Keynote: Rethinking measurement for accountable assessment

Abstract

The underlying model for most formal educational measurement (e.g. standardised tests) is based on a very simple model: the student takes a test (possibly alongside other students). The complications of there being an instructional plan, actual instruction, interpretation of the outcome, and formulation of next steps, are all bypassed in considering how to model the process of measurement. There are some standard exceptions, of course: a pre-test/post-test context will involve two measurements, and attention to gain score, or similar. However, if we wish to design measurement to hold to Lehrer’s (2021) definition of ‘accountable assessment’ – as ‘actionable information for improving classroom instruction’ – then this narrow conceptualisation must be extended. In this presentation, I will posit a simple model that reflects the simple one-test context described above, and then elaborate on it by adding in a) a framework for design of the assessments that is keyed to educational interpretation, b) further rounds of data collection that can indicate changes in a student’s underlying ability, and c) provision for varied assessment modes that will allow for i) classroom-independent tasks that operate at the summative and meso levels, and ii) classroom-dependent tasks that operate at the micro level. The former are designed to provide a basis for triangulating student responses across different contexts, and the latter are designed to closely track the variation of student performance over time in a classroom instructional context. This framing will be exemplified in a in a K–5 elementary school that is seeking to improve the quality of instruction and students’ understandings of measure and arithmetic. The different levels of data collection will be instantiated by two different pieces of software, which operate at the micro level and the meso/summative levels respectively.