Bias in research comes in many different forms and can influence your findings, perceptions and behaviour when conducting and facilitating research. While juggling the multiple components of research, it is of utmost importance that researchers are aware of bias and how it can influence user research.
We’ve handpicked six common types of bias and share our tips to overcome them:
1. Confirmation bias
Confirmation bias is when data is analysed and interpreted to confirm hypotheses and expectations. It is one of the most detrimental biases found in user research. A researcher’s role is to be a source of truth, even if the findings contradict expectations. Remember that all findings, positive or negative, are valid and should be reported on.
2. The Hawthorne effect
The Hawthorne effect occurs when participants are aware they are being observed and can result in them deviating away from natural behaviour. It is difficult but possible to mitigate against the Hawthorne effect. Clearly communicate that the research is not testing the participant in any way and that there are no right or wrong answers. Create a calm space in which participants feel relaxed and less conscious of being observed to reveal their natural behaviour.
3. Implicit bias
Implicit bias is when our stereotypes and attitudes influence how we behave and perceive others. It is a difficult form of bias to recognise. It is of paramount importance that a researcher remains neutral during the research process to provide a standard and impartial experience for every participant: a fair test. One way in which researchers can seek to remain neutral towards participants is by only collecting the essential information about participants before the session, which will reduce the potential for unconscious preconceived perceptions to surface.
4. Expectancy bias
Expectancy bias, also commonly known as observe bias is similar to the Hawthorne effect in that research participants may change their behaviour to please the researcher and provide the answers they think you want. Often, participants may seek subtle cues to suggest they are doing the right thing. This is why it is important to remain neutral in language and nonverbal behaviour. Avoid nodding or agreeing with what participants say and maintain a neutral expression.
5. Leading Language
Leading language bias is when a question or task uses certain wording or terminology that provides a hint for a particular response or behaviour. It is an easy bias to spot and overcome through practice and experience. A researcher’s script should contain open questions and be proofread by a colleague. Proofreading and piloting your research session is the best way to overcome leading language as your pilot test respondent can highlight any words that could lead them to a response.
6. Recall bias
A portion of user research requires participants to recall past events or tasks to understand how they behaved at the time. This can incur what is known as recall bias, as some memories may be more prominent than others, meaning some behaviours or events are not mentioned. Mitigate against this by applying best practice: review your questions carefully to ensure they are understood and allow participants enough time to recall long-term memories. Factor in additional time to allow for this. Use intentional silence and ask good questions to surface natural thoughts and feelings.
Becoming aware of bias and its potential to influence research is the first step to overcome it. Do not shy away from it, rather develop an awareness and understanding of it. Remember that the role of the researcher is to become a neutral observer. Step aside from personal influences, feelings and opinions. Ground findings in what is witnessed and observed.
You may find it helpful to write out all your expectations or hypotheses beforehand so you can name them and put them to one side. Later, analyse the data and refer to your expectations and hypotheses to notice any differences.