University of Cambridge, United Kingdom
Dr. Aiden is a Psychometrician at The Psychometrics Centre, University of Cambridge. He completed his PhD in Psychology at the University of Cambridge in 2018, having graduated in 2014 with a Distinction MPhil in Social and Developmental Psychology. Aiden is involved in modelling latent factors in psychological scales and educational assessment using IRT and MIRT models. His primary research looks at ways of automating the design of new items with predictable characteristics through approaches such as crowdsourcing, rule-based and machine learning algorithms. Aiden is concerned with understanding the intersection of human and machine intelligence and assessing the impact of explainable algorithms. He has conducted extensive research on explanatory IRT models.
Professional item writers usually create educational items manually. Even though there are strict guidelines in the item development process, a large percentage of items are often discarded after empirical analysis, which is costly and time-consuming for test publishers. Furthermore, the increased application of remote online testing via modern day technology (e.g. computer, smartphone, tablets) have made it easier than ever for test-takers to compromise test security and increase the risk of exposing the items. How can we improve the efficiency of the item development process and ensure that there are sufficient items with excellent psychometric properties in the item bank without incurring a higher significant cost? Automatic Item Generation (AIG) is an emergent discipline that uses computational means to create new items on-the-fly with the goal of producing items with acceptable psychometric properties. In this talk, I will briefly cover a variety of AIG techniques and the challenges of using such approaches. I will showcase several automatic item generators that have been developed at the Cambridge Psychometric Centre which are used for psychological and educational testing purposes. I will discuss the empirical methods (exploratory IRT) that are employed to analyse items developed through such means. Finally, I will conclude by suggesting future directions in AIG research such as the involvement of crowd-sourcing and machine learning models.
Dates below are not final ones:
Submission Deadline - January 01, 2020
Author Notification - January 25, 2020
Final Manuscript - February 07, 2020
Submission Deadline Second Cycle - February 25, 2020
Conference - 27th to 29th March 2020
The ICFE conference proceedings will be submitted to Scopus for indexing (ISSN: 2631-8458)