Standardization

[OLA]

Does anyone know how standardization is done?

Is age always against older children?

[SunlampVexesEel]

Does anyone know how standardization is done?

Yes http://www.nfer.ac.uk/research-areas/as ... sation.cfm

Is age always against older children?

Standardisation does not negatively impact older children. Not At all. Not one bit.

What standardisation does is ensure that when comparing raw scores that candidates are compared against those of the same age (normally by month) not against the whole cohort.

Regards
SVE

[OLA]

Thanks SVE
I will check that site in a minute, but a quick question. Can it be used to weight papers in terms of level of difficulty in other words 9/10 in a more difficult paper will translate into a higher standardized score?

[SunlampVexesEel]

Yes. The process does also normalise.

So regardless of how easy or hard the paper was in the end the average candidate will score 100 and the top candidate 140.

Published standardised scores usually range from 70 to 140 or from 69 to 141

Actually this does make for an interesting observation... if the group is self-selecting (i.e. only people with a very good chance of doing well apply) then a score of 70 shouldn't be too disheartening... a score of 70 might be equivalent to a much higher score if tested against the whole population.

This is why it is never possible to extrapolate from test/tutor results to the real test as the level of candidates being tested may be different and hence the normalisation will work out different.

Regards
SVE

[Looking for help]

SVE

Can you tell me whether the standardisation takes account of only the children sitting the test, eg the person scoring the least after standardisation scores 69 say (or whatever the lowest possible score on the test is) , while the highest standardised score is 141?

Or do the standardised scores reflect the whole population, and therefore the lowest person could one year get 100?

Does this make sense or am I completely confused
Thanks
LFH

[SunlampVexesEel]

Can you tell me whether the standardisation takes account of only the children sitting the test

Exactly. The only input is the raw scores from the actual candidates.

, eg the person scoring the least after standardisation scores 69 say (or whatever the lowest possible score on the test is) , while the highest standardised score is 141?

Exactly. Someone will score 69 and someone will score 141; with the rest being distributed (using a normal distribution) around an average of 100.

The distribution is classic bell shaped... most people around the middle with less at either end. The slope of the graph is steep so around the cut-off selection point a very small difference in raw score has a big impact.

(Another reason I maintain the luck/performance on the day factors make the whole test unfair for borderline candidates... but that's another topic).

Regards
SVE

[Looking for help]

Thank you for clarifying this for me,SVE.

[OLA]

Thanks SVE.
I've been to that site and read about the standardisation. A lot of the things make sence to me now than before maybe because of all the 11+ business.
Taking the sample of children that sit a test in a particular location. I'm just trying to appreciate what you said about the slope of the graph being steep and a very small difference in raw score having a big impact.

My son had 48/50 in the Maths Reading test and came up with a standardised score of 117. As the highest scorer gets a standardised score of 141 this means 49 could have a standardised score of about 130? How good do they need to get? This scares me.

[SunlampVexesEel]

My son had 48/50 in the Maths Reading test and came up with a standardised score of 117. As the highest scorer gets a standardised score of 141 this means 49 could have a standardised score of about 130? How good do they need to get? This scares me.

IMHO This is an example of a test that was too easy or didn't have enough questions to seperate out the abilities well. It looks like the average students were probably getting around 47/50 so there was only 3 marks separating the best from the average.

How good do they need to get?.

Depends on the cutoff for your area but probably in top 20% of those tested on the day. If 20% in this example got 50/50 then that's what you need, but IMHO it was not a good test.

I would hope a real test would have more questions and be harder so that the ability differences are more apparent.

Regards
SVE

[mattsurf]

My son had 48/50 in the Maths Reading test and came up with a standardised score of 117. As the highest scorer gets a standardised score of 141 this means 49 could have a standardised score of about 130? How good do they need to get? This scares me.

The scores seem very perplexing - in order for them to be correct, against the others that took this test in the same age bracked, the mean score was 46.5/50 - your son scored 48/50, so was in the 87th percentile - in order to have a standardised score of 130, you would need a score of 49.1/50 (49/50 is a standardised score of 128.5).

I would say that this is not a particularly good test as the mean is much too high and the standard deviation much too low, as a result, single marks make a massive amount of difference