Score standardisation

[Tinkers]

By only comparing the children born in the same month, it's actually fairer.

If they just had more marks added the younger they were, how do you decide how many marks to add?

By comparing each month, you can see the effect of age on the scores. If there's very little effect then younger ones don't 'need marks added'. If there's a big difference then they 'need more marks added'. I've seen scores this year from one area (can't remember which) where there is very little difference in the maths score over the age range but a fair bit for English. The standardised scores take this into account. Simply just adding marks wouldn't.

[KenR]

"Standardised by age" means that they are only compared to others of the same age. Therefore if, for example, the children born in March all score very highly, the raw score for 121 might be higher for that group. See the posts from yesterday where it is explained (several times).

In theory yes but in practice due to the statistics of a large cohort on average older children always perform better than younger candidates. That's the whole rationale for Age Standarisation. Following the statistical Age Standarisation process by Durham CEM an Age Standardisation Table will be produced to convert raw scores to age standardised scores for each "block" of questions. (this could be by section type or paper - the method depends on the approach selected by the particular LEA - Bucks is by weighed question type I recall?). As stated previously the columns in the table relate to children's ages within 1 month. However it is incorrect that the top 30% of each monthly age group are selected - viz. that each candidate is only competing against children of the same age group on the date of the exam.

I've seen some actual Durham CEM Age Standardisation tables for some older Birmingham Consortium exams an the standardised age adjustments between oldest and youngest candidates can be quite significant.

For example, in the Birmingham KE VR Tests in 2004 , to achieve a Standardised Score of 118 a child aged 11.18 years on the exam day would have to achieve score 71/100, whereas a child who was only 10.22 years would only have to score 65/100. Quite a difference.

Note: 3 x 118 was the minimum pass mark required to get into KECHB that particular year

[streathammum]

Thanks KenR. This looks more like what I would have expected - ie that many tests over many years are evaluated to assess average differences between children of different ages, and that this is used to derive a general standardisation table. This is then applied in a linear fashion across all the children taking the test.

If the average standardisation is correct, this should mean that a broadly even spread of birth months passes each year, but still allows for occasional bunching of stronger or weaker candidates in individual months.

[Guest55]

It will not necessarily be linear ...

It still doesn't help premature births ... they are the group that really miss out.

[anotherdad]

It still doesn't help premature births ... they are the group that really miss out.

What are you saying?! Coming to the forum soon..."My expectant wife has hypertension and the doctors want to induce her. Is it worth her holding on for another few weeks to improve our chances of a securing a place at AGS in eleven years' time? My unborn son will get marks added that way, won't he?"

[streathammum]

Yes Guest55, not necessarily linear - but surely under the method KenR outlines, there would be no possibility of younger children having to get higher raw scores than older children in order to get the same standardised score? Which is not the case under the method that ScaryMum and others have said happens in Bucks.

Agree re premature children.

[Guest55]

This scenario:

Baby due November - born in August 12 weeks premature - the child takes the Selection Test a whole year earlier than they would have had they been born even 10 weeks early.

I think the truth about standardisation is hard to find and, in theory, I've seen data that implies younger children may have to score more highly if their birth month cohort are particularly able. I know of Primary school data [in RAISEonline reports] that show some summer born cohorts do sometimes out perform autumn borns now they are all at school for the same amount of time.

[Lillie]

There was a report in the FT (albeit a few years ago), which showed that without age standardisation, grammar schools were disproportionately filled with Autumn born children.


With the CEM test, no one knows the facts about how the standardisation process works, though I think it would be highly unlikely the cohort of July or August born children would have raw scores higher than the January cohort, for example. Of course, individual children would score much more widely in each group.

In terms of school years, although the summer borns might start formal school at the same time as the Autumn born, they will have up to a year less in a nursery environment. My own summer born DC only had 1 year of pre-school before she had to start school. Her Autumn born friends (some less than two months older) had another year of nursery. All the slow learning through play which underpins formal education takes place there and it seemed such a rush to start school at that time.

[Aethel]

Guest55, you could say that about any child.

My dd2 was born in the last week of August, she'll take the test at 10 years plus two weeks. It doesn't matter if they are premature, on time, or overdue, they are still young for the school year. So comparing just the August children with each other means no-one "misses out".

DD1 (Early Sept) took it at 11 years and a few days. So she was compared with other older-for-year Sept children.

Remember how unfair any cohort exam is by definition.

I explained it to the children this way: if you take this exam with 99 Einsteins, you will not pass it. But neither will 69 of the Einsteins. Only the top 30 Einsteins will pass.
Which is daft, but that's the way the marking works.

If you take this exam with 99 incredibly stupid ogres, you will pass it.
But so will 29 incredibly stupid ogres! And the 70 thickest ogres will not pass.
Just to make it more unfair, about 10 of the ogres who don't make it were just unlucky on the day with the questions and really they should have got a good enough mark to pass.

[Guest55]

Guest55, you could say that about any child.

So are you claiming that being 12 weeks premature is no obstacle? The child should be 9 years and 9 months and in Year 5.