Policy makers know that value-added measures (VAM) are a powerful tool in their effort to discern teachers' effectiveness. However, a new paper points out that policies that include VAMs often assume that a teacher's value added score in one subject mirrors her score in another - an assumption that has rarely been scrutinized.
Dan Goldhaber and colleagues investigated whether this "one score fits all subjects" assumption holds for elementary teachers, who often teach multiple subjects. Is an elementary teacher who teaches math effectively also an effective reading teacher, and vice versa?
Using seven years of North Carolina data and applying multiple statistical models, Goldhaber and team found that value-added scores across subjects are a pretty close match. A teacher's efficacy in one subject does not perfectly parallel her efficacy in the other, but it generally comes pretty close. In fact, the correlation between VAM across math and reading was 0.6 within the same year, and 0.3-0.4 for comparisons across consecutive years. The correlation was even stronger when corrections were made for sampling error.
This finding suggests that while teachers' efficacy is largely consistent, it is not exactly the same across all subjects. Applying a VAM score from one subject to encompass a teacher's overall ability may be right most of the time, but stronger policies will recognize that this supposed VAM mirror image has some imperfections.