Voodoo Statistics

Today the DfE has proposed a new method of identifying what it calls “Ordinary Working Families”. The reason for this, according to the DfE is to ensure that children in those “just about managing” families are properly served by the education system.

The reality of course, as exposed by the interviews Justine Greening has been giving this morning, is that the purpose behind the new “statistic” is to give cover to the No 10 policy of introducing new Grammar Schools and to enable their spurious statistics to be used without the UK Statistics Authority telling them off. Again.

The statistics are clear. There are a relatively miniscule number of disadvantaged children accepted into Grammar Schools. The existence of Grammar Schools leads to an overall lower outcome across the population.

In the face of this the government intend to do what governments have always done.

Change the statistics.

The new methodology of calculating family income was published yesterday. Do read it, it is quite interesting. It is, however built on very shaky foundations, including a LA wide “housing cost” adjustment to income. Anyone involved in the creation of this data set needs to take a long hard look at themselves before they next call themselves a statistician. The number of children left out of the data set gives you an indication of how poor it is. Whole groups are excluded (or have their income misrepresented), including the self-employed. To then add a housing cost adjustment, which works on an LA wide basis, is astonishing. There comes a point with any data analysis process where you have to say “You know, this isn’t working”. I can imagine the statisticians did say this at one point, but were told to get on with it.

What does this do? It depresses the income of families living in areas with high housing costs. Now, remind me, which areas are Grammar schools located?

No. 10 is convinced they can increase the number of disadvantaged children at Grammar Schools and that those schools can educate those children better than they are currently educated in the comprehensive school system. By increasing the numbers of disadvantaged children the intention is to make the policy more palatable. Not surprisingly, for that was its purpose, this new set of “statistics” shows that the proportion of “ordinary working families” that have children at Grammar Schools is higher than you might think. A classic case of when the facts don’t fit the theory, change the facts. Job done. But only by redefining the meaning of disadvantaged to include people who aren’t disadvantaged. This is statistical voodoo.

Here’s my challenge to them.

You have a test bed. You have 162 existing Grammar Schools, predominantly in a small number of geographically locations. Currently, they do not do what you suggest they can. You say you have changes to the system that will make them work for all children. Go on then, make your changes. But for those schools only. Let them prove they can work. Let them prove that all children, no matter which school they are educated at, get a better education under their regime than a comprehensive area. After all, you claim to be running evidence-based policy. This is your chance to show it.

Sadly I doubt this offer will be taken up. Simply because, as these “statistical” changes show, the government isn’t interested in improvement for all. It isn’t interested in evidence-based policy. It is just interested in implementing its ideological policy of expanding Grammar Schools. For which they have no mandate.

I agree with Tony.