The model-predicted merit value (Swedish: modellberäknat meritvärde) is the average merit value that the SALSA model calculates a school "should" achieve, given the students' background variables. It is the expected result based on statistical correlations.
How is it calculated?
The SALSA model uses linear regression with three background variables: parents' education level, proportion of recently immigrated students and gender distribution. Based on these variables, the model calculates the average merit value that schools with a similar student composition typically achieve.
Example
A school in a municipality with a high proportion of university-educated parents and few recently immigrated students receives a high model-predicted value — perhaps 250 points. If the school's actual merit value is 260, it has a positive residual of +10 points. If the actual value is instead 240, it has a negative residual of −10 points.
Interpretation
The model-predicted value should not be interpreted as a target that the school "should" reach. It is a statistical reference point. The residual shows whether the school performs better or worse than what is typical for schools with a similar student composition — but it does not say why. Factors such as the school's leadership, resource allocation and local conditions also affect the result.
Data are sourced from Skolverket's SALSA publication and updated annually.