A working paper from the National Bureau of Economic Research reports on an experiment that tried to establish just that. In Houston, Texas, 50 elementary schools were randomized into treatment or control groups. Treatment schools altered their timetables to have teachers specialize in subjects such as math, science, social studies, and reading based on each teacher's strengths (assessed by the school principal). A class might be taught by one teacher for math and science, and another for reading and social studies. Other classes had one teacher for math, another for reading, and a third for science and social studies.

The results were negative. In the first year, schools with specialist teachers saw an effect size of -0.07 on math and reading achievement. Over the first two years, the effect size was -0.05 for math and -0.04 for reading, with the math result statistically significant. For children in special education, the results were even worse, with an impact of -0.15 for reading and -0.20 for math.

A teacher survey measured views on lesson planning, teacher relationships with students, enjoyment of teaching, and teaching strategies. Teachers in treatment schools were significantly less likely to report providing tailored instruction for their students. All other survey outcomes on teaching strategy were statistically identical between treatment and control.