Scaled score assessments

[DebbieFiderer]

Sorry if that's not what they are actually called, I'm not sure what the proper term is.

My daughter's school do assessments with the kids every term for reading and maths. The scores seem to be done in a similar way to the KS2 Sats, with 100 being the baseline, and 115 being exceeding expectations (or something similar, I can't remember their exact term). Can anyone give me more information about what these tests might be and how the scores are calculated? From something the teacher said I'm curious to know how the raw scores get converted to the scaled score (again, sorry if that's not the right term). She is Y5 in an English state school if it makes a difference.

[Redlocks28]

Do you mean standardised scores?

[DebbieFiderer]

Possibly, I don't really know what they are called.

[spanieleyes]

Could be anything, there are lots of commercial tests that provide scaled scores. The companies that produce them all claim that they are standardised in line with SATS or similar. So , in effect, they take someone who achieves 100 in SATS, gets them to sit their own test and the score they achieve is given a standardised score of 100 too( not quite how it works as usually they use thousands of children in their samples bu the principal is the same!). The actual score needed to achieve a scaled core of 100 can vary, as it does in the real SATS, depending on how easy the cohort find the test. If it's an easy test you will need to score higher to achieve the same scaled score that you would if the test was a difficult one.

[spanieleyes]

Mixed up scaled and standardised there! The scaled score is the score marked against te expected standard, the standardised score is set against cohort averages!

[XenakisCarter]

I don’t know the maths behind how they work it out but 90-110 is average so some will be higher, some lower. 120 is a good score, mid-120s very good but, as we live in a grammar area, my DC need to average at least 130-135 to get in.

[pastapestoparmesan]

@XenakisCarter

I don’t know the maths behind how they work it out but 90-110 is average so some will be higher, some lower. 120 is a good score, mid-120s very good but, as we live in a grammar area, my DC need to average at least 130-135 to get in.

Those tests are not Y6 SATs. The highest score possible is 120.

[DominaShantotto]

I'd just ask the school what the name of the test scheme is - then you can research it from there. There are that many different assessments with different methods of scoring things that you'd go bonkers trying to work it out from a vague description.

[Reader1984]

100 is the average of that year's children nationally. The overall score for the Reading, SPAG and Maths SATs will therefore differ each year. But it combines these 3 tests and then the average is 100. 110+ and your child will be considered 'higher band' in secondary.

[DebbieFiderer]

Thanks all. I guess I would need to know the exact test to be able to answer my question, the teacher (and actually last year's teacher too) made a throwaway comment related to how the score gets worked out which just made me curious to find out more. I'm a bit of a data geek so like to understand the underlying numbers for things like this. Does anyone know anything about how the raw scores get converted in tests like this? It it a direct correlation, ie full marks on the test = 130, dropping down the scale for each mark lost, or is it more complicated?

[onemouseplace]

Our school does Pira and Puma tests every term - could be those. I have no idea on how the scores work though as the school only uses the results for internal purposes and don't give parents any more information other than whether their child is working towards, achieving or exceeding expectations.

[Redlocks28]

What did the teacher say that made you curious?

I don’t think anyone here will be able to tell you anything useful about the scoring/marking mechanism without knowing exactly which tests they are.

[Yellowmellow2]

It’s probably NFER or PIRA/PUMA

[Tiredforfive45]

The tests all come with conversion tables.

If you google sats conversion tables you can get an idea of how they work. Some raw scores have the same standardised score. So 34 and 35 out of 50 could both have a S.S of 101 (for example).

[Karwomannghia]

Raw scores get converted using a table where you look down the side for the child’s age and across the top for the raw score and look up the standard (or scaled) score and it gives you a score that tells you how a child of that age scores compares to children of the same age.

Standard scores have the average at 100 (anything ranging between 90-110) falls into the same average bracket. Percentiles link to standard scores and 100 on standard score is at the 50th percentile.

Scaled scores are often from sub tests which have an average of 10 but are done on the same principle as standard scores.

The reason they are there is because a 5 year old and a 10 year old both getting, say, a raw score of 25 doesn’t tell you much about the achievement of either. It could be that the older child has learning difficulties or that the younger one is highly intelligent. A standard score tells you how they’re getting on in comparison to peers.

Standard scores are calculated when the test is produced by giving the test to a huge sample of children and working out the distribution of scores for each age range. These norms have to be updated every so often to reflect changes in the population.

Standard scores are also helpful to look at progress over time for a single child, so a child attaining a standard score of say 95 every year means they’re progressing at the expected rate; their raw scores would however be increasing each year.

[DebbieFiderer]

Is there a maximum and a minimum standard score? And would that maximum/minimum be the same for all children in the year group or would it differ according to birthday?

[Karwomannghia]

Yes there is a minimum and maximum standard score and it would be the same curve for all year groups. Standard scores are just a way of showing how scores are distributed, so the huge bulk are in the middle and a few get higher or really high or low or really low. If a child got full marks in raw scores they would get different standard scores depending on age. So if full marks is average for a 10 year old they’d get a standard score of 100 even though they’d got everything right. This is called a ceiling. They may have potential to get higher but the test doesn’t have the scope to show it.

But children of different ages would get different raw scores to get the same standard score. So a 10 yo with a standard score of 110 is at the same level in their age group as a 5 yo with a standard score of 110. They’re both high average. But they will have got different raw scores.

Different tests will have different standard scores though, some come across as ‘easier’. It depends on when they normed the test, how many they did it on and the background of the children.

Y6 SATs standard scores are confusing because 100 is ‘expected standard’ so there’s like a value judgement on there and they’ll say 60% achieved expected standard etc whereas with proper statistical standard scores there is average, above average etc and it’s just a reflection of how scores are and there will always be 50% above and 50% below.

[spanieleyes]

thirdspacelearning.com/blog/sats-scores-explained-ks2-ks1/

has a good explanation of SATS scores. These aren't affected by age. But different tests will have different conversion rates/scores. Some use age weighted scores, some don't, some have both! You really need to know the tests used!

[Karwomannghia]

SATs results could just be raw scores really but because they’re out of different amounts eg SPAG is out of 35 and maths 50 or whatever instead of them saying 20/35 is expected they’re converted them all to be above or below 100 so one point on the SPAG is about 3 ‘standard score’ points. Nothing to do with age.

[Karwomannghia]

This might be useful as well www.gl-education.com/media/1757/guide-to-standardised-tests.pdf