Is the future of education: computers interacting with students in English, classrooms liberated from traditional constraints, students given free access to unlimited information via global communications networks? These were the visions presented at the first Conference on Natural Language Processing Techniques and Technology in Assessment and Education, held by ETS, Princeton, May 18-19, 1994.
We were excited, but puzzled. If schools will look like this, then what will happen to assessment? Will technological solutions become the seductive siren song of post-MCQ assessment? New technologies will enable us to record interesting observations that impinge on the assessments we want to make. They will provide novel ways of eliciting and recording a variety of student behavior: capturing not only product and content, but also process. But they also threaten to swamp us with a tidal wave of information, which we won't know how to use, but won't be able to reject, because "the tiniest piece might be crucial."
Empirically-based methods will be overwhelmed by an exponentiating stream of new sources of data. The only way to survive is to have a theory-driven viewpoint that focuses attention on relevant information, enables meaningful condensation, and motivates subsequent action. Turning observations into measures, then meaning, then progress, will require a framework far beyond and far simpler than mere technology.
Such a framework requires answers to questions like:
What do we intend students to learn when we teach a subject? (How important is learning how to "do" science. How important is content?)
What constructs describe a subject and its framework for assessment? (Are we constructing a single dimension or a bundle of loosely linked elements?)
What technologies usefully record the evidence required to assess each subject?
How are an assortment of behaviors, recorded as a myriad of data points, to be transformed into measures of ability on relevant constructs?
How are measures and their accompanying wealth of diagnostic information to be communicated usefully and meaningfully?
How will we validate such sophisticated and complex testing procedures?
As the siren song of mindless technology strikes our ears, let this be our motto: "Don't do anything, if you don't know what you're doing or why you're doing it."
Neil Jones 1994 RMT 8:1 p. 336
University of Cambridge Local Examination Syndicate
Technology: good servant or bad master? Jones N. … Rasch Measurement Transactions, 1994, 8:1 p.336
The URL of this page is www.rasch.org/rmt/rmt81c.htm
Website: www.rasch.org/rmt/contents.htm