Survivorship bias compared to Pareto principle
Reflexion about survivorship biasis compared to Pareto principle​
What is the survivorship biasis ?
Let's look at aircraft that came back from the battlefield. Bullet holes are examined to determine where to add armor on the plane. The problem here is that those planes came back, and the aicraft that should be analyzed are the ones that didn't return.
Armor plate will so be most efficient where bullets didn't touch the returning planes.
In conclusion, survivorship biasis occurs when the analysis of data or observations is skewed because only few are taken into account. It can leads to false conclusions in statistics, economics and analysis.
About data
The important thing to remember is that bias is about to happen, mistakes is about to be made, a data that seems correct will not give the expected result while compared with other data and the reason will not be easy to find.
If you ask an AI the approximate hour of the average clock in the world, it will likely answer 10:10 or close to 10:10.
Why ? Because his database is full of watch pictures that shows mostly 10:10 on the clock.
Why ? Marketing, 10:10 or close to 10:10 looks good.
You can determine the average hour of every clock in the world using your algorythm, however, the result is false.
Expend the database ? The result will remain the same.
Using digital watch ? Fun fact, same result.
What is pareto principle ?
It's a concept named after the italian economist Vilfredo Pareto. He observed that almost 80% of the wealth in Italy was owned by 20% of the population.
This principle applied for various fields today such as business in time management, quality management, risk management etc... It states that roughly 80% of outcomes result from 20% of cause and so, to grow to have to focus on the 80% wasted.
In conclusion, a minority of inputs generates the majority of results.
Why the comparison ?
While company today are working hard for determined how to maximize profit, it can be an easy thing to look at Pareto and estimate that only the top 20% should be looked at for incresing benefits. However, it's important to keep in mind that from all the various sectors that represent the company, some data might be biased, even more if the company is extended in the international.
And by looking at what should be improved in those 80% we often forgot data that are important to understand what is working now and why it is working on different sectors.
What do we got at the end ?
While Pareto can be effective finding issues it globally become a "armored the bullet hole part" solution, and history shows us how ineffective is this way of thinking. Details are good, but sometimes having a wilder view on process rather than on specific part of it can bring in mind better solutions that will often become part of the work of change management.
Company, for further improvement should focus on what is working globally and investigate for the root causes of problems before they happen in order to maximize the mitigation when issues are going to occured.