Measuring primary writing: part two, 2019–20

This is a three part blog series that explains how you can compare our primary scores from 2018–19, 2019–20 and 2020–21. This is part two, and deals with 2019–20.

In the previous blog, I explained how in 2018–19, we linked each of our six primary writing assessments to each other, so we could compare all the year groups together, like so.

We also placed the results of our 2019–20 assessments on this same scale, but we did not use linking to do so. We used a different process called equating. Equating is a technical process which means that we made the assumption that each year group distribution from 2019–20 would be the same as the equivalent year group from 2018–19. So, once the judging had been completed in each window in 2019–20, we set the shape of the distribution for each year group to match that of 2018–19.

We did check after the judging was completed to see if this assumption was justified, and it was. When we did link the tasks, we found that there were some small fluctuations, but in five of the six year groups these changes were not statistically significant.

Why did we make this assumption — why didn’t we just link the tasks to see if, say, the year 3s in 2019–20 were doing better than the year 3s the year before, and update the distributions and scores accordingly?

First, consistency. In normal circumstances, it is very unlikely that there will be dramatic fluctuations in performance from one year to the next. Generally you would want to see a few years of evidence before you assume that performance really is improving or declining from one year to the next.

Second, ease of interpretation: keeping the distributions the same allows us to keep both the grade boundaries and the percentage reaching each grade the same, making it much easier to compare between years.

However, we are not in normal circumstances any more! For 20–21 we are going to move away from this process of equating, and are going to move back to linking. This is because, given the circumstances of lockdown, it is not reasonable to assume that average cohort performance will have stayed the same from one year to the next.

Our 2020–21 plans are explained fully in the next blog.

Learn more about Assessing Primary Writing.