Robert Rosenthal

Introduction

Robert Rosenthal (March 2, 1933 – January 5, 2024) was a German-born American psychologist whose career rested on a single, unsettling idea: that the person running an experiment, teaching a class or treating a patient can transmit what they expect, and that the expectation then produces the result that appeared to confirm it. He called these interpersonal expectancy effects, and the classroom version — the Pygmalion effect — became one of the handful of psychological findings that reached the general public.

He was equally important as a methodologist. Faced with a sprawling, contradictory literature on his own phenomenon, he helped build the machinery of meta-analysis, named the file drawer problem, and pushed the behavioural sciences toward double-blind procedures and effect-size reporting. He spent thirty-seven years at Harvard, latterly as Edgar Pierce Professor of Psychology, then moved to the University of California, Riverside, where he was a Distinguished Professor and University Professor until retiring in 2018.

Early Life and Education

Rosenthal was born in Giessen, Germany, the son of Hermine (née Kahn) and Julius Rosenthal, and grew up in Limburg, where his father co-owned a dry-goods factory. The factory was seized by the Nazis in 1938, and the family fled to Cologne hoping to conceal their Jewish identity in the anonymity of a large city. They had a quota number permitting emigration to the United States, but someone stole it; Julius Rosenthal's brothers, living in the British colony of Southern Rhodesia, helped the family reach Africa, obtain a visa, and settle in New York in 1940. The family later moved to Los Angeles.

Rosenthal had a habit of turning biography into self-deprecating material. He said that German children harassed him for being Jewish and that British children in Rhodesia beat him on the way home from school for being German; he also joked that by persuading his teachers to let him leave school slightly early each day he became "the psychologist who invented the head start program."

He enrolled at the University of California, Los Angeles, majoring in psychology and working in the stadium hot-dog stand, taking his B.A. in 1953 and his Ph.D. in 1956 while still a graduate student in clinical psychology. His dissertation, supervised by Bruno Klopfer — a student of Jung and a developer of the Rorschach — compared projection in paranoid schizophrenic patients and normal controls after an experimental manipulation. Examining his pre-test data, he found that the groups already differed in the direction that supported his hypothesis, before any manipulation had taken place. That irritation became his life's work.

Academic and Professional Career

Rosenthal's first academic post was at the University of North Dakota, where he directed clinical training in the Ph.D. programme for five years. He moved to Harvard in 1962 as a lecturer in clinical psychology and then as a professor of social psychology, chaired the Department of Psychology from 1992 to 1995, and was Edgar Pierce Professor of Psychology from 1995 to 1999. He retired from Harvard in 1999 for the University of California, Riverside, where he worked for twenty-five years and became a University Professor — the highest honour in the statewide UC system — in 2008.

One detail he liked to cite as evidence for his own thesis: when he moved from North Dakota to Harvard, manuscripts of his that had been rejected or were still under review were suddenly accepted, now carrying the Harvard letterhead. Nothing about the papers had changed.

Major Contributions to Psychology

Experimenter expectancy

With Kermit Fode in the early 1960s, Rosenthal gave student experimenters ordinary laboratory rats and told half of them their animals had been bred to be "maze-bright" and the other half that theirs were "maze-dull." The rats were assigned at random and did not differ. The maze-bright animals nonetheless learned faster — apparently because the students handled them more and recorded their behaviour more carefully. The finding generalised: if an experimenter's belief can shift the behaviour of a rat, it can certainly shift the behaviour of a human participant.

Pygmalion in the classroom

In 1964, at the invitation of an elementary school principal, Lenore Jacobson, Rosenthal ran a study at Spruce Elementary School in South San Francisco. All children were given a standard IQ test that the researchers presented as the "Harvard Test of Inflected Acquisition," a fictitious instrument said to identify children about to show a spurt in measured intelligence. Roughly twenty per cent were then selected at random and named to their teachers as likely bloomers. Eight months later, the designated children showed larger gains — dramatically so in the youngest grades, where the first-grade "spurters" gained an average of more than twenty-seven IQ points against about twelve for their classmates.

The book, Pygmalion in the Classroom (1968), argued that teachers had conveyed the expectation through ordinary interaction — more attention, warmer climate, harder material, more detailed feedback — and that the children had risen to it.

Meta-analysis and the file drawer

Because the expectancy literature refused to settle, Rosenthal turned to method. Working with the Harvard statistician Donald Rubin and others, he developed and formalised the procedures that became modern meta-analysis: converting results to a common effect-size metric, testing heterogeneity, and modelling study-level moderators. He also identified the file drawer problem — that journals preferentially publish significant results, so the published record overstates effects — and in a 1979 paper proposed the "fail-safe N," the number of unpublished null studies that would be needed to overturn a finding. By 1978 he and Rubin had counted 345 studies of interpersonal expectancy across settings.

Nonverbal communication and research ethics

To find out how expectancies leak, Rosenthal studied nonverbal channels — face, voice, body movement — developing measures of nonverbal sensitivity including the Profile of Nonverbal Sensitivity. He also argued, less famously, that badly designed research is itself an ethical failure: it wastes participants' time, consumes resources that could fund a better study, and can produce findings that misdirect policy. That argument was part of the push toward double-blind procedures in medical and behavioural research.

Major Publications

  • Experimenter Effects in Behavioral Research (1966) — the systematic treatment of how an investigator's expectations contaminate data.
  • Pygmalion in the Classroom (with Lenore Jacobson, 1968; expanded edition 1992) — the Oak School study.
  • "The 'file drawer problem' and tolerance for null results," Psychological Bulletin 86 (1979): 638–641 — one of the most cited methodological papers in psychology.
  • Meta-analytic Procedures for Social Research (1984) — the standard reference for the technique he helped create.
  • Essentials of Behavioral Research: Methods and Data Analysis (with Ralph Rosnow) — a widely used methods text.

Impact and Legacy

Rosenthal received the American Psychological Association's Gold Medal Award for Life Achievement in the Science of Psychology in 2003, the AAAS Prize for Behavioral Science Research in 1960, and election to the American Academy of Arts and Sciences. A 2002 survey of eminence placed him eighty-fourth among the most cited psychologists of the twentieth century. He died of an aneurysm in Riverside on January 5, 2024, at the age of ninety, and Harvard's memorial minute described him as one of the most influential psychologists of the preceding sixty years.

His legacy is double-edged in an instructive way. The Pygmalion effect entered education, management and medicine as a moral lesson — expect more, get more — while the methodological work he did partly in response to its weaknesses became a permanent part of how psychology assesses evidence.

The criticism that reshaped the finding

The instrument was broken. Robert L. Thorndike's 1968 review pointed out that the baseline scores were implausible: some whole classes averaged "reasoning IQs" far below anything a real cohort would produce, in some cases below the floor of a test calibrated for ages six to ten. Gains measured from an artificially low starting point are hard to distinguish from regression toward the mean.

The effect depended on a few children. Janet Elashoff and Richard Snow's book-length reanalysis, Pygmalion Reconsidered (1971), confirmed that the headline result was concentrated in first and second grade and that some of the "bloomer" gains traced to small numbers of children with extreme score swings. Fleming and Anttonen's large 1971 study failed to reproduce the effect.

Which way does the arrow point? Charles West and Thomas Anderson raised the problem of preponderant causation in 1976: when a teacher expects a lot and a child does well, nothing in the design establishes that the expectation did the work rather than the child's prior ability producing both the expectation and the outcome.

Meta-analysis shrank it. Stephen Raudenbush's 1984 synthesis of eighteen experiments found a pooled effect well under 0.2 standard deviations, with later reanalysis of the nineteen-study dataset giving about 0.078 with a standard error of 0.052 — not statistically distinguishable from zero. Crucially, the effect was moderated by how long the teacher had already known the child: it was largest when teachers received the bogus information before forming their own impressions and near zero when they had not. That moderator is a direct problem for the practical lesson drawn from the study, because real teachers already know their pupils.

Teachers are usually right. Jussim and Harber's 2005 review, the most thorough modern treatment, concluded that teacher-expectancy effects in real classrooms are real but small — typically d of 0.1 to 0.3 — largest for young children, attenuating sharply with age, and heavily confounded with accurate teacher perception of pre-existing differences. Brophy's work made the same point: everyday teacher expectations track actual performance closely, so self-fulfilling distortion adds only a modest increment, and Brophy and Good had already noted that Rosenthal's four mediating channels — climate, input, output opportunity, feedback — are so highly intercorrelated in real settings that they cannot be separated for intervention.

The historiography. Sam Wineburg argued in 1987 that the public version of Pygmalion is far more confident than the data justify, prompting an exchange with Rosenthal in consecutive pages of the same journal issue. The deeper point, echoed by later historians of the controversy, is that a fragile result that everybody repeats stops being treated as fragile.

And the ethics. The study deceived teachers and, in effect, gave one group of children an advantage and withheld it from others. The complementary experiment — inducing deliberately low expectations to test the "golem" effect — cannot be run at all.

Learn More

References

  • Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom. The Urban Review, 3(1), 16-20.

Last updated: 2026-10-20

Menu
Loading