Economics

Hard Data Help, a Little Bit

See allHide authors and affiliations

Science  12 Oct 2012:
Vol. 338, Issue 6104, pp. 173
DOI: 10.1126/science.338.6104.173-c

Performance appraisals are generally not high on anyone's list of likes and are often points of friction between employers and employees. One educational program in the United States, Race to the Top, has encouraged the use of student outcomes—in the form of scores on standardized tests—in the evaluations of teachers. Rockoff et al. have analyzed the results of a randomized trial carried out in the New York City public schools, where the difference between treatment and control groups of school principals was the provision of value-added measures of individual teacher performance to the former. They found that the hard data did correlate with the principals' prior subjective evaluations of their teachers; moreover, the agreement was greater when the hard data were more precise or when the principals had been observing the teachers for more years—as one would hope and expect. Also, the principals weighted the hard data more strongly in cases where their subjective evaluations were less reliable, and teachers who received lower-value-added reports were more likely to be let go. On the critical issue of student performance, however, the outcomes were modest: Math achievement went up a little bit, and English achievement stayed the same.

Am. Econ. Rev., in press (2012); www.nber.org/papers/w16240.

Navigate This Article