We believe in assessment. We believe it improves the processes we use for assembling the right student body in admission, for supporting student learning, and for continuously improving our educational programs.
But we shouldn’t take this on faith - it is worthwhile to test our beliefs and consider whether there is evidence to support them. For instance, is there evidence that collecting and using assessments of student learning really does promote improved student learning outcomes? Obviously, this is the premise of federal and state education policy, such as NCLB; to a lesser extent it is the premise of independent schools using standardized testing. But is there evidence on whether it works and how it can work best?
These questions formed the backbone of the agenda for a research analysis undertaken by the RAND Corporation last year, funded by the Hewlett Foundation and its Deeper Learning program. Entitled “New Assessments, Better Instruction? Designing Assessment Systems to Promote Instructional Improvement,” (Faxon-Mills, Hamilton, Rudnick and Stecher, 2013), it’s an accessible, succinct (40 pages), free, thoughtful, and very useful read for any educator intending to do this kind of work.
The RAND researchers conducted a broad literature review to make their conclusions, and found that under the right circumstances, assessment in the form of outside testing can influence curriculum. It changes what is taught, as one would expect - often, tests which measure student proficiency on basic skills result in a narrowing of the curriculum. This research confirms the common concern we have about NCLB style assessment.
But the RAND group also found the opposite to be true—that if we measure “deeper learning,” such as critical thinking, complex problem-solving, and authentic student expression, we can make it more frequently included in the curriculum. For example, they write of research establishing that “teachers implementing performance assessments shifted their curriculum away from basic, isolated, and routine content toward mathematical problem solving and communication.” Sadly, however, the effect of testing in this direction is “less potent” than in the effect in the direction of narrowing the curriculum to basic skills.
While is harder to have the “testing effect” we hope for when assessing higher-order thinking, it becomes all the more important to attend to what conditions make the effect most likely, which RAND has helpfully uncovered and shared. Five things are most important:
- “Teachers receive training and support to interpret scores effectively;”
- Test scores “matter” but important consequences do not follow from test scores alone;
- Accountability metrics should value growth in achievement, not just status;
- Metrics are a part of an integrated assessment system that includes formative and summative components;
- Metrics are “one component of a broader systemic reform effort.”
- The testing has “face validity;” in other words, when teachers and administrators view the assessment tool, they can recognize and respect that it mirrors both high quality instruction and in-class testing, and see it as an effective technique for its purposes and goals.
Educational leaders and administrators in our schools, including admission directors participating in school-wide discussions on improving the use of assessments in their school, would do well to consider the above list when evaluating their programs.