Sample size for PeerCompare (Greater London)

[CestMoi]

We started doing CEM prep and my child seemed to be getting on well, scoring 13,13,14 in the first three comprehension tests. I used PeerCompare (which seemed buggy, at least from an iPad) and learned she was 47% percentile rising to 75% percentile, definitely nowhere near the standard required.

Is it possible to get sample sizes on the report?

I'm a little disheartened as this suggests that 100% is needed across the board.

Thoughts welcome.

[Guest55]

Surely this comparison is to those taking the online tests and not the whole cohort sitting the 'real thing'?

I would think the 'peers' are not typical ...

[PurpleDuck]

If that's of any help, we went through a few FPTP books and I registered for the 'peer compare' but ended up never using the facility, it just seemed to much hassle. There are probably a lot of other people not using it - none of the mums I know did - so I wouldn't rely on it as a representative sample. The sample size is likely to increase closer to the real tests, when children have worked through more of the tests from the books but we don't know whether parents of children with lower scores are as likely to use 'peer compare' as those who score very highly and if they don't, the 'peer compare' sample would not be representative, whatever its size.

Another thing is that a number of people may have used it in a speculative manner, i.e. entered theoretical scores just to see how high their children would have to score in a test to be in the top 10% of the cohort. It may sound a silly thing to go, but I did it myself in order to have a benchmark, so the theoretical score I entered (which had nothing to do with reality) would have artificially skewed the average. Thinking about it now, if other people did a similar thing, that benchmark was pretty much meaningless. In short - don't worry about 'peer compare' too much; for comparison against a cohort, mock tests are by far more useful.

[CestMoi]

Thanks for your responses. I think you have both hit the nail on the head in that people self select to enter results, which skews the numbers in the way suggested. I have now tried it on a range of papers and books, and it is the same on all - only 100% is enough to be in the "right centile".

So either the FPTT books are too easy for CEM or the people entering results are only bothering to enter high scores.

It's a shame as it could be a great resource if used..