Teach to One earns promising marks in math learning
Personalized-learning models powered by technology posted more promising gains in the 2012-13 school year, according to a newly released Columbia Teachers College study on the first-year impact of New Classrooms’ Teach to One blended-learning model.
Many news articles have by now covered the study, which showed that the 2,200 students in the Teach to One program across seven different schools experienced, on average, nearly 20 percent more growth than the national average in math on the Northwest Evaluation Association’s (NWEA) Measures of Academic Progress (MAP) assessment.
The study is careful to note that the results cannot be taken as definitive proof of Teach to One’s causal impact because the study lacked an experimental design. That said, the report suggests many reasons to be optimistic about Teach to One’s impact along a number of dimensions, including that Teach to One achieved these results while serving a disproportionately high number of minority, low-income, and special-education students who started the academic year significantly below national averages. As a result, there is good reason to believe that the gains between Teach to One’s students and students from similar backgrounds nationally would be even greater, but unfortunately the MAP results are not broken out by subgroup. Furthermore, students who started with the weakest mathematics skills made the greatest gains—50 percent higher than the national average.
There are some reasons to be cautious, however, about the results. Although the average gains of Teach to One students in most demographic sub-groups outperformed national norms—including for language minority, special education, and low-income students—that did not hold true for black students. The study reiterates that Teach to One students are not representative of same-age students nationally, so people should be wary of interpreting the below-average gains. But if a goal is to close the achievement gap, then the result requires more investigation. Second, student gains were uneven across Teach to One’s schools; one school in particular stood out with student-learning gains significantly higher than those posted in other schools, which begs the question of what is causing the performance gains—the Teach to One model or some other factor.
Lastly, interestingly enough, the students who started the school year off with the strongest mathematics skills experienced gains that were slightly below average. Joel Rose, CEO of New Classrooms, has told me that the reason is likely because the team was too conservative in challenging high-performing students, as they wanted to be sure that students truly mastered something regardless of what the assessments might have suggested. This is a great example of how research can help the innovation process, as the results inform how the team is tweaking the model.
Beyond the results though, there are a couple things worth pointing out.
First, the education technology news cycle has been dominated in recent weeks by some high-profile struggles in implementing technology in schools. What I hope people take away from New Classrooms’ promising results is that the real insight behind what it is doing is that it is not about the technology first and foremost. It’s about the learning model itself, and technology then acts in service of that model. What New Classrooms has done with Teach to One is create an Individual Rotation blended-learning model that provides an individualized approach for each student to learn. It does this by assigning students to one of a number of instructional approaches—from teacher-led instruction to student collaboration and from virtual tutoring to math software—based on what their assessment results suggest they need. The process provides teachers with real-time information about student performance and, importantly, frees their time to support individual and small groups of students. This approach stands in stark contrast to many education technology implementations that ignore the need to think through what will the learning model look like first—as in, what will students and teachers do and how will they use their time on a daily basis—which then determines the technology needs. Too often the technology leads.
Second, the results point to the need for more research to understand more deeply what is causing the results and to gain a deeper understanding of what is working for whom and under what circumstances. My colleague Julia Freeland wrote recently about why this matters. Although New Classroom’s model is explicitly designed to reach each student by in essence asking this question and then mixing and matching the right set of approaches for each student, its results suggest that although it may be doing this well for many students, it likely is not doing it for all yet. Additionally, although the results by subgroup help illustrate this phenomenon, it’s almost certain that just because an individual student falls within a certain subgroup, it does not mean that we can predict that student’s results in Teach to One. Gaining this understanding will require further research.
Perhaps most encouraging is that this is research New Classrooms remains committed to do; remains committed to share publicly the results so that we can all learn; and remains committed to learning the lessons from the results and improving itself. That makes for a sound innovation process, which is yet another lesson education could learn.