human behaviour
July 17, 2023

Human behaviour is more complex than too much shallow analysis

By Paul Laughlin

I am writing this opinion piece to capture the thinking I’ve been doing since completing reading “The Model Thinker” which reminded me of the complexity of human behaviour.

As well as being impressed with the depth & practical relevance of that book, I was reminded of too much simplistic work in the past. In both teams that I have led and in other organisations, all too often I see behavioural analytics reduced to simple observation. Seeking to identify common patterns or trends, even predictive variables for past behaviours of interest.

Yet, reading that book reminded me to think again about the reality of what is really going on. Reflecting on my own behaviour (for instance when buying products) as well as what I observe in others. Multiple influences (including social pressures, personal networks & decision-making biases) come into play. The real world is much more complex than a simple data viz of past behaviour or a simple regression model.

What could you be missing by simplistic behavioural analysis?

First, let me explain what I mean by simplistic analysis. Past experience and some clients with whom I’ve worked tell me that much “understanding of customer behaviour” is a gross simplification of reality. Traditionally this is often multi-variate analysis to spot typical/majority past behaviour, perhaps early indicators or ‘triggers‘ for the actions of interest & perhaps clustering by differential behaviour patterns. Building upon this exploration of that data, many analysts (or data scientists) will then perform regression analyses to both identify predictive variables for target behaviour & deliver a propensity model.

But, let’s step back a moment and think about our past behaviour. How much could your past buying (or other actions) decisions be explained by just a few observable variables? For many of us, much of the time, more factors are at work. We are influenced by family & friends. Behavioural biases like recency, temporal discounting, anchoring or Status Quo bias come into play. Plus, our decision-making is influenced by viral marketing, our values, societal trends, competitor & other media messages plus where we happen to be when needs occur to us.

So, before you or your team rush into yet another cursory analysis of behaviour observable by your organisation, I encourage you to pause. Perhaps take some time to read “The Model Thinker” or “How to make the world add up“. Take that time to consider other potential approaches that could explore more of the full dynamics at play. To help prompt your thinking, let me share three potential perspectives that have occurred to me…

1) Considering networks of influence

A number of the chapters in “The Model Thinker” explore potential network models that can help us. The author introduces different network topologies, formation, function & robustness. More importantly for our present discussion, he outlines where they can be relevant. We see how such models can be applied to different types of important social networks (family & friends, political organisations or online influencers). This topic is especially important for those working in B2B insight teams, as buying behaviour by organisations is really a complex network of individual relationships & influence.

2) Systemic influences and competitive behaviour

All too often as well as analysing our customers as if they are not influenced by our competitor’s behaviour we also view them in isolation. What I mean is we do not consider how their behaviour might be in response to social or organisational norms for them or the behaviour of others. Once again, “The Model Thinker” can help us consider other perspectives. In chapters on both Systems Dynamic models & Game Theory models we explore how the dynamics of both the wider systems in which a person operates and how they view the behaviour of others influence their actions. Aren’t you too influenced by historic norms of behaviour in your family or employer? Can you spot times when you were really competing with or reacting to the choices or behaviour of others?

3) Considering evolving understanding

It’s amusing to consider that we might be giving more time to consider the power of AI than better understanding people. What I mean is that many mathematical approaches to modelling human behaviour ignore the potential for those people to learn. We talk a lot about new machine learning models, but when did your analysis or people apply learning models to proxy their behaviour? In chapters on Learning Models & Rugged Landscape Models, Scott Page demonstrates their relevance. How different models of reinforcement learning can help explain human behaviour. Taking into account people’s learnt experience as to what gives them the greatest pleasure, alignment with values or social good. If you’ve never considered such an algorithmic approach I recommend it.

Is this challenge relevant for you or your team?

There, I feel sated from the thoughts that have been buzzing around my head. I hope the above challenge was helpful. I’m interested to hear from those who have taken such an approach or see its relevance for their work. Feel free to disagree as well, it would be great to start a conversation on this.

I was encouraged earlier today to read a very similar perspective. This was in a recent post on the very active & helpful hub Data Science Central. In his post, Bill Schmarzo (an author I’ve recommended before) outlines the importance of data scientists thinking more like economists:

Next-Gen Data Scientist: Thinking Like an Economist –

Generative AI (GenAI) products like OpenAI ChatGPT, Microsoft Bing, and Google Bard are disrupting the roles of data engineers and data scientists. According to a recent report by McKinsey, these GenAI products could potentially automate up to 40% of the tasks performed by data science teams by 2025.

For those wanting to explore further, I recommend four books to help you take your analysis of human behaviour to the next level:

  1. To ensure you have a statistical foundation: “The Art of Statistics” by Prof David Speigelhalter
  2. For the mindset you need for such an approach: “How to make the world add up” by Tim Harford
  3. Explaining & applying all the modelling approaches above: “The Model Thinker” by Scott Page
  4. If what you really need to explore is causation: “The Book of Why” by Judea Pearl

I wish you well for such an exciting learning journey. Please let me know what helps & how.