Reasonably matched

SHEU have been publishing from our accumulated databanks in our 'Young People in...' series since 1986, and more recently in our 'Trends' series. We have been offering a little 'health warning' with these reports, as follows:

Each year we produce a report in the Young People series, and however careful we are to describe the populations involved in the surveys, the total picture is often referred to by the media as ‘national data’.

The surveys that give rise to the data are large, numerous, and from many parts of the United Kingdom, but they do not form a deliberately-selected sample. The origin and structure of these surveys is described very carefully and fully on subsequent pages.

Despite this difference, the picture produced by our annual data set typically matches survey outcomes from other data-collection agencies using orthodox strategies such as stratified random sampling. On pages xx–xxvii we draw attention to evidence supporting this claim.

We have recently had an opportunity to assess if there is in fact a bias in the SHEU databanks, and if so, how large or important is it.

OFSTED hold a variety of pieces of contextual information about schools.
In order to test how representative the SHEU samples of schools might be of the country as a whole, OFSTED were asked to compare the SHEU schools with the national sample on some convenient measures. The results are below:

National SHEU
National SHEU
National SHEU
National SHEU
National SHEU
Pupils eligible for free
school meals
18% 16% 18% 16% 18% 15% 17% 14% 17% 14% 17% (1) 14% (2)
Pupils with SEN


3% 3% 3% 3% 3% 2% 3% 2% 3% 2% 3% 2%
Pupils with SEN but
without statements
18% 17% 18% 18% 17% 16% 14% 13% 14% 14% 15% 14%
First language is

known or believed to
be other than English

9% 7% 9% 7% 9% 8% 9% 9% 10% 9% 10% 9%
White (UK/other) 83.6% 86.5% 83.4% 86.0% 82.5% 87.9% 80.3% 82.7% 81.2% 83.7% 81.6% 83.6%

(1) 2005 Sample schools : Upper quartile = 27% Lower Quartile = 8%

(2) 2005 National figures : Upper quartile = 22% Lower Quartile = 6%


  • Overall the analysis seems to show a reasonable match between the SHEU sample averages and the national
  • For Free School Meals the SHEU sample seems to be slightly more disadvantaged and this is borne out by the position of the quartiles
  • SHEU samples have a slightly lower proportion of ethnically White pupils; there are in particular more Black pupils, although these are spread across the two main sub-groups (Caribbean and African)
  • So, our conclusion is two-fold: yes, there is a bias, but it is not a very large or important one.
    SHEU would like to thank David Howarth of OFSTED for his time and diligence in producing this analysis.