MP warns Icelandic school tests could mislead

Municipalities should establish what conclusions they can safely draw from Icelandic and mathematics test results before using them to track pupil achievement, Independence Party MP Jón Pétur Zimsen argues in an opinion piece published by Morgunblaðið.

Zimsen says decisions about improving schools must rest on results that can withstand scrutiny. He identifies four risks in the Matsferill assessment programme: repeated use of test questions, a testing window of up to four weeks, pressure created by public comparisons between schools, and tests that measure only a limited part of the curriculum.

A March 17, 2025 memo from the Ministry of Education and Children envisages using the same tests from year to year, he writes. If pupils become familiar with the questions, schools could prepare them for the specific tasks that will appear, potentially affecting how school performance is assessed.

The testing window creates another risk because pupils sit the tests at different times. Zimsen cites a 2018 study by Wang and colleagues that discusses the possibility that pupils who take a test earlier may pass questions on to those who take it later. Randomising the questions, he says, would not eliminate the risk of pupils knowing them in advance.

In 2025, the Centre for Education and School Services, known as MMS, cited the absence of competitive use as one reason the risk of leaks was low, Zimsen writes. A decision has since been made to publish comparisons between schools from 2027. He says the change calls for a new risk assessment and stronger safeguards to ensure the tests are secure and the results valid.

The tests are based on selected curriculum competency criteria, according to a March 10, 2025 memo from MMS. Zimsen argues that they therefore offer only limited information about achievement: better results do not, by themselves, demonstrate improvement across an entire subject or in school performance as a whole.

He warns that public comparisons could put pressure on schools and municipalities to show progress. Combined with repeated questions, a lengthy testing window and a narrow range of criteria, that pressure could raise scores without a corresponding increase in pupils’ skills, creating false confidence at the expense of students.

As evidence that the concern is not far-fetched, Zimsen cites a 2019 experiment by Toton and Maynes. Participants who received questions and answers in advance scored an average of 11.33 out of 12, compared with 7.67 in a comparison group. The researchers found no significant difference on questions participants had not seen beforehand.

Zimsen calls for unknown test questions, standardised administration and tests that cover a broader range of the curriculum. He says pupils should take well-designed assessments several times during their schooling, measuring both their current achievement and their progress, to provide a reliable basis for decisions about education.

He also asks what evidence shows that results from Lesferill, a related assessment programme that has been in use for about a decade, have been used to improve reading instruction and that this follow-up has improved pupils’ reading skills. Zimsen is a member of parliament for the Independence Party.