Skip to content

Information for students, faculty and staff regarding COVID-19. (Updated: 27 November 2020)

printicon

Innovative statistical methods for standardized achievement tests to ensure comparability and fairness of test scores

Research project Standardized achievement tests are used in educational contexts such as university admission and school grading. Test results are often crucial for people’s future life choices. Thus, it is important that the tests are reliable and fair for all.

Standardized achievement tests are typically distributed on a large number of occasions and test forms are designed in different versions to prevent that items become known in advance for the test takers. Ideally, all test forms should be equivalent, but they might differ in degree of difficulty, even if they are pretested.

Head of project

Project overview

Project period:

2020-01-01 2025-12-31

Funding

Marianne och Marcus Wallenberg: 6 000 000 SEK

Participating departments and units at Umeå University

Faculty of Social Sciences, Umeå School of Business, Economics and Statistics

Research subject

Statistics

Project description

The aim of this project is to develop innovative statistical methods to ascertain valid and reliable test scores over time and to ensure fairness among the test takers. The research is subdivided into three themes: 1) Extend the observed-score equating framework. 2) Examining the effect of pretesting items on the quality of standardized achievement tests. 3) Develop quality control methods and propose quality measures to guide choices for standardized achievement tests. The themes aim to develop innovative statistical methods in different test situations where there are knowledge gaps today.

The project team consist of leading international experts in the educational testing field, with access to unique test data from their respective countries. Statistics and educational testing theories will be used with both simulated and real test data. Simulated test data allow examination of different scenarios and real test data make sure that the methods are useful in practice in actual educational contexts. The research will make strong theoretical and practical contributions to the psychometric field, and the results will be extremely valuable for test constructors, policymakers, teachers, researchers and every test taker who use standardized achievement tests. The goal is to get a fairer test processes and minimize possible errors in both admissions to universities and achievement comparisons between students.