Laying the groundwork for creating statistically valid personas.
Photo by cottonbro studio
Personas are more than just UX and product design concepts; they are strongly anchored in psychology ideas about personality types. The significant overlap between ‘persona’ and ‘personality’ is not coincidental. Alan Cooper, credited with pioneering the incorporation of personas into UX research, may not have been completely aware of the psychological underpinnings, but the parallels are enormous and unmistakable. Personas, as used in psychology and UX research, refer to individuals as full beings.
Taking this holistic perspective allows UX researchers to shed light on similar groups of users who differ fundamentally. Recognizing these qualitatively different groups provides product managers and UX designers with a more meticulous lens through which they can predict and observe user behavior.
As a UX researcher who has developed statistically validated personas for an app with more than 7 million daily users, I’ve realized that beyond a certain point in an app’s evolution, it’s important to remember that users are more than just numbers and metrics. It’s easy to become absorbed by the vast array of data and to try summarizing it using descriptive statistics. However, as UX visionaries, we have a mission to remind businesses that users are more than these numbers. We must acknowledge that, given the large scale of the user base, it is impossible to meet the needs of every single user. Nonetheless, this fact should not be seen as a burden; rather, the app should strive to do its best in this regard. To this end, we, as UX ambassadors, are tasked with analyzing and introducing groups of users who are, in one way or another, similar to each other. While every user is unique, this uniqueness has its limits, and there are seldom outliers who cannot be grouped on a larger scale. It is our duty to identify and introduce these broader groups to businesses, ensuring that the needs of particularly special and unique user segments are not overlooked.
To this end, I aim to introduce a relatively new approach in the field of quantitative UX research, albeit an established one in psychology: the person-oriented approach. Adopting this approach can significantly impact the strategies businesses develop based on our deliverables as UX researchers.
Exploring approaches in UX research
Variable-Oriented approach
The variable-oriented approach in quantitative UX research investigates and reports on variables such as user satisfaction, NPS, etc., in an isolated context where they are reported based solely on their descriptives, such as average and standard deviation. These provide a general overview of the user base but do not extend beyond that point. Even when researchers aim to identify general patterns within this approach, they may utilize analyses like regression, correlation, and ANOVA solely to test their hypotheses and predict general trends for the future, as well as to investigate possible reasons for the observed metrics.
Person-Oriented approach
Contrary to the common variable-oriented approach, there is person-oriented approach which doesn’t take users as the sum of the variables we can measure. But instead, it introduces users as human beings as unbreakable wholes which should not only be studied in their entirety, but also should be appraised and investigated regarding their unique beings. As Bergman and Trost have written:
According to the person-oriented paradigm, the individual is an active agent in the person-environment system. The system is hierarchical and must be studied by carefully separating its different levels (from the molecular to the global). At each level, the system functions as an integrated, organized totality that is formed by the interactions among the elements, and the totality derives its meaning from these interactions and all elements considered simultaneously (Magnusson, 1990). Magnusson also introduced the concept of the person-oriented approach and developed it theoretically together with Bergman (Bergman & Magnusson, 1997; Magnusson, 1988, 1999).
Key features:
Holistic analysis:
Holism, as opposed to reductionism, provides researchers with a method to study how the unique characteristics of individuals (or personas) and the interactions among these characteristics within an individual (or persona) define who they are. This is not just theoretical; there are also tools and analytical methods that support this perspective.
Adopting a holistic approach in studying users — known as the person-oriented approach — allows us to view them as complex systems. One of the fundamental characteristics of complex systems is the unpredictability of their future behaviors when analyzed using common linear statistical methods. This aligns with a key principle in UX research, which is to focus on asking retrospective questions and observing past or current behaviors, rather than attempting to predict future behaviors of users. After all, users are humans, and humans are complex systems, to say the least. This unpredictability in the behavior of a complex system is referred to as emergence. In other words, human (or user) behavior emerges from a complex mix of characteristics, context, and their interactions, which we may perceive as a single, complete entity. While it’s not impossible to predict what might emerge from a given set of characteristics and interactions, the methods required for such predictions are far from those currently used in UX research. I will explore these methods and discuss how we, as UX researchers, could employ them in a future article.
Respecting, instead of eliminating, outliers and uniqueness:
Traditional statistical methods have the tendency to remove outliers and unique data points to achieve a semblance of stability, creating an illusion of control and predictability in an inherently unpredictable world full of uncertainties. This approach has helped humans and businesses survive for a long time but has also contributed to significant downfalls in the history of economics and corporate ventures. Despite our desire to grasp reality with simple numbers like averages and medians, the truth is more complex than we imagine.
In UX research, users exemplify this phenomenon. No matter how much we attempt to summarize user behavior using averages and descriptive metrics, they consistently surprise us. This element of surprise is what truly keeps the field of UX research vibrant; without it, the role of UX researchers would become redundant, as mere data collection would suffice. Thus, the uniqueness of users is not just a challenge; it is what sustains us. As quantitative UX researchers, we must never underestimate this factor by relying solely on traditional linear statistical methods. Instead, the statistical methods employed in the person-oriented approach not only preserve outliers and unique elements in the data and analyses but also strive to integrate them meaningfully into the broader context of the entire user community.
Persona Template From visme.co
User personas as an outcome:
Adopting a holistic perspective and respecting users as human beings, and thus complex systems, leads to the revelation of another crucial characteristic of users’ collective behavior: self-organization. This phenomenon can be studied at two different scales within user communities and societies.
Firstly, it can be analyzed in terms of how the entire user base, viewed as a single entity, behaves and interacts with our product. Secondly, it often reveals how this cohesive entity can be segmented into similar clusters or groups of users, known as Personas in the field of UX. These personas are self-organized, emergent groups that are unique and qualitatively distinct from each other, yet each group can be studied as a cohesive whole, akin to a single person. Understanding these dynamics might enhance our ability to thrive in the unpredictable world we inhabit.
Common statistical methods:
As quantitative UX researchers, we often handle categorical or ordinal datasets. The Person-Oriented approach introduces two key methods for analyzing such data:
Latent class analysis: This technique is utilized to identify self-organized groups of users (personas) based on categorical data, such as that obtained from a one-time survey. It helps in pinpointing distinct user segments within the data.Latent transition analysis: This method tracks how personas evolve over time. It requires data gathered from surveys conducted at regular intervals, allowing researchers to observe changes in user behavior and group dynamics.
I plan to explore these methods and demonstrate how to conduct them to develop statistically valid personas in a future article.
Should we eliminate the variable-oriented approach from our toolkit as UX researchers?
TLDR; NO, it depends!
When to use a variable-oriented approach
Your product is new.You don’t know your user-base at all and want to gain a general overview.You don’t want to take any strategic decisions.You want to perform simplified comparisons.
When to take a person-oriented approach
Your product is well-established in the market.You want to enhance user engagement with the addition of new features.You want to address deficiencies in your product.You want to understand how the interactions of your users with your product change over time.You want to achieve a detailed understanding of your users.Image Created by Myself
The need for a balanced research strategy
To maximize the effectiveness of UX research, it is essential to employ both variable-oriented and person-oriented approaches. While the former provides a broad overview of user behaviors, the latter offers a deep, personalized understanding of user interactions. Together, they equip UX researchers and product managers with the comprehensive insights needed to make informed strategic and product decisions, ensuring that both broad and specific user needs are met.
This balanced approach not only enhances user satisfaction but also drives product success by ensuring that user experiences are both broadly appealing and deeply personalized.
How to balance the 2 research approaches
The current landscape of UX research significantly lacks awareness of alternative approaches for investigating users. This is why I have emphasized the benefits of adopting the person-oriented approach in this article. However, in a scenario where UX researchers are well-informed about the available options, it becomes logical to utilize these two approaches where they are most suitable. Specifically, the variable-oriented approach is ideal for gaining a broad overview of user behaviors, while the person-oriented approach is best suited for achieving a deep, personalized understanding of users and their interactions with the product. By strategically applying both approaches, products can potentially maximize engagement by fully respecting and addressing the needs, frustrations, and goals of their users.
Introducing the person-oriented approach in UX Research was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.
