Data never lies...
I am sure we have all heard that phrase before. It is especially common for it to pop up in arguments. You are discussing with a friend about a hot topic or perhaps its a business meeting and you are debating which supplier you will go with. That is where somebody wanting to reinforce their argument will show some data that reinforces his point and expect that this will change your mind too. After all, its science and science can't be wrong ... right?
People do lie
In fact, there are multiple ways such arguments could be much weaker in practice. First of all, the data itself might be wrong or biased. But even assuming you can trust the source, data without proper scaling is of little importance. Always pay attention to starting points of diagrams and the relative sizes of shapes, because while the data might be true, the visualization might be optimizing for storytelling.
Also good piece of advice: ask for anchor values. A sales number of €3 million might sound impressive, until you realize that the company deals with €1 million dollar products and has revenue of more than €3 billion. Put data into perspective
The magic is in the interpretation
But more importantly, data and science both are very good at giving us answers but not posing questions. We should always be aware of what question is being asked and what are the implications.
Here are three practical examples:
- Our company makes two products. After analysis, it turns out the profitability per unit for product A is €100 and for product B its -€20. A "data based approach" might hint to ditching product B and going all in on A. But B might be a product that satisfied a customer so much that ends up giving us more deals of product A.
- The ministry of health wants to pay doctors by their efficiency. In a remote hospital, there are 2 doctors with a mortality rate of 0.1% and 15%. Its tempting to say that the 0.1% doctor is better, but think of what is a doctor... Sometimes, they are forced to take on cases where the patient is in a bad shape because of age or smoking. A brave doctor that accepts all cases will thus inherently have a worse score, despite potential being a better doctor.
- Considering the GDP of a modern country, power accounts to roughly 3%. Selling or losing that, wouldn't mean that the GDP of the country would fall only by 3% though, since all modern infrastructure and production relies on power. Hidden dependencies might obscure an otherwise clear data point/
In conclusion, while data may never "lie", it is still capable of hiding the truth. So ask better questions!
Comments
Post a Comment