Via Twitter, I'm seeing a lot of anxiety about the part of President Obama's new universal preschool proposal that calls for "comprehensive data and assessment systems" to track student progress and program quality. I know what you're thinking: Standardized tests for 4-year olds? And I agree, it is absurd to imagine toddlers filling in Scantron bubbles. But that isn't at all what the administration has ever meant when it talks about assessing pre-K quality. Instead, a big part of the Obama/Duncan vision is for statewide data systems that link students' early elementary school performance back to the preschool programs they attended, so those programs (not individual teachers or students) can be judged on whether they adequately prepare children for school.
As I reported for Slate in 2011, the potential best part of this Big Data for Little Kids push is that the accepted best practices for early childhood assessment include "testing" not only literacy and numeracy skills, but also the sorts of social, emotional, motor, and creativity skills that have gotten such short shrift since the onset of No Child Left Behind. For the youngest students, those skills include whether a child is confident asking questions, whether she can fasten her Velcro shoes and zip her jacket, whether she can hum a song and draw, and whether she plays cooperatively with peers. One widely-respected model for early-childhood assessment is the Work Sampling System, in which teachers observe students over the course of weeks or months, and then assign them a numeric rating in each of these categories. Again: Unlike assessment systems in use for older kids, these numbers are not meant to punish or reward individual teachers or students. Instead, the idea is to use the data to help teachers identify where students need to grow – and to help schools, districts, and states figure out how to improve teacher training and program management at the system level.
Testing for older students, I think, has a lot to learn from the best practices in early-childhood assessment. One lesson is that we have to look at the whole child over time, not just at his reading and math scores during a specific hour. Another lesson is that, as psychometricians have been warning us, the most accurate use of a test is as a diagnostic tool for the person who is taking it, not as a human resources tool to punish or reward the test-taker's teacher or principal.
Now. I don't want to be too optimistic, because early-childhood assessment can certainly get screwed up. In the Slate piece, I describe how Ohio managed to turn a good idea for longterm holistic assessment into a 10 to 15-minute, one-time, sit-down, scored literacy test for 5-year-olds. Most child development experts laugh at the idea that you can tell very much about any kindergarten student in a specific, short block of time like that, since 5-year olds don't typically have the self-control to put aside immediate concerns like hunger or boredom to focus on a Very Important Task.
A colleague of mine is working on a large paper about early-childhood assessment, so there will be much more to report in the coming weeks and months. Stay tuned.