A new approach to educational assessments without testing: Automated predictive systems in the prediction of educational outcomes in reading and writing.

MUSSO, M.; CASCALLAR, E. C.

Bulgaria

Conferencia; 9th annual Conference, Assessment Educational Association- Europe; 2008

Assessment Educational Association- Europe

A growing body of literature addresses the changes in assessment practice and the development of new modes of assessment.  Recent developments, facilitated by new methodologies and technologies, have focused on the possibility of assessments which use a wide range of data or student productions to evaluate their performance without the need of traditional testing.  These tools should be sensitive enough to accrue information about the level of performance that the students have reached so far in the domain of study. This approach should also include the prediction of the expected outcomes that best capture the students’ current level of learning. There are a series of predictive approaches that examine in detail the multiple elements involved in these phenomena.  In particular these series of studies use a predictive systems approach, utilizing neural networks in a stream analysis to accomplish this goal.  In particular preliminary results from two pilot studies to detect and classify students with potential reading and writing problems will be presented and explained.  Back propagation algorithm were used, due to their excellence in generalization, and their ability to classify extreme cases despite lack of data. In the current two pilot studies, with 450 and 1500 students participating, the models examined predicted between 89% -96% accuracy the classification of students in their actual levels of performance, utilizing variables from a complex model of self-regulation in learning, and a large number of contextual variables.  These measures and procedures, as well as the implications of the results, will be explained during the presentation.