As individuals, we all have a unique set of heuristics and biases that show up in our daily lives, whether we notice them or not. This means that for a lot of the decisions we make, whether personal or professional, there is an underlying, subconscious process that our brains go through that affects the outcome of those decisions. So, let’s explore some of these heuristics and biases and how to overcome them in this modern age of analytics.
Think of heuristics as a mental “shortcut” your brain uses to try and win a race. The human brain only has so much attention that it can give, much like a runner only has so much physical energy they can expend, so we utilize shortcuts to help us get to the finish line faster, while also conserving valuable energy. Although this method may be rational, that does not mean it is effective. Shortcuts may save energy, but they also rob you of perspective as they cause you to miss out on the full race and what is required to complete it, while completing the full race will give you the full picture, and lead to a real result, not a manipulated one. To make sure you see the full picture, here are a few heuristics to watch out for:
- Affect Heuristic — Or the “trust your gut” heuristic. Results feel right so they must be right.
- Anchoring Heuristic — Also known as “focalism.” This is the tendency to attach or “anchor” our perspective on the first piece of information we receive, even if that information is incorrect or irrelevant.
- Representativeness Heuristic — Or “stereotyping.” This is the tendency to see seemingly similar events, people, or objects as having equal likelihood of occurring or being the same.
Think of biases more like seeing everyone skip a water station, so you do too even though you’re thirsty. Like heuristics, biases typically occur subconsciously. Over time, biases can create the heuristics we use to cut the course. Here are a few biases to watch out for:
- Belief Bias — Or “that sounds right to me.” Results are accepted or rejected based on how believable we find them to be rather than how logical they are.
- Hindsight Bias — Or the “I knew it all along” phenomenon. This is the tendency we have to see past events as more predictable than they actually are.
- Planning Fallacy — Or “I can do that in a jiff.” This is the tendency we have to do the opposite of under promise and over deliver. We tend to underestimate how quickly we can complete a task regardless of our experience of it taking longer.
Overriding Our Thoughts
Knowing that heuristics and biases exist is the first step in overriding them. Below are some ways you can overtake them in the race:
Validate your thoughts with data — even the small stuff. There is a reason a lot of us will bust out the calculator for basic arithmetic. Validation is comforting and a “sure thing” is better than messing up the small stuff in your analysis.
Do not base your perceptions of a problem solely on the newest/smallest/highest piece of information, do the correct hypothesis testing. Imagine you go into a dealership that is selling a Prius for $38,000, but they are willing to make you a deal and sell it for only $35,000. Sounds pretty good, right? First, ask yourself, is an alternative hypothesis possibly true? Is it possible that a Prius does not equal $35,000? Instead of anchoring to that first piece of information, do some research. If you do, you’ll find that the Kelley Blue Book value of a fully loaded 2021 Prius is only around $32,000.
Remember, correlation is not causation (aka spurious correlation). The total revenue generated by arcades is highly correlated (.99, where 1=exact same movement) to the number of computer science doctorates awarded. It is reasonable to assume that PhDs are not causing that much influence on arcade sales. Although the numbers may move in a similar manner, that does not mean they are related. Further research would be needed to confirm if there was actual causation between the two.
Trust the processes and utilize proper model validation on your results. Sometimes results of a model can seem correct, but that does not mean they are. Validate your results with test sets, check for multicollinearity, heteroscedasticity, and whatever else is appropriate for your model. Confirm your results are representative, not just believable. You might recognize this as the person who always has something to say. Being loud and consistent makes people tend to believe them, but that doesn’t mean what they are saying is true or meaningful.
Create robust business cases for your work and know your limits. You may know how something can be done, but can your team preform the tasks involved? Does your team have the right tools for the job? If so, how long will it take them and what is the tangible cost/benefit for the business? Excel is a great tool, but do not expect your team to build complex models with that application.
Heuristics and biases may change the way we run the race subconsciously but being aware they exist should help us all run a more complete race. Taking the time to step back when it matters and challenge yourself on the items in this article will help you make more holistic decisions.