Panel Recommends New Breed of Assessments for Science Learning
Laying out a new vision for science assessments, a panel of the National Research Council Tuesday proposed that states design testing systems that integrate several key types of science learning, and blend classroom-based assessments with state-level "monitoring" tests and gauges of students' "opportunity to learn."
The proposal, detailed in a 256-page report, offers an expert panel's ideas on how testing should change to fully reflect the Next Generation Science Standards adopted by eight states so far. The picture it paints departs markedly from current assessment practice, which tilts heavily toward students' knowledge of science facts, and typically takes place in one large-scale statewide exam each spring.
Instead, to gauge student learning, the panel recommends that states obtain feedback from three sources. One is ongoing, classroom-based, or "formative," assessments, which would draw students into building and refining scientific models, generating and analyzing data, and creating oral and written arguments about what they're learning.These could take the form of curricular units, student-work portfolios, tasks drawn from a district's bank of items, or other activities.
One example of classroom-based assessment mentioned in the report shows how a teacher might ask 6th grade students to build models of air particles, and then lead them in discussion so she can ascertain what they did—and didn't—understand about the scientific practices used to build them. Another shows how a 5th grade teacher could oversee an extended unit on biodiversity in the schoolyard, guiding students as they gather data, analyze it, and build arguments to interpret it.
The second source of information would be state-level "monitoring" tests that would be aimed at measuring how well students have learned the material over the course of a year, and that could be used to meet states' accountability needs. The expert panel suggests that while these tests would include multiple-choice and short-essay items, they should lean as heavily as possible on performance tasks—or, at the very least, "multicomponent tasks." The panel urges states to consider using a matrix-sampling design for parts of their tests, in which group-level results are drawn from students taking portions, rather than the entirety, of the test.
Finally, the panel says states should collect school-level information about resources that affect students' chances "to learn science in the ways laid out in the [NRC] framework and the [new science standards]," such as access to good instructional materials, the level of teachers' subject-matter expertise, and instructional approaches that allow students "of varying cultural and linguistic backgrounds" to access the material. The NRC framework is a document developed by a team of experts that was designed to help guide the development of the Next Generation Science Standards.
The three dimensions of the new science standards"core ideas" of the sciences, the "practices" scientists use to do their work, and "cross-cutting concepts" that connect the science disciplines—should all be integrated into curriculum and instruction, but also into science assessment, the panel says.
'Thorough Rethinking' of Assessment Needed
The NRC report notes that the framework underlying the new science standards "proposed a dramatic rethinking of science education," and "established goals that cannot be achieved through tinkering," so "a thorough rethinking" of assessment is required as well. "Measuring the learning described in the NGSS will require assessments that are significantly different from those in current use," the report says.
"It will not be feasible to assess all of the performance expectations for a given grade level during a single assessment occasion," the report cautions. "Students will need multiple—and varied—assessment opportunities to demonstrate their competence on the performance expectations for a given grade level."
Putting such a new assessment system into practice will take time, and should start from the "bottom up," with the classroom-based assessments, rather than from the "top down," with the state-level tests, the report says. States must pay particular attention to professional development as they think about creating these new approaches to testing, it says.
The report was written by a panel of 17 national assessment and science experts drawn largerly from universities, along with a few from the private sector and from state or local education agencies. It was co-chaired by James W. Pellegrino of the University of Illinois-Chicago and Mark R. Wilson of the University of California at Berkeley.
In a series of meetings, the panel drew on input from science-instruction and assessment experts in state departments of education; leaders of PARCC and Smarter Balanced, the two federally-funded assessment consortia; and assessment organizations including WestEd, the College Board, and the National Assessment Government Board, which administers NAEP.
Funding for the report was provided by the William and Flora Hewlett Foundation, which also supports Education Week's coverage of "deeper learning"; the S.D. Bechtel, Jr. Foundation; and the Carnegie Corporation of New York.
For a great look at the challenges ahead in implementing the new science standards and designing assessments for them, take a look at a two-story package that my colleagues Erik Robelen and Sarah Sparks produced earlier this year.