Demand characteristics refer to clues or signals in an experimental setting that hint to participants about the experimenter’s expectations, leading them to behave in a certain way to match these expectations, potentially biasing the results.
Research participants often act in ways that are inconsistent with how they would normally behave in the real world. This is a result of demand characteristics (Orne, 1959): subtle hints that suggest to a participant what the experimenter predicts or hopes to find.
Noticing these hints motivates the participant to conform to the experimenter’s expectations, changing the outcome of the experiment (Orne, 2009).
Participants may change their behavior consciously or unconsciously. Either way, the change may or may not match what the experimenter was actually hoping for.
Demand characteristics can have a huge impact on the results of any psychological study.
Psychology experiments are different from experiments in fields that rely on cells or bacteria samples. They require participation from real, everyday people.
Participants often have to answer questionnaires or take part in simulations. Researchers use these to draw broader conclusions about how the human brain operates and how people act in certain situations.
For that reason, participants must be truthful and accurate when filling out a self-report questionnaire, being interviewed, or taking part in a role-play situation.
Otherwise, they risk skewing the data. The researcher might draw faulty conclusions.
Types of demand characteristics
Many different types of demand characteristics can bear weight on the outcome of an experiment. These characteristics provide clues regarding the overall research hypothesis.
Here are some of the most common types of demand characteristics:
Study rumors
Information relevant to the experiment but learned about outside of the experiment itself (Orne, 1962).
For example, a study may involve reading a story first. Unbeknownst to the participant, the study may later prompt them to recall details from that story after a series of unrelated tasks.
If rumors circulate that this is the procedure, a participant may pay extra close attention and make an effort to remember the details, thus skewing the results of the study.
Lab setting
The specific location in which the experiment is performed is itself a cue (Orne, 1962). A sparse room with one-way mirrors and a clipboard-carrying researcher signals “this is science” in a way a school gym or a living room does not.
That signal alone changes how someone behaves. A study conducted in an artificial laboratory setting, rather than a natural one, places hidden demands on the participant to respond in a certain way.
Order of procedure
The order of the actual questions in the experiment may bias the participants to answer in a certain way (Orne, 1962).
For example, if demographic questions are asked at the beginning, a participant becomes more aware of their own demographics, such as their race, gender, or socioeconomic status. They may then subconsciously, or consciously, respond to later questions in a way that adheres to stereotypes or expectations about that group.
In psychology, this is called a self-fulfilling prophecy (Merton, 1948). Exposure to beliefs about a group’s expectations leads a person to act in ways that conform to them.
Explicit or implicit communication
Any form of communication, intentional or not, can affect the outcome of the study (Orne, 1962). Even a smile counts.
For example, a researcher might consciously or unconsciously smile or frown during the experiment. This can suggest to the participant that their answers are what the researcher hoped for. It also affects their future responses.
These four types of demand characteristics are by no means the only ones that occur in research studies. Even the title of an experiment, or a tool such as a video camera, can create hidden demands for how a participant should act.
With all these cues in mind, it is worth considering how they may affect a participant and, as a result, the study’s findings.
Orne’s Classic Demonstrations
Martin Orne first outlined these ideas in a 1959 conference presentation and developed them fully in a landmark 1962 paper.
Aim: Orne (1962) wanted to show that participants do not behave as passive responders to a stimulus. Instead, they actively investigate the real purpose.
Method: Orne gave volunteers sheets of random digits to add, then told them to tear each completed sheet into 32 pieces before starting the next, with no explanation why. In separate work, his “real–simulator” design tested unhypnotized volunteers pretending to be hypnotized alongside genuinely hypnotized participants.
Results: Participants in the tearing task persisted for hours at an obviously meaningless activity, simply because it had been framed as “an experiment.” Simulators behaved indistinguishably from genuinely hypnotized participants.
Conclusion: Orne concluded that participants treat “being in an experiment” as a social contract in which almost any instruction becomes legitimate. That compliance is itself a source of bias masquerading as a genuine effect.
The real–simulator design also became a quasi-control technique. Simulators face the same cues as real participants but cannot be affected by the manipulation itself. Any behavior common to both groups can therefore be attributed to demand characteristics rather than to the manipulation.
How do demand characteristics affect participants?
Beyond just being present in the study, demand characteristics bear weight on the performance of participants.
There’s a famous psychology study called the white bear problem, where participants were told explicitly not to think of a white bear. This made them think of it even more. Wegner (1994) called this ironic process theory.
In other words, conscious attempts to suppress certain thoughts make them more likely to surface. And this is what can happen with demand characteristics. Once a participant becomes aware of their presence, it becomes increasingly hard to ignore them.
For example, if a participant realizes that the experimenter is smiling, it’s hard to stop noticing it and try to act in an honest manner.
Here are four specific ways in which demand characteristics affect the behavior of participants once they become aware of the demand characteristics:
Good-participant role
These participants try to assist the researcher with their findings, acting in ways that support what they believe to be the hypothesis. Being a “good subject” who contributes usefully to science is rewarding in itself (Nichols & Maner, 2008).
For example, a participant who believes a memory experiment is testing whether calming music improves recall may try unusually hard to memorize the list in the music condition specifically.
Negative-participant role
These participants do the opposite. Having guessed the hypothesis, they deliberately behave in ways that disconfirm or embarrass the researcher, such as by lying in their responses or answering inconsistently (Weber & Cook, 1972).
This “screw-you” behavior can arise from resentment at being used as a research subject. Sometimes it is simply contrarian mischief.
For example, a participant who suspects a researcher wants to see attitudes toward recycling improve might deliberately report a less favorable attitude after reading a persuasive message. They do this purely to spite the study’s perceived aim, not because their view actually changed.
Apprehensive subject
These participants try to produce the most socially desirable answers to avoid being judged by the experimenter.
This phenomenon is called the social desirability bias (Paulhus, 1984), whereby participants over-report more socially desirable attitudes and behaviors (Weber & Cook, 1972).
For example, if filling out a self-worth questionnaire, participants may report more positive feelings about themself than they actually have.
Faithful subject
This participant follows the instructions of the study exactly as presented, trying not to let any guess about the hypothesis affect their behavior (Weber & Cook, 1972). In principle, this is the ideal participant.
In practice, it is the hardest role to sustain. Once a participant has guessed the hypothesis, even trying not to be influenced by that guess is itself a response to it.
Investigator (Experimenter) Effects
Demand characteristics are not the only source of bias.
The researcher’s own expectations can also shape a study, through investigator (or experimenter) effects (Rosenthal & Fode, 1963).
This happens in two main ways.
During data collection, a researcher who expects a particular result may unintentionally treat participants in different conditions differently, such as with a warmer tone or a longer pause.
During interpretation, a researcher’s prior belief about what “should” be found can subtly steer judgment calls, such as how an ambiguous behavior gets scored (Harris & Rosenthal, 1985).
Rosenthal and Fode’s (1963) Rat Study
Aim: Rosenthal and Fode (1963) tested whether an experimenter’s expectations about a subject’s ability could measurably affect that subject’s performance. They used rats to rule out the chance that human participants were simply complying with social cues.
Method: Student experimenters were each given rats from the same genetic stock and told, falsely, that their rats had been bred to be either “maze-bright” or “maze-dull.” The students ran the maze trials and recorded performance.
Results: Rats believed to be “maze-bright” performed significantly better than rats believed to be “maze-dull,” despite there being no genuine difference between the two groups.
Conclusion: Rats cannot consciously infer or comply with a hypothesis. The result therefore suggested that the experimenters’ expectations shaped the rats’ actual training conditions, such as through subtle differences in handling or timing.
This strengthened the case that investigator effects are a genuinely distinct mechanism from demand characteristics, not just another name for the same thing.
Rosenthal and Jacobson’s (1968) Pygmalion Study
Aim: Rosenthal and Jacobson (1966, 1968) tested whether the same expectancy effect shown with rats would also apply to how teachers’ expectations shape their pupils’ intellectual development.
Method: Pupils in 18 classrooms were given a disguised IQ test, described to teachers as identifying children about to show an unusual “spurt” in growth. About 20% of children in each classroom were then randomly labeled to their teachers as “spurters.”
There was no real basis for this in the scores.
Results: Children randomly labeled as “spurters” showed significantly greater IQ gains over the following months than their classmates, despite no real difference between the groups at the outset. The effect was strongest among the youngest children.
Conclusion: Teachers’ expectations about a pupil, however arbitrarily formed, can become self-fulfilling. Teachers may unknowingly give “expected” high achievers more encouragement, and those pupils may respond by performing better.
This finding, sometimes called the Pygmalion effect, has since been supported across many later studies of interpersonal expectancy effects (Harris & Rosenthal, 1985).
Why do demand characteristics matter?
Demand characteristics matter because they can alter the results of a research study. Once a participant becomes aware of these hidden demands, it becomes incredibly difficult not to let them affect their behavior and responses.
Demand characteristics are a type of extraneous variable in a psychology experiment: any variable, other than the one being tested, that could affect the results if left uncontrolled.
Thus, these characteristics pose a threat to both the internal and external validity of the study (Spencer, 1978).
With these demands present, it becomes harder to say whether the independent variable is truly responsible for the change in the dependent variable. The participant’s own altered behavior may have played a role instead, affecting the internal validity of the study.
Internal validity is how confidently a result can be attributed to the independent variable rather than some other factor.
Similarly, these demand characteristics limit generalizability too. The findings cannot be generalized to populations outside of the research study itself, impacting the study’s external validity.
How To Reduce Demand Characteristics
Don’t give up hope just yet. Even though demand characteristics are relatively prevalent in psychology studies, it does not mean we can’t combat them.
There are many methods researchers can take to help reduce the impact these demand characteristics have on the results of a study, such as:
Deception to conceal demand characteristics
Trying to deceive participants about the study’s true meaning helps prevent them from behaving in a certain way (Orne, 2009). But it comes with a cost.
Deception can sometimes be an unethical practice, so researchers must debrief participants on the study’s actual purpose at the end. Researchers also often include filler questions to distract participants from the true research purpose.
For example, they might insert a memory test midway. This makes participants think the researcher is studying memory, when in reality the experiment has nothing to do with it.
Between-subjects design
In a between-subjects design, participants are either part of the experiment or control group instead of receiving both treatments in a within-groups design.
A within-groups design makes it easier for a participant to discern what the study is about, because they can see the differences between the two groups. A between-subjects design is therefore better for combating demand characteristics.
Implicit measures
Implicit, or indirect, measures can help reduce the impact of demand characteristics (Orne, 2009). Reaction-time tasks are a common example: they tap a response participants cannot consciously control.
These tasks work below conscious awareness. Participants therefore have far less control over their answers. They cannot deliberately try to either confirm or sabotage the researcher’s hypothesis.
These measures also help reduce social desirability bias, since participants cannot explicitly respond in ways they think will look good.
A reaction-time task usually asks for a response in a fraction of a second. That is well before anyone could consciously fake it.
That speed is what makes the measure genuinely hard to fake in either the good-participant or the negativistic direction of demand characteristics.
Single-blind and double-blind procedures
A single-blind study keeps participants unaware of which condition they are in, such as whether they received a real drug or a placebo. A placebo is an inactive substance made to look identical to the real treatment.
Expectations still matter, though.
This reduces the risk that a participant’s own expectations distort the outcome. It does not, on its own, control the researcher’s side of the equation.
A double-blind design goes further, withholding the same information from whoever is running or scoring the study. An experimenter’s expectations can act as a confound too. Keeping the person delivering the treatment blind removes that channel of bias as well.
Double-blind, placebo-controlled, randomized designs are often called the methodological “gold standard” for testing causal claims. Some interventions, such as surgery or psychotherapy, cannot realistically be disguised from either party, which limits how far blinding can be pushed.
Critical Evaluation
Demand characteristics are a real but often overstated risk. Orne’s (1962) and Rosenthal and Fode’s (1963) foundational studies were compelling, but they were informal or small-scale by modern standards.
For decades, the field had no systematic estimate of how large or consistent these effects actually are across studies. Modern meta-analytic evidence now supplies that estimate (see Contemporary Research, below), and it paints a more mixed picture than the classic narrative implies.
Deception creates a genuine ethical tension too. Concealing a study’s true purpose is one of the most effective ways to stop participants guessing the hypothesis, but it conflicts with the right to informed consent (Orne, 2009).
Debriefing participants afterwards mitigates, but does not eliminate, this tension.
Contemporary Research
Coles, Wyatt, and Frank (2025) carried out a three-level meta-analysis pooling 253 effect sizes from 53 studies that had experimentally manipulated explicit demand characteristics.
For example, some studies directly hinted to one group of participants what the study’s hypothesis was, then compared their responses with a group given no hint.
Averaged across all 53 studies, explicit demand characteristics produced a small but real shift toward hypothesis-consistent responding (Hedges’ g = 0.21). The effect was also highly inconsistent.
The 95% prediction interval for a new, similar study ran from a large increase in hypothesis-consistent responding down to a moderate shift in the opposite direction.
This is direct, quantitative confirmation of a pattern researchers had described qualitatively for decades: demand characteristics do not push behavior reliably in one direction. They can inflate, have no effect on, or even reverse a study’s apparent findings, depending on the sample and paradigm.
This picture is reinforced by a different research tradition. Mummolo and Peterson (2019) tested for demand effects directly in large-scale online survey experiments, across more than 12,000 participants and five separate replicated designs. Even direct financial incentives to respond in the “expected” direction rarely produced a detectable shift.
Together, these findings suggest a more nuanced conclusion than either “demand characteristics are a pervasive threat” or “demand characteristics are a myth.”
The risk appears real on average. It concentrates in paradigms with obvious manipulations and repeated-measures designs, and shrinks in more anonymous, incentive-structured designs such as many modern online surveys.
Frequently Asked Questions
Is participant bias the same as demand characteristics?
Participant bias occurs when a participant consciously or unconsciously responds in a way that they think the researcher wants them to (Brito, 2017).
As such, demand characteristics are often the cause of participant bias, placing hidden demands that biases the participant and alters their behavior.
Is participant bias the same as expectancy effects?
Participant bias and expectancy effects are similar but not the exact same. Participant bias, on the one hand, involves the participant changing their behavior to what they think the researcher wants rather than acting as they normally would.
Expectancy effects, on the other hand, occur when the researcher’s own cognitive biases affect the way they interact with the participant (Harris & Rosenthal, 1985). This, in turn, may produce participant bias, altering the participant’s behavior whether knowingly or unknowingly.
Is participant bias the same as the placebo effect?
The placebo effect occurs when the participant thinks they had the treatment effect when, in fact, they did not (Sliwinski & Elkins, 2013). As such, they alter their behavior under the wrongful assumption that they have been assigned a certain condition in the experiment.
This is a type of participant bias, where participants change their behavior because of the specific reason that they think they are in the treatment or experimental group when they are actually in the control group.
Is participant bias the same as reactivity (observer effect)?
In psychology, expectancy effects are another name for the observer effect, both of which are a form of reactivity (Harris & Rosenthal, 1985).
As such, while participant bias refers to the participant responding in a way they think the researcher wants, the observer effect occurs when the researcher themselves acts in a way that biases the participant’s behavior.
Is the Hawthorne Effect the same as demand characteristics?
The Hawthorne Effect occurs when a participant alters their behavior after becoming aware that they are actively being observed (Landsberger, 1958).
A common example in the literature demonstrated that when medical workers were aware that they were being watched, they were much more likely to regularly use antiseptic hand rub when washing their hands (Eckmanns et al., 2006).
While the Hawthorne effect is more about observation, demand characteristics are the clues that reveal what the experimenter wants.
The key difference is that the Hawthorne effect doesn’t necessarily result in the behavior that the researcher wants. It just results in altered behavior as a result of being watched.
What is the difference between demand characteristics and social desirability?
As mentioned, social desirability refers to the bias of participants to act or respond in a way that they perceive to be socially desirable.
While certain demand characteristics may produce social desirability bias, the key difference is that demand characteristics typically involve the participant trying to ascertain the research hypothesis, whereas social desirability bias can result in behavior that may or may not coincide with the researcher’s aims.
Are demand characteristics a confounding variable?
Confounding variables are variables other than the independent variable that affect the outcome of the dependent variable (VanderWeele & Shpitser, 2013).
As we know with demand characteristics, the experiment has caused them to implicitly or explicitly alter their behavior from what it would have been without the demand characteristics present.
Thus, demand characteristics are a type of confounding variable that has an impact on the results of the study beyond the independent variable.
Are demand characteristics extraneous variables?
Demand characteristics are a type of extraneous variable that can affect the outcome of a study. They can serve to invalidate the results of a study altogether by providing an alternative explanation for the observed results (Kalton, 1968).
Demand characteristics can be classified as both an extraneous and a confounding variable that can greatly impact the outcome of an experiment. But luckily, we’ve discussed ways to help combat the impact demand characteristics can have on experiments.
Do demand characteristics affect internal validity?
As mentioned earlier, demand characteristics make it extremely hard to know whether the independent variable truly is responsible for the change in the dependent variable or if the participant’s subsequent altered behavior played a role.
Therefore, demand characteristics can pose a great threat to the internal validity of an experiment and can, in extreme cases, completely invalidate the results altogether.
Do demand characteristics affect ecological validity?
Ecological validity refers to the extent to which the design of an experiment matches the participant’s real-world context.
Consequently, certain demand characteristics can affect the ecological validity of an experiment, such as an experimenter’s implicit or explicit behavior (and there isn’t ever an experimenter administering a research task in real life!).
The most obvious demand characteristic that can bear weight on the ecological validity of an experiment is the lab setting in which most experiments are conducted (Orne, 1968). As we discussed, the very environment of a study can signal certain clues that affect the way a participant behaves.
And a lab setting certainly does not mirror that of a real-world environment. Thus it becomes clear that demand characteristics can often (and in almost all cases do) affect ecological validity.
Key Takeaways
- Two-Sided Problem: Demand characteristics (participants guessing the hypothesis) and investigator effects (a researcher’s own expectations shaping the study) are two sides of the same social situation, not one single bias.
- Participant Roles: Once a participant suspects the hypothesis, they may try to help confirm it (the good-participant role), try to disconfirm it (the negativistic role), or simply follow instructions and try to ignore the guess (the faithful-subject role).
- Hawthorne Effect: The Hawthorne effect is a related but distinct phenomenon: a change in behavior that follows from being watched, whether or not the participant has guessed the hypothesis.
- Controls: Single- and double-blind procedures, standardized scripts, and covert or naturalistic observation are the standard defenses against both demand characteristics and investigator effects.
- Effect Size: A 2025 meta-analysis of 53 studies found a real but small and highly inconsistent effect, useful evidence that the risk is genuine without being catastrophic in every study.
References
Brito, C. F. (2017). Demonstrating experimenter and participant bias.
Coles, N. A., Wyatt, M., & Frank, M. C. (2025). A meta-analysis of the impact and heterogeneity of explicit demand characteristics. Collabra: Psychology, 11(1), Article 143005. https://doi.org/10.1525/collabra.143005
Eckmanns, T., Bessert, J., Behnke, M., Gastmeier, P., & Rüden, H. (2006). Compliance with antiseptic hand rub use in intensive care units the Hawthorne effect. Infection Control & Hospital Epidemiology, 27(9), 931-934.
Harris, M. J., & Rosenthal, R. (1985). Mediation of interpersonal expectancy effects: 31 meta-analyses. Psychological bulletin, 97(3), 363.
Kalton, G. (1968). Standardization: A technique to control for extraneous variables. Journal of the Royal Statistical Society: Series C (Applied Statistics), 17(2), 118-136.
Kihlstrom, J. F. (2021). Ecological validity and “ecological validity”. Perspectives on Psychological Science, 16(2), 466-471.
Landsberger, H. A. (1958). Hawthorne Revisited: Management and the Worker, Its Critics, and Developments in Human Relations in Industry.
Levitt, S. D., & List, J. A. (2011). Was there really a Hawthorne effect at the Hawthorne plant? An analysis of the original illumination experiments. American Economic Journal: Applied Economics, 3(1), 224-238.
McCambridge, J., Witton, J., & Elbourne, D. R. (2014). Systematic review of the Hawthorne effect: New concepts are needed to study research participation effects. Journal of Clinical Epidemiology, 67(3), 267-277.
Merton, R. K. (1948). The self-fulfilling prophecy. The antioch review, 8(2), 193-210.
Mummolo, J., & Peterson, E. (2019). Demand effects in survey experiments: An empirical assessment. American Political Science Review, 113(2), 517-529. https://doi.org/10.1017/S0003055418000837
Nichols, A. L., & Maner, J. K. (2008). The good-subject effect: Investigating participant demand characteristics. The Journal of general psychology, 135(2), 151-166.
Orne, M. T. (1959). The demand characteristics of an experimental design and their implications. American Psychological Association, Cincinnati.
Orne M.T. (1962) On the social psychology of the psychological experiment: With particular reference to demand characteristics and their implications. American Psychologist 17: 776–783.
Orne, M. T., & Holland, C. H. (1968). On the ecological validity of laboratory deceptions. International Journal of Psychiatry, 6(4), 282-293.
Orne, M. T. (2009). Demand characteristics and the concept of quasi-controls. Artifacts in behavioral research: Robert Rosenthal and Ralph L. Rosnow’s classic books, 110, 110-137.
Paulhus, D. L. (1984). Two-component models of socially desirable responding. Journal of personality and social psychology, 46(3), 598.
Roethlisberger, F. J., & Dickson, W. J. (1939). Management and the Worker: An Account of a Research Program Conducted by the Western Electric Company, Hawthorne Works, Chicago. Harvard University Press.
Rosenthal, R., & Fode, K. L. (1963). The effect of experimenter bias on the performance of the albino rat. Behavioral Science, 8(3), 183-189. https://doi.org/10.1002/bs.3830080302
Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the Classroom: Teacher Expectation and Pupils’ Intellectual Development. Holt, Rinehart & Winston.
Spencer, C. D. (1978). Two types of role playing: Threats to internal and external validity. American Psychologist, 33(3), 265.
VanderWeele, T. J., & Shpitser, I. (2013). On the definition of a confounder. Annals of statistics, 41(1), 196.
Weber, S. J., & Cook, T. D. (1972). Subject effects in laboratory research: an examination of subject roles, demand characteristics, and valid inference. Psychological bulletin, 77(4), 273.
Wegner, D. M. (1994). Ironic processes of mental control. Psychological review, 101(1), 34.
