The idea that a manager's belief in someone becomes a self-fulfilling prophecy is close to an article of faith in leadership writing. It has a research pedigree, a definitive review, and a conclusion that almost nobody quotes.
Key Takeaway
After 35 years of empirical research: "Self-fulfilling prophecies in the classroom do occur, but these effects are typically small, they do not accumulate greatly across perceivers or over time, and they may be more likely to dissipate than accumulate." And, fourth: "teacher expectations may predict student outcomes more because these expectations are accurate than because they are self-fulfilling."[1]
Our Grades For These Claims
Applying the scheme from the first article in this series.
Grade A that self-fulfilling prophecies occur. The review's own first conclusion, from thirty-five years of research, states it plainly.
Grade A that the effects are typically small and do not accumulate greatly. Same sentence, same authority.
Grade B that expectations may predict outcomes more through accuracy than through causation, which the review states with the hedge may and which we keep.
The management application is ungraded, because we located a meta-analysis on it and did not obtain its findings.
Our position: this is a case where the careful answer is genuinely more useful than either the enthusiastic version or the dismissal.
A Note On Method
Everything here is verified to August 2026.
We obtained the published abstract in full from the publisher and from three further reproductions, all identical[1][2][3], plus the opening page from the second author's own institutional copy[4].
We did not obtain the full review, and report no effect sizes, because the abstract states none.
We did not obtain Rosenthal and Jacobson (1968), the management meta-analysis, or any of the individual studies, and report citations and characterisations only.
One conclusion concerns stigmatised social groups. We report it in the authors' own words, without extension, because paraphrasing it would risk changing what they said.
All arithmetic is ours and uses invented figures.
This article reviews education and social psychology research. It is not management or human resources advice.
The Claim As Business Writing Makes It
What is usually asserted. This section is our own characterisation.
The management version runs roughly: a manager's expectations are communicated through behaviour, the employee responds, performance moves toward the expectation, and the process compounds. Believe in someone and you build them; write someone off and you break them.
Two observations.
The compounding step is load-bearing. Without it, the claim is that a manager's belief produces a small one-off shift, which is a far less interesting proposition and would not justify the emphasis it receives.
And the claim is flattering in a specific way: it makes a manager's inner state a lever on other people's capability, at no cost and requiring no resources. That is a property worth noticing in any idea that spreads easily, and this series has now noted it several times.
The Concept Predates The Study
A piece of intellectual history that matters for attribution.
The review's opening attributes the concept of the self-fulfilling prophecy to Merton (1948)[4], and reference lists give the citation as Merton, R. K. (1948), The self-fulfilling prophecy, Antioch Review, 8, 193–210[1].
Two observations, ours.
That is twenty years before the study everyone cites. The idea is a sociological concept that a later experiment attempted to demonstrate, not something the experiment discovered.
Which matters, because a finding that fails to replicate does not automatically retire the concept it was testing. The two can come apart, and in this literature they have.
The Study Everybody Cites
The original.
Rosenthal and Jacobson published Pygmalion in the Classroom in 1968[1]. We did not obtain it.
A citing paper describes the design and result: "Even though they had been made up, the test results shaped how some teachers treated the students and affected the students' educational outcomes."[5]
Two observations, ours.
The fabricated test results are the design's strength. Because the labels were random, any difference in outcome cannot be attributed to the students, which makes it a genuine causal test rather than a correlation.
And note the hedge in that description: some teachers. Even the sympathetic summary does not claim it worked uniformly.
Appallingly Flawed
The reception, in the reviewers' own words, and it is unusually blunt for an academic opening.
The review's first page records two opposed camps. On one side, the term "has been known to inspire righteous indignation for teachers' supposed role in creating inequalities", on the basis that erroneous teacher expectations may lead students to perform at levels consistent with those expectations[4].
On the other: "Other researchers, equally righteous, condemn some teacher expectation research as appallingly flawed and have expressed dismay at how it has been misinterpreted and has captured the popular" imagination, with our source truncating[4].
And the review's own position: "This review conveys why 35 years of research on teacher expectations leads to conclusions that fall between these extremes."[4]
Two observations, ours.
The authors are explicitly positioning between two committed camps, which is the posture most likely to produce a usable answer and least likely to produce a memorable one.
And appallingly flawed is a quotation of other researchers' view, not the reviewers'. We report it as their characterisation of a position in the field, not as a verdict.
The Review
The paper this article is built on.
Jussim and Harber published Teacher Expectations and Self-Fulfilling Prophecies: Knowns and Unknowns, Resolved and Unresolved Controversies in Personality and Social Psychology Review, 9(2), 131–155, in 2005[1].
It offers four numbered conclusions, which is unusually structured, and this article takes them in turn.
One observation, ours: the title itself, with its knowns and unknowns, resolved and unresolved, tells you the answer will be partitioned rather than unified. That is a good sign in a review and a bad sign for anyone wanting a slogan.
Conclusion A
The first and most quoted, in full.
"(a) Self-fulfilling prophecies in the classroom do occur, but these effects are typically small, they do not accumulate greatly across perceivers or over time, and they may be more likely to dissipate than accumulate."[1]
Four separate claims in one sentence, and this parsing is ours.
They do occur. That is a positive finding and the review is not a debunking.
They are typically small.
They do not accumulate greatly across perceivers, meaning one teacher's expectation is not reinforced by the next teacher's.
And they may be more likely to dissipate than accumulate, meaning the default trajectory is decay rather than growth.
The third and fourth are the ones business writing never quotes, and they are the ones that matter.
Why Accumulation Decides Everything
The reason that clause is the whole article. Our own analysis.
A small effect is uninteresting on its own. What makes the Pygmalion story compelling is the implied feedback loop: expectation raises performance, higher performance raises expectation, and the two climb together.
Three consequences.
If effects compound, a trivial per-period nudge becomes a large difference over a career, and a manager's belief is genuinely a lever.
If effects merely persist, you get simple addition, which is smaller but still meaningful.
If effects dissipate, each period's small push decays before the next arrives, the process settles at a low plateau, and nothing accumulates at all.
The review's own words point at the third. That is not a small qualification to the popular story. It is the removal of the mechanism that made the story worth telling.
The Arithmetic Of Three Assumptions
Making that concrete. The figures below are invented by us to demonstrate a mechanism and come from no study.
Take a per-period effect of 0.05 standard deviations, which is small, and follow it across twelve periods under three assumptions.
If it compounds, it reaches about 0.80 standard deviations by period twelve.
If it simply persists and adds, about 0.60.
If it decays at thirty percent per period, it settles at about 0.16 by period six and stays there.
Two observations, ours.
The same per-period effect is either transformative or negligible depending entirely on the accumulation assumption. Nothing about the size of the individual effect distinguishes the three columns.
And the review states that the evidence points at the third column. Which means anyone citing this literature for a claim about compounding is citing it against its own conclusion.
Conclusion D
The fourth conclusion, and on our reading the most consequential in the paper.
"(d) teacher expectations may predict student outcomes more because these expectations are accurate than because they are self-fulfilling."[1]
Two observations, ours.
This is a claim about why the correlation exists. Teachers' expectations do correlate with outcomes. The standard reading is that the expectation caused the outcome. This says the expectation may simply have been a good estimate of something that was going to happen anyway.
Note the hedge may, which is the authors' and which we keep. This is offered as the better-supported reading, not as established.
Accuracy Against Causation
Why that distinction is hard to see and easy to get wrong. Ours.
Suppose a manager forms accurate expectations. High expectations of people who will do well, low of people who will not.
Three consequences.
The correlation between expectation and outcome will be strong, and it will look exactly like a self-fulfilling prophecy in any dataset that only measures both.
Distinguishing the two requires making the expectation wrong on purpose, which is what the fabricated test scores in the original study did. That is why the experimental design matters and why observational correlations cannot settle it.
And the practical implication reverses. If expectations are largely accurate, then a manager who talks themselves into believing something untrue about an employee has not created a prophecy; they have made an error, and the review's first conclusion says the resulting effect would be small and fading anyway.
Conclusion B
The second conclusion, reported in the authors' own words without extension.
"(b) powerful self-fulfilling prophecies may selectively occur among students from stigmatized social groups."[1]
Three observations, ours, and we are deliberately restrained here.
The word powerful is doing work, and it contrasts directly with typically small in conclusion (a). The review is saying the average conceals a subgroup where the effect is larger.
That is the same structural point this series has reached repeatedly: the third article found choice overload appearing only under four conditions, the twentieth found a happiness plateau only among the least happy. An average across heterogeneous groups describes nobody in particular.
And we did not obtain the underlying evidence for this conclusion, we do not extend it, and we report it because omitting a numbered conclusion from a four-part finding would misrepresent the paper.
Conclusion C
The third, which is a declaration of ignorance and is worth respecting as such.
"(c) whether self-fulfilling prophecies affect intelligence, and whether they in general do more harm than good, remains unclear."[1]
Two observations, ours.
After thirty-five years, a careful review says we do not know whether the net effect is positive or negative. That is a striking admission and it should temper any confident prescription in either direction.
And the intelligence question is the original claim's most ambitious version. A reference list we obtained includes a paper titled Beleaguered Pygmalion: A history of the controversy over claims that teacher expectancy raises intelligence, in Intelligence, 27(3), 199–234[6]. We did not obtain it and our source does not clearly give its author, so we report the title only, and the word beleaguered conveys the state of that debate.
A Counterpoint The Same Team Found
Evidence that accumulation can occur, reported because it cuts against this article's main line.
An indexed description of work involving one of the review's authors records: "Findings suggest that mothers' inaccurate beliefs repeatedly assessed across time had additive self-fulfilling effects on their children's subsequent alcohol use assessed at a single later point in time, which served to exacerbate differences in the alcohol use of children who had been consistently exposed to unfavorable" beliefs, with our source truncating[3].
We did not obtain this study and report only this description.
Three observations, ours.
The word additive is the point. This describes effects that did accumulate, which is the opposite of conclusion (a)'s general finding.
But look at the condition: the same perceiver, repeatedly, over time, with consistent exposure. That is a mother, not a series of teachers. Conclusion (a) specifically says effects do not accumulate across perceivers.
Which suggests a reconciliation, and it is ours rather than the review's: accumulation may require one persistent source rather than a succession of independent ones. That would make the management case closer to the mother than to the school, because an employee has one manager for years.
The Management Literature
The application to work, and our gap on it.
The review's reference list includes McNatt, D. B. (2000), Ancient Pygmalion joins contemporary management: A meta-analysis of the result, in the Journal of Applied Psychology, 85, 314–322[1].
We did not obtain this paper and report no findings from it.
Two observations, ours.
That is a meta-analysis specifically on the management version, published in a leading journal, and it is the single most relevant document to a business reader. We could not get it, and a reader whose decision depends on this should.
And we note that the review we did obtain concerns teachers and students. Everything this article says about managers and employees is an extension, and the workplace setting differs in duration, in voluntariness, and in the presence of pay.
What Survives For A Manager
The honest residue. Ours.
Four propositions we think hold on the evidence obtained.
Expectations do influence outcomes. Conclusion (a) says so plainly, and this is not a debunking.
The influence is small and probably fades. Same sentence.
Much of the correlation may be accuracy rather than causation, on conclusion (d), which means the manager who reads a person well is doing something real that is not a prophecy.
And the subgroup finding matters more than the average, on conclusion (b), though we did not obtain its evidence.
What does not survive is the compounding story, which is the version that makes the idea worth a chapter in a leadership book.
The Honest Version Of The Advice
What we would say instead. Ours, and it needs no research support.
The usual advice is: believe in your people and they will rise.
Three problems with it, all following from the above.
It asks you to hold a belief rather than take an action, and beliefs are not directly choosable.
It implies the belief works through some transmission rather than through anything observable, which makes it unfalsifiable in practice.
And on conclusion (d), it may have the causation backwards, in which case manufacturing a belief you do not hold is manufacturing an inaccuracy.
Our own replacement, which is about behaviour rather than belief: give people the work that would demonstrate the capability you are unsure about. That is observable, it produces evidence either way, and it does not require you to believe anything in advance. If expectations are largely accurate, the way to improve them is better information, and the way to get better information is to create the opportunity for it.
What To Do
Drop the compounding claim. The review states effects do not accumulate greatly across perceivers or over time, and may be more likely to dissipate than accumulate.
Keep the underlying finding. Self-fulfilling prophecies do occur. This literature is not a debunking, and the effect is real if small.
Consider that your expectations may be accurate. On the review's fourth conclusion, the correlation may exist because managers read people correctly rather than because they shape them.
Notice that the average hides a subgroup. The review reports powerful effects occurring selectively, which means an aggregate figure describes nobody in particular.
Do not try to believe something. Beliefs are not directly choosable, and on the accuracy reading, manufacturing one you do not hold is manufacturing an error.
Give the work instead. Assigning something that would demonstrate a capability produces evidence either way, requires no prior belief, and is observable.
Watch for one persistent source. The one study we found showing additive effects involved the same perceiver repeatedly, which describes a manager better than it describes a school.
Obtain the management meta-analysis if this matters to you. It exists, it is in a leading journal, and we could not get it.
The Limits Of This Analysis
Several caveats matter. This article reviews education and social psychology research and is not management or human resources advice. Everything is verified to August 2026. We obtained the review's published abstract in full and its opening page, but not the full paper, and report no effect sizes, because the abstract states none. We did not obtain Rosenthal and Jacobson (1968), and describe it only as a citing paper characterises it. We did not obtain the management meta-analysis, which is the single most relevant document to a business reader, and report no findings from it. We did not obtain the evidence behind conclusion (b), which we report in the authors' own words without extension, nor the paper on the intelligence controversy, whose author our source does not clearly give. The counterpoint study on additive effects reaches us as an indexed description that truncates, and our proposed reconciliation between it and conclusion (a) is our own conjecture, not stated by any source. All arithmetic is ours, uses invented figures, and demonstrates a mechanism rather than estimating anything. The review concerns teachers and students; every application to managers and employees in this article is our own extension to a setting with different duration, voluntariness and incentives. The characterisation of the business claim, the accuracy-versus-causation analysis, and the replacement advice are our own reasoning, not findings. Roughly twenty-one years of subsequent literature was not reviewed.
Frequently Asked Questions
Is the Pygmalion effect real?
Why does the accumulation point matter so much?
What is the fourth conclusion?
So should I stop believing in my team?
Is there evidence effects ever accumulate?
Does this apply to workplaces at all?
References
- Jussim, L., & Harber, K. D. (2005). Teacher Expectations and Self-Fulfilling Prophecies: Knowns and Unknowns, Resolved and Unresolved Controversies. Personality and Social Psychology Review, 9(2), 131–155. DOI 10.1207/s15327957pspr0902_3, publisher record reproducing the abstract in full, on 35 years of empirical research on teacher expectations justifying four conclusions: that self-fulfilling prophecies in the classroom do occur but these effects are typically small, do not accumulate greatly across perceivers or over time, and may be more likely to dissipate than accumulate; that powerful self-fulfilling prophecies may selectively occur among students from stigmatized social groups; that whether self-fulfilling prophecies affect intelligence, and whether they in general do more harm than good, remains unclear; and that teacher expectations may predict student outcomes more because these expectations are accurate than because they are self-fulfilling; together with the paper's reference list identifying Merton, R. K. (1948), The self-fulfilling prophecy, Antioch Review, 8, 193–210, and McNatt, D. B. (2000), Ancient Pygmalion joins contemporary management: A meta-analysis of the result, Journal of Applied Psychology, 85, 314–322. Note: the publisher's own record. We obtained the abstract in full but not the paper, and it states no effect sizes. journals.sagepub.com
- Repository copy of Jussim and Harber (2005) reproducing the abstract identically, and its reference list identifying Madon, S. J., Smith, A., Jussim, L., Russell, D. W., Eccles, J., Palumbo, P., and colleagues (2001), Am I as you see me or do you see me as I am: Self-fulfilling prophecies versus self-verification, Personality and Social Psychology Bulletin, 27, 1214–1224. Note: a repository copy, used as independent corroboration of the abstract text. academia.edu
- Semantic Scholar record for Jussim and Harber (2005), reproducing the abstract, and carrying indexed descriptions of related work including a study whose findings suggest that mothers' inaccurate beliefs repeatedly assessed across time had additive self-fulfilling effects on their children's subsequent alcohol use assessed at a single later point in time, which served to exacerbate differences in the alcohol use of children consistently exposed to unfavorable beliefs; and a description of Jussim (2012), Social Perception and Social Reality. Note: a bibliographic record with third-party indexed summaries. The study on additive effects reaches us as a description that truncates, and we did not obtain it. semanticscholar.org
- Second author's institutional copy of Jussim and Harber (2005), opening page, on the term teacher expectations having been known to inspire righteous indignation for teachers' supposed role in creating inequalities; on the key reason being the self-fulfilling prophecy, attributed to Merton (1948), under which erroneous teacher expectations may lead students to perform at levels consistent with those expectations, citing Brophy and Good (1974) and Rosenthal and Jacobson (1968); on it not being clear that the evidence justifies condemnations of teachers for their supposed role in creating injustices; on other researchers, equally righteous, condemning some teacher expectation research as appallingly flawed and expressing dismay at how it has been misinterpreted and has captured the popular imagination, our source truncating; and on the review conveying why 35 years of research leads to conclusions that fall between these extremes. Note: an author's own institutional PDF; we obtained the first page only. The phrase "appallingly flawed" is the authors' characterisation of a position held by other researchers, not their own verdict. nwkpsych.rutgers.edu
- Publisher record for Jussim and Harber (2005) with third-party citing text, on the test results in Rosenthal and Jacobson (1968) having been made up yet having shaped how some teachers treated the students and affected the students' educational outcomes; on the phenomenon becoming known as the Pygmalion effect or teacher expectation effect; and on teachers' expectations being communicated through differential behaviour. Note: third-party citing text on a publisher page; we did not obtain the 1968 study. researchgate.net
- Reference list on a decision-science reference site, identifying a paper titled Beleaguered Pygmalion: A history of the controversy over claims that teacher expectancy raises intelligence, Intelligence, 27(3), 199–234, DOI 10.1016/S0160-2896(99)00026-4; and confirming the citation for Jussim, L., & Harber, K. D. (2005), Personality and Social Psychology Review, 9(2), 131–155. Note: a commercial reference website. We did not obtain the Beleaguered Pygmalion paper, and our source does not clearly give its author, so we report the title and journal details only. thedecisionlab.com
This article reviews education and social psychology research and is not management or human resources advice. The review was not obtained in full and no effect sizes are reported. The 1968 study, the management meta-analysis, and the evidence behind the review's second conclusion were not obtained. The counterpoint study reaches this article as a truncated indexed description and the reconciliation offered is the authors' own conjecture. All arithmetic is the authors' own and uses invented figures. Every application to managers and employees is an extension from research on teachers and students.