Quo vadis assessment?

2019-11-14T01:44:58Z (GMT) by Gavin Brown
This is the video of my keynote presentation at the biennial EARLI conference in 2019 at Aachen Germany. Introduction is by Prof. Pat Alexander, U. Maryland.

Educational assessment began in the 20th century with the development of highly reliable statistical methods for scoring tests. Amazing refinements have been produced in statistical models, marking methods, administration, etc. With the development of personal computing, tests are now delivered on-screen with adjustments for student performance and an ever-increasing verisimilitude of real-world processes. We now have a vast array of statistical methods to model student responses for almost any test-like product. However, this success has been achieved by neglecting complicating factors such as schooling processes and intra- and inter-personal factors involved in learning and instruction.

Assessment ‘for’ learning has drawn attention to the interactive classroom processes of question and answer, feedback, micro-adjustment to instruction, and the involvement of students in assessment. Unfortunately, these dynamic processes are extremely difficult to model mathematically because they are so context dependent. Thus, they remain largely outside the abilities of psychometric statistics. Further, there is strong resistance in many quarters to reducing the complexity of classroom action to something that could be expressed as a statistical model. Thus, we know that assessment is much more than a product, but we have little grasp of how to capture these complexities in methods suited to the evaluation of assessment products.

For a long time, the domains of educational, social, and cultural psychologies related to learning and instruction have been neglected in assessment. We are now on the cusp of significant and substantial development in educational assessment as greater emphasis on the psychology of assessment is brought into the world of testing both products and processes. Herein lies the future for our field: integration of psychological theory and research with statistics and technology to understand processes that work for learning, identify how well students have learned, and what further teaching and learning is needed.