Midway through a seven-week clerical course in the Israel Defense Forces, several instructors were unexpectedly relieved and replaced. The newcomers were not told what the old ones had been told.
The setup was simple. Sixty men in their first half-year of service were split into five training groups. Each instructor was told that a quarter of their trainees had high “success potential.” That quarter had been picked at random. Another random quarter heard it straight from a psychologist, in a short personal interview. Everyone else was a control.
The researchers, Dov Eden and Gad Ravid, chose not to repeat the message for the replacements. The gap kept going anyway. Through the rest of the course, both groups of “high potential” men kept outscoring the controls, on weekly written exams and in weekly instructor ratings. The men in both high-expectancy groups also reported feeling overrewarded, which the authors thought probably drove them to put in more.
The idea that what a teacher expects shapes what a student does had come from schools. Eden, of Tel Aviv University, wanted to know whether it held for adults at work. His first try, with Abraham Shani and published the same year, used 105 male trainees. They were matched on aptitude and sorted at random into three groups: instructors were led to expect better performance from one, regular performance from another, and were told nothing about the third.
The first group scored significantly higher on objective achievement tests. They also had more positive attitudes and described their instructors as better leaders. By the paper’s own count, instructor expectancy explained 73 percent of the variance in performance. That is a very large share, and it comes from one study of 105 men.
In 1990 Eden ran it on whole units. Twenty-nine platoons were split at random, and the leaders of one group were told their soldiers, on average, had unusually high command potential. Those platoons outscored the others.
Then the effect started to thin out. A 1995 study by Eden and two colleagues noted that every published confirmation among adults so far had involved men. Among female officer cadets led by women, the false briefing raised expectations but changed none of the performance measures. In a course with separate men’s and women’s platoons, it worked for men led by a man and for women led by a man, but not for women led by a woman. A 2000 meta-analysis of 17 studies, by D. B. McNatt, found a strong average effect that varied widely: stronger in the military, with men, and for people of whom little had been expected at first. McNatt called for more research in ongoing jobs and in companies, rather than in training courses and the military.
These experiments relied on giving leaders false information about their people, and a critique the same year by S. S. White and Edwin Locke raised the ethics of that deception. The honest version would be to train managers to expect more. Eden’s team tried it in seven field experiments, building the workshop up from a one-day introduction to a three-day program with follow-up sessions.
The average effect was small. In the authors’ words, “There was little evidence that the workshops influenced leaders or followers.”