Why do we change our behavior when we’re being watched?

Observer Expectancy Effect

, explained.
Bias

What is the observer expectancy effect?

The observer expectancy effect, also known as the experimenter expectancy effect, refers to how the perceived expectations of an observer can influence the people being observed. This term is usually used in the context of research, to describe how the presence of a researcher can influence the behavior of participants in their study.

Where this bias occurs

If you have ever worked in a research lab, you’re probably quite familiar with this phenomenon. A key factor of research design is figuring out how to avoid accidentally influencing participants. For example, if you’re running a study examining the effects of a certain new medication on participants’ stress levels, you’ll probably expect participants receiving the medication to be less stressed than those receiving a placebo pill. Even if the participants all think they’re receiving the actual medication, you might unconsciously treat the two groups differently. Since you expect the placebo group to be more stressed, you might treat them like they are more stressed, which may cue them to act more tense than they normally would. On the other hand, you’ll expect participants getting the real medication to be less stressed and treat them like they’re more relaxed or ask them leading questions that hint that they should be more relaxed. In this way, you may affect the behaviors of both groups, ultimately compromising the accuracy and generalizability of your results.

Debias Your Organization

Most of us work & live in environments that aren’t optimized for solid decision-making. We work with organizations of all kinds to identify sources of cognitive bias & develop tailored solutions.

Learn about our work

Individual effects

As a researcher, failure to take the observer expectancy effect into consideration could be detrimental to your credibility, as it is evidence of poor research design. It can also negatively affect participants. Many of the surveys and questionnaires used in research offer participants the opportunity to think critically about themselves and their opinions. If they are attempting to adhere to the experimenter’s expectations, they may answer in a way that is not completely accurate, which can cause them to miss out on that valuable chance for self-reflection. Moreover, participants in certain clinical trials may not improve as much as they have the potential to if they are in the control group and the researcher does not expect them to change. As such, it benefits everyone for the observer expectancy effect to be considered when a study is being designed.

Systemic effects

The observer expectancy effect is something that academics must take into consideration when designing a study. If measures are not taken to prevent it, the study could garner biased results. This may lead researchers to draw inaccurate conclusions. Specifically, since the observer expectancy effect is characterized by participants being influenced by the researcher’s expectations, it may lead the research team to conclude that their hypothesis was correct. False positives in research can have serious implications, particularly in fields where there is a lot at stake, such as medicine. Not only that, but research is quite expensive to conduct, so it is crucial that researchers are certain that their study design allows for the most accurate results possible before testing can begin.

Why it happens

The observer expectancy effect arises due to demand characteristics, which are subtle cues given by the researcher to the participant about the nature of the study, as well as confirmation bias, which is when the researcher collects and interprets data in a way that confirms their hypothesis and ignores information that contradicts it.

Demand characteristics

Demand characteristics are a form of response bias that may give rise to the observer expectancy effect. Typically seen in psychological research, demand characteristics are subtle cues from the experimenter that may give the participants some idea of what the study is about. While some information about the study must be divulged to the participants for ethical reasons, it is ideal for participants to know as little as possible about the nature of the research being done. The more participants know, the more likely it is that they will try to “help” the researcher by behaving in the way they think the researcher wants them to. Unfortunately, when this happens, the data collected is inaccurate and therefore not very informative.

Many things can act as demand characteristics. Any verbal or non-verbal communication between the participant and the experimenter, the appearance of the room where the study is being held, and any knowledge the participant might have of the kind of work the lab does could all suggest certain behaviors to the participants. Naturally, demand characteristics cannot be eliminated completely, but their effects can be minimized.

Participants may provide biased responses in studies due to social desirability – wanting to give a good impression of themselves. However, it has been shown that, when participants have some knowledge of the researcher’s hypothesis, they are more likely to respond in a way that they think will benefit the researcher, whether or not it makes them look good. They do not want to provide “bad” information that would ruin the study or disprove the researcher’s hypothesis.1

Despite the good intentions of many participants, demand characteristics give rise to the observer expectancy effect, which compromises the accuracy of the study. Accurate data that yield no significant results are more informative and more valuable than inaccurate data that yield significant results.

Confirmation bias

Another factor that contributes to the observer expectancy effect is confirmation bias. Researchers are highly motivated to find evidence in support of their hypothesis, particularly now, when it is so difficult to get papers published in reputable journals. The intense motivation to collect data in support of a hypothesis can cause researchers to selectively notice and remember information that aligns with their hypothesis. This biased interpretation of data is referred to as confirmation bias.

The motivated search for information that confirms their hypothesis can lead experimenters to interpret participants’ behavior in a way that confirms their hypothesis. However, confirmation bias not only affects how we interpret data; it influences how we collect the data in the first place. As such, researchers may ask participants leading questions, which prompt a specific response, or even treat participants in a way that elicits the desired behavior. Of course, the data collected under such conditions is not accurate and therefore not helpful in the progression of knowledge.

Why it is important

Knowledge of the observer expectancy effect is necessary for the express purpose of avoiding it. A good researcher is aware of the factors that may compromise their results, such as sampling, leading questions, and, of course, the observer expectancy effect. Understanding how these issues arise allows academics to structure studies in a way that minimizes their influence. Furthermore, knowledge of the possible consequences of factors like the observer expectancy effect not only motivates researchers to avoid it, but also allows them to evaluate the accuracy of research conducted by other academics.

How to avoid it

A double-blind research design can be effective in preventing the observer expectancy effect. In studies with this design, neither the participants nor the researchers know which participants are in the experimental group or which are in the control group. The experimental group is the group that differs from baseline, for example, in a drug trial, they would be the ones receiving the medication. The control group is the baseline comparison group, so in the case of the drug trial, they would differ from the experimental group only in that they are not receiving the drug. A double-blind approach is useful in avoiding biased results, as it prevents the researcher from projecting their expectations onto the participant.

How it all started

Robert Rosenthal is one of the researchers most closely associated with the observer expectancy effect. He wrote many papers on the topic, the first of which was written in collaboration with Kermit L. Fode in 1963, and titled “The effect of experimenter bias on the performance of the albino rat”.2 The aim of this paper was to demonstrate the ease with which an experimenter can influence a participant to exhibit a certain behavior. Rosenthal and Fode provided evidence for this through a now highly renowned study. The participants were experimental psychology students, each of whom was given a group of five rats, which they were supposed to teach to navigate a maze to reach the darker of two platforms. Each participant was then told that they were either working with particularly bright or particularly dull rats, although there were no significant differences between any of the rats. At a later point, when the rats were tested on their ability to navigate the maze, those who had been randomly labeled as “bright” performed better than those who had been randomly labeled as “dull”. The students working with the rats had been biased by these labels. This caused them to treat the rats differently, ultimately resulting in the animals conforming to their expectations in a sort of self-fulfilling prophecy. This early example of the effects of experimenter bias prompted further research on the subject and helped raise awareness for the effect among investigators.

Example 1 - Clever Hans

The classic example of the observer expectancy effect is the study of Clever Hans. Hans wasn’t your typical participant, mostly because he wasn’t human. Hans was a horse living in Germany in the late 19th and early 20th centuries. What earned him the title of “Clever” were his owner’s claims that he had near-human intelligence. Hans and his owner, Wilhelm von Osten, gave several performances, in which Hans displayed his many impressive abilities. He was seemingly able to perform basic arithmetic, identify colors, read, and recognize musical notes.3

Interestingly, von Osten never displayed any explicit signals cueing Hans to answer in any particular way. This caused many people to be drawn in by the act and believe that Hans was exactly as clever as he was cracked up to be.4 However, not everyone was so sure. German biologist and psychologist, Oscar Pfungst, launched a study into the so-called “Clever Hans Phenomenon” in 1907, and found that Hans only answered correctly when the questioner, usually his owner, van Osten, knew the correct answer.5 He concluded that Hans was not answering the question, but simply responding to subtle cues, likely unconsciously made, from the questioner.6

This is an exemplary instance of the observer expectancy effect, as Hans’ questioner influenced the horse to behave in a way that conformed with his expectations. In a situation where the questioner was not present, Hans would not have responded as he did in their presence. As such, this is a good example of how experimenter expectations can heavily influence participant behavior.

Example 2 - Teacher expectations

While the experimenter expectancy effect is usually used in the context of research, it can also be applied to classroom settings; while teachers are not experimenters, they do administer tests to their students. Rosenthal and Fode showed that when university students believed the rats they were teaching to be “dull”, the rats performed more poorly on the maze task than the rats labeled as “bright” – even though the “dull” and “bright” labels were given randomly.7 Similar effects can be seen in classroom settings. Whether based on comments from past teachers, report cards from previous years, or observed schoolyard behavior, a teacher may expect a certain student to either fail or excel in their course. As a result, teachers may consciously or unconsciously treat students differently. This may influence the students to behave in ways that align with their teacher’s expectations, even though they may have otherwise performed differently. Unfortunately, this means that children with poor academic records may not be given the chance to improve, while children with excellent academic records may struggle with the pressure of living up to their teachers’ high expectations.

Summary

What it is

The observer expectancy effect describes how the perceived expectations of an observer can influence the people being observed, particularly in the context of research.

Why it happens

The observer expectancy effect arises due to demand characteristics, which are subtle cues given by the researcher to the participant about the nature of the study, as well as confirmation bias, which is when the researcher collects and interprets data in a way that confirms their hypothesis and ignores information that contradicts it.

Example 1 – Clever Hans

One of the earliest recorded cases of the observer expectancy effect was that of Clever Hans, a German horse in the late 19th and early 20th centuries. Many people believed he had near-human intelligence because he was able to answer mathematical and vocabulary questions. However, it was later shown that Hans only ever answered correctly when the person asking the question also knew the response; he was responding to subtle cues from the questioner and not actually answering the question. This illustrated how the expectations of the experimenter can heavily influence participant behavior.

Example 2 – Teacher expectations

Teachers often form expectations for how their students will perform throughout the school year. This can result in a self-fulfilling prophecy where the teacher consciously or unconsciously influences their students to behave in a way that aligns with their expectations.

How to avoid it

Researchers can avoid the observer expectancy effect by using a double-blind design, in which neither the participants nor the experimenters know which participants are in the experimental condition and which are in the control condition. This way, the experimenter’s expectations will not influence participant behavior.

Sources

  1. Nichols, A. L., Maner, J.K. (2008). The Good-Subject Effect: Investigating Participant Demand Characteristics. The Journal of General Psychology. 135(2), 151-165.
  2. Rosenthal, R., & Fode, K. L. (1963). The effect of experimenter bias on the performance of the albino rat. Behavioral Science8(3), 183–189. doi: 10.1002/bs.3830080302
  3. Ferguson, P.M. (2019). Clever Hans. Encyclopaedia Britannica. https://www.britannica.com/topic/Clever-Hans
  4. See 3
  5. Samhita, L. and Gross, H.J. (2013). The “Clever Hans Phenomenon” revisited. Communicative and Integrative Biology. 6(6). doi: 10.4161/cib.27122
  6. See 3
  7. See 2

About the Authors

Dan Pilat's portrait

Dan Pilat

Dan is a Co-Founder and Managing Director at The Decision Lab. He is a bestselling author of Intention - a book he wrote with Wiley on the mindful application of behavioral science in organizations. Dan has a background in organizational decision making, with a BComm in Decision & Information Systems from McGill University. He has worked on enterprise-level behavioral architecture at TD Securities and BMO Capital Markets, where he advised management on the implementation of systems processing billions of dollars per week. Driven by an appetite for the latest in technology, Dan created a course on business intelligence and lectured at McGill University, and has applied behavioral science to topics such as augmented and virtual reality.

Sekoul Krastev's portrait

Dr. Sekoul Krastev

Sekoul is a Co-Founder and Managing Director at The Decision Lab. He is a bestselling author of Intention - a book he wrote with Wiley on the mindful application of behavioral science in organizations. A decision scientist with a PhD in Decision Neuroscience from McGill University, Sekoul's work has been featured in peer-reviewed journals and has been presented at conferences around the world. Sekoul previously advised management on innovation and engagement strategy at The Boston Consulting Group as well as on online media strategy at Google. He has a deep interest in the applications of behavioral science to new technology and has published on these topics in places such as the Huffington Post and Strategy & Business.

Notes illustration

Eager to learn about how behavioral science can help your organization?