5 useful tips to reducing bias in UX research: a beginner’s guide.
What is cognitive bias and how does it affect our work as user researchers and design decisions.
INTRODUCTION
Becoming a better UX researcher requires practice and a lot of thinking and analysis. This becomes tricky because the human brain is powerful, but it also has limitations. One of these limitations is the “cognitive bias”. As UX practitioners, it is good to be aware of our bias and take steps to mitigate it to reduce its effects on critical aspects of product development.
WHAT IS BIAS AND HOW DOES IT IMPACT USER RESEARCH
Cognitive bias are mental errors that are difficult to access and are likely to exist deep within our minds. They affect everyone and occur when the brain tries to make decision-making easier. They do, however, influence how we think and act. Through user research, we can better understand users and their needs. Unintentionally, bias creeps into user research studies, influencing the results and skewing them in favour of certain users or groups.
Here are four examples of common bias in user research, as well as some personal experiences with it.
- Confirmation Bias: The tendency to select and favour ideas that confirm one’s own pre-existing beliefs is known as confirmation bias. Confirmation bias is dangerous for Uxers because it prevents us from being open-minded and eliminates the possibility of other viewpoints. For example, I wanted to learn about voters’ perceptions of voting and elections in a specific region, and I assumed voters would find the experience stressful, so I was surprised to hear voters say they enjoyed it.
- Framing Effect: The Framing Effect: People respond differently to the same questions depending on how they are phrased. The way we ask questions and analyze data reflects this bias. We can avoid this by refraining from asking leading questions. For example, rather than asking directly about pain points or “joy points”, I try to ask “How was your experience using this product” when conducting research to gain insights into the users’ experience to avoid this bias. With 1, 037 UX practitioners, the NN/g investigated this bias and how it affects our research findings. Participants were randomly assigned to see each version of the hypothetical study results, according to NN/g:
- 4 out of 20 users could not find the search function on the website.
- 16 out of 20 users found the search function on the website.
Half saw the negative version and half saw the positive. All were asked the same follow-up question: “Should the search function be redesigned? Of course, everyone had a different reaction to this question, and while there is no right or wrong answer, the survey revealed how framing can influence design decisions. Check the study out
![A graph of showing the research by the NN/g](https://miro.medium.com/v2/resize:fit:488/0*x3sfkHlyldqnhoO5.jpg)
- Recency Bias: People tend to give their most recent experiences more weight, which can have a positive or negative impact on your data. When responding to survey questions, respondents are more likely to base their decisions on their most recent experiences rather than their overall experience. For example, When I use survey responses, I always try to get a large sample size because it helps to cancel out some of the bias.
- False Consensus Bias: We tend to assume that others will share our viewpoint. People have different perspectives and interact with products in different ways. Instead of assuming that you know what your target audience wants and needs, you should work with them and test your ideas so you don’t invest time and effort developing products users don’t need. For example, when I conducted a usability test, I assumed that users would prefer to prioritize one feature over others because it would save them time. The usability tests, on the other hand, revealed that different users interacted with the app in different ways.
HOW MIGHT WE REDUCE BIAS WHILE CONDUCTING USER RESEARCH?
Here are five useful tips for reducing bias in research:
1. Be aware of your own personal biases.
As researchers, we must be aware of our own biases, and one strategy I recommend is being open about our assumptions before beginning research. This makes it easier to be open-minded and self-aware of your assumptions after spending time looking at data before beginning research. I use the “5W1h” methodology to break down
- who I believe my target users are
- what problems my solution solves for them
- when they would use my product (during a commute, drive, swim, etc. )
- where they would use it (car, mobile phone, desktop, watch)
- how they would use it ( on an app, web interface, physical product etc.).
2. Create a research plan:
Create a research plan that includes your research objectives, desired participants, their characteristics, the interview script and the metric you want to use to evaluate your study. To avoid the framing effect, it’s also a good idea to be introspective by reflecting on each step.
This allows you to be
- More specific about your target users — while also ensuring diversity and inclusivity.
- It also allows you to structure your user questions. You can avoid the framing effect by not asking leading questions.
- Define your method for analyzing results and user behaviour, including (i.e., ideal time-on-task, expected bounce rate, etc.) and how you’ll categorize user behaviour quantitatively (i.e., what percentage of test participants should complete the flow successfully).
3. Listen to your participants
Interrupting your participant while they are answering your questions or interacting with your product is not a good idea. Listen to them and create a safe environment in which there are no right or wrong answers, only different points of view. Break the ice, be friendly and ask clarifying follow-up questions like “Why do you think so” or even a simple “How so” instead of leading questions and interpretations of user feedback. Keep an eye on your body language and try to remain as neutral as possible.
4. Have a large sample size
This is particularly useful when conducting research through surveys. One of the things that can make your research unreliable is small sample sizes. Obtain a large sample size of users who represent all segments of your target market. You won’t be locked into one group’s perceptions this way, and you’ll be less prone to biases like social desirability and false consensus.
5. Make results of your research:
Reflect on the data and ask someone from your team to look at it with fresh eyes and a different perspective for each piece of feedback you receive and group it into insights. This will give you a more objective and well-rounded view of what the data means for your product.
CONCLUSION
It’s easy to believe that these biases only affect other people. However, it is natural to have them. It is beneficial for us to be aware of these biases and how to reduce them so that they do not have such a negative impact on our analytical and critical thinking that it affects design decisions and product goals. Which ultimately reduces what we as UX practitioners optimize for — designing human-centred experiences.