In many of the schools where we work, the whole premise of the personalized learning system created by New Classrooms, the nonprofit I help run, runs into trouble when the winter chill sets in.

In the fall semester, most schools use Teach to One: Math to “meet students where they are.” Students are matched with the math concepts they are ready to learn, and grouped with students ready for the same skills.

But as winter begins, we face a tremendous amount of understandable pressure to push students to study concepts that they may not be ready for but will be included on their end-of-year state exams, which are designed to measure grade-level proficiency. One national testmaking group, the Smarter Balanced consortium, is trying to address this by including some below-grade level questions on its computer-based test. But generally, if a seventh grader needs to study the fourth-grade skill of how to find the area of a rectangle, there’s little external incentive for a teacher to devote class time to filling this gap.

So we face an annual dilemma: Do we design students’ individualized curricula around the skills they need — the point of our learning model, which has already demonstrated early positive results — or push them into grade-level material that will be on the test regardless of their readiness?

The annual tests that undermine our best efforts were required by the federal No Child Left Behind law that went into effect in 2002. So I was optimistic when Congress overhauled the law in December, and disappointed when I realized that ESSA might not change this dynamic.

ESSA did give power to states to add measures of student academic growth, including adaptive computer assessments. But the law still requires testing of students every year from grades 3-8 on grade-level proficiency. Adaptive assessments would be supplemental, meaning that developing a richer picture of student learning will likely lead to more testing, rather than smarter testing.

I see some reasons for hope. A provision under the new law will allow up to seven states to create alternative assessment systems. I’m cautiously optimistic that this will show the power and effectiveness of more holistic, competency-based assessments.

For now, though, we’re still working with a testing framework that is trying to do too much. States that want to understand student growth will have to assess both student growth and grade level proficiency. We have experienced this dual design, and the teachers implementing Teach to One: Math can attest to the complexity that this creates.

This is really difficult work. We want assessments to offer students a holistic understanding of themselves as learners. We also want teachers and families to be able to get a comprehensive perspective of a school. History teaches us that schools need rigorous standards to strive toward, and that schools need to be held accountable for having high expectations for all students. Balancing all of these needs is certainly complex.

One solution could be to create new ways to measure student growth from grades 5-7, and then have a proficiency-based eighth grade exit exam. Another is to use the kinds of short, daily, adaptive assessments we use in Teach to One to create a cumulative gauge of student understanding instead of using end-of-year snapshot assessments at all.

ESSA attempts to make everyone happy, but no one will get what they think is most important if annual proficiency assessments persist instead of a more nuanced, balanced system. For now, teachers across the country, including the ones that Teach to One works with, will continue to do their best to balance these multiple sets of expectations while weathering the winter chill. We await the results.

Disclosure: Chalkbeat rents space from New Classrooms.