« Remembrances | Gene Expression Front Page | fMRI detects Gene Expression »
March 20, 2005

Education and Poverty

In recent posts here and here I looked at the data on educational achievement by ethnic groups in England. This showed substantial differences in attainment by different groups, from Black Caribbeans at the bottom to Chinese at the top, as well as the familiar superiority of girls over boys.

I didn’t say much about the causes of these differences, for the simple reason that I don’t know the answer.

However, others are more confident, and the Economist magazine on [correction] 10 March (this issue no longer available free online) gave its own analysis, saying:

Why do black children do so badly at school? … Poverty of aspirations probably counts for less than simple poverty. Last year, 26% of Afro-Caribbean children received free school meals because their parents were poor or on benefits. They did better in exams than the 11% of white British children in the same situation. Blacks seem to do badly largely because they tend to live in bad neighbourhoods with bad schools. In Lambeth and Lewisham—the two districts with the highest concentrations of Afro-Caribbeans in Britain—just 35% of white children earned five or more good GCSEs in 2003. That was 16 points below the English average and not much better than the 30% of Afro-Caribbean and mixed-race children who achieved the same in those areas. Black children do not perform nearly as well as other ethnic minority groups, nor as well as they should. But the explanations—and, hence, the solutions—have less to do with bad attitudes among teachers and pupils and more to do with the old difficulties of poverty and place.


So there you have it: the explanation is poverty, bad neighbourhoods, and bad schools. Voilà! Nothing that can’t be solved by a liberal application of taxpayers’ cash.

But let’s look more closely at the data on free school meals (FSMs)...

FSMs are available to children whose parents are receiving social security benefits or otherwise are below a certain threshold of household income. FSMs are the standard criterion of ’poverty’ in English educational statistics, and the Education Department provides data on performance by FSM and non-FSM children in various age groups as an Excel file available here (click on the second Excel file of ‘Additional Information’ and save to disk). I have extracted the following data for the proportion of children attaining the target level of achievement in GCSEs at age 15/16 - the figures used by the Economist.

______________1.________2.________3.______4.
__________% on FSMs___FSM____Non-FSM____All
W___________11.6______22.2______56.3_____52.3
W/BC________28.6______24.5______45.6_____39.7
W/BA________28.6______30.7______53.5_____47.1
W/A_________16.7______36.4______71.7_____65.7
A____________31.1_____41.9______61.3_____55.3
I____________12.9_____49.2______69.2_____66.6
P____________39.1_____36.8______50.8_____45.2
Ba___________62.3_____46.3______51.6_____48.4
Bl____________32.5_____29.6______43.6_____39.0
BC___________26.3_____26.0______39.1_____35.7
BA___________40.0_____32.4______50.5_____43.3
C____________9.5______68.3______75.0_____74.2
All___________14.0_____26.1_______56.1_____51.9


Column 1 shows the proportion of children in each ethnic group who receive FSMs. Column 2 shows the proportion of FSM children aged 15/16 in each group who attained 5 ’good’ GCSEs in 2004. Column 3 shows the proportion who attained 5 ’good’ GCSEs among children not receiving FSMs, and Column 4 shows the proportion for all children in the group. The ethnic groups are those discussed in previous posts:

W = White
W/BC = Mixed White and Black Caribbean
W/BA = Mixed White and Black African
W/A = Mixed White and Asian
A = Asian
I = Indian
P = Pakistani
Ba = Bangladeshi
Bl = Black
BC = Black Caribbean
BA = Black African
C = Chinese
All = all children in the age group.


It will be evident that there is some connection between poverty and educational performance. In all ethnic groups the performance of FSM children is below that of non-FSM children, often over 20 percentage points lower. And the Economist is right to point out that the performance of white FSM children is below that of FSM Blacks. However, this tells us little about either the strength of the connection or the reasons for it. Even at a glance, the statistics should warn us that a simple formula ‘poverty causes educational failure’ won’t stand up. There are three obvious difficulties:

- Even after allowing for FSM status, there are big differences in performance between different ethnic groups, whether we look at those receiving FSMs or those who are not. For example, Chinese children on FSMs do much better than all other groups on FSMs, and indeed better than most children not on FSMs. Indians and Bangladeshis on FSMs also do respectably well - almost as well as the average for all whites, and better than Black Caribbeans who are not on FSMs. While FSM is a relatively crude measure, which may conceal wide differences of income and living standards, it is the only measure we have, and the onus of proof is on those who rely on poverty as the explanation of low attainment to show that these differences can be accounted for.

- There are also large differences in the proportions of children in different ethnic groups who are receiving FSMs. This should at least raise the question whether the causes of poverty are the same across all groups, and whether the ability range of children covered by FSMs varies according to the causes of poverty.

- There also seem to be differences in the impact of poverty (if any) in different groups, as judged by the difference between columns 2 and 3. For example FSM status makes a very big difference in the case of whites, but comparatively little in the case of Chinese and Bangladeshis. If poverty were all-powerful, we would expect its impact to be more uniform.

I don’t have a confident alternative explanation for differences in attainment, but I suggest the following speculative points:

1. There is a loose correlation between educational attainment and parental income, both between and within ethnic groups. Within each group, those who are on FSMs will have below average attainment, but since the correlation is not perfect, there will be regression toward the mean: that is, the attainment of FSM children will be closer to the mean of their ethnic group than the mean of their parents’ income is to the mean of all parental income in that group.

2. The strength of the correlation within a group will vary according to the range of children receiving FSMs. In cases where FSM children make up a large proportion of the ethnic group, the difference between FSM and non-FSM children is likely to be relatively small. Conversely, when FSM children are a small proportion of the ethnic group, we would expect the difference to be relatively large. In other words, there should be a negative correlation between the proportions in column (1) and the difference between columns (2) and (3). I have calculated that there is in fact a negative correlation of about -.5. However, I would not put much weight on this, as the negative covariance comes mainly from just two groups: White and Bangladeshis. If we omitted Bangladeshis from the analysis the correlation would be much smaller. Also, the Chinese group goes against the trend: it has a small proportion of children on FSMs, but also a small difference between FSM and non-FSM children. This is awkward for my hypothesis.

3. The causes of poverty in the different groups are probably different. This would need proper sociological investigation, but I suggest that in the White group FSM status (which covers only 11.6% of White children) is usually associated with personal problems and deficiencies of the parents. The parents are likely (no doubt with many exceptions) to have such problems as alcoholism, drug abuse, low IQ, lack of educational qualifications, mental illness, criminal records, and violent and abusive behaviour. The mother will often be a single parent (single parents account for about half of all ‘child poverty‘ in Britain), and the father unknown or absent. To some extent these factors will apply also in the Black Caribbean group, but there is a Caribbean culture of matriarchal single-parent families, and Caribbean single mothers may still be well-qualified, hard-working and aspirational. FSM status among Black Caribbeans is therefore a weaker indicator of ’underclass’ qualities than among Whites. It is probably an even weaker indicator among Asians. I would guess that FSM children among Asians usually come from families with two married parents, several children, and the father either ill, disabled, or in a low-paid job - poor but honest. I do not know much about Black Africans in England (themselves a very diverse group), but I speculate that they are closer to the Asian groups in this respect. As for the Chinese, there is the complication that recent Chinese immigrants are often asylum seekers, whose children would get FSMs automatically. If these speculations are anywhere near the mark, it is pointless to generalise about the effects of 'poverty' on these children in different groups, as 'poverty' is a very diverse phenomenon.

4. I also suggest that the direction of causation is not simply from poverty to low educational attainment. The key causal variable is not poverty as such but parental characteristics, which affect the children by both nature and nurture.

Finally, a word about ‘bad neighbourhoods’ and ‘bad schools’. What makes a school or neighbourhood ‘bad’? In the case of schools, it is a not a lack of resources, as ‘bad’ schools have extra money thrown at them by the Government. In the case of neighbourhoods, it is not (usually) a matter of bad housing and physical environment, as there are many cases of ’good’ neighbourhoods that go ’bad’, and sometimes then go ’good’ again, according to the pressures of the housing market. What makes a neighbourhood or school good or bad is primarily the people in it. If the people in a neighbourhood are prone to criminality, vandalism, and anti-social behaviour, there will be a vicious circle, as respectable people move out, and more and more anti-social individuals and families move in. (In England, local authority housing policies reinforce this trend, as certain areas become known as ’sink estates’, and are used as dumping grounds for the mentally ill, drug addicts, released prisoners, and asylum seekers.) Housing and the physical environment then deteriorate due to vandalism and neglect. The schools will become prey to indiscipline and disruptive behaviour, standards will fall, good pupils will leave as their parents move away, teachers will be demoralised, staff turnover will rise, and so on and on.

This leaves unresolved the question how far poor school performance is due to factors of heredity and environment. What I dispute is the facile assumption that it can be explained simply by poverty. In modern western countries poverty is seldom a random blow of fate, but rather a consequence of individual characteristics, which in turn affect the child and his or her educational prospects.

Posted by David B at 04:51 AM