Robert L. Thorndike

Who they are

Thorndike's review appeared in the American Educational Research Journal in November 1968 (volume 5, number 4, 708-711), the same year the book was published. It is short and specific.

He pointed to a class of nineteen pupils whose control group had an average reasoning IQ of about thirty-one, which the authors used without comment. He judged the first-grade pretest close to worthless, because its mean implied a raw score of about two on a test where random marking would earn five or six correct answers. He found that six pupils selected as spurters in one second-grade class had a posttest mean IQ of about 150 with an implausibly large spread.

The authors reported no raw scores and no pupil ages, so Thorndike reconstructed figures from the test's published norms. His verdict was that the conclusions might be correct but were not adequately supported by the data, so that if they were right it would be a fortunate coincidence.

This page rests on the review alone.

Key papers

Thorndike, R. L. (1968). Review of Pygmalion in the Classroom, by R. Rosenthal and L. Jacobson. American Educational Research Journal, 5(4), 708-711. https://doi.org/10.3102/00028312005004708

Key ideas

Measurement before mechanism. A striking difference between groups is only as good as the test that produced it. Thorndike's complaint was about floor effects and implausible scores, not about the idea of expectancy.

The quiet half of his verdict. He allowed that a self-fulfilling expectation effect can exist. Later experiments, pooled by Stephen Raudenbush in 1984, found a small average effect, which fits that reading.

---