Supporting Families
  • banner2
  • banner10
  • banner13
  • banner11
  • banner7
  • banner8
  • banner12
  • banner4
  • banner6
  • banner3
Friday, 15 January 2021 18:32

Kent Test Results by Birth Month 2018-20: Sharp Decline through the Year

Children born in the first quarter of the school year 1 September 2009 and 31 August 2010, have performed much better in the 2020 Kent selection procedure than those born in the fourth quarter between June and August.

It is some years since I previously analysed Kent Test results by the month of birth of children sitting the Test and found little difference at that time between performance across different ages. Given the built-in disadvantages for some children brought about by the Coronavirus crisis this year, the decline in the pass rate was no great surprise, except that the difference was almost the same in 2018. The reason for the fall in performance is therefore not to do with Coronavirus as I initially suspected, but appears to be caused by inherent problems with the Kent Test age standardisation, which is surely neither fair nor acceptable.

The differences in performance by birth quartiles across the school year are illustrated in the table below, a Kent selective assessment being made up of a combination of Test performance (18% of the Year Six cohort pass the Kent Test on average) and Headteacher Assessment (a further 6% found selective). 

Kent Grammar Assessment by Birth Month
2020 2019 2018
Took Test
% of Roll
Took Test
% of Roll
Took Test
% of Roll
Sep -Nov 46% 27% 52% 26% 51% 26%
Dec-Feb 47% 25% 48% 24% 51% 25%
Mar-May 44% 23% 48% 25% 49% 22%
Jun-Aug 40% 22% 45% 24% 46% 21%
To quote GL Assessment the company which sets the Kent Test:
Age-standardised score: An age standardised score converts a pupil’s raw score to a standardised score which takes into account the pupil’s age in years and months and gives an indication of how the pupil is performing relative to a national sample of pupils of the same age. The average score is 100. A higher score is above average and a lower score is below average. The SAS is key to benchmarking and tracking progress and is the fairest way to compare performance of different pupils within a year group or across a year group.

 In other words, if similar samples of children born in each month take the Kent Test, they should produce a similar proportion of passes. A complication arises in the Kent test because the younger the child the less likely they are to take the Kent Test (an average of 6% between the first and fourth quartiles across the three years), but it is reasonable to assume that most of the difference comes from children who would not have passed in the first place.

The Bottom Line
In October 2020, 1214 pupils were found selective in the first age quartile of 4557 Kent state school pupils, 117 more than would be expected, from the county average of 1093.  995 of the 4578 pupils in the fourth quartile were found selective, 98 fewer than one should expect. It is therefore evident that for some reason around 100 of the youngest children are discriminated against in the Kent Selection Process. There is a similar outcome for 2018.

There was an opportunity to rectify this through the further process of Headteacher Assessment which in practice identifies on average an additional 6% of the Year Six cohort as suitable for grammar school. This is done through the scrutiny of past work, performance in assessments over the previous year, scores in the Kent Test and a recommendation by the primary headteacher including reference to special circumstances. Whilst this would have been an opportunity to correct the age imbalance between first and fourth quartile, in practice, it has actually increased it as shown in the table below, with a similar pattern in 2018.

Kent Grammar Assessment by Birth Month
Took Test
% of Roll
Kent Test
Sep -Nov 46% 19.2% 7.4%
Dec-Feb 47% 19.8% 5.3%
Mar-May 44% 18.5% 4.7%
Jun-Aug 40% 16.2% 5.5%
 What Can/Should be done to correct  Age Discrimination?
       School Appeals 2021
This is the only area where 2020 outcomes can be challenged. Only for pupils with test marks close to a pass, parents may wish to use the data provided to support their case, especially as with two lockdowns many children from state schools will have scant academic evidence from their primary school to support their case.  Appeal panellists and primary headteachers writing references also need to be made aware of the issue and I know of a number from both categories who follow this site regularly.
Headteacher Assessment
For entry in 2021 onwards, HTA Panels also need to be made aware of the discrimination in coming to their decisions, and primary schools need to be aware of this when putting forward candidates to be considered.
Age standardisation
I believe it is clear that the age standardisation is flawed. It may be a coincidence that KCC uses the words: ‘A slight adjustment is made to take account of each child's age so that the youngest are not at a disadvantage’. It is, therefore, my view that KCC needs to go back to GL Assessment which sets the Kent Test and devises the age standardisation tables, and ask for them to be revised for future years, making the adjustment less slight.   
Data is derived from an FOI Request to KCC, published on the What Do They Know website. 
Last modified on Monday, 18 January 2021 20:34

1 comment

  • Comment Link Sunday, 17 January 2021 22:05 posted by Sorrel

    The 2019 data shows very little difference between the groups though, with the youngest quartile doing as well as the second. Is 2017 data available? The tests do vary a bit from year to year (inc in 2019 the English paper needed a much higher raw score than the maths for a top score, as many did well on it). PETER: This is why I chose to contrast Quartiles One and Four. As noted at the foot of the article, I obtained the data through a third party FOI which did not include 2017.However, the requester has just asked for 2014-2017, so I will amend the data when this is published.

Leave a comment

Make sure you enter the (*) required information where indicated.
Basic HTML code is allowed.