Cut-off scores

[tiffinboys]

Thanks to informeria. She/he has put the following tool in Berks, probably nfer based.

Change the number of applicants to whatever you think and get an idea of cut-off.

http://kendrickcalculator.weebly.com/" onclick="window.open(this.href);return false;

Slight caution: Nfer works to max score of 140; this table goes a little beyond. But it would give fairly close result.

[ClappedOutMum]

Thanks to informeria. She/he has put the following tool in Berks, probably nfer based.

Change the number of applicants to whatever you think and get an idea of cut-off.

http://kendrickcalculator.weebly.com/" onclick="window.open(this.href);return false;

Slight caution: Nfer works to max score of 140; this table goes a little beyond. But it would give fairly close result.

Leaving aside the caution of the Max of 140, I crunched some numbers to see if (from past few years) we can extrapolate what the likely cut-off will be this year for Tiffin with 2000 applicants.

As far as I can see if you take the number of Applicants and Initial Qualifying Score you can see where in the ranking out of the number of applicants the last person was.

So, for example, in 2013 1644 candidates, initial cut-off score of 231 (divide by two to use the Kendrick calculator to get 115.5), meant that a person of score 115.5 was at 15% (approx) ie 250th in rank. Meaning that 110 higher scoring candidates had Tiffin not as first successful preference. A percentage of 44% preferred elsewhere.

Doing the same for other recent years I got

2013: 1644: 231: 15%: 250: 44%
2012: 1685: 231: 15%: 250: 44%
2011: 1367: 228: 17%: 240: 41%
2010: 1314: 228: 17.5%: 230: 39%

So each year approximately 39 - 44% preferred elsewhere. [I know that in past years there were more offers than the PAN made initially due to tied-scores, but I have ignored this.]

And with a new PAN of 150, assuming say 44% percentage will prefer elsewhere for 2014, you'd need to be in the top 340 this year to get an initial offer.

A predicted initial cut-off of 231.

So - in conclusion - I know a lot of assumptions begin made in this - the increase of 350 in number of applicants is just about offset by the increase of 10 in PAN.

Happy to have this critiqued by statistician or anyone else for that matters.

[sys4cad]

We should consider other factor such as OOC:InCatchment if school applies catchment area. ( i.e. Nonsuch and Wallington girls )
However, we have historical cut-off scores as reference.
So it is still possible to make a rough estimation at least.

[tiffinboys]

Also worth noting is that 2000 is the numbers registered for the test. The number actually who takes the test is some times much lower. I am told by some one usually reliable that last years number of candidates registered for the test was 1800+. however, we know that only 1644 chose to sit for these.

This year, Tiffin is the last test. So can't be the free mock (except may be for indies). We have to wait & see, how many appears for these tests.

About the increase in PAN, actually there is no material change. As Tiffin School always admitted all at the tied score, the actual numbers admitted were around 148-150. From this year, all at the tied score would not be admitted. Distance will count for the cut-off score.

[ClappedOutMum]

Thanks, TiffinBoys, since you are our resident Tiffin expert - do you then not think there will much of a difference in cut-off this year?

[tiffinboys]

It will be quite clear as soon as the results come out i.e. around 21 Oct.

[ClappedOutMum]

How is that so? Not being cheeky, just don't understand! Until forms are filled and preferences expressed how will the school know what the cut-off will be?

[tiffinboys]

Hopefully results will tell the score and the number of candidates, then perhaps we can guesstimate likely cut-off score. Parents can then work out the likely outcomes. What we will not know is the preference pattern.

That Schools will know in November and we will have to wait till 3 March.

[One_Shot]

Happy to have this critiqued by statistician or anyone else for that matters.

Disclaimer: Not a statistician.

I think there is a minor flaw in the argument. 115% might represent 1 sigma deviation and top 250 for a single test but when there are two tests and if a student gets 115% in both, then the rank is likely to be higher that 250. Something to do with correlation and covariance or some such thing. if I recall from days long gone by, the means of two standardised sets can be added but not their std deviations.

So, I think that the rank is likely to be somewhere between 150 and 250 depending on how strongly correlated the performance in the two sets are.

[One_Shot]

BTW, this 100 as mean and 15 as std deviation probably does not hold good.

There are some great posts by "Tree" which are very interesting reads.

http://www.elevenplusexams.co.uk/forum/ ... 12&t=18108" onclick="window.open(this.href);return false;
http://www.elevenplusexams.co.uk/forum/ ... &hilit=exp" onclick="window.open(this.href);return false;