Tuesday, 5 September 2017

Education and inequality — geography

In the last post we saw that Scotland has some high percentages of pupils living in one council area but attending secondary school in an adjacent area. There were two notable examples: pupils commuting from Glasgow to its leafy suburbs — East Dunbartonshire and East Renfrewshire — and also from Clackmannanshire to Stirlingshire. In this post we'll look at geographical associations that may yield clues as to what's behind this.

You, like me, may have an intuition informed by anecdote about what's happening, but I'd encourage you to put that aside for the time being because the point of this series of posts is to let the data speak for itself and step carefully through quantitative evidence.

Population density

First, let's take a look at the population densities of councils, shown on the map below on the right. The population density is the number of people living in a council divided by its area in square kilometres (sq km). The left-hand map is the one from the last post (with a few visual improvements).

Click to enlarge. Council names are omitted for clarity, but you can see them on this map.

At first glance they seem quite similar. Population density is low across Scotland and, in most places, 90% or more of pupils attend a school in the council area where they live. Both maps have hot spots around Glasgow which has the country's highest population density (3521 people/sq km).

But, population density cannot be the only factor involved. East Dunbartonshire (616 people/sq km) and East Renfrewshire (538 people /sq km) do not seem much different to Glasgow's other neighbours: Renfrewshire (673 people /sq km), West Dunbartonshire (566 people /sq km) or North Lanarkshire (722 people /sq km). Only semi-rural South Lanarkshire stands out as having a significantly different population density (179 people /sq km).

Also, although Edinburgh (1926 people /sq km) has a lower population density than Glasgow, it is still very high relative to its neighbours, West Lothian (421 people /sq km), Fife (279 people /sq km) and Midlothian (251 people /sq km), but there is much less movement of pupils across its city boundary than for Glasgow.

Nor does population density provide an obvious explanation for why there is a flow of pupils from Clackmannanshire (323 people /sq km) to Stirlingshire (43 people /sq km). Like Glasgow, it is from a high population density area to a low one, but there are several other pairs of adjacent areas where this is the case but there is no flow.

Scotland's overall population density is 69 people/sq km and so all the councils mentioned above, apart from Stirlingshire, are relatively densely populated. At the other end of the scale, the lowest mainland population densities are the Highland council area at 9 people/sq km and Argyll & Bute at 13 people/sq km.  With this in mind, and being careful not to speculate too far beyond what the data is telling us, the most we can say is that high percentages of pupils commuting across council boundaries occur where population density is high, but population density alone cannot be used to predict it.

House prices

Home prices offer a measure of the affluence of an area because they reflect both wealth and, through mortgages, income levels. The map on the left is again the one from the previous post, but the one on the right now shows the median residential property price for each council in Scotland: half of properties cost more than the median price, half less.

Click to enlarge. Council names are omitted for clarity, but you can see them on this map.

Now we can see that East Dunbartonshire and East Renfrewshire stand out as the most affluent areas in Greater Glasgow. This is, of course, no surprise. And the differences are not small median home prices for both these suburbs are double that for North Lanarkshire.

This contrasts with the situation around Edinburgh where there is less of a difference between the city and its neighbouring councils, which are also significantly lower in population density than those around Glasgow.

It's also clear that Clackmannanshire is less affluent than Stirlingshire and, as with Glasgow and its leafy suburbs, there is a flow of pupils towards the more affluent area.

We can tentatively conclude that given a sufficiently high population density, which is actually a proxy for schools being close to one another, then there is a draw for pupils living in poorer areas to attend school in a nearby wealthier area. And this draw seems larger for greater differences in affluence, at least as measured by home prices.

But this association is incomplete, because it would also predict that pupils living in Clackmannanshire should be drawn to schools in neighbouring Perth & Kinross which is slightly more affluent than Stirlingshire. In other words, why do so many pupils residing in Clackmannanshire go to Stirlingshire instead of Perth & Kinross?

The most likely explanation is that Clackmannanshire has no Catholic secondary school and St Modan's in the city of Stirling is considerably closer than those in Perth (see this map with council boundaries turned on).

Taking stock

So, in summary, there does seem to be an association between wealth and/or income inequality in high population density areas and the choice of which schools children attend. Other local factors, such as faith schooling, can be important too. No doubt there will also be an effect for private (i.e. non-state funded) schools but these are not included in data considered here.

But I'd urge caution in that we have only found associations, not causes. We cannot say whether it is a pre-existing difference in affluence that somehow creates a social incentive for parents to send their children to other councils for their education, or if it is the perceived standing of certain schools that is pushing up house prices and increasing the affluence of certain areas. Showing causation rather than correlation is more difficult, but the next post may yield some clues on this point by looking at recent trends in the data.