Information Technology Reference
In-Depth Information
feedback. At this point, the new teaser can undergo a first round of testing. Two to
three rounds of testing are required to refine coding menus, check the accuracy of
the learning sequence, evaluate item functioning, and optimize reliability.
Unlike conventional tests, DiscoTests can be used indefinitely; students can take
the same teaser several times without exhausting its potential to help them gain an
increasingly sophisticated understanding of targeted concepts. This is because the
items are deliberately constructed to be answerable at several different levels of
sophistication. Moreover, because the primary role of DiscoTests is educative (and
items do not have single correct answers) concerns about cheating are minimal.
Furthermore, DiscoTests are both educative and standardized. All performances are
placed on the same, domain independent, general scale. This makes it possible to
compare learning across any range of subjects or contexts. Finally, DiscoTests dou-
ble as data collection instruments. Eventually, they will yield large longitudinal
databases that will allow researchers to construct increasingly refined accounts of
the pathways through which students learn important skills and concepts.
New Tools Foster New Values: Revisioning Education and Testing
We have discussed the history of our contemporary testing infrastructure and
explained the need for new approaches grounded in the science of learning. We
have also provided an overview of one new approach that combines advances in
basic research about learning with new techniques in psychometrics to build embed-
ded formative assessments that are both standardized and richly educative. The
DiscoTest initiative is engaged in building
a mass-customized testing infrastruc-
ture
wherein metrics that are informed by learning theory and research are used in
the design of standardized tests that fit the needs of specific curricula and support
learning. This way of designing tests has broad implications for what is considered
possible and preferable for standardized testing infrastructures.
This approach to test design also allows us to transcend the dichotomies men-
tioned in the introduction and third section. First, by using a psychometrically
refined developmental measure to research specific learning sequences and then
using that same measure to assess student performance relative to the researched
learning sequences, we transcend the dichotomy between educative and standard-
ized assessment. Student performances are evaluated both in terms of where their
performance is in relation to the Skill Scale and in relation to the learning sequence
being assessed, a measurement that is simultaneously standardized and education-
ally relevant. Knowing where the student performs on the General Skill Scale
provides all the benefits (and liabilities) of standardized testing (e.g., allowing for
comparison between students, or between subject areas for the same student, or
between groups of students). But knowing where a student performs relative to
an empirically grounded learning sequence provides all the benefits of formative
assessment; with rich information about what the student understands and could
best benefit from learning next. With the right measures, research approach, and
