This study is focusing on students at compulsory level and their performance in national exam tests when the students first language is taken into account. Students in Swedish schools with non-Swedish speaking parents commonly follow language classes for limited Swedish proficient students. How does this affect their results when tested in other subjects as physics,biology and chemistry? A general claim is that national tests can provide data for balancedand fair grades. This view is challenged in this paper, through Principal Component Analysisof test results that shows a difference between schools when the schools social position and share of limited Swedish proficient students are measured. Standardized tests are often culturally biased, according to Iris Marion Young and the test themselves are rituals designed to show who holds the key to power and control in the educational system, in the view of Michel Foucault. Guidelines for compulsory level education specifically states that schools should be sensitive to students with non-Swedish cultural experience and respond withadequate measures to students who have special needs. The finding of this study is that the current tests procedures contradict such policies. One reason being the weak instructions when making adaptations of test procedures for limited Swedish proficient students. If adaptations are made in a hap-hazardous fashion they jeopardize the whole purpose of thetests and make them unfair.