Predictive models of question difficulty - A critical review of the literature
Recent decades have seen a proliferation of research into the identification and manipulation of question difficulty factors.
This paper evaluates three predictive models of question difficulty, each of which provides valuable insights into some of the successes and potential pitfalls involved in the process of delivering an examination question at a specific level of difficulty.
In the light of these insights, the viability of developing a unified model of question difficulty applicable across item types and subject domains is assessed. The paper concludes that for any such model to be useful it may have to sacrifice a degree of rule-based precision in favour of flexibility and responsiveness.