States Can Afford Better Tests by Restructuring Costs, Study Says
States can afford higher-quality assessments by reallocating the money they currently spend on tests, according to a new report.
"Developing Assessments of Deeper Learning: The Costs and Benefits of Using Tests That Help Students Learn," released yesterday, argues that states can replace as many as half the multiple-choice items on their current tests with essays and performance items without spending more than they currently do on testing, and they'd get assessments that offer good learning experiences for students and valuable feedback for teachers.
That argument might seem counterintuitive, since the big rap on essays and performance items—and much of what drove states away from them in recent years—has been the higher cost of scoring them. It's bedeviled the field, since many educators see these longer, more complex items and tasks not only as a more nuanced, authentic way to gauge student learning, but valuable experiences of learning in and of themselves, assuming they are designed well.
In the new study, co-authors Linda Darling-Hammond and Frank Adamson of Stanford University detail several kinds of cost savings that could be mined to support an assessment system that leans more heavily on performance tasks. One would be efficiencies realized through state collaborations like the two consortia that are developing tests for the common standards. Another would be computer delivery and scoring.
Another source of cost savings would be shifting funds currently spent on interim and benchmark testing, test-prep materials, and related data-management systemswhich the co-authors argue have been driven by No Child Left Behind's high-stakes accountability pressures—into more-heavily-performance-based assessment systems that include formative/diagnostic tools as well as tests that measure the results of learning.
Additional cost savings could be realized by using teachers, rather than vendors, to score parts of the tests that aren't well suited for computer scoring, Darling-Hammond and Adamson argue. And that involvement would have another, important benefit, they say: It would provide valuable professional development for the teachers.
The Assessment Solutions Group, for instance, found that states spend, on average, $20 to $25 per student on the summative tests required by No Child Left Behind in math and English/language arts. Testing in additional subjects and grades can add another $10 per student. And districts spend an additional $15 to $20 per student on interim and benchmark tests, pushing total per-student costs on testing to $35 to $55, ASG found.
ASG calculated that a state could develop and administer a "high-quality assessment"—defined as replacing half its multiple-choice items with essays and performance tasks—for $55.67, but taking advantage of a variety of cost-savings mechanisms, such as computer delivery and scoring, teacher scoring, and collaboration with other states, could bring that price down to $21.19.
Darling-Hammond and Adamson argue that the price estimates from the two assessment consortia, Smarter Balanced—which Darling-Hammond advises—and PARCC are affordable to states because their initial pricing estimates were $18 to $25 per student for the summative tests and $7 to $8 for the additional, optional interim/diagnostic tools.
Those consortia pricing estimates, by the way, come from the Assessment Solutions Group's 2012 report. The sourcing for those figures lacked citations, however, so I called lead author Barry Topol. He told me they were the original cost estimates ASG created for the two consortia's applications for federal Race to the Top funding, back in 2010.
PARCC officials tell me they have not updated their pricing estimates since then, but they anticipate doing so later this spring. Smarter Balanced, whose original estimates were $19.81 for summative and $7.50 for formative/diagnostic, now projects $22 per student for the summative tests only, and $5 more for a package that includes summative tests and interim/formative tools.