4.2.3.1 - Demand characteristics and investigator effects

Why these effects matter

Good research is not just about choosing the right method. It is also about stopping the research situation itself from changing the behaviour being measured. In AQA research methods, demand characteristics and investigator effects are two important threats to validity because they can create a false difference between conditions or hide a real one.

This first threat starts with how the participant reads the situation.

The difference is the direction of the influence. Demand characteristics are usually participant-side: the participant notices clues and reacts. Investigator effects are usually researcher-side: the investigator's expectations, appearance, behaviour or decisions shape what happens. In practice, they can work together. A researcher who smiles more in one condition may create a demand characteristic because the participant picks up the clue and behaves differently.

Both matter because they can reduce internal validity. If memory scores are higher in the "caffeine" condition, the researcher wants to conclude that caffeine caused the improvement. If the participants guessed the aim and tried harder, or the researcher gave that group more encouragement, the independent variable may not be the true cause.

For exams, the skill is not simply naming the effect. You need to identify the clue or researcher influence, explain how it changes behaviour or measurement, and link this to validity.

An extraneous variable could affect the outcome; a confounding variable varies systematically between the IV conditions and offers an alternative explanation for the DV. For example, encouragement given only to the treatment group may confound the treatment effect. A bias affecting both groups equally can distort measurement without necessarily creating a difference between groups.

Demand characteristics

Martin Orne (1962) argued that a psychology experiment is a special kind of social situation. Participants do not simply arrive, behave naturally and leave. They try to work out what sort of situation they are in, what the researcher wants, what counts as a "good" response and whether they are being judged.

Common demand characteristics include:

• the title or cover story of the study
• instructions that reveal the likely aim
• the order of tasks in a repeated measures design
• obvious differences between experimental and control conditions
• researcher tone, facial expression or body language
• the laboratory setting itself
• what participants overhear from other participants

Once participants think they know the aim, they may try to help by behaving in the expected direction, deliberately do the opposite, or become self-conscious and behave differently. These reactions need not produce the result the researcher expected. In self-report research, a related problem is social desirability bias, where participants give answers that make them look sensible, healthy or morally acceptable.

Imagine a study on whether revision music improves concentration. Mia is put in the music condition. The researcher says, "This playlist has helped lots of students focus," then gives her a memory test. Mia now has a strong clue that music is expected to help, so she tries unusually hard. If she scores highly, the result may reflect the demand characteristic rather than the music.

Demand characteristics are especially likely when the aim is easy to guess. A repeated measures design can be vulnerable because the same participants experience both conditions and may infer the real comparison. A laboratory study can also be vulnerable because the setting feels artificial and participants may search for hidden meaning in every instruction. This does not mean repeated measures or laboratory experiments are automatically weak; it means researchers must control the risk.

Ways to reduce demand characteristics include using a credible cover story, giving standardised instructions, preventing participants from discussing the study, keeping participants unaware of key condition information, and including filler tasks where appropriate. Each method has a trade-off. For example, a cover story can protect validity but may involve deception, so it needs careful ethical handling and debriefing.

Investigator effects

Investigator effects are not limited to a researcher deliberately biasing a study. In many cases the influence is unconscious. A researcher may expect one condition to perform better and, without meaning to, speak more warmly to those participants, give them more time, look more encouraging, explain instructions more clearly or record ambiguous answers in a more favourable way.

Investigator effects can occur at several stages:

• Design: choosing materials that favour one condition, writing leading instructions, or creating unequal tasks.
• Sampling and allocation: selecting participants in a way that favours a desired outcome, or allocating easier participants to one condition.
• Interaction: using different tone, facial expression, prompts or encouragement with different groups.
• Recording and interpretation: hearing, coding or scoring ambiguous responses in line with expectations.