Welsh-medium school pupils underperform in tests despite more advantaged backgrounds
Secondary schools in Wales that teach pupils through the medium of Welsh are outperformed by their English-speaking counterparts in maths, reading and science tests, according to a new study by Lancaster University.
The average results of pupils attending Welsh-language secondary schools are markedly lower than pupils in English-language schools. This is despite Welsh-medium school pupils having more books available at home, spending more time on their studies outside of school and far fewer qualifying for free school meals.
New research, published today in the Wales Journal of Education, uses Programme for International Student Assessment (PISA) data from 2015 which capture results of standardised tests for 15 year olds across 80 countries to compare results in maths, reading and science.
Within Wales, PISA data reveal the average score for Welsh-language school pupils in the maths test was 476, compared to an English-language school pupil of 485. In reading tests, Welsh-medium pupils scored an average of 469, compared to an average of 494 achieved by English-medium school pupils. In science tests, Welsh-medium pupils scored 484 on average, compared to English-medium pupils averaging 499.
Geraint Johnes, a Professor of Economics at Lancaster University Management School, authored the study. He said: “Recent concerns about the standards of education in Wales have prompted reforms but while there is a great deal of focus on the education system, there has been little attention paid to the comparative performance of English-language and Welsh-language schools.
“Despite Welsh-medium schools being regarded very highly and attracting wealthier families, data reveal that secondary school pupils achieve lower scores in reading, maths and science tests when compared with those in English-medium schools. Considering the pupils are coming from more privileged homes, you may expect them to achieve the same scores or perhaps even higher – but this is not the case.”
Around 200,000 pupils are taught in Wales in just over 200 secondary schools - 24 per cent of these are Welsh-medium, catering for around 20% of pupils. Of the 200, 140 secondary schools participated in the 2015 PISA tests – 18 of which were Welsh-medium schools.
Data reveal Welsh-language schools spend less time in class reading and studying mathematics, but spend slightly more time studying science:
PISA data for 15 year old pupils in Wales
The new study also looks at how advantaged or disadvantaged each pupil may be in terms of family background, including measures such as household wealth. Using the method of ‘data envelopment analysis’, the study establishes how ‘efficiently’ students transform these background factors into results.
Professor Johnes continues: “In addition to measuring test scores, I looked at how ‘efficient’ pupils were in terms of their performance, considering their background and socio-economic status. For example, pupils from poorer families with less time to study outside of school but who manage to achieve fantastic results are classed as highly efficient. I found that the best, most efficient pupils in Welsh-medium schools were still around 10 per cent less efficient than those attending English-medium schools.
“There are a few plausible explanations for the difference in results that we see. Welsh speaking schools face tougher challenges when recruiting teaching staff, resulting in them fishing in a more limited talent pool. There is also a chance that there are systematic differences in how schools approach these types of standardised tests – they may not taken as seriously by some teachers and pupils as they are by others. To eliminate any doubt, further advances need to be made in terms of releasing data sets about secondary schools in Wales so additional in-depth analysis can be done.”
The paper, ‘Medium efficiency: Comparing inputs and outputs by language of instruction in secondary schools in Wales’, is published in the Wales Journal of Education https://doi.org/10.16922/wje.22.2.3
Back to News