« Loser Pays | Gene Expression Front Page | Ladies' Choice Poll Results »
March 17, 2005

More boring statistics...

In my recent post on Education and ethnic groups in England I noted that there were some gaps in the statistics. I have been trying to fill these, with partial success…

I mentioned three main gaps:

a) the statistics covered only state schools, and not private ones

b) for education after age 16, the statistics did not cover specialist post-16 colleges

c) the statistics on performance at age 18 (GCE A-level and equivalent qualifications) were expressed as points scored by those who entered the examinations, and not as averages for the whole population at that age.

A partial solution to these problems is provided by the Youth Cohort Study, a representative survey of young people over 16. The 2004 YCS report on young people aged 16 (after leaving compulsory education) includes some data with an ethnic breakdown of participation in full-time education, from which I have extracted the following:

________________State school___Private___Post-16 colls.___Total f/t educ.*

__Other Asian________40________10_________37__________91

*The total includes the first three columns plus a small proportion in other types of full-time education.

Note that the ethnic classification is not the same as in the earlier data: there are no ’mixed’ categories, ’Black’ is not divided between African and Caribbean, and there is no separate category for Chinese. (I do not know whether Chinese in the sample have gone into ‘Other Asian’ or ‘Other’. If the Census categories are applied, they should go into ‘Other‘.)

The table gives the percentages of each ethnic group who are continuing in full-time education (the total in the final column), and in state schools, private schools, and post-16 colleges respectively. It will be seen that all the non-white groups have a higher proportion of young people continuing in full-time education after age 16 than whites. This is well-known from earlier data: see my old post for discussion. The data also confirm that there are some differences in the proportions of different ethnic groups in private schools. As I pointed out earlier, the state school data will be somewhat depleted in high-ability children, and this is likely to affect some groups (particularly Indians) more than others, but as the differences are only a few percentage points, this does not affect the big picture. A more serious problem revealed by the YCS data is the large difference between ethnic groups in the proportions attending post-16 colleges. These are particularly high among the Black and Bangladeshi groups. The omission of these colleges from the data in my previous post could therefore be a serious defect in the data on post-16 (A-level and equivalent) attainment by different ethnic groups.

This problem could be overcome if we assume that the average attainment level within each ethnic group is the same whether the childern are in schools or in post-16 colleges. We could then simply gross up the school data to allow for the relative numbers in post-16 colleges, and get a result that would be valid for comparison of attainment of ethnic groups across the entire population of the age group. Unfortunately the assumption would be invalid. Post-16 colleges fall into two main types: specialist 16-18 schools (usually called ’6th Form Colleges’ - roughly equivalent to Senior High in America), and Further Education Colleges, which offer education and vocational training to the whole over-16 population. The 6th Form Colleges have characteristics and performance similar to those of ordinary High Schools - it is just a matter of local organisation whether secondary education is provided in single schools covering the 11-18 age group, or divided into two or three separate schools, including a 6th Form College. FE Colleges, on the other hand, are very different. Not all the students are aiming for A-level or equivalent qualifications, and of those who are, the average attainment is much lower than in schools or 6th Form Colleges. Simple ’grossing up’ would therefore be misleading.

I toyed with the idea of using more complicated adjustment factors, e.g. based on the difference in overall performance between schools and FE colleges, but concluded that, apart from being tedious, the results would be invalid for the comparison of ethnic groups. For this purpose we really need adjustment factors specific to the ethnic groups concerned, and I have not found sufficient data for this purpose. Suppose for example that average A-level point scores in FE colleges are only half of those in schools (across all ethnic groups), it would not be safe to use this as an adjustment factor for Bangladeshis in particular, because their relative performance in schools and FE colleges might be quite different from the average. So I have not pursued this approach.

The data on post-16 attainment therefore remain imperfect. For 16-18 children remaining in schools, the data in my last post give an indication of relative performance, but this may either over- or under-state the attainments of an ethnic group in the population as a whole. For the latter purpose the only evidence I can find for this age group is the Youth Cohort Study at age 18, which includes an analysis of the highest qualification achieved at that age. Since these people are not old enough to have attained Level 4 qualifications (university degrees, etc), this shows the proportions who have obtained some Level 3 qualifications (A-levels, etc). In 2004 these were as follows:

__Other Asian________n/a

(n/a indicates numbers in sample too small for a meaningful percentage).

The lack of information on Chinese people, and the lack of breakdown between Pakistanis and Bangladeshis, is unfortunate, but the general rank order of Indian>White>Pak/Bang>Black is consistent with most of the other data.

I stress again that I am not expressing a view on the causation of these differences. In my next post I hope to examine the hypothesis that they can be explained by ‘poverty’.

Incidentally, while rooting around in the DFES statistics I came across this recent research report on ethnicity and education, which usefully updates the study by Bhattacharyya et al which I linked to in an older post.

Posted by David B at 04:07 AM