Leaving Cert Standardisation

[Duke of Marmalade]

If it is a true statement that the school a student attends affects their results then the lack of modeling for a school effect will of course reduce accuracy.

The current model is based on averaging in all national past achievements between jc and lc.

That's the bottom line. The standardisation gave no credit for the fact that the school you went to may have been a factor in your progress from JC to LC - that schools actually matter. Of course there are those who would welcome this, in whose view it is basically unfair to be able to "buy" a better chance at LC. These same people would have welcomed the temporary suspension of the advantage for paying for private health insurance. At least this latter group got their money back!

 

[joe sod]

I think the shenanigans over predictive grading has shown one thing we should never ditch the leaving cert and a big exam at the end of school , it is the only fair way to assess students. We risk completely downgrading our education system and demotivating good students when the system downgrades them based on statistical models than on their actual ability and knowledge
The government took the path of least resistance, they should have continued with the exam in June as planned rather than the silly initial idea to have it at end of July.

 

[llgon]

There are basically two inputs:
(1) The Teachers' Assessments
(2) Predictions from Junior cycle f

Great analysis Duke. I just have one question for you; did the marks the teachers gave their pupils have any effect at all on the results?

From my reading of it the pupil's subject ranking within the school appears to the most significant factor before standardisation. I'm not clear of the significance, if any, of the mark given.

 

[Duke of Marmalade]

Great analysis Duke. I just have one question for you; did the marks the teachers gave their pupils have any effect at all on the results?

From my reading of it the pupil's subject ranking within the school appears to the most significant factor before standardisation. I'm not clear of the significance, if any, of the mark given.

You’re right that the rank decided everything. But I think the rank followed the marks.

 

[llgon]

So the teachers/schools could have given a rank without a mark and the outcome would be the same. Most of the discussion in the media is about how the teachers' marks differed from the results where the reality seems to be that the teachers marks are pretty much irrelevant.

It is somewhat ironic that the students were given their marks today but not their ranking.

 

[SPC100]

I don't think that's true.

I guess the model accepted anything within a certain window of performance (based on your classes jc results), and only adjusted marks if outside that window.

Most grades were not changed.

 

[Duke of Marmalade]

So the teachers/schools could have given a rank without a mark and the outcome would be the same.

No. Let me try and explain why ranking was needed to allocate the final calculated scores.
Let's say it is thought appropriate to give a school 50% recognition for its teachers' assessments (TA) and 50% recognition for its Junior Cert prediction (JCP).
The obvious first attempt would be to give each student the average of their own TA and their own JCP. Two problems with that. Firstly not everybody will have a JCP. But much more seriously each student's LC outcome would be 50% tied to how they did in JC. This was rejected as wholly unfair and inappropriate and I fully agree with that.
But this would not be unfair in aggregate at the school level if the school was sufficiently big. So what they did was assume that in aggregate the school would perform 50% in line with the TA and 50% in line with the JCP. But not just in average outcome but in the whole shape of how individual scores would be spread about that average. This is what the statisticians call a "distribution".
Having thus determined the school's aggregate calculated distribution it remains to allocate calculated marks to individual students. This is done by slotting them into the distribution based on how they ranked in their TA.
So note that the inflation inherent in TA remains in this model at least to the extent of 50%. That inflation comes from marks not from ranking.
Of course one might have contemplated using 100% JCP, in which case you would be correct and the teacher ranking would be all that would be needed. It would certainly have eliminated grade inflation. Perhaps one for COVID 20

 

[NiallSparky]

We risk completely downgrading our education system and demotivating good students when the system downgrades them based on statistical models than on their actual ability and knowledge

This essentially happens every leaving cert year anyway. Results are made to fit a normal distribution, so students get moved around based on statistical methods.

 

[Duke of Marmalade]

This essentially happens every leaving cert year anyway. Results are made to fit a normal distribution, so students get moved around based on statistical methods.

That is not correct. There is an element of standardisation of the initial marks so as to keep a reasonably level standard year on year - students are not allowed to benefit en masse from an easy maths paper, say. But this standardisation is done on a uniform basis with no attempt, intended or otherwise, at social engineering.