The Experimenter Effect: Can an Investigator's Expectations Influence Research Results?

Introduction

Two investigators enter the same room. One has been told that previous visitors reported a figure near the window; the other has not. Later, both are asked whether they noticed anything unusual.

Their answers may differ—not necessarily because either person is dishonest, but because expectations can influence attention, interpretation and reporting.

The experimenter effect refers broadly to ways in which a researcher's expectations or behaviour can influence participants, observations or the conduct of a study. In paranormal research, where many observations are ambiguous, reducing this influence is essential.

1. What Is Observer or Experimenter Bias?

Observer bias occurs when an observer's expectations influence how information is noticed, recorded or interpreted. Experimenter effects can also arise when a researcher's behaviour unintentionally affects participants.

Examples include:

  • Asking leading questions.
  • Reacting visibly to a sound or movement.
  • Highlighting a location where an event is expected.
  • Recording confirming observations more carefully than ordinary ones.
  • Selecting clips that fit an expected interpretation.
  • Changing analysis thresholds after seeing results.

These processes can occur without deliberate manipulation.

2. How Expectations Influence an Investigation

Expectations may affect several stages:

Attention: An investigator may monitor a particular doorway more closely after hearing a story about it.

Interpretation: An ambiguous sound may be more likely to be described as a voice when a voice is expected.

Memory: Details that fit an interpretation may be easier to recall than details that do not.

Questioning: A question such as "Did you hear the footsteps?" assumes that footsteps occurred. A neutral question asks, "What did you hear, if anything?"

Analysis: A reviewer who knows the alleged event time may interpret a noisy segment differently from a reviewer who does not.

3. Blinding as a Bias-Reduction Tool

Blinding means withholding selected information from participants, observers or analysts so that expectations are less likely to influence a measurement.

Examples:

  • An audio reviewer receives anonymised clips without knowing which are alleged EVP recordings.
  • A video analyst reviews footage without being told where a witness reported a figure.
  • A participant is not told which room is considered the "active" location.
  • The person scoring results does not know which condition was experimental or control until scoring is complete.

Blinding is not always possible, and it does not remove every source of bias. It should be used where practical and documented clearly.

4. Neutral Witness Interviews

Use open-ended questions:

  • "Please describe what happened."
  • "What did you notice first?"
  • "Where were you looking?"
  • "How long did it last?"
  • "What else was happening at the time?"
  • "How certain are you about that detail?"

Avoid:

  • "Was that a ghost?"
  • "Did you hear the woman crying?"
  • "You saw the shadow too, didn't you?"

Record the witness's original words. Separate direct observations from interpretations and do not correct a witness into the team's preferred terminology.

5. Independent Observation and Analysis

AERM can reduce shared expectation by asking team members to record observations independently before discussion.

For recordings:

  1. Preserve the complete original file.
  2. Create anonymised review copies.
  3. Randomise clip order.
  4. Include control clips.
  5. Ask reviewers to describe what they hear or see without suggested labels.
  6. Record agreement and disagreement.
  7. Reveal context only after initial scoring.
  8. Preserve the initial independent assessments.

Agreement is informative, but it should be evaluated in light of the number of reviewers, the clarity of the event and the possibility of shared cues.

6. Predefine Analysis Rules

If an investigator repeatedly adjusts a threshold until a sensor graph displays a dramatic spike, the result may reflect analytical flexibility rather than a robust discovery.

Before reviewing data, specify:

  • Primary question.
  • Event definition.
  • Measurement window.
  • Thresholds and scoring rules.
  • Exclusion criteria.
  • Planned comparisons.
  • How missing data will be handled.

Exploratory analysis is legitimate, but it should be labelled exploratory. A new pattern discovered after examining the data should be tested using fresh data where possible.

7. A Simple Demonstration of Expectation Bias

Prepare a set of ambiguous audio clips containing ordinary environmental sounds. Randomly assign participants to one of two groups. One group receives neutral instructions; the other receives a suggestion that some clips may contain speech.

Ask everyone to describe the clips independently. Compare the frequency and content of speech reports.

The study must be ethically designed, avoid distress and use a clear protocol. The outcome would demonstrate how instructions may influence interpretation in that particular task; it would not prove that all unusual reports are caused by suggestion.

Conclusion

Investigators are part of the measurement process. Their expectations can influence what they notice, how they question witnesses and how they interpret ambiguous evidence. Blinding, neutral interviews, independent observations, predefined analysis rules and transparent reporting can reduce these risks.

The goal is not to eliminate human judgement, which is impossible, but to make its influence visible and manageable. AERM should be willing to test evidence against its own expectations as rigorously as it tests alternative explanations.

AERM principle: No anomaly is considered paranormal until environmental, psychological, technical, and scientific explanations have been systematically evaluated.

References and Further Reading

  1. Rosenthal, R. (1966). Experimenter effects in behavioral research. Appleton-Century-Crofts.
  2. Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2(2), 175–220. https://doi.org/10.1037/1089-2680.2.2.175
  3. Klein, R. A., et al. (2014). Investigating variation in replicability. Social Psychology, 45, 142–152. https://doi.org/10.1027/1864-9335/a000178
  4. National Academies of Sciences, Engineering, and Medicine. (2019). Reproducibility and Replicability in Science. https://doi.org/10.17226/25303
  5. Nosek, B. A., et al. (2018). The preregistration revolution. Proceedings of the National Academy of Sciences, 115(11), 2600–2606. https://doi.org/10.1073/pnas.1708274114

About AERM

AERM is an independent research and media initiative focused on the systematic evaluation of unusual reports. Paranormal Frame communicates evidence-based investigation methods through accessible documentary and educational content.

Educational Disclaimer

This article is educational and is not a substitute for formal research-ethics review. Studies involving participants should use informed consent, protect privacy and minimise the risk of distress.