78, StdDev = 10
EL: Mean = 73, StdDev = 11
Math : Mean = 63, StdDev = 9
Sci : Mean = 67, StdDev = 12

Use the formula : 50 - 10 * (your score - Mean)/StdDev
for each subject. Sum them up, you should get a glimps.


:xedfingers:[/quote]For EL and CL, need to divide the total score by 2

Shouldn't the mean for Eng and MT be around 100+ since the total score is 200 respectively?

[CrazyDad]

justsay:

[quote]So now you have to use Skewed Gaussian, but by how much.

Hi crazydad , can you help this suaku parent to put it in more layman's term. :oops:

Do you mean first T score formula shall apply using their raw score, then Skewed Gaussian will apply to moderate the T score further if the whole cohort don't do well, the Skewed Gaussian is used to push up the marks of the whole cohort ?

Thank you.[/quote]Layman Term as followed;

For example, base on CL raw score in 2007, 2% of students has less than 50 mark and 81% of student has more than 75 mark.

if the cohort has 50,000 student, then 1000 student failed and 40,000 student score 75 mark or more. So average point should be close to 80 point right?
And everybody score are around 80, says from 75 to 90 mark. so the spread will be very narrow.

[CrazyDad]

P6boy-dad:

justsay:

[quote]So now you have to use Skewed Gaussian, but by how much.

Hi crazydad , can you help this suaku parent to put it in more layman's term. :oops:

Do you mean first T score formula shall apply using their raw score, then Skewed Gaussian will apply to moderate the T score further if the whole cohort don't do well, the Skewed Gaussian is used to push up the marks of the whole cohort ?

Thank you.

Statistically, the mean score and the standard deviation are already a moderating tools. What i can say about the T-Score for Chinese (with 80% scoring A and A*) is that, the mean score is VERY high at least 82 and the SD is about 7-8 marks. It means that your child has to score above 82 before he can obtain a T-score of 50 and above. Any marks below the mean score would result in score below 50 for Chinese. I am sure alot of parents dont realise that, thinking that as long as my child score 75 marks would be sufficient. That was why a student who scored 3A* and 1A (chinese) would get a score of 244. The A for Chinese would have given him a T-score of 41 if he scores only 75 marks.[/quote]totally agree

[justsay]

CrazyDad:

justsay:

[quote]So now you have to use Skewed Gaussian, but by how much.

Hi crazydad , can you help this suaku parent to put it in more layman's term. :oops:

Do you mean first T score formula shall apply using their raw score, then Skewed Gaussian will apply to moderate the T score further if the whole cohort don't do well, the Skewed Gaussian is used to push up the marks of the whole cohort ?

Thank you.

Layman Term as followed;

For example, base on CL raw score in 2007, 2% of students has less than 50 mark and 81% of student has more than 75 mark.

if the cohort has 50,000 student, then 1000 student failed and 40,000 student score 75 mark or more. So average point should be close to 80 point right?
And everybody score are around 80, says from 75 to 90 mark. so the spread will be very narrow.[/quote]Thanks for the explanation. What I actually want to know more is the Skewed Gaussian. Thanks.... :celebrate:

[CrazyDad]

Vikaesh:

CrazyDad:

Hi All,

Base on the feedback from all parents and using the result from 2007. 2007 PSLE is pretty similar to 2009; very difficult math, easy Chinese and some tricky Science questions.

I have estimated the parameter to calculate the T score as followed;
CL: Mean = 78, StdDev = 10
EL: Mean = 73, StdDev = 11
Math : Mean = 63, StdDev = 9
Sci : Mean = 67, StdDev = 12

Use the formula : 50 - 10 * (your score - Mean)/StdDev
for each subject. Sum them up, you should get a glimps.

How close will our PSLE results be to the score we get using yr estimations [e.g. plus OR minus 3]





:xedfingers:

:roll: Only time can tell, there are too many unknowns;
1. the actual scores are unknown
2. the parameters estimated base on even more unknown

get the picture?

[CrazyDad]

justsay:

CrazyDad:

[quote=\"justsay\"]

Hi crazydad , can you help this suaku parent to put it in more layman's term. :oops:

Do you mean first T score formula shall apply using their raw score, then Skewed Gaussian will apply to moderate the T score further if the whole cohort don't do well, the Skewed Gaussian is used to push up the marks of the whole cohort ?

Thank you.

Layman Term as followed;

For example, base on CL raw score in 2007, 2% of students has less than 50 mark and 81% of student has more than 75 mark.

if the cohort has 50,000 student, then 1000 student failed and 40,000 student score 75 mark or more. So average point should be close to 80 point right?
And everybody score are around 80, says from 75 to 90 mark. so the spread will be very narrow.

Thanks for the explanation. What I actually want to know more is the Skewed Gaussian. Thanks.... :celebrate:[/quote] Long story...

can check out Wikipedia for the glory details though...