Assuming the budgets and timelines are comparable, which research approach is less risky?
You want to conduct research that helps you determine whether your target customers are interested in your new product idea.
A. Field one large survey with 1,000 customers.
B. Field three smaller studies with 50 customers each.
Correct Answer: B
How can it be less risky to talk to 150 customers than 1,000?
Conducting a single research study—whether with 10 people or a million people—is inherently high-risk for one major reason: bias!
Common types of research bias
Bias can creep into a research study in many, many ways:
- Design bias: Choosing the wrong research methodology for the objective or poorly designing the study. Fun fact: there are more than 100 types of research methods! Different methods answer different types of research questions.
- Selection bias: Selecting the wrong people to participate in the study, or not including different audiences with alternative experiences or perceptions.
- Question bias: Using leading questions, double-barrelled questions, ambiguous questions, loaded questions, or does not provide “does not apply” or “other” responses.
- Social desirability bias: Using questions or methodologies where respondents feel pressured to select “socially desirable” responses, such as not wanting to hurt a professor’s feelings or not admitting to cheating on an assignment.
- Participant bias: Respondents answer questions based on what they think the researcher wants to hear, misrepresents themselves, or straightline due to fatigue (uninteresting or repetitive questions)—usually to obtain an incentive
- Confirmation bias: This occurs when researchers have preconceived ideas or expectations that influence their interpretation of the data, leading them to focus on information that supports their beliefs and ignore contradictory evidence.
- Observer bias: The researcher’s own beliefs or experiences influence the way they observe and interpret the participants’ behavior or responses.
- Hawthorne effect: Participants modify their behavior or responses because they know they are being studied, rather than acting naturally.
- And many more…
Strategies for avoiding bias in UX research
By continuously reflecting on our research approach at every stage in the project—from planning to reporting—and taking deliberate steps to avoid bias, qualitative UX researchers (and their clients) can be more confident the research is as unbiased and rigorous as possible.
Triangulation
Simply, triangulation means using multiple sources of data or methods to cross-check and verify findings. This helps to ensure that the research is accurate, reliable, and robust, and that the conclusions are based on a strong foundation of evidence.
Reflexivity
Researchers reflect on their own biases by keeping a research journal to record their thoughts, feelings, and reactions to the data. This exercise can help us become more self-aware of bias and how we might be unintentionally influencing the research.
Collaborative analysis
Involving multiple people in the analysis process is an effective way to avoid individual biases. Create an analysis team or use participatory action research methods, which involves collaborating with participants and other stakeholders to analyze and interpret the data.
Mobile ethnography
Researchers can ask participants to use their smartphone or social media account to collect data in real-time and in natural settings. This can help avoid biases that might arise from asking participants to recall their experiences after the fact.
Disruptive research
Deliberately seek out participants who are not representative of the typical population being studied. By intentionally seeking out diverse perspectives, researchers can challenge their own biases and assumptions and gain a deeper and different understanding of the research topic.