Demand characteristics: why participants guess the hypothesis, and how to control it

Demand characteristics are cues in an experiment that tell participants what the researcher expects. Martin Orne named them in 1962 and defined them as the totality of cues that convey the experimental hypothesis. Exams often use the term for the behavior change that follows, but Orne meant the cues themselves. The cues threaten validity because they give an alternative explanation for the result. A 2025 meta-analysis of 53 studies found that explicit demand shifts answers toward the hypothesis by a small amount on average, with wide variation across studies.

What counts as a demand characteristic

A cue counts when it helps a participant infer the hypothesis and adjust behavior. Orne listed rumors about the study, the solicitation message, the experimenter, the lab setting, explicit and implicit communication during the study, and the procedure itself (Orne, M. T. (1962). On the social psychology of the psychological experiment. American Psychologist.).

Common cue groups are easy to spot once you know them.

• Rumors outside the lab. A classmate says the second test measures memory, so you rehearse during the filler task.
• The setting. A sparse room with a one-way mirror and a clipboard signals science and careful performance.
• Communication from the researcher. A smile, a longer pause, or a hint in the instructions suggests which answer fits.
• The procedure itself. A test given before and after a treatment tells even an inattentive student that change is expected, above all when the test matches the treatment.

If the cues are clear, most participants perceive the same hypothesis. If the procedure is ambiguous, different participants infer different aims and add noise.

The study that named it

Orne tried to find a task that waking volunteers would refuse, to compare waking compliance with hypnotic compliance. He asked volunteers to add each adjacent pair of numbers on sheets of random digits. One sheet needed 224 additions. He gave each person a stack of about 2000 sheets, took away watches, and said to continue to work because he would return eventually. He gave up after five and one half hours. Performance stayed stable for hours.

Then he made the task more frustrating. After each sheet, the person drew a card that told them to tear the finished sheet into at least 32 pieces and start the next sheet. Every card said the same thing. Subjects still persisted for hours with little hostility. When asked afterward, they called it an endurance test or similar. The phrase this is an experiment had legitimized an otherwise meaningless request (Orne, M. T. (1962). On the social psychology of the psychological experiment. American Psychologist.).

Orne drew two conclusions that still hold. Participants act as active problem solvers who infer the purpose and respond in a way that supports it. Demand characteristics cannot be removed, but their effects can be studied against the experimental variables.

Their real-simulator design became a quasi-control. Volunteers who only pretended to be hypnotized faced the same cues as hypnotized participants. Behavior common to both groups could then be assigned to demand rather than to hypnosis.

The four roles participants take

Once a participant infers the hypothesis, behavior often follows one of four roles described by Weber and Cook and later work (Ruhl, C. (2026). Demand Characteristics In Psychology. Simply Psychology.).

The good role is also called the please-you effect. The negativistic role is also called the screw-you effect. The apprehensive role overlaps with social desirability bias. The faithful role rarely holds, because trying not to be influenced by a guess is itself a response to it.

The other side of the bench

Demand characteristics describe the participant side. Investigator effects describe the researcher side. The two often appear together, but they have different sources.

Robert Rosenthal and Kermit Fode showed the researcher side in 1963 with rats. Twelve students each received five rats from the same stock, falsely labeled maze bright or maze dull. Each student ran an elevated T-maze with ten trials per day for five days. Students with bright labels recorded better learning and faster times. Watchers found no false entries. The authors proposed handling differences as the likely channel (Rosenthal, R., and Fode, K. L. (1963). The effect of experimenter bias on the performance of the albino rat. Behavioral Science.).

Rosenthal and Jacobson extended the same logic to classrooms. Pupils in 18 classrooms took a disguised IQ test. About 20 percent in each room were named as spurters at random. Those children later showed larger IQ gains, above all the youngest. The classroom version is often called the Pygmalion effect, a form of self-fulfilling prophecy in which a false expectation evokes behavior that makes it come true (Ruhl, C. (2026). Demand Characteristics In Psychology. Simply Psychology.).

The practical point is simple. Participant cues and researcher expectations are two sides of one social situation. Controls must address both.

Why it matters for validity

Demand characteristics threaten internal and external validity. Internal validity asks whether the independent variable caused the change. External validity asks whether the result generalizes beyond the study. Demand gives a rival answer to both.

It makes it hard to tell whether the independent variable caused the change. If participants in the treatment group try harder because they guessed the aim, the dependent variable may move without any real treatment effect. If conditions differ in how obvious they are, the bias differs by condition and confounds the comparison.

It also limits generalization. A lab with clear cues does not match real life, where no researcher watches and no hypothesis waits to be guessed. That gap is an ecological validity problem. Textbooks call demand an extraneous variable, and when it differs by condition, a confounding variable.

How researchers reduce it

No control removes every cue. Each control removes one channel at a cost.

Between-subjects designs deserve special attention. Within-subjects designs show each person all conditions, which reveals the manipulation. Tutor2u notes the same risk ranking (Tutor2u. (2021). Demand Characteristics.).

Demand characteristics against lookalikes

Exams test sorting more than recall. The four terms below often appear together.

Placebo differs again. Placebo is improvement from believing one received treatment. Reactivity is the broader term for behavior change from being studied.

What new evidence adds

The classic story says demand ruins studies. New quantitative work says the effect is real on average, small, and inconsistent.

Coles, Wyatt, and Frank pooled 253 effect sizes from 53 studies that manipulated explicit demand in 2025. Explicit demand shifted responses toward the hypothesis by Hedges g equal to 0.21, with a 95 percent interval from 0.11 to 0.30. The prediction interval for a future study ran from minus 0.48 to 0.89. About 62 percent of true effects pointed toward the hypothesis, 20 percent were negligible, and 18 percent pointed away from it (Coles, N. A., Wyatt, M., and Frank, M. C. (2025). A Meta-Analysis of the Impact and Heterogeneity of Explicit Demand Characteristics. Collabra: Psychology.).

Two nuances matter. Mummolo and Peterson tested demand in large online surveys with more than 12000 participants across five designs (Mummolo, J., and Peterson, E. (2019). Demand Effects in Survey Experiments. American Political Science Review.). Even payment for hypothesis-consistent answers rarely moved results. McCambridge and colleagues reviewed non-laboratory studies in 2012 and found only seven eligible studies with weak designs (McCambridge, J., de Bruin, M., and Witton, J. (2012). The Effects of Demand Characteristics on Research Participant Behaviours in Non-Laboratory Settings. PLoS ONE.).

For exams, keep the sentence small. Demand shifts answers toward the perceived hypothesis on average, but a single study can show no shift or a reverse shift.

How to remember it for exams

Sort by trigger, not by example. If the stem mentions cues, rumors, order effects, or guessing the aim, choose demand characteristics. If it mentions only being observed, choose Hawthorne effect. If it mentions looking good with no hypothesis cue, choose social desirability bias. If it mentions researcher handling or scoring, choose investigator effects. If the design shows all conditions to each person, name demand as a flaw and propose between-subjects or deception with debrief as the fix.

Save demand characteristics plus Hawthorne effect, social desirability bias, investigator expectancy effect, and double-blind to one bookmark list in Psychology World and drill that list with the Question game until the sorting feels automatic.