Implicit bias in design
Recognizing and understanding our hidden biases.
Bias is a double-edged sword, one that benefits us by streamlining and automating our thought processes, but also one that risks the development of problematic and discriminatory beliefs that are difficult to unlearn. While there have been many articles written about the different types of biases prevalent in design, there was one particular bias that I felt hasn’t been talked about enough: implicit/unconscious bias.
What is Implicit Bias?
While it can be true for most other biases, implicit biases are best framed as the mental equivalent of a reflex: a form of thinking that is quick, requires virtually no effort, and resides in our unconscious. This is different from the more widely known explicit biases (e.g. political affiliations, racial supremacy, etc.) in that individuals may not be aware of the implicit biases that motivate their behavior. Furthermore, despite what we consciously believe, our implicit biases may run contrary to our beliefs.
To test for implicit biases, psychologists have been using the Implicit Association Test (IAT), which presents users with a binary option and measures which option they prefer more. It’s particularly important to acknowledge that the IAT doesn’t indicate an absolute belief (e.g. I like meat and hate vegetables), but a relative one (e.g. I like meat more than vegetables), as many media articles have framed the test as something capable of revealing our hidden racism. This has consequentially led to some in the psychology community to question the validity of the IAT and whether implicit biases can significantly influence an individual’s behavior. While it should be acknowledged that the IAT is an imperfect test, there has been plenty of evidence that demonstrates the consequences of implicit biases on a societal level.
In fields where impartiality and objectivity are important, the presence of implicit bias has repeatedly, and continually, led to severe consequences. In medicine, there has been an issue where certain races, particularly African Americans, have received insufficient aid due to a medical professional’s belief in a myth that certain races have higher levels of pain tolerance. In law enforcement, while most officers may want to uphold peace and protect their community, implicit biases can affect their judgments in stopping “suspicious” individuals as well as their split-second fight-or-flight response to shoot a perceived offender. It’s become such a problem that some police departments have established implicit bias training programs, although its efficacy has been questionable at best.
While the stakes are considerably lower in UX, the pervasiveness of implicit bias on an individual and societal level undeniably affects us as designers and researchers, therefore our intentions towards building accessible and inclusive designs may not be as accessible or inclusive as we might think. Luckily, there’s decades of research into identifying and addressing implicit bias, and it all begins with a little introspection.
Implicit Bias in UX Research
While it should be acknowledged that scientific research and UX research (UXR) are handled differently and have different purposes, they do share two common traits:
- A goal to establish a truth.
- The quality of the research components directly affects the quality of the truth produced by the research process.
The first trait is essentially a given: we do research to fill a gap in our knowledge base, and we use our newfound knowledge to continue chipping away at a much larger gap in knowledge or to use it in a practical application. The second trait, however, speaks on the quality of the components that make up the research process — mainly the researcher(s), research participants, and the collection, analyzation, and interpretation of data. This is where biases, especially implicit biases, can skew our findings and lead us to make misinformed decisions despite having the best intentions. If the research process is flawed, everything that follows is flawed.
When it comes to UX, there are many points during the process that ours, or our participant’s, biases can corrupt our research conclusions:
- Focus groups (i.e. user groups) may consist of people who fit a set criterion first, rather than a group that reflects the population (i.e. availability bias).
- Participants may be swayed to respond based on what they think they want the researcher to hear (i.e. participant bias).
- As researchers, we may inadvertently favor findings that confirm what we initially thought rather than accept what the data represents. Even worse, we may construct our research around what may most likely confirm our hypothesis (i.e. confirmation bias).
While certain biases may only exist at a single point during the research process, implicit bias is prevalent throughout the entirety of the process. We may be selecting users based on who we believe the ideal user would be, treat participants in a way that would alter their responses, and carry a perception of the ideal outcome that may not be what’s best for the user or our clients. If we are unaware of what drives us to do what we do, we run the risk of creating a problematic habit that we may never think to challenge.
Personas
This ultimately leads to what initially pushed me to write about implicit bias in the first place: personas. Coming from an academic research background, personas were always one of the main UXR components that puzzled me the most: why are people anthropomorphizing data? Initially, it seemed like a way to organize raw data into something that UX researchers and their clients could understand, but looking deeper, it’s a flawed system because it cannot encapsulate the raw data in its entirety nor would it necessarily correlate with real-world behaviors. Furthermore, personas potentially facilitate the cherry-picking of data so that an idealized character can be crafted and used as the focal point for the remainder of the design process — a prime opportunity for implicit bias to thrive.
Beyond the risk of a researcher selecting and omitting data for the sake of a persona, another deficit is that personas often don’t represent people with specific needs and disabilities. While there are certainly some widely known accessibility standards (e.g. alt text for screen readers), they often seem like an afterthought or add-on once the core design has been completed. If personas are meant to help us empathize with our users, why aren’t there more personas that help us to empathize with people who requires a bit more aid? This is implicit bias at work: while we consciously support and strive for accessibility and inclusivity, our methods don’t reflect that commitment.
Where we go from here
As someone who’s relatively new to the field of UX, I’ve been fortunate enough to be able to apply an outsider’s perspective to a field I’m continually learning from, but I’ll inevitably lose this ability the more I become immersed in the UX community. This in itself is another kind of implicit bias that we should be aware of: in-group/out-group bias — an instinctive bias where individuals take on a group’s ideology and become more retaliatory towards opposing groups and beliefs.
Addressing implicit bias, and any other biases, on a personal and societal level is no easy feat, and there’s no quick-fix solution due to how ingrained our biases are in our psyche. The first step, however, is to recognize where we can go wrong and to bring our problematic unconscious biases into our consciousness. Once we are aware of what holds us back, we can do what we do best as a field: problem-solve through vigilant discussion and reiterative feedback.
