Meet the Experts

Interview to Steven Johnson: at the intersection of Data Science and Behavioral Science

“The reason why I’m so passionate about behavioral science and why I got into this area in the first place is because it represents a massive opportunity for us to have a positive impact on the world”.

How can Data Science be used to derive insights on human behavior?

Fundamentally what Data Science has done in the context of behavioral insights, is that it has massively extended our ability to identify behavioral patterns. The whole enterprise of Behavioral Science is focused on that: we are looking for regularities in past human behavior, so that we can predict future human behavior. If we can understand what predicts human behavior, we can then use our knowledge to build interventions and products that make certain desired services more likely to happen.

Data Science sits within a whole ecosystem of what I would consider behavioral innovation: given the computing power and the size of datasets that we have now, this opens the doors to new and bigger opportunities to identify those behavioral patterns and make predictions.

I noted with my own work that clients found it useful to understand the application of Data Science on three levels:

You mentioned that traditional nudge theory uses a “one-fits-all” approach to implementing interventions, could you give us an idea of the emerging technologies and methods that Data Science uses in understanding human behavior?

What we found with our clients is that a lot of organizations have been collecting plenty of customer data for a very long time, but they just don’t do anything with it. It’s mainly data that their operational systems sweep up as a matter of course as the business runs. Therefore, I think it is fundamental in the first place to make the most of the data you already have.

We are also seeing a proliferation of the use of RCTs (randomized controlled trials), which generate directly observed behavioral data. Furthermore, given that there are more commercial organizations as well as academic institutions using RCTs, we have the opportunity to build big datasets with RCT data and use an aggregated approach to analysis.

Another example are megatrials (example on vaccination hesitancy) with huge datasets and massive amount of (diversified) data. Then there is the Psychological Science Accelerator, that is trying to assemble global networks of research teams to run trials across many different cultures and geographies – all with the aim of addressing the WEIRD bias in the existing evidence base.

In a sense, many things that are going on are just an extension of what has already been done. The game changer, though, is the digital space and the amount of behavioral data we are starting to accumulate through mobile devices, Internet of Things and Surveillance.

In an article for the Guardian (2013), you emphasized how, given that people are not that good at understanding themselves and their choices, the use of self-reported measures for understanding consumer behavior, such as surveys, should decrease, while the use of approaches based on ethnographic or co-creation principles should be preferred. Could you introduce us these two approaches?   

Quantitative research, allows us to understand the What, and, in my opinion, it is even more important that we understand the Why; that is where the qualitative research comes in. At the time, the qualitative research that was used predominantly included focus groups, interviews, and other self-reporting methodologies. For a long time now, though, Behavioral Science has been telling us that, given the amount of processing and activity that goes on outside of conscious awareness, human beings are not necessarily best placed to reliably report their own needs, wants, and desires.

Ethnography and co-creation are two techniques we designed to gain direct access to people’s lived experience and answer the Why question.

Ethnography, or participant observation, involves spending time with the people you are designing for. You may talk to them, listen to them, ask them questions, but the most important part is observing their behavior within a particular context. This unlocks deeper insights than a retrospective memory of what somebody did, felt, or intended.

Co-creation, or co-design, follows the principle of designing solutions directly with the people who will be using certain products or services. Rather than asking questions to people, we are collaboratively engaged in design-related tasks; we are actively working together to solve a problem.

Moving forward in the discipline requires greater diversity in the practitioner community, research community, and the communities we are researching with. Together, this pushes us to move towards personalized interventions. In my opinion, the use of co-design is probably one of the best ways of thoroughly understanding how an intervention needs to respond to the needs of the people we are dealing with, rather than taking a one-size-fits-all approach based on the average overall impact size of interventions within an RCT format.

As you also specialize in human-centered design, could you introduce us to the field and its present-day relevance? 

Human-centered design can be seen as a technical discipline (with defined processes, frameworks, and tools), but it can also be seen as a mindset. Whether or not you are well-versed, technically competent or qualified as a human-centered designer, you can still adopt a human-centered design mindset that will enrich your practice and research. In general, what underpins both the mindset and the technical discipline is ensuring that everything we do is rooted in an understanding of the human condition.

Human-centered design starts with the issues facing human beings and tries to use that understanding to drive the research element and the design elements of a process. A lot of the things that we have previously talked about, like the use of research techniques such as ethnography and co-design, could be considered characteristic of the technical discipline that sits under the human-centered mindset.

The importance of human-centered design now must be seen in terms of the progress the Behavioral Science discipline needs to make towards more personalized interventions in line with a greater diversity of lived experience. Human-centered design brings this element to complement Data Science.

In the study of Behavioral Science, we often refer to Nudge theory as developed by Richard Thaler and Cass Sunstein. With advancements in Data Science, Sunstein is concerned about the rise of ‘hypernudges’. Could tell us a bit more about ‘hypernudges’ and your opinion on it? 

The use of hypernudges seems to be very similar to that of nudges but within choice architecture and within a real-time data dynamic environment. Karen Yeung wrote a brilliant paper on Hypernudges, which is useful to outline the space. As we have now behavioral data at such a scale and in such a real time (working dynamically with the observed behavior), we can design and constantly refine choice architectures in a very small and subtle way. As a consequence, the role of Behavioral Science as a means of influencing human behavior has massively increased.

In general, the reason why I’m so passionate about Behavioral Science and why I got into this area in the first place is because it represents a massive opportunity for us to have a positive impact on the world.

I think the level of complexity and sophistication that we’ve got to through hypernudges and other Data Science-related technologies previously introduced, could be massively positive if used in the right way. Of course, though, there is also an opportunity to drive commercial growth for a concentrated set of interest, and maybe even issues around State control, surveillance practices, and the use of data from the public realm. As with any technology, it has all to do with how they are used: with great power comes great responsibility.

Any behavioral change tool begsan ethical question. This could be a situation where our political and ethical philosophies aren’t able to keep up with pace of technological change. In fact, touching upon the ethical issues surrounding the use of hypernudges and similar approaches, I don’t think that our current understanding of informed consent is really fit for purpose in a world of hypernudges, as the notion of consent related to the use private data is very outdated.

When the recent Netflix documentary “The Social Dilemma” was released, many concerned were raised about addictive app design. How can apps be better designed to cater to the well-being of its users?

Firstly, I think we need to take in consideration that there is a whole host of apps that are designed to increase people’s satisfaction, happiness, and wellbeing (e.g. mindfulness apps, fitness trackers, etc.), though I’m not quite clear on the evidence based for how effective they are. Furthermore, I think we might need to understand if it is the addiction per se or the harm that certain apps may bring with them that we are interested in.

For instance, in terms of the addiction to technology, one could argue that we have become addicted to cars. We have been on a long journey in our relationship to cars and a lot of positive things have come with that, but a lot of harm has come too (environment, deaths, etc.). It has taken us a while to work out what those harms are and then mitigate them while still allowing us to benefit from the positive things the technology brings. Over time, we’ve got closer and closer to a relationship with cars that is more sustainable and healthy.

Another example might be the obesity problem: we are reformulating products so that they give us all the same nutritional values and same taste, but without the harm associated with the high levels of fat, salt, and sugar. On the same level, I think the journey we will go on with apps might be something similar. First, we need to get clearer on what is the harm that is being caused and by which apps (we mainly think about social media). Once we have done that, we should try to parse out the negative harms from the positive benefit (for example social media, especially in the last year, have been fundamental in keeping us in touch and closer to people).

How did you get in the field and do you have any advice for people who are looking to get in the same field?

There is a whole debate going on that I’m really interested in: do you need a Behavioral Science degree or to be a behavioral scientist to be ‘doing’ Behavioral Science?

I came from a design background, and what I’ve done is assimilate an understanding of Behavioral Science and Data Science into my practive. I’ve done this for quite a while, and I think that that combination unlocks a huge amount of power for innovation.

I think we should start seeing Behavioral Science and Data Science as just two sides of the same coin. Can we really separate the two things apart or do we want to? That doesn’t mean to say that everyone who wants to be a behavioural scientist or is exploring Behavioural Science, needs to be able to program in R or Python, build models, or get to that level of technical complexity. However, having a conceptual understanding of the topic is fundamental in order to work productively with data scientists. Together, they can define behavioral problems in the right way and seek the right tools in the first place.

Similarly, for a data scientist, I don’t expect them to be steeped in the behavioral economic literature and understand all the cognitive biases, but, as they’re dealing with behavioral data, they need to have some conceptual understanding of how that data can then be used, how the value of that data can be unpacked.

Whether it is Behavioral Science or Data Science, or whether it is what I like to call behavioral innovation, I think to a certain extent we’re all going to become more interested in how to unlock the power of Behavioral Science to create things that have an amplified positive impact on the world.

In general, I believe in the notion of T-shaped people and I find that to be a really useful guiding principle. It means having a deep specialism in one field or discipline (might be Behavioral Science, Data Science, or other), together with a shallower, but broad understanding (at a conceptual level) of all those other disciplines. This allows for multidisciplinary collaboration different pockets of knowledge, can come together to productively address some of society’s most pressing issues.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s