Tag

data science Archives - SUE | Behavioural Design

Zombies - conspiracy

Why data scientists and conspiracy theorist have a lot in common

By | All, Behaviour in Organisations

I want to share a couple of thoughts and insights on how data produces Fata Morgana’s . I have become a bit obsessed lately with how easy it is to be fooled by data. In this blog want to argue that many researchers and data scientists fall for exactly the same mistakes as conspiracy theorists.

The psychology behind Conspiracy Theorists

You probably heard of Qanon. It’s a conspiracy theory about liberals running secret Satan-worshipping, child molesting, blood-drinking networks. The Qanon theory spread like wildfire on the internet in the last couple of years. In a brilliant post on Medium a while ago, a game designer argued that the nature of Qanon is strikingly similar to a well designed Alternate Reality game.

Alternate reality games (ARG’s) are designed for you to look for cues to solve a puzzle. One of the problems that game designers often encounter is a phenomenon called “apophenia”. Apophenia is : “the tendency to perceive a connection or meaningful pattern between unrelated or random things (such as objects or ideas)”. 

Better said: Once you are searching for patterns, you will start finding them everywhere. Your players might encounter some scraps of wood on the floor that accidentally form an arrow, and they will become convinced that this must be a clue and can’t be a coincidence.

The same mechanisms are at play in the alternate reality of conspiracy theorists. The thrill of being a Qanonist is that cues are everywhere. Once you are sucked into the community of like-minded truth seekers’, you will stumble upon cues that are so convincing that they must be true. The addictive part is the fact that your fellow conspiracists will challenge you to connect the dots for yourself. “Wake up! Open your Eyes!” Nobody tells you what to think or believe, but once you connect the dots, the truth will reveal itself. Cracking the puzzle is similar to the dopamine rush you get from solving a game puzzle.

Of course, the problem is: Once you start looking for patterns, you will always find some. You don’t believe in Illuminati? Well, what about all these pictures of Hollywood stars who use the Illuminati “one eye” symbol? Coincidence? I don’t think so?

Still not convinced that liberals run satanic networks? Well, why do all these Hollywood stars use the 666-symbol, the number of Satan? Once you start looking for it, it’s so damn obvious! How can we all have missed this?

666

The Full Fundamentals Course experience.
All from your Home or Office.

You can now access the FULL Behavioural Design Fundamentals Course live from your home or office. Watch a live trainer and get real-time feedback & coaching. Interact with fellow peers. Work together in virtual break-out rooms. Don’t let COVID slow you down!

((Image Courtesy: Curious Institute)

Of course, all of this is an illusion. An illusion fostered by wishful thinking, selective attention and the addictive thrill of finding patterns.

Data Scientists are a bit like conspiracy theorists.

Today I read a piece in the Dutch newspaper NRC about recent research amongst Dutch voters. We’re having elections here within five weeks. The study delved deep into the wants and beliefs of the Dutch electorate. And lo and behold, it discovered some fascinating patterns: “About 30% of the population are culturally conservative but economically liberal”. Or “There’s still an untapped potential if far-right parties would embrace more leftwing policies” or “Although 70% are in favour of a big government and income redistribution, progressive parties are suffering from a steady decline. This loss can be attributed to the fact that only a minor group of those people (15%) favour progressive themes as abortion, euthanasia, multi-culturalism and European unification”.

The problem with all the above: Sounds reasonable, but it’s bullshit. Real people don’t change their voting behaviour based on these issues. They answer a different question in the voting booth: Which leader or team do I trust the most to fight for the things that threaten my way of living? To whom do I sympathize?

Under the article, NRC posted a series of short portraits of voters and their consideration. The first portrait was featuring an entrepreneur, aged 35. Every time he filled in a voting configurator online, the Dutch Liberals came out as the party that best matches his beliefs and values. Yet he categorically decided not to vote for the liberals because he chose to answer a more powerful different question: He feels it’s time for a system reboot. So he feels more sympathy for the challenger parties, some to the far left, some to the far right. He voted for the FvD (a far-right party), but only because he felt sympathy for one of the (ex) leaders’ fighting spirit, even though he despises their racist whistleblowing.

 

Why you should have a healthy distrust for data.

The problem with quantitative research is that numbers and graphs signal objectivity and power. If you make the case with solid data, you are more convincing. But the problem is that the patterns we find are often a mirage. A fata morgana that the dataset produced. In the case of Qanon, the fata morgana is produced by combining random pictures that suggest a secret code. In the research above, the fata morgana is created by asking for beliefs and values within the voter base of parties. But it only takes a simple look beyond to data to realize that parties’ rise and fall have everything to do with the rise and fall of their leaders.

Tom De Bruyne
Co-Founder SUE Behavioural Design

PS: If you like this post, don’t forget to subscribe to our free Behavioural Design newsletter, in which we look at the world around us to decode how influence works.

Want to learn more?

Suppose you want to learn more about how influence works. In that case, you might want to consider joining our Behavioural Design Academy, our officially accredited educational institution that already trained 2500+ people from 40+ countries in applied Behavioural Design. Or book an in-company program or workshop for your team. In our top-notch training, we teach the Behavioural Design Method© and the Influence Framework©. Two powerful frames to make behavioural change happen in practice.

You can also hire SUE to help you to bring an innovative perspective on your product, service, policy or marketing. In a Behavioural Design Sprint, we help you shape choice and desired behaviours using a mix of behavioural psychology and creativity.

You can download the Academy brochure here, contact us here or subscribe to Behavioural Design Digest at the bottom of this page. This is our weekly newsletter in which we deconstruct how influence works in work, life and society.

Or maybe, you’re just curious about SUE | Behavioural Design. Here’s where you can read our backstory.

sue behavioural design

How Behavioural Science and Data Science should collaborate

By | All, Behavioural Science

Last week I had the honour to speak and lead a data science hackathon at the I-com conference in Malaga. I have never been surrounded by so many brilliant people in my life. What fascinated me was how much behavioural designers and data-scientist have in common, and yet, how little both disciplines know about each other, or even collaborate with each other.

Why I fell in love with Data Science

What data-scientists do is they look at data sets, look for patterns in that data and use that understanding to build a model that could predict behaviour. Their models predict things like: “What products will new mums buy more or less?” or “when will you watch what kind of content on which device?”, or “in which region can you stop distributing product catalogues, without hurting sales?”. I saw teams in the 24-hours hackathon come up with mindblowing predictions. And I felt really stupid for not understanding a single bit of how they build their models. But the thing is: I don’t need to because the computer simply calculates the predictive power of the model, so there’s no cheating or bullshitting possible. Fascinating stuff.

The blind spot of Data Science

There is, however, a very big limitation to this approach. And that’s the fact that we’re dealing with humans. It’s not because I can predict to a certain extent your future behaviour, based on data-analysis, that I wouldn’t be able to influence you to make different choices. After all, our choices are heavily influenced by how choices are being presented to us.

I can make you reconsider the choice of buying a camera, by letting you choose between three, instead of two models. If I present you with three models, you are far more likely to chose the middle one. And you wouldn’t have chosen it if I only had shown you two. If I play loud music in a restaurant, you will buy more unhealthy food. If I would prime you with images of Italy, you will buy more Italian food. If I add urgency and scarcity to buy the last items of a sales promotion, you will probably choose differently. The means for screwing up the behaviour that is predicted by the model are endless.

Meaning versus Reach

The challenge with the classic behavioural design method of interviewing and observation is that it’s very rich on meaning but poor on reach. And data science has exactly the opposite problem: the insights are scalable, but they are not rich in meaning. The fact that new mums buy much more housecare products, but buy far less mouthcare products , doesn’t explain why they do that. If you can understand why people do the things they do, you can easily figure out new ways to optimize your marketing.

How to collaborate?

The answer is simple: Design a creative process that leverages the best of both worlds. I would always follow these steps:
Let the whole team do interviews to develop a deeper understanding of how the target audience thinks, feels and behaves. This helps them to overcome their own biases, assumptions and prejudices and helps them to build interesting hypotheses.

Analyse the data you have with the hypothesis you’ve just formed. Try to figure out which hypothesis actually predict behaviour. But also: dare to go back: if you find interesting other patterns (e.g. non-parents buy way more deodorants than new parents), try to see if this insight could help you to revise your deep understanding of the drivers of the behaviour of your audience

Once you’ve developed a predictive model, based on qualitative and qualitative results, use Behavioral Design principles to come up with ideas for improvements of the customer journey
The fun part: Prototype, test and measure. Design experiments, measure results, improve your overall plan. Aggressive experimentation is what sets apart the truly innovative companies from the laggards.

This is such an interesting time to really make a measurable impact. But everyone’s struggling with the HOW-question: how to turn a deeper understanding of behaviour into business value. The creative method is the answer.

#justsaying 🙂
Have a great day,
Tom

Want to learn more?

If you want to master the science of influence yourself, you could consider enrolling in our two-day course Behavioural Design at our SUE | Behavioural Design Academy. You can download the Academy brochure.

Or maybe you currently have a challenge in which you want to influence choice or change behaviour. Please, take a look at our Behavioural Design Sprint. It might be the answer you’re looking for.

Or could be you just would like to get to know us a little better. We happily introduce ourselves here.