Four ways of asking the same question, drawn on identical axes so the lines can be compared directly. Horizontal means staffing makes no difference. The steeper the line falls, the more a worse ratio goes with a worse teaching score. Staff to student ratios are from the Guardian University Guide 2027, teaching scores from NSS 2026.
1. Same subject, different universities
Each dot is one subject at one university, against the average for that subject
2. Same university, different subjects
Each dot is one subject at one university, against the average for that university
3. Universities compared with each other
Each dot is one whole university, against the average university
4. Subjects compared with each other
Each dot is one subject nationally, against the average subject
| Comparison | Ppt per +5 students | r | n | p |
|---|
| Subject | Students per staff | Teaching score | Universities |
|---|
Teaching score is NSS Theme 1, the percentage of final-year students answering positively. Subject cells need at least 25 responses, whole universities at least 100. Guardian subjects are mapped to the NSS CAH2 groups, averaging where several Guardian subjects sit inside one group. The first two panels use the same 1,942 subject cells, measured against different averages. Correlations on averaged groups are always stronger than the same relationship measured inside them, which is why the fourth panel should be read as the loosest comparison rather than the most powerful finding.