Tuesday, 7 October 2014

Streaming in schools: author replies


On September 26 I wrote a post about streaming by ability in schools, based on a paper written by Samantha Parsons & Sue Hallam. The link is given below.


I followed my normal practice of sending the link to the authors and inviting comments, and Prof Susan Hallam of the Institute of Education, University of London has kindly replied:

I have been working in this field for over 20 years now and it is clear that there are no simple practical solutions for schools.

All we try to do is present evidence which will help schools make decisions based on their aims and the intake that they have.

What the evidence suggests from a range of research in the UK and elsewhere is that the gaps in attainment widen when children are placed in highly structured groups.

Once placed in a particular group it is very hard for children to move to a higher group because of the ways that schools operate. Movement between groups is negligible. This then has an impact on long term educational outcomes, for instance, the level of GCSE entry, etc. We also know that children who have the same scores on tests but are placed in different groups because of class size issues also show a divergence in outcomes - those in the higher group do better.

While we appreciate your comments on our paper we feel that it needs to be considered within this wider context.


  1. I have this question for Professor Hallam: why did you not use propensity score modelling to analyse the data, given its well-known advantages over standard multiple regression/ANCOVA for determining causality from observational data? If ever a data-set called for propensity score modelling this is it; non-random assignment to treatments, a fairly small experimental group, and a vast pool of potential controls available to draw the matched pairs from.

    My apologies in advance if I have misunderstood something obvious and this is a stupid question.

  2. What a disturbing reply. It's hard to be certain whether Susan simply takes for granted that you share her assumption of how to evaluate outcomes, or whether she's daring you to misstep by contradicting her. But, I suspect the former.

  3. Well that was anti-climatic...