Mental Models Mindset
The simplest definition of mental models is that they describe the way the world works. They influence how we think, understand, and form beliefs. Let’s look at a few examples of mental models commonly employed when using advanced analytics, how they reveal flaws in our thinking, and how they can be used as corrective measures.
First principles thinking: First principles reasoning helps clarify complicated problems by separating the underlying ideas or facts from any assumptions based on them. In other words, it’s a way to expose assumptions underlying your thinking and challenge what you think you know about a problem. This process requires you to keep digging, sweeping away unproven assumptions until you arrive at the facts.
One method, called the “five whys,” requires challenging each outcome with the simple question “why?” This technique, first formally used by Toyota as part of their Lean manufacturing process, is now a standard method for getting to cause and effect relationships.
Leading vs. lagging indicators: One way to think about leading indicator metrics is they measure the activities that lead to results. Amazon, a leader in using analytics to drive decisions, refers to them as “controllable input metrics.” By identifying, defining, measuring, and monitoring leading indicators, you can anticipate problems and intervene before it’s too late to fix them. You rely less on postmortem processes like the “five whys” and more on real-time monitoring, intervening, and implementing course corrections.
This method is an excellent way to operationalize a mental model. Rather than periodically sitting down and challenging assumptions underlying past decisions (e.g., postmortem), you set up metrics that challenge assumptions continuously. In other words, the metrics you set up constantly answer the “what” questions (and perhaps the “why” questions) in near real-time. And to the extent they don’t, modify what you are measuring or how you measure it.
For all you data scientists still reading, it should sound familiar, as it’s much like monitoring a model you’ve built!
Probabilistic thinking: Probabilistic thinking is the process by which you determine the likelihood of any specific outcome happening in the future. We engage in this thinking whenever we check the weather to see if it will rain or speculate about the next Super Bowl winner.
But we are not particularly good at understanding probabilities in our personal or professional lives. We tend to use imprecise language to describe the likelihood of an outcome and are overly optimistic about our future predictions. Being right and making correct predictions is important, but knowing why you were right is essential. Adopting practices like Bayesian updating can help maintain your outside view by constantly adding new information to your existing data to get closer to the ground truth.