03 · What You Need to Know
How Recruitment Messages Can Shape Participants and Research Outcomes
What Is Expectation Bias in Research Recruitment?
Expectation bias is a broad expression for distortions associated with beliefs about what is likely to happen or what researchers expect to observe. In recruitment, these beliefs may develop before a participant encounters the intervention or completes a measurement.
The term should not be treated as one precisely defined statistical mechanism. Several related processes may operate:
-
Expectation-related self-selection: Individuals with particular beliefs may be more likely to respond to the invitation.
-
Demand characteristics: Recruitment language may communicate what participants believe researchers want them to report or demonstrate.
-
Expectancy effects: Beliefs about an intervention may influence experiences, behavior, or outcomes.
-
Response bias: Participants may report experiences in ways influenced by expectations, perceived norms, or social desirability.
These processes are related but should not be treated as interchangeable. A recruitment advertisement may affect the sample composition without changing anyone's subsequent responses. Alternatively, it may influence expectations among enrolled participants even if the sample composition remains unchanged.
Researchers should identify the mechanism relevant to their study rather than attributing every recruitment-related difference to a single generic bias.
Advertisements Can Influence Who Chooses to Participate
Recruitment advertisements do more than communicate eligibility and logistical information. Their wording may signal which experiences or attitudes are especially welcome.
Consider an advertisement stating that a study seeks people who have benefited from mindfulness training. Individuals with positive experiences may regard the invitation as relevant, while those who experienced no benefit may assume they are unsuitable.
Even when the study intends to examine the full range of experiences, the advertisement may disproportionately attract participants who interpret themselves as fitting its implied narrative.
This is a selection problem. The advertisement influences who enters the study rather than necessarily changing outcomes after enrollment.
The problem overlaps with selection involving highly motivated volunteers, particularly when advertisements appeal to enthusiasm, dissatisfaction, or strong personal commitments.
Researchers should distinguish actual eligibility requirements from persuasive language that unintentionally creates additional perceived requirements.
Advertisements Can Shape Expectations About an Intervention
Participants may form beliefs about an intervention before receiving it.
For example, describing a learning activity as likely to improve memory could lead participants to anticipate better performance. Describing the same activity as potentially frustrating or cognitively demanding could create different expectations.
In some research contexts, expectations can influence subjective experiences, effort, attention, or responses.
Placebo and nocebo research provides evidence that expectations and contextual information can influence certain outcomes, particularly subjective symptoms and experiences. However, the magnitude and mechanisms vary across interventions and outcome types.
It would be inappropriate to infer that promotional recruitment language necessarily improves objective performance or that every negative description produces a nocebo effect.
Instead, researchers should recognize participant expectations as a plausible influence and assess whether the study design makes that influence consequential.
Demand Characteristics Can Begin Before the Experiment
Demand characteristics arise when features of the research situation provide cues about the behavior or responses participants believe are expected.
Orne's foundational discussion of demand characteristics emphasized that participants interpret research situations and may respond to perceived experimental expectations.
Recruitment advertisements can be among the earliest cues participants encounter.
For instance, an advertisement asking students to help demonstrate the effectiveness of an AI writing assistant may communicate that successful participants should benefit from the tool.
Some participants might consequently try harder when using it, report more favorable experiences, or interpret ambiguous outcomes positively.
Others might react against the perceived expectation. Demand characteristics do not imply that every participant attempts to confirm the researcher's hypothesis.
The relevant concern is that the advertisement may introduce a systematic influence not adequately accounted for in the study design.
Expectation Effects and Selection Bias Must Be Distinguished
Researchers should identify when the advertisement's influence occurs.
| Mechanism |
When it operates |
Possible consequence |
| Recruitment self-selection |
Before enrollment |
People with particular beliefs or experiences participate disproportionately |
| Pre-intervention expectations |
After exposure to the advertisement, before treatment or activity |
Participants enter the study with different anticipated outcomes |
| Demand characteristics |
During participation or measurement |
Behavior or responses may reflect perceived researcher expectations |
| Response framing |
During interpretation of questions or experiences |
Self-reports may be influenced by earlier descriptions |
These mechanisms can coexist, but they have different implications for design and analysis.
For example, random assignment after recruitment can help protect the internal validity of a treatment comparison, but it does not necessarily eliminate expectations established before randomization.
If everyone receives the same advertisement, its influence may affect both randomized groups. Whether this biases the estimated treatment contrast depends on the intervention, the outcomes, and how expectations interact with assignment.
In an observational study, advertisement-related selection may affect both descriptive estimates and associations between variables.
Positive and Negative Framing Can Both Be Problematic
Researchers often recognize the risk of overly positive recruitment messages but overlook the influence of negative framing.
An advertisement describing a technology as harmful may attract individuals concerned about its risks. One describing it as transformative may attract enthusiasts.
Both can influence recruitment and expectations.
Balanced wording does not require pretending that established evidence is uncertain. If an intervention has known risks or benefits, these should be communicated accurately when relevant to informed consent.
The concern is unsupported or unnecessarily persuasive framing, particularly when the study is designed to investigate the very claim being advertised.
Researchers should not remove important safety information merely to create symmetrical or emotionally neutral wording.
Images and Design Can Communicate Expectations Without Explicit Claims
Expectations may arise from visual elements as well as written text.
For example, an advertisement showing only smiling participants receiving an intervention may imply positive experiences. Images emphasizing distress may suggest that the study primarily concerns negative outcomes.
Testimonials, endorsements, institutional logos, and promotional graphics can also influence how credible, beneficial, or demanding a study appears.
These elements are not inherently inappropriate. Institutional identification may be necessary for transparency, and accessible visual communication can improve understanding.
Nevertheless, researchers should examine whether the overall presentation implies benefits, risks, or preferred experiences that the study has not established.
Recruitment materials should be reviewed as complete communications rather than evaluating wording independently of imagery and design.
Advertisements Should Not Present Research Participation as Treatment
Clinical recruitment creates particular concerns when advertisements imply that participation will provide effective treatment.
Potential participants may misunderstand the difference between receiving clinical care and entering research designed to evaluate an intervention.
This concern is related to therapeutic misconception, in which participants may fail to appreciate important differences between research objectives and individualized clinical care.
Researchers should avoid advertising experimental interventions as established treatments or implying guaranteed personal benefit.
Accurate descriptions of possible benefits must remain consistent with the evidence and applicable consent requirements.
The U.S. Food and Drug Administration's guidance on recruiting study subjects recognizes recruitment advertising as part of the informed consent and subject-selection process for FDA-regulated clinical investigations. It cautions against promotional claims that imply investigational products are safe or effective when such claims are not established.
Can Neutral Wording Reduce Expectation-Related Bias?
Neutral wording can reduce unnecessary evaluative cues.
For example, researchers may describe a study as comparing two approaches to learning rather than testing whether a new approach is superior.
However, neutrality is not the same as withholding important information.
Participants should receive accurate information about the research purpose, procedures, foreseeable risks, potential benefits, and voluntary participation, as required by the applicable ethical framework.
The distinction between describing the purpose and announcing the predicted outcome is particularly relevant when deciding whether recruitment materials should reveal the research hypothesis.
Researchers should also avoid assuming that a neutral advertisement is expectation-free. Participants may bring strong prior beliefs to the study or infer expectations from the intervention itself.
When Should Researchers Measure Participant Expectations?
Measuring expectations may be useful when they are theoretically relevant to the outcome or could influence interpretation.
For example, a study of a digital learning intervention might ask participants about anticipated usefulness before the intervention begins.
However, expectation measurement introduces methodological decisions of its own.
Asking participants what they expect may make particular outcomes more salient. Expectations measured after exposure to an advertisement may already reflect the advertisement's influence.
Moreover, adjusting statistically for expectations is not automatically appropriate. If expectations are affected by the recruitment message or intervention, they may be part of a causal pathway rather than an ordinary baseline confounder.
Researchers should determine why expectations are being measured and how the information will be used before adding expectation scales to the protocol.
Blinding and Standardization May Help, but They Do Not Solve Every Problem
Where feasible, blinding can reduce knowledge of treatment assignment or study hypotheses that might influence participants, investigators, or outcome assessors.
Standardized instructions and consistent procedures can also reduce avoidable differences in how research conditions are presented.
Yet some interventions cannot be fully blinded. Participants in educational research may know whether they are using an AI tool, attending an online class, or receiving a particular teaching approach.
In such cases, researchers may use outcome measures less susceptible to subjective expectations, independent assessors, or other appropriate safeguards.
Objective measurement is not automatically free from expectancy effects. Expectations may influence effort or behavior that subsequently affects an objectively scored outcome.
The appropriate safeguards depend on the specific mechanism through which expectations could influence the result.
Different Advertisements Across Recruitment Channels Can Complicate Interpretation
When participants are recruited through several channels, differences in advertisement framing may become entangled with differences in participant characteristics.
Suppose a social media advertisement emphasizes the benefits of a learning intervention while an institutional email describes a neutral comparison study.
Participants recruited through the two channels may differ both in prior beliefs and in expectations created by the recruitment messages.
If their outcomes differ, researchers may struggle to distinguish selection, framing, and other channel-related mechanisms.
Using substantively consistent recruitment information can reduce this avoidable source of variation.
When differences remain, researchers should consider whether recruitment-related subgroup differences warrant analytical attention.
Statistical adjustment cannot automatically separate mechanisms that are perfectly confounded by the recruitment design.
What Does the Evidence Actually Support?
Research on demand characteristics, expectancy effects, and placebo or nocebo mechanisms provides a scientific basis for considering how information and expectations may influence research behavior or outcomes.
However, evidence about these broader mechanisms should not be treated as proof that a particular advertisement caused bias in a particular study.
Advertisement effects depend on message content, participant characteristics, research context, and outcome measurement.
Researchers should therefore describe expectation-related bias as a plausible threat when direct evidence is unavailable, rather than asserting that every promotional advertisement necessarily distorts findings.
If the study is designed to investigate advertisement framing itself, researchers may experimentally compare ethically acceptable messages. Such a design can provide stronger evidence about message effects than an observational comparison of participants recruited through different advertisements.
Watch Out
Do not attempt to eliminate expectations by concealing material risks, misrepresenting the intervention, or providing misleading recruitment information. Scientific validity must be pursued within the applicable requirements for truthful communication and informed consent.
