The report is one of the most complex I've seen, for a range of reasons. First, it is difficult to identify exactly who is a native speaker. Next, there is a statistical method of "informing" A level grade boundaries on the basis of average GCSE grades, which does not take account of the particular problem of progress from GCSE in the language. The current GCSE syllabus has been discredited for its lack of attention to grammar, and the widespread cheating in controlled assessment. Next, we have the problem with self-declaration as a native speaker, and the fact that only a quarter of schools contacted provided a reply. Experience tells me that schools would not reply to a questionnaire if they felt it might be to their disadvantage, so I don't trust the statistical base. It would have been better to rely on the DfE's own figures - available for around 60% of the entry - than to go for self-reporting, that took a lot of time and money, and resulted in a 25% response.
In German, 17% of entrants are identified as native speakers, and are taking around half of the A* grades. This is serious enough to put potential Oxbridge candidates off taking German A level, and Cambridge is finding it hard to meet its recruitment target for state schools. A table later in the report estimates that a native French speaker is five times more likely to achieve an A* than an English speaker. These estimations, though are only part of the information we need. Pupils are getting lower A level results in languages than in other subjects. Why? Some people say grading is severe, but that is a matter of statistics rather than the candidates' actual knowledge of the language. We are suffering from the knock-on effect of the weaknesses of GCSE, which often mean that proper language study begins in Y12. The current reforms to GCSE are an essential step towards restoring language teaching at A level.
The following paragraphs from the report (p4) show that the issue goes beyond German
Model 1 shows that after accounting for prior attainment, gender and school type, the association between whether students are native speakers or not and overall performance is statistically significant and quite large, ranging from nearly 44 UMS marks for Spanish, to nearly 56 UMS marks for German. This suggests that whether students are native speakers or not has a significant effect on performance, with native speakers outperforming their non-native speaker counterparts. The effect of gender and school type are also statistically significant, with girls and independent school students outperforming boys and state school students. However, these effects are much smaller.
Model 2 shows that for each language, there is also a significant interaction between mean GCSE score and whether students are native speakers. The interaction term is negative, indicating that as the mean GCSE score of a student increases, the difference in performance between native and non-native speakers decreases. In other words, the role of being a native speaker is larger for students with lower prior attainment. This is summarised in Table 11, which reports the effect of being a native speaker on UMS marks for students with a mean GCSE score of 5 (representing a typical grade C student at GCSE) and a mean GCSE of 7 (representing a typical grade A student at GCSE).
It is not surprising that non-native English speakers do not do as well at GCSE as English native speakers. This suggests that their relative advantage at A level is even greater than the raw results suggest.