Teacher expectations are teachers’ judgments about students’ likely learning, performance, progress, and educational possibilities. They matter because expectations organize how a teacher interprets evidence, allocates instructional opportunities, responds to mistakes, and anticipates improvement. Research identifies measurable relationships between teacher expectations and student achievement, including genuine self-fulfilling effects. The size and meaning of these relationships depend on whether an expectation accurately reflects prior performance, introduces perceptual bias, or changes the instruction a student receives. (Jussim and Harber, 2005) (Wang et al., 2018)
A central result of the literature is that teachers often make reasonably informative judgments while also sometimes overestimating or underestimating individual students. A favorable expectation can be educationally consequential when it produces additional access to challenging work, supportive feedback, and opportunities to demonstrate competence. A low expectation can become consequential when it narrows those opportunities. Neither sequence is automatic; institutional context, teacher beliefs, prior knowledge, and student responses shape its development. (Südkamp et al., 2012) (Rubie-Davies and Hattie, 2025)
What Teacher Expectations Mean in Educational Psychology
Expectations, beliefs, judgments, and aspirations
An expectation is a forecast: what level of achievement or improvement is anticipated for a student or class over a specified period. Teachers may expect a student to complete a reading task successfully, master a mathematical concept, enter an advanced course, or need continuing assistance. A belief is a broader proposition concerning ability, intelligence, effort, instruction, or the capacity of students to improve. Beliefs can influence the predictions a teacher forms and the decisions made in response to them. Aspirations or standards identify desirable destinations, while expectations estimate what seems likely to occur. (Rubie-Davies and Hattie, 2025) (Timmermans et al., 2016)
These distinctions matter in practice. A teacher may accurately recognize that a student currently reads below grade level while maintaining a demanding, supported learning goal for the next term. Conversely, a teacher may express the aspiration that all students succeed yet routinely assign less challenging work to particular students. Research on expectations therefore requires measures of both teachers’ predictions and their enacted instructional practices. Statements of optimism alone cannot establish that opportunities are equitable. (Rubie-Davies, 2007) (Rubie-Davies and Hattie, 2025)
Individual expectations and class-level expectations
An individual expectation concerns a particular student. A class-level expectation reflects the teacher’s judgment that the students in an entire group can make substantial progress. The distinction alters the practical target of improvement. An intervention that changes teachers’ beliefs about who can succeed may affect grouping, curriculum access, questioning, and feedback for many learners simultaneously. Studies of high-expectation teachers describe classroom environments in which ambitious goals are combined with flexible grouping and continuing academic support. (Rubie-Davies et al., 2015) (Rubie-Davies and Hattie, 2025)
Researchers also distinguish relative expectations—who is expected to outperform peers—from absolute expectations—what level of mastery is expected against an instructional standard. A teacher can rank students accurately yet underestimate the growth possible for the whole class. For this reason, a productive analysis attends to predictive accuracy, the demands of the tasks provided, and the improvement students actually demonstrate over time. (Rubie-Davies and Hattie, 2025)
Three Explanations for the Expectation–Achievement Relationship
Accuracy: expectations reflect existing evidence
Students differ in prior achievement, prior knowledge, attendance, opportunities, and engagement. Teachers observe some of these indicators repeatedly. An expectation can therefore predict later achievement because the teacher has noticed information that itself predicts subsequent performance. This is the accuracy explanation. It is central to interpretation because a simple correlation between initial teacher expectations and later test scores cannot show that the expectation produced the outcome. (Jussim, 1989) (Jussim and Harber, 2005)
The meta-analysis by Südkamp, Kaiser, and Möller combined 75 studies comparing teachers’ achievement judgments with standardized performance measures and reported an average correspondence of approximately r = .63. This is a measure of judgment accuracy, not an effect of changing expectations. The review found that informed judgments corresponded more closely to measured achievement than uninformed ones. The result supports the practical value of teacher observation while leaving room for systematic errors and substantial variation between settings. (Südkamp et al., 2012)
Perceptual bias: expectations change evaluation
An expectation can influence what an evaluator notices or how ambiguous work is interpreted. Two students may demonstrate similar performance but receive different grades, explanations, or interpretations when an evaluator begins with different prior beliefs about them. This mechanism is a perceptual or evaluative bias. It changes the recorded assessment of achievement even when the student’s underlying learning has not changed in the same way. Longitudinal evidence has detected modest bias in grades alongside evidence of predictive accuracy and smaller self-fulfilling effects. (Jussim, 1989)
The distinction between subjective grades and independently scored achievement tests is especially important. If the outcome is a teacher-assigned grade, the same person may both hold the expectation and evaluate the final result. A grade difference can partly represent a change in the teacher’s judgment. If the outcome is a blinded external assessment, the interpretation is stronger for actual learning, although confounding and selection remain possible in observational designs. (Jussim, 1989) (Wang et al., 2018)
Self-fulfilling prophecy: expectations change opportunities and learning
A self-fulfilling prophecy occurs when an initially inaccurate expectation contributes to behaviors or conditions that make the predicted outcome more likely. In education, a teacher’s expectation may influence how much time, feedback, academic challenge, participation, or support a student receives; the student’s responses then affect later achievement. The causal sequence requires change in student outcomes attributable, at least partly, to expectations operating through classroom experience. A prediction that merely recognizes preexisting differences does not satisfy this definition. (Jussim and Harber, 2005) (Gentrup et al., 2020)
The three explanations may operate simultaneously. A teacher may be mostly accurate about current skill, biased in the way certain work is graded, and causally influential through the opportunities offered next. Treating accuracy, bias, and self-fulfilling causation as separate questions makes the literature coherent and makes an equity audit more informative. (Jussim, 1989) (Jussim and Harber, 2005)
From Pygmalion to Contemporary Expectancy Research
The original historical question
The idea that one person’s belief can affect another person’s performance became widely known in education through Rosenthal and Jacobson’s 1968 Pygmalion in the Classroom study. The researchers sought to alter teachers’ expectations by identifying some children as likely intellectual bloomers, even though that designation was experimentally assigned. The provocative hypothesis was that teacher belief could influence later performance independently of children’s original capacities. The study attracted substantial attention and immediate methodological criticism. (Good et al., 2018) (Raudenbush, 1984)
The historical achievement of the Pygmalion program was to make teacher expectation effects experimentally testable and to stimulate decades of observational, developmental, and intervention studies. Its original intelligence-gain claims, measurement controversies, and uneven replications do not justify the popular formula that simply expecting more causes large increases in intelligence. Later research identified more specific psychological and instructional processes, sometimes finding effects, sometimes null results, and often smaller effects than popular retellings imply. (Good et al., 2018) (Jussim and Harber, 2005)
Experimental credibility and prior contact
Raudenbush’s synthesis of 18 experiments investigated why induced expectations produced different effects on pupil IQ. Effects tended to be smaller when teachers already knew their students well before the experimental information was supplied. This moderator matters because a weak or implausible external label may be overridden by direct classroom experience. It also means that an experiment in which a teacher knows almost nothing about a student has limited direct equivalence to an established school-year relationship. (Raudenbush, 1984)
Later narrative assessments of the first two decades of research concluded that differential teacher expectations and the communication of those expectations are real phenomena. They also recognized that studies failed to replicate uniformly and that the translation from manipulated expectations to broad claims about intelligence remains disputed. Historical Pygmalion experiments should therefore be read together with contemporary tests of achievement, instructional mechanisms, and teacher professional development. (Good et al., 2018) (Jussim and Harber, 2005)
How Expectation Effects May Develop
Formation of expectations
Teachers form expectations from test results, prior grades, student work, classroom participation, referrals, previous teacher reports, family communication, and daily interactions. These signals differ in validity. A recent correctly scored assessment can be informative for the skill it measures, whereas a reputation conveyed by another adult can carry inaccuracies or social stereotypes. Expectation formation is thus an evidence-integration problem in which useful observations and irrelevant cues can enter the same judgment. (Timmermans et al., 2016) (Timmermans et al., 2015)
Perceptions of effort, confidence, classroom behavior, and the teacher–student relationship can also influence predicted academic achievement. The study by Timmermans, de Boer, and van der Werf examined several such perceived student attributes. This matters because behavioral conformity and academic competence are separate dimensions: a quiet student may struggle academically, while a talkative or behaviorally challenging student may possess considerable subject knowledge. Educational judgments improve when observers keep the target construct explicit. (Timmermans et al., 2016)
Input: the difficulty and breadth of work students receive
Expectations can shape instructional input: the complexity of tasks, exposure to advanced material, explanations offered, and opportunities for revision. Teachers may inadvertently provide students considered highly capable with open-ended problems and richer discussions while giving others only repetitive work. Different tasks then produce different opportunities to develop complex understanding, creating a pathway by which initially modest differences can persist. Observational research on high- and low-expectation classrooms identifies systematic differences in the learning opportunities teachers organize. (Rubie-Davies, 2007) (Rubie-Davies and Hattie, 2025)
Appropriate instructional adaptation does not require identical assignments. A student who needs an accessible entry point may benefit from temporary scaffolding, explicit explanations, and additional practice. The critical issue is whether adaptation preserves access to ambitious learning goals and whether support is revised as competence develops. A fixed assignment of students to low-demand activities based on an old expectation can restrict their opportunities to demonstrate improvement. (Rubie-Davies, 2007) (Rubie-Davies et al., 2015)
Output: opportunities to participate and demonstrate knowledge
Classroom output includes being called on, answering open questions, presenting reasoning, attempting challenging problems, and receiving time to complete an explanation. Higher expectations may increase the number and difficulty of these opportunities. Participation also gives teachers new evidence about what students know, so restricted participation can reduce both learning opportunities and the accuracy of future judgments. Studies of classroom interactions identify questioning patterns as one of the practices differentiating high-expectation and low-expectation teaching. (Rubie-Davies, 2007) (Rubie-Davies and Hattie, 2025)
Participation quality matters more than an indiscriminate requirement that every student speak the same number of times. Learners vary in language proficiency, anxiety, disability-related access needs, and preparation. Equitable participation means offering multiple valid ways to show thinking, sufficient wait time, clear success criteria, and responses that advance understanding. The scientific claim is about access to meaningful opportunities, while the exact format should respond to educational goals and students’ needs. (Rubie-Davies and Hattie, 2025)
Climate: warmth, credibility, respect, and belonging
Expectations can also be communicated through social climate. A student may infer what a teacher believes from tone, patience, responsiveness to mistakes, nonverbal behavior, and willingness to invest time. Warmth alone is insufficient when instructional challenge is weak; demanding lessons can be discouraging when relational support is absent. The high-expectation classroom literature emphasizes their combination: serious academic goals embedded in a respectful, supportive classroom environment. (Rubie-Davies and Hattie, 2025) (Rubie-Davies, 2007)
Students are active interpreters of these signals. Two students may respond differently to the same question or correction because of previous experiences with adults, peer expectations, and confidence in the subject. The familiar three-step account—teacher expectation, differential treatment, student response—therefore describes a plausible process with moderators at every stage, rather than a deterministic psychological law. (Jussim and Harber, 2005) (Wang et al., 2018)
Feedback: what counts as an informative response
Feedback can carry different educational messages. A response directed at the substance of a solution identifies a mathematical error, explains why a claim needs evidence, or suggests a next step. Feedback directed primarily at compliance can miss the opportunity to develop knowledge. If teachers systematically provide detailed academic feedback to some students while offering others mainly behavioral praise or brief evaluation, students receive different information for improvement. (Gentrup et al., 2020)
Nevertheless, feedback mediation should be tested rather than presumed. Gentrup and colleagues followed 1,026 first graders in 64 German classrooms and videotaped interactions in a smaller subsample of 354 students across 19 classes. Higher expectations were associated with differences in feedback; however, the feedback indicators studied did not substantially mediate the relationship between expectations and year-end achievement. This directly limits the claim that measured feedback differences alone explain the full mechanism. Other aspects of instruction, unmeasured interactions, and measurement constraints remain relevant possibilities. (Gentrup et al., 2020)
How Strong Are Teacher Expectancy Effects?
Average effects versus exceptional classrooms
Jussim and Harber’s major review concluded that self-fulfilling effects occur but are usually modest in ordinary classroom research. They found limited evidence that such effects necessarily compound year after year, while noting that greater effects may arise in particular circumstances, including some stigmatized groups. Their account distinguished an expectation that predicts an outcome from one that changes that outcome. The conclusion should restrain dramatic universal claims about what teacher belief can accomplish by itself. (Jussim and Harber, 2005)
A more recent narrative review by Rubie-Davies and Hattie focused on teachers whose underlying beliefs and classroom practices differ systematically. It reported large contrasts between selected high- and low-bias, differentiating, and class-level-expectation teacher groups, with reported standardized differences around d = 0.85–0.92. These are contrasts across carefully selected research groupings and analytic comparisons; they are not a universal causal effect of telling all teachers to believe more strongly, nor a directly interchangeable estimate with a typical longitudinal expectancy coefficient. (Rubie-Davies and Hattie, 2025)
These apparently different summaries answer different questions. A typical student-level expectation discrepancy may have a small prospective effect, while a coordinated set of classroom practices associated with a high-expectation teaching style may correspond to a much larger group difference. To compare them responsibly, researchers must identify the exposure, comparator, outcome, unit of analysis, adjustment variables, and design. A standardized difference without that context cannot establish how much improvement a school will obtain from a new policy. (Jussim and Harber, 2005) (Rubie-Davies and Hattie, 2025)
Why a correlation is not an intervention effect
A correlation measures association, and a regression coefficient after adjustment measures conditional association. A randomized intervention estimates an effect of assignment to a particular intervention, subject to attrition, adherence, measurement, and study design. The distinction matters because changing teachers’ beliefs also frequently changes questioning, grouping, coaching, and expectations of student effort. An intervention can improve achievement through these practices without isolating a pure belief-only effect. (de Boer et al., 2018) (Rubie-Davies et al., 2015)
Similarly, an estimate derived from teachers rating students within one school system should not be automatically transported to another country, grade, or subject. Different systems use different assessment rules, tracks, student supports, and teacher responsibilities. A credible evidence synthesis separates predictive accuracy, classroom interaction mechanisms, experimental efficacy, and effects on long-term educational trajectories. (Timmermans et al., 2015) (Wang et al., 2018)
Evidence From Longitudinal and Classroom Studies
Early expectations and first-grade achievement
Gentrup and colleagues assessed teachers close to the beginning of students’ formal schooling, reducing the chance that the teacher–student relationship had already created the baseline outcomes. Controlling for prior achievement, cognitive abilities, motivation, and background, inaccuracies in expectations predicted later reading and mathematics outcomes in some directions. Inaccurately high expectations were associated with higher achievement in reading and mathematics; inaccurately low expectations predicted lower reading achievement, while the corresponding mathematics result was less consistent. (Gentrup et al., 2020)
These results are useful because they test prospective relations while recording relevant baseline characteristics. They remain observational with respect to the naturally occurring expectation discrepancies: initial expectations were not randomly assigned to teachers. Unobserved instructional variables could contribute to the association. The study also illustrates why the reading and mathematics effects cannot be compressed into one universal expectancy coefficient. (Gentrup et al., 2020)
Expectations and longer-term educational pathways
Using data on roughly 11,000 students tracked over five years, de Boer, Bosker, and van der Werf examined expectation bias in relation to later educational performance. Their longitudinal analysis found that expectation bias was associated with subsequent performance and partly mediated relationships involving student characteristics. The breadth and duration of this sample strengthen the relevance of expectation judgments to educational trajectories, including systems with consequential transitions. (de Boer et al., 2010)
Even a large multiyear observational dataset cannot by itself establish every causal link. School placements, previous attainment, family resources, peer contexts, and subsequent teacher assignments may jointly shape both expectations and outcomes. The study therefore supports the importance of monitoring expectation accuracy and institutional consequences, while a claim that a specific belief intervention would reproduce these long-term associations needs separate experimental evidence. (de Boer et al., 2010)
Classroom observation and teacher heterogeneity
Observational studies of teachers with contrasting expectation profiles have reported variation in lesson organization, questioning, collaborative learning, emotional support, and use of flexible groups. Teachers characterized by high class-level expectations commonly combine challenging tasks with frequent attention to learning progress. These practices suggest concrete ways expectations can be translated into opportunities; they also raise the possibility that teachers’ overall instructional skill contributes to group differences. (Rubie-Davies, 2007) (Rubie-Davies and Hattie, 2025)
Teacher heterogeneity is a central empirical issue. Averaging results across every teacher in a dataset can obscure small effects for most teachers and larger patterns within subsets whose beliefs lead to pronounced differential treatment. Yet identifying a distinctive subgroup after seeing outcomes can inflate estimated contrasts or reduce generalizability. The 2025 narrative review emphasizes belief moderators and notes how few researchers had repeatedly studied the same teacher-belief moderators across independent investigations. (Rubie-Davies and Hattie, 2025)
Expectations, Stereotypes, and Educational Equity
Group differences in expectations and treatment
Teacher expectations may reflect achievement information and social signals at the same time. The meta-analyses by Tenenbaum and Ruck examined differences in expectations, referrals, and teacher speech toward ethnic-minority students compared with European American students in the studies available at the time. Average differences favored European American students relative to Latino and African American students on several measures, while the pattern differed for Asian American students and for negative speech. These results document heterogeneity across groups and outcomes, not a uniform bias held by every teacher. (Tenenbaum and Ruck, 2007)
Ethnic or racial identity, family income, gender, disability status, accent, and language background can be associated with educational experiences through multiple pathways. Causal interpretation must consider previous opportunity, systemic tracking, assessment design, and how achievement was measured. Group mean differences neither diagnose bias in any one teacher nor justify using demographic membership to forecast an individual student’s academic potential. They do justify examining whether expectations exceed or fall below predictions based on relevant learning evidence. (Timmermans et al., 2015) (Tenenbaum and Ruck, 2007)
Implicit attitudes and explicitly reported beliefs
Some research distinguishes attitudes measured through explicit teacher reports from more automatic associations measured by implicit tasks. In the study by van den Bergh and colleagues, teachers’ implicit prejudiced attitudes were related to educational expectation patterns and achievement gaps. These associations support studying mechanisms beyond what teachers consciously report about their beliefs. They do not establish that an implicit-score instrument is an individual diagnostic test for misconduct or that a single score can explain a particular student’s trajectory. (van den Bergh et al., 2010)
An approach to educational equity must examine actual access to curriculum, feedback, placement and assessment, along with expectation judgments. Training that asks teachers to state inclusive values can be useful, but observable practices and student outcomes provide stronger evidence of what changed. Avoiding fixed ability labels, using current work samples, and reviewing low-stakes instructional decisions create opportunities for expectations to be revised when students demonstrate growth. (Rubie-Davies and Hattie, 2025) (Timmermans et al., 2016)
Streaming, grouping, and recommendation decisions
In systems with tracks or selective programs, teacher recommendations may affect access to later instruction. Timmermans, Kuyper, and van der Werf studied 500 teachers and 7,550 students at the end of Dutch primary school and found between-teacher differences in expectation bias linked to student background and class context. Such decisions deserve particular scrutiny because an expectation may influence not only how a student is taught today but also what educational options become available tomorrow. (Timmermans et al., 2015)
The study took place in a particular national tracking system. Its estimates and specific background categories should not be applied mechanically to U.S. course recommendations, special education placement, or college admissions. The more general implication is methodological: consequential judgments should be based on transparent learning evidence, calibrated against common standards, and examined for systematic discrepancies. School psychology procedures and diagnostic accommodations remain governed by their own professional frameworks. (Timmermans et al., 2015)
Motivation, Self-Concept, and Student Responses
An expectation can be transmitted to a learner as a message about competence, future prospects, and the value of effort. Students may interpret demanding tasks and constructive correction as signals that a teacher believes they can improve. They may also read restricted access to advanced work as evidence that further effort will not matter. These interpretations can influence participation and self-perceptions, although they vary by student and classroom. The strongest claims need studies that directly measure both students’ interpretations and later motivational outcomes. (Rubie-Davies and Hattie, 2025) (Wang et al., 2018)
Teacher expectations intersect with self-efficacy, belonging, and autonomous motivation but remain conceptually distinct. Self-efficacy concerns a learner’s judgment of their own capacity to perform tasks; teacher expectations concern the teacher’s forecast. A student can be confident despite a teacher’s pessimism or uncertain despite encouraging instruction. The broader motivational mechanisms are treated in the published scientific chapter Academic Motivation: Self-Determination Theory in Education, while the present article retains ownership of expectation formation and expectancy effects. (Rubie-Davies and Hattie, 2025)
Student agency also complicates linear models. Learners seek help, select peers, invest effort, resist labels, and form judgments about teachers’ reliability. A teacher’s initial expectation can be revised after surprising performance or maintained despite new evidence. Longitudinal study designs that repeatedly measure both parties are better suited than one-time questionnaires to test such reciprocal development. (Wang et al., 2018) (Jussim and Harber, 2005)
Can Teacher Expectations Be Changed?
What intervention meta-analysis shows
The review and meta-analysis by de Boer, Timmermans, and van der Werf examined 19 teacher-expectation interventions. They grouped approaches into changing teacher behavior, increasing awareness of expectancy effects, and addressing beliefs underlying expectations. Meta-analytic summary effects were approximately Hedges’ g = 0.38 for teacher expectations and g = 0.30 for student achievement. These values suggest potentially useful changes but do not imply that all programs or school systems will achieve the same gains. (de Boer et al., 2018)
Only a subset of the studies supplied data suitable for each quantitative synthesis. The authors identified heterogeneity and variation in methodological rigor, making program type comparisons uncertain. The narrative findings suggested that teacher support for implementation might matter, but the available data did not permit definitive moderator testing. Thus intervention effects are promising rather than a license to adopt a generic motivational workshop as a proven universal solution. (de Boer et al., 2018)
Randomized professional development targeting classroom practices
Rubie-Davies and colleagues randomly allocated 84 teachers to an intervention or control condition. Intervention teachers attended four workshops examining high-expectation classroom practices, planned changes in instructional routines, and received follow-up visits supporting implementation. Students taught by intervention teachers showed greater mathematics improvement over one year than those taught by control teachers. The trial provides a stronger causal basis for the effect of this specific professional-development package than an observational comparison of naturally high- and low-expectation teachers. (Rubie-Davies et al., 2015)
The trial also illustrates the limitation of mechanism attribution. Its intervention combined teacher learning, new classroom behaviors, planning, and implementation support. The study cannot assign all observed improvement to an isolated psychological belief independent of practice. Generalization should consider subject area, duration, number of teachers, adherence, comparison condition, and whether later studies replicate the gain. The practical intervention is a coordinated way of teaching, with expectations functioning as one component. (Rubie-Davies et al., 2015)
Training content that fits the evidence
A defensible expectation intervention helps teachers notice prediction errors, use fresh assessment information, reconsider assumptions about who can master complex material, and extend meaningful participation to all learners. It includes opportunities to practice feedback and questioning, observe classroom implementation, and evaluate what students learn. The intervention literature supports efforts to change the instructional environments accompanying expectations; the exact relative contribution of every component remains a research question. (de Boer et al., 2018) (Rubie-Davies et al., 2015)
An ethical intervention replaces fixed labels with testable predictions: What prerequisite knowledge has this student demonstrated? What teaching response is available? What evidence would change my estimate? What demanding task can be reached with support? This approach permits realistic diagnosis of instructional needs while keeping educational opportunities responsive to new evidence. It also reduces the risk that high expectations are confused with excessive pressure or denial of student support needs. (Rubie-Davies and Hattie, 2025)
Measuring Expectations and Their Effects
Measuring the forecast precisely
Researchers ask teachers to estimate likely grades, future test performance, track placement, academic improvement, or success on domain-specific tasks. A clear study identifies the forecast horizon, scale, reference group, and subject area. “How intelligent is this student?” and “How well will this student do on a reading comprehension assessment in three months?” are different questions. They cannot be treated as interchangeable measures of teacher expectancy. (Südkamp et al., 2012) (Timmermans et al., 2015)
The timing of measurement matters. An expectation recorded after months of instruction may already incorporate learning that teachers helped create, making causal order difficult to establish. A baseline expectation recorded before substantial contact better captures initial judgment, though it may still reflect previous records or socially transmitted information. Studies should preserve the distinction between expected level, expected growth, and the discrepancy between expectation and well-specified baseline predictors. (Gentrup et al., 2020)
Measuring bias and calibration
One way to estimate expectation bias is to compare a teacher’s prediction with a statistical prediction built from students’ prior achievement and other defensible antecedent variables. A positive residual indicates higher expectations than predicted by the selected model; a negative residual indicates lower expectations. This approach depends heavily on what predictors the model includes. Omitting relevant performance information can make an accurate teacher appear biased; including variables that encode historical inequality can make disadvantage appear normatively expected. (de Boer et al., 2010) (Timmermans et al., 2015)
Calibration concerns whether predictions of a specified outcome match outcomes over repeated cases, including whether a given probability corresponds to the observed frequency. Ranking accuracy, absolute error, and calibration tell different stories. A teacher who reliably ranks students may still underestimate every student’s capacity for growth. For educational equity, analysts must also examine calibration by group and decision threshold, while guarding against small samples and unstable comparisons. (Südkamp et al., 2012) (Timmermans et al., 2015)
Measuring student outcomes
Relevant outcomes include independently scored achievement, task mastery, persistence, course access, graduation pathways, academic self-concept, and participation. They operate on different time scales. A short-term test improvement does not establish persistent transfer, while higher teacher-assigned grades may partly reflect grading bias. A study should predefine which outcome would constitute success, measure both benefits and unintended costs, and distinguish a change in rating from a change in knowledge or skill. (Jussim, 1989) (Wang et al., 2018)
What Makes Causal Research Credible?
Randomly inducing a teacher expectation can isolate some aspects of belief, but it must be credible to the teacher and ethically managed. Randomly assigning teachers to a professional-development program estimates the effect of that entire package, including implementation. Natural experiments, repeated measures, and careful multilevel analyses can contribute evidence when randomization is impractical, but each design has specific assumptions. The Pygmalion history illustrates why experimental manipulation quality and measurement validity are essential. (Raudenbush, 1984) (de Boer et al., 2018)
Students are nested in classrooms and schools. Their outcomes are correlated because they share teachers, curricula, peers, and resources. Analyses that treat every student observation as independent risk overstating precision. Baseline imbalance, teacher selection, attrition, reporting bias, teacher awareness of the measurement, and publication selection can further distort apparent expectancy effects. Research quality therefore depends on the correct unit of randomization and analysis, preregistered outcomes when available, transparent missing-data handling, and replication. (Timmermans et al., 2015) (Wang et al., 2018)
Mediation claims require particularly careful designs. To say that teacher feedback explains an expectancy effect, studies must show appropriate temporal ordering and evaluate whether feedback mediates the association after relevant confounding is addressed. Gentrup and colleagues found a link between expectations and feedback while obtaining little evidence that their measured feedback indices explained the expectancy–achievement relationship. A plausible mechanism and a demonstrated statistical mediator are different levels of evidence. (Gentrup et al., 2020)
Meta-analyses pool estimates whose meanings may differ. Effect-size labels must state whether the statistic is a correlation, standardized mean difference, regression coefficient, or treatment effect. Study weighting, heterogeneity, outcome selection, small-study effects, and the possibility of overlapping samples deserve attention. This is especially important when juxtaposing Jussim and Harber’s average-effects synthesis, the intervention meta-analysis, and narrative comparisons of teacher subtypes. (Jussim and Harber, 2005) (de Boer et al., 2018) (Rubie-Davies and Hattie, 2025)
Age, Subject, and Cultural Context
The evidence spans elementary schooling, secondary transitions, and a range of national systems, but samples are not distributed evenly across all contexts. An effect found among first graders in German reading classes should not be assumed identical in university engineering, adult education, or multilingual online instruction. Expectations may be communicated differently where teachers see students daily compared with short-term or asynchronous courses. (Gentrup et al., 2020) (Wang et al., 2018)
Age and prior contact can alter both the content and influence of a teacher’s forecast. Younger learners may have fewer independent academic records and less prior contact with a new teacher; older students may already have consolidated self-beliefs, past grades, and institutional pathways. Raudenbush’s experimental review found weaker expectancy-induction effects when teachers knew students better, reminding researchers to distinguish new impressions from established teacher knowledge. (Raudenbush, 1984)
Different subjects offer different opportunities for visible performance and error correction. Mathematics problems may yield relatively immediate right-or-wrong information, whereas literacy performance can include more open judgments about reasoning and expression. In the German first-grade study, the associations linked to inaccurately low expectations differed between reading and mathematics. Researchers should not assume the same mechanism, size, or policy implications across disciplines. (Gentrup et al., 2020)
Cultural context shapes classroom authority, ability grouping, parent–teacher relationships, and the meaning of teacher recommendations. The Dutch tracking studies illustrate consequences in one institutional structure, while the U.S.-focused race and ethnicity meta-analyses address different categories and educational histories. Comparing them demands attention to how categories were defined and measured rather than importing labels across societies. (Timmermans et al., 2015) (Tenenbaum and Ruck, 2007)
A Research-Grounded Classroom Approach
Set ambitious goals and identify the path to them
High expectations become instructionally meaningful when a teacher identifies challenging, learnable goals and specifies the knowledge and support needed to reach them. A teacher can acknowledge current difficulties directly and still design a progression toward demanding work. The research on high-expectation teaching describes flexible grouping, mastery-focused goals, student choice, and systematic progress monitoring as part of a supportive instructional environment. (Rubie-Davies, 2007) (Rubie-Davies and Hattie, 2025)
A practical planning question is: Which students are currently excluded from the most cognitively demanding parts of a lesson, and what would make participation possible? This question can lead to scaffolds, modeling, additional practice, differentiated entry points, and opportunities to revise work. It avoids replacing evidence-informed support with a uniform task that is inaccessible to some learners. (Rubie-Davies et al., 2015)
Make participation and response quality visible
Teachers can inspect whom they call on for explanation, who receives follow-up questions, whose incorrect responses are explored, and whose work receives substantive feedback. A simple observation protocol can sample interactions across several lessons rather than relying on memory. The aim is to check whether students with comparable learning needs receive comparable access to discussion and corrective information, while accommodating individual participation needs. (Rubie-Davies and Hattie, 2025) (Rubie-Davies, 2007)
When a student’s answer is incomplete, the next step can focus on the underlying concept instead of using the response as confirmation that the student is incapable. Good questioning permits teachers to gather new evidence, which can revise their forecasts. Related instructional processes, including the role of evidence in adapting instruction, are examined in Formative Assessment: Using Evidence to Improve Learning. (Gentrup et al., 2020)
Recalibrate expectations using new student work
Expectations should be provisional and updateable. Teachers can compare predictions with independently assessed work, document growth on a transparent rubric, and revise instructional plans in response to new performance. Reviewing predictions by achievement band and student group can reveal patterns worth investigating. A responsible audit also asks whether results were affected by uneven opportunities to learn the assessed material. (Südkamp et al., 2012) (Timmermans et al., 2015)
School leaders can support this process through moderated grading, shared rubrics, common assessments, observation of instructional opportunities, and professional discussion of evidence. These are practical implications of the accuracy-and-bias literature, not claims that one standardized audit has already been proven to remove all expectancy effects. Interventions should be evaluated with appropriate comparison groups and student outcomes. (de Boer et al., 2018)
Combine high challenge with psychological support
The instructional goal is a climate in which errors provide useful information, students receive serious tasks, and support remains available. A belief in student improvement is educationally productive when translated into time, resources, feedback, and repeated opportunities to learn. Teachers also need to identify genuine barriers and adapt teaching rather than treating positive expectation as a substitute for instruction. (Rubie-Davies et al., 2015) (Rubie-Davies and Hattie, 2025)
For students experiencing academic anxiety, explicit expectations should be communicated with clear success criteria, manageable learning steps, and predictable support. The separate article Test Anxiety in Students: Symptoms, Performance, and Support reviews the evidence on test-related symptoms and educational support. Expectancy research alone does not establish a mental-health diagnosis or justify a clinical intervention. (Wang et al., 2018)
What Schools and Policymakers Can Infer
School policy can treat expectations as part of instructional quality, assessment reliability, and educational equity. Professional development should demonstrate how a teacher can provide ambitious tasks and responsive support, and evaluations should look for changes in classroom opportunities and student learning. A schoolwide slogan about believing in every child is inadequate as an outcome measure. Randomized professional-development evidence supports the possibility of improvement when teaching practice changes, though the magnitude varies by implementation. (Rubie-Davies et al., 2015) (de Boer et al., 2018)
Placement and recommendation systems should document why a forecast was made, what student work supports it, and how it can be challenged or revised. Where demographic gaps appear, analysts must examine the quality of the measures, differences in opportunity to learn, class context, and the effects of decisions themselves. Statistical adjustment helps identify possible discrepancies but cannot mechanically determine a just outcome without institutional context. (Timmermans et al., 2015) (Tenenbaum and Ruck, 2007)
Teacher preparation programs can distinguish evidence of current performance from beliefs about stable ability and create opportunities for trainees to practice challenging interactions equitably. Coaching can target specific, observable behaviors: distribution of demanding questions, elaboration after mistakes, curriculum access, and revision opportunities. These targets align more closely with the mechanisms investigated than does instruction to maintain permanent optimism regardless of evidence. (Rubie-Davies, 2007) (Rubie-Davies and Hattie, 2025)
Common Misinterpretations of the Pygmalion Effect
The claim that expectations automatically create achievement confuses correlations, changes in grading, and effects on learning. The claim that every teacher’s expectation is biased overlooks evidence that teachers’ judgments often contain predictive information. The opposite claim that accurate forecasts can never cause harm ignores the possibility that a partly accurate forecast may become fixed and restrict subsequent opportunity. The strongest formulation is conditional: expectations can contribute to learning outcomes when they shape meaningful teacher behavior and student experience. (Jussim and Harber, 2005) (Südkamp et al., 2012)
Large standardized differences reported for selected teacher profiles cannot be presented as the typical result of a one-hour workshop. Effect sizes from different study designs and outcomes answer different questions. Teacher-expectation interventions show encouraging but heterogeneous average effects, and the best-known early experiment encountered replication and measurement disputes. Sound communication of the research states these features directly. (Rubie-Davies and Hattie, 2025) (de Boer et al., 2018) (Raudenbush, 1984)
Finally, high expectations do not mean demanding identical work, minimizing disability-related barriers, or using harsh criticism to force achievement. A strong educational expectation concerns the possibility of learning with effective instruction and support. Equitable challenge requires accessible pathways, accurate assessment, appropriate accommodations, and a willingness to update what is expected when students reveal new knowledge. (Rubie-Davies et al., 2015)
Frequently Asked Questions
Do teacher expectations actually affect student achievement?
Yes. Longitudinal studies and intervention research support teacher expectancy effects, including changes in achievement after programs targeting teachers’ expectations and practices. The average self-fulfilling component in ordinary observational research is often modest; effects differ by setting, teacher behavior, outcome, and research design. The key is identifying what expectations cause teachers to do differently and which students consequently learn more. (Jussim and Harber, 2005) (de Boer et al., 2018)
Is the Pygmalion effect scientifically proven?
The general existence of some expectancy effects is supported by decades of research. The broad claim that raising any teacher’s expectations reliably causes large gains in intelligence is not established. The original Pygmalion study generated criticism, and follow-up experiments showed important moderators, including how well teachers already knew their students. Classroom and intervention studies provide a more precise account than the simplified historical slogan. (Good et al., 2018) (Raudenbush, 1984)
Are teachers usually accurate about student ability?
Teachers often provide moderately informative academic-achievement judgments. A meta-analysis of 75 studies found an average teacher-judgment correspondence of about r = .63 with standardized achievement measures. That statistic refers to achievement judgment, not a complete or fixed measure of intelligence, and it does not exclude systematic errors for individual students or groups. (Südkamp et al., 2012)
Do low expectations always damage learning?
No single expectation determines a student’s trajectory. Low expectations can be harmful when they reduce challenging input, participation, and responsive feedback, but the extent depends on teacher actions and student context. Some low forecasts accurately reflect current performance; teachers can respond to such evidence by providing stronger instruction and maintaining demanding growth goals. (Jussim and Harber, 2005) (Rubie-Davies and Hattie, 2025)
Can professional development improve expectation effects?
Some programs have improved teacher expectations and student achievement. The 2018 intervention meta-analysis found positive mean effects with heterogeneous designs, while a randomized study trained teachers in high-expectation classroom practices and detected improved mathematics outcomes. Implementation, teacher engagement, and the components of the intervention matter. (de Boer et al., 2018) (Rubie-Davies et al., 2015)
How can a teacher avoid unfair expectations?
Use specific recent evidence, distinguish academic mastery from compliance, monitor who receives demanding tasks and explanatory feedback, and review predictions when student performance changes. Where placement decisions are consequential, use transparent criteria and check for systematic patterns across groups. These are evidence-informed practices grounded in research on accuracy, bias, and high-expectation instruction; no brief checklist guarantees the elimination of every form of inequity. (Timmermans et al., 2015) (Rubie-Davies and Hattie, 2025)
Conclusion
Teacher expectations are judgments that can describe learning accurately, bias evaluation, and influence learning through classroom experience. Their educational significance lies in the interaction among evidence, belief, instructional opportunity, and students’ developing performance. The central scientific finding is conditional and actionable: teachers often possess useful information about learners, but expectations can also change what learners are allowed and encouraged to accomplish. Research supports high challenge with responsive support, continual revision of judgments, and assessment of real learning outcomes. (Jussim and Harber, 2005) (Rubie-Davies et al., 2015) (Rubie-Davies and Hattie, 2025)
Related Articles
Cluster Learning and Educational Psychology
Formative Assessment: Using Evidence to Improve Learning
Academic Motivation: Self-Determination Theory in Education
Learning Theories in Education: Behaviorism, Cognitivism, Constructivism, and Sociocultural Learning
Test Anxiety in Students: Symptoms, Performance, and Support
Differentiated Instruction: Benefits, Evidence, and Implementation — Examine differentiated teaching by learner needs, design constraints, and evidence on educational outcomes.
Classroom Management: Learning Climate and Evidence-Based Practices — Analyze routines, supportive classroom practices, learning time, and behavior approaches that aid instruction.
Cooperative Learning and Peer Instruction: Evidence and Conditions — Assess structured collaboration, peer explanation, accountability, and effects on achievement and engagement.
References
de Boer, H., Bosker, R. J., and van der Werf, M. P. C. (2010). Sustainability of teacher expectation bias effects on long-term student performance. Journal of Educational Psychology, 102(1), 168–179. https://doi.org/10.1037/a0017289
de Boer, H., Timmermans, A. C., and van der Werf, M. P. C. (2018). The effects of teacher expectation interventions on teachers’ expectations and student achievement: Narrative review and meta-analysis. Educational Research and Evaluation, 24(3–5), 180–200. https://doi.org/10.1080/13803611.2018.1550834
Gentrup, S., Lorenz, G., Kristen, C., and Kogan, I. (2020). Self-fulfilling prophecies in the classroom: Teacher expectations, teacher feedback and student achievement. Learning and Instruction, 66, 101296. https://doi.org/10.1016/j.learninstruc.2019.101296
Good, T. L., Sterzinger, N., and Lavigne, A. (2018). Expectation effects: Pygmalion and the initial 20 years of research. Educational Research and Evaluation, 24(3–5), 99–123. https://doi.org/10.1080/13803611.2018.1548817
Jussim, L. (1989). Teacher expectations: Self-fulfilling prophecies, perceptual biases, and accuracy. Journal of Personality and Social Psychology, 57, 469–480. https://doi.org/10.1037/0022-3514.57.3.469
Jussim, L., and 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. https://doi.org/10.1207/s15327957pspr0902_3
Raudenbush, S. W. (1984). Magnitude of teacher expectancy effects on pupil IQ as a function of the credibility of expectancy induction: A synthesis of findings from 18 experiments. Journal of Educational Psychology, 76(1), 85–97. https://doi.org/10.1037/0022-0663.76.1.85
Rubie-Davies, C. M. (2007). Classroom interactions: Exploring the practices of high and low expectation teachers. British Journal of Educational Psychology, 77, 289–306. https://doi.org/10.1348/000709906X101601
Rubie-Davies, C. M., and Hattie, J. A. (2025). The powerful impact of teacher expectations: A narrative review. Journal of the Royal Society of New Zealand, 55(2), 343–371. Published online August 26, 2024. https://doi.org/10.1080/03036758.2024.2393296
Rubie-Davies, C. M., Peterson, E. R., Sibley, C. G., and Rosenthal, R. (2015). A teacher expectation intervention: Modelling the practices of high expectation teachers. Contemporary Educational Psychology, 40, 72–85. https://doi.org/10.1016/j.cedpsych.2014.03.003
Südkamp, A., Kaiser, J., and Möller, J. (2012). Accuracy of teachers’ judgments of students’ academic achievement: A meta-analysis. Journal of Educational Psychology, 104(3), 743–762. https://doi.org/10.1037/a0027627
Tenenbaum, H. R., and Ruck, M. D. (2007). Are teachers’ expectations different for racial minority than for European American students? A meta-analysis. Journal of Educational Psychology, 99(2), 253–273. https://doi.org/10.1037/0022-0663.99.2.253
Timmermans, A. C., de Boer, H., and van der Werf, M. P. C. (2016). An investigation of the relationship between teachers’ expectations and teachers’ perceptions of student attributes. Social Psychology of Education, 19, 217–240. https://doi.org/10.1007/s11218-015-9326-6
Timmermans, A. C., Kuyper, H., and van der Werf, G. (2015). Accurate, inaccurate, or biased teacher expectations: Do Dutch teachers differ in their expectations at the end of primary education? British Journal of Educational Psychology, 85(4), 459–478. https://doi.org/10.1111/bjep.12087
van den Bergh, L., Denessen, E., Hornstra, L., Voeten, M., and Holland, R. W. (2010). The implicit prejudiced attitudes of teachers: Relations to teacher expectations and the ethnic achievement gap. American Educational Research Journal, 47(2), 497–527. https://doi.org/10.3102/0002831209353594
Wang, S., Rubie-Davies, C. M., and Meissel, K. (2018). A systematic review of the teacher expectation literature over the past 30 years. Educational Research and Evaluation, 24, 124–179. https://doi.org/10.1080/13803611.2018.1548798