Arnhem, 26 April 2007

Angela Verschoor awarded doctorate for her thesis Genetic Algorithms for Automated Test Assesmbly

If there is one development characterising the history of test construction over the last two decades, it is the increased use of item banks. Several models for test assembly have been formulated, and advanced systems have been developed for storage of test items, together with their psychometric and content properties. While these developments open possibilities for complex test assembly models, current solution methods are often inadequate to deal with these models. In her PhD thesis, Angela Verschoor shows how genetic algorithms provide solutions for these models. Today, Angela Verschoor defends her thesis, entitled Genetic Algorithms for Automated Test Assembly, at the University of Twente.

Genetic Algorithms

Angela Verschoor has explored to what degree genetic algorithms can be used to solve complex test construction problems. By imitating processes in biology, she shows that genetic algorithms offer promising solutions for the construction of tests that meet content as well as psychometric requirements.

Subdividing the Item Bank

Since intensive use of item banks may result in their exhaustion, Angela Verschoor also explored the long-term effects of computerised test construction by comparing several strategies for preventing item bank exhaustion. Subdivision of the entire item bank, using only one part for test construction in each period, proved to be a simple and effective measure.

Doctoral Candidate

Angela Verschoor has worked at Cito’s Psychometric Research Centre since 1991. She had previously worked as a consultant with the World Bank in Washington DC. Angela Verschoor’s thesis supervisor is Professor W.J. van der Linden.