Sammendrag
Do non-standardized, publisher-provided tests for lower secondary school provide valid and reliable
measures of mathematical competence? We analysed a sample of items pooled from tests
accompanying three different Norwegian textbooks using Rasch analysis. The pooled sample of items
was found to be sufficiently unidimensional for measuring function competence, with four strands of
sub-competencies in accordance with theory. The competence associated with an increasing difficulty
of items could be qualitatively characterised by shifts from a) identifying through constructing to
reasoning about representations, b) using visual to using algebraic representations, and c) local to
global interpretations of functions. While the individual tests differed substantially in the distribution
of items across strands of mathematical competence, minor adjustments to the combined instrument
were sufficient for providing a valid and reliable measure of mathematical competence.
Vis fullstendig beskrivelse
