Statistics, Not Anecdotes

Scientific Literacy In Business

By Santi on Jun 25, 2019



Information, Not Opinions

Steven Arthur Pinker is a Canadian-American experimental psychologist, cognitive scientist, linguist and popular science author. He wrote 2 of Bill Gates' favorite books: "The Better Angels of Our Nature" and "Enlightenment Now", which argue, based on statistics, that the condition of the human species has been improving over the years. Violence and poverty are declining, while health, life expectancy, freedom, and education are on the rise. He's not without his critics, though, who characterize him as an unjustified optimist.

If he was able to change one thing about the world, he told The Harvard Gazette, he would like society to focus more on data and statistics:

How do we change this destructive statistical illiteracy and disdain for data? We need to make “factfulness” [...] an inherent part of the culture of education, journalism, commentary, and politics.

This factfulness (the habit of only carrying opinions with strong supporting facts) should also be the driver of business decisions. Companies need to ensure that proper data is gathered before decision-making, not after the fact to defend unjustified courses of action. The world of business management has its fair share of heuristics which, while useful sometimes, can lead to failure without proper analysis. Better to focus on the story the charts tell.

No Better Way

The term scientific management has earned a bad reputation because of its mechanistic view of workers and its bad science:

[...] The committee’s heavily publicized work concluded with a report criticizing Taylorism for its dehumanizing effect on workers — for treating them as automatons whose movements must be preordained by a higher intelligence. It ridiculed Taylor’s pretensions of rigor — a charge later echoed by scholars who found his reports on labor productivity to have about the same scientific quotient as a sermon.

The company is a highly interconnected and complex system. Any management theory should accommodate that fact. The problem was not too much science, it was its terrible quality.

Management is art and science. Thus, proper scientific reasoning has a lot to bring to the discipline. The academic flavor of scientific research is not very appropriate for the fast-paced markets we have nowadays, but more logical (biases, fallacies), statistical (distributions, chart reading) and technological (algorithms, modeling) literacy can be a catalyst for institutional growth.

Thinking consumes time and increases cognitive load (it tires us), so heuristics and traditions will always have their place among information sources -albeit not scientific ones-. There are just too many decisions to be taken given the hours in a day and our limited mental resources. Besides, trying to make every decision scientific can lead us astray, for example, through analysis paralysis.

Bridgewater Associates, an investment management firm, is running an interesting experiment. It implements a complex decision system through a software tool: the Dot Collector. Every employee is rated by his colleagues on his performance across different criteria, such as brightness and common sense. Decisions are then weighted "based on people's merits". Whether this system reduces bias or amplifies it (for example due to the halo effect or the green beard effect) is hard to tell.

Another issue is that the way data is interpreted can vary widely. Given an objective observation, scientist often disagree on which framework best explains it, and the conclusions and predictions that can be inferred. What good practitioners don't do, however, is ignoring the facts, and in the scientific community, consensus if often reached where multiple conflicting views existed, even if considerable time is needed. This is because science is about contrasting hypothesis with data, so new data can lead to new conclusions.

The alternative to science is superstition. At the end of the day, decisions need to be properly informed, otherwise, they are based on anecdote.


References

Hating Pinker | Chronicle.com
Pinker | Harvard Gazette
Scientific Management | The Economist
Taylorism | HBR
How to Build a Company Where the Best Ideas Win | TED

‹ Previous: On Data Projects Next: Dark Mode ›