Definition
SALSA (Skolverkets Arbetsverktyg för Lokala SambandsAnalyser — the Agency's tool for local correlation analyses) is a regression model that controls for three background variables:
- Parents' level of education
- Proportion of students with a foreign background
- Gender distribution
The residual is the difference between the school's actual merit value and the value predicted by the model given the student composition.
- Positive residual → the school performs better than expected
- Negative residual → the school performs worse than expected
SALSA is recalibrated annually by Skolverket with new coefficients. This means residuals are not directly comparable across years.
How to interpret
The residual shows "added value" — not absolute quality. A school with a residual of +10 and a merit value of 200 may contribute more to student learning than a school with a residual of −5 and a merit value of 280. The first outperforms expectations; the second underperforms.
Compare residuals, not absolute merit values. If you want to know which school adds the most — given the students' circumstances — the SALSA score is the right measure. See SALSA: Actual vs Expected.
Combine with other indicators. SALSA only controls for three variables. Other factors (special educational support, leadership, student health services) are not captured. Read the residual alongside school survey results and teacher certification.
Common mistakes
- Assuming a positive residual = a good school. The residual can be noise, especially for schools with few students. A small school with 20 students may score +15 one year and −10 the next.
- Ignoring that the control variables are limited. SALSA controls for three variables. Factors such as parental income, neighbourhood and the school's selection mechanisms are not captured.
- Comparing residuals across years. The model is recalibrated annually — coefficients change. A residual of +5 in 2023 and +5 in 2024 are not equivalent.
- Interpreting SALSA for schools with fewer than 30 students. Small samples make the residual statistically unstable. See the guide on small schools.
Statistical uncertainty in the SALSA score
The SALSA score has a standard uncertainty that depends on school size. Rule of thumb: SE ≈ σ / √n, where σ ≈ 15 (typical merit value standard deviation) and n = number of pupils.
| Pupils | Approx. SE (±) | Assessment |
|---|---|---|
| <30 | ±2.7 or more | Low certainty |
| 30–100 | ±1.5–2.7 | Low–moderate |
| 100–250 | ±0.9–1.5 | Moderate |
| >250 | <±0.9 | High certainty |
On school pages, a confidence indicator is shown: High certainty (>100 pupils), Moderate certainty (100–250) or Low certainty (<100). Residuals for small schools should be interpreted with great caution.
Schools where SALSA is often misleading
- Small schools (<30 students in year 9): Individual students' results have a disproportionate effect on the residual.
- Special and resource schools: Atypical student composition that the model's three variables do not capture.
- Selective admissions: Schools with competitive entry requirements (e.g. profile programmes, advanced tracks) — selection creates a homogeneous group that the model does not adjust for.
- Profile schools (music, sports, Montessori): The selection effect means students are not representative of the model's assumptions.