National Test Results: Representations and Misrepresentation Challenges for Municipal and Local School Administrations in Norway

• In Event: Data Use in Education: Alluring Attributes and Productive Processes

Abstract

This paper addresses the publication of results from national test in primary schools, with the aim to address the usability of national test results for Norwegian municipalities. In Norway, public primary education is governed at the municipal level. We ask to what extent is the information made available to school owners and principals by the official website useful to them? Is ther a discrepancy between the assumed usability of the tests for school owners and principals and their actual usefulness, especially for small municipalities.

In the Norwegian context, the implementation of the national quality assurance system is characterised by low-stakes testing and follow-up (Skedsmo, 2011). However, how this is done varies, and may be linked to the data presented simply being inadequate for the purpose of follow-up of school quality.

The main discussion in the paper is related to interpretation of aggregated data at the municipal and school levels. Although national tests include all 5th graders, at the municipal or school level the cohort sample for a single municipality or school may be quite small. Even though the test is administered to the population of 5th graders across the country each year, it is still possible to view a single municipality or school in one particular year as a sample of a population across a period of several years. Hence, assuming representativeness might not be appropriate, as suggested through ‘the law of small numbers’ (Kahneman & Tversky, 1972; Tversky & Kahneman, 1971). Due to many small municipalities and schools, this type of argument is also relevant to national test results in Norway.


This paper focus on the presentation of data from national tests, made available to the public through a website, Skoleporten.no, which is run by the Norwegian Directorate for Education and Training. In order to investigate how the issue of size plays out in presentations of aggregated data from national tests, we show how data has been presented on the website.


The paper uses graphical presentations of data at two points in time, in 2013 and 2016, as presented on the website. The research-based evaluation of national tests in 2013 forms the baseline (Seland et al., 2013). This comparison forms the basis for a discussion of the usefulness of average test results presented at municipal or school level.


Over time the presentation of aggregated test results have improved, in the sense that the data presented in 2016 contain more information, and also displays margins of error. However, the challenge to using data for small schools and municipalities is still present, as this is linked to geography and the population pattern throughout the country.


National tests are assumed to be able to provide usable information at both national, municipal and school level. However, this might not be the case for small municipalities and schools. In a country such as Norway, the challenge of few students in small schools and municipalities remain, and the usability of the information provided by aggregated national test results for these groups are limited.

Author

• Elisabeth Hovdhaugen, University of Oslo