CEM tears and fears

[ginx]

Okanagan, I'm shocked by that. So if my dd2 had been born before Dec, she wouldn't have got a place at gs. I suppose she was lucky to be born in May. 8 points seems a lot, even just as an indication.

Do more summer born dc go to gs? Perhaps I'd better ask dd2 to do survey of birthdays; there have been five so far in her class.

[FelixJ]

what about age discrimination? How would that stand in an appeal?

[Bob1892]

Okanagan, I'm shocked by that. So if my dd2 had been born before Dec, she wouldn't have got a place at gs. I suppose she was lucky to be born in May. 8 points seems a lot, even just as an indication.

Do more summer born dc go to gs? Perhaps I'd better ask dd2 to do survey of birthdays; there have been five so far in her class.

I read somewhere that autumn born dc tend to gain more gs places than summer born in a Cem Test because of the higher level of maturity required to tackle a cem test. I hope I'm proved wrong for my lad.

[paperino]

The way that raw scores are transformed into standardised scores puzzles me a lot. There are in particular two points. First the standardisation is completely endogenous and could be manipulated. Let me give an example, let us say I have a child born in October, if I want to give an advantage to my child I can persuade lots of "low academically able" October born children to take the test. This will lower the average performance of the October court and therefore they will get a bonus. Second, from what I understand from a previous post, in order to calculate the "bonus" by month of birth, the raw scores of children taking the test are averaged by month, and the differences in the monthly averages are given as "bonus". But what if these averages are not statistically different from each others, is any test carried out? The standard deviation may be enormous.
I am not a statistician but I believe that a real one would be quite horrified by this procedure.
Does anybody have any information of alternative standardization methods ?

[FelixJ]

Yes you are right there in theory. Blimey, but I guess it probably wouldn't happen that way.

I must say when DD sat 11 plus there were dc in tears at the end, was a real shame to see

[paperino]

Maybe it doesn't happen this way, but it could.
I believe that someone could easily base the appeal on this (quite rightly).
There are obviously differences between someone born at the beginning of the school year and someone towards the summer. But at the same time there are also some well known studies on gender bias, for example, showing that boys develop later than girls, or social bias, showing that children from middle class background develop earlier than children from poor background, and so on.
My point is that the age standardisation correction, as it is know, appears quite an arbitrary exercise. For example looking at the latest available figure (2012), between October and November there is a difference of 2 points, but between November and December of zero points. So how can CEM seriously claim that a there is quite a big difference between two children born one in October and one in November, but no difference at all between two children one born in November and the other in December?
Maybe only an astrologist could explain this!!!!

[Brum Mum]

I believe (and Okanagan will provide the definitive explanation ) that age standardarisation isn't standard each year - its based on the cohort that took the test that year and then broken down by those in each age group within that. I think (hope) I have that right

There are many studies which show a marked difference in the maturity / learning ability of those who are September born and those that are August born...there are lots of threads on here discussing it as you can imagine! I don't know how this compares in relation to gender bias or background bias though.

[Brum Mum]

Okanagan, I'm shocked by that. So if my dd2 had been born before Dec, she wouldn't have got a place at gs. I suppose she was lucky to be born in May. 8 points seems a lot, even just as an indication.

Do more summer born dc go to gs? Perhaps I'd better ask dd2 to do survey of birthdays; there have been five so far in her class.

Ginx - Littleginx would have been older, had more development time etc - I don't think it works like that.

My understanding is that older i.e. September / October borns are more likely to get places (hence standardisation?).

I'm sure Okanagan will be along shortly to put us right!

[Okanagan]

For example looking at the latest available figure (2012), between October and November there is a difference of 2 points, but between November and December of zero points. So how can CEM seriously claim that a there is quite a big difference between two children born one in October and one in November, but no difference at all between two children one born in November and the other in December?

It doesn't quite work like that. The figures I gave (because I don't have a breakdown) were the overall effect for one particular combination of scores across the three subjects. It ISN'T a blanket adjustment applied to all children born in those months, and it might have looked a bit different with a different combination of scores, even for the same overall total.

Each subject area is standardised separately, and the overall total comes from adding the three together. What you see in the aggregate is the cumulative effect of that - in this case 11 points for 11 months. But that might not have come equally from all three sections (in fact it almost certainly doesn't). Clearly if there was an adjustment every month for each subject area the minimum adjustment would be 33 points - so there almost has to be a zero in there somewhere for any age adjustment to be a realistic reflection of the difference in scores between children, who are, although there ages might vary slightly, all at the same point in terms of their educational journey, so have at least in theory covered the same curriculum!

To get a better idea have a look at full age standardisation table - e.g. here are some from the 2010 KS2 SATS - and you'll see that for a particular raw score there isn't an adjustment for every month. It's not a case of CEM claiming anything which is any different to ay other age adjusted educational scoring. Choose two different raw scores and work across a table changing the actual figures to a +0 or +1 relative to the month before and it should be obvious that although there's a general trend it doesn't mean that there is an adjustment for every additional month, or even the same month on month adjustment for two different scores.

[MSD]

I might have some more stats on this. My son is one of the older boys in the class being a middle of September birth. I requested his detailed result last year quoting the FOI act.

Here it is:

Total standardised score : 377
Weighted standardised score : 371

So I can only presume that the reduction of 6 standardised points was down to his age. Unless, this weighted score had nothing to do with age but the actual weightage based on each subject area.

MSD