Automated AUT Scoring

Alternative Uses Task (AUT)

The Alternative Uses Task (AUT)¬†is a commonly used divergent thinking task used to measure creativity. In this task, people have to come up with as many creative uses for an everyday object, such as a brick, as they can. In order to score a person’s creativity on the AUT there are three frequently used methods: (1) fluency: the number of given uses, (2) uniqueness: how rare a given use is, and (3) expert ratings using the “Consensual Assessment Technique” (CAT). With the CAT, multiple experts rate creativity on a Likert- scale. A person’s creativity score consists of the overall mean of expert ratings. Compared to other ways of rating the level of creativity, CAT ratings are good predictors of creative solutions and are independent of the number of uses a person supplied. At the same time, inter-rater reliability of CAT is often low and only sufficient when raters have a high level of expertise.

Abbas Foundation Test Development Funds

In 2016 Claire Stevenson, Matthijs Baas and Han van der Maas received funding from the Abbas Foundation to develop a database and automated scoring algorithm for the Alternative Uses Task.

This project brings together psychometric research and machine learning methods to provide a big-data CAT solution for AUT scoring. Our AUT database which was used to train and teach our automated scoring algorithm is currently filled with AUT responses and scores provided by different researchers in the Netherlands.

Technical report of our automated AUT scoring method

This document contains the final technical report of our automated AUT scoring method supported by the Abbas Foundation Test Development Funds.