{"id":1693,"date":"2024-01-03T14:46:27","date_gmt":"2024-01-03T14:46:27","guid":{"rendered":"https:\/\/staticalmo.com\/?p=1693"},"modified":"2024-10-24T08:26:24","modified_gmt":"2024-10-24T08:26:24","slug":"non-basta-la-statistica-per-prendere-decisioni-migliori","status":"publish","type":"post","link":"https:\/\/staticalmo.com\/en\/non-basta-la-statistica-per-prendere-decisioni-migliori\/","title":{"rendered":"Statistics is not enough to make better decisions"},"content":{"rendered":"

In itself it is optimistic to say, as I do in my videos, that various domains of statistics, or data science, help in making more rigorous decisions, which lead to increasing turnover and\/or decreasing costs. This is because we do not live in an era where we have automated decision makers. We don't even have this situation in industrial chemical plants, although there we have pumps and valves that \"decide\" to turn on\/off when they receive certain information.<\/span><\/p>\n

Entrepreneurs and managers, as human beings, have a personality and consequently it is not enough to communicate the results of certain statistical analyzes in order to <\/span>convince them <\/b>to make certain decisions, or make them change certain repeated decisions, i.e. a strategy.<\/span><\/p>\n

Unfortunately I am delving into psychology and will end up in psychometrics because statistics, from being an auxiliary science, inevitably ends up in other areas.<\/span><\/p>\n

What cognitive limits block entrepreneurs and managers? Why don't they act based on statistical evidence?<\/b><\/h2>\n

I will deal with a topic that has been known for about ten years now: the contents of the book \"Thinking, Fast and Slow\", by a Nobel Prize winner in economics who essentially demolished a strong hypothesis of the old economic models: the theory of the efficient market, and therefore the existence of rational participants. In this case entrepreneurs and managers.<\/span><\/p>\n

When I read it, in my fourth year of high school, I was shocked by a sentence from the author, who says that knowing the existence of cognitive biases, i.e. prejudices, distortions or cognitive limits,<\/span> DOES NOT <\/b>defends against committing them. So useless reading? Largely not. In other words, it seems like knowing about the existence of mosquitoes and being able to do almost nothing to defend yourself. An almost fatalism that smacks of an uncomfortable truth.<\/span><\/p>\n

Some recurring cognitive limitations of entrepreneurs and managers<\/b><\/h2>\n

confirmation bias<\/b><\/h3>\n

analyzes only aimed at confirming the hypothesis of the entrepreneur or manager. In other words, paying the statistician to not stay out of your comfort zone.<\/span> A sort of mirror of desires.<\/span><\/a> The statistician is reduced to a mere executor, a technician, canceling them out as a scientist. In fact, some call data scientists \u201cbusiness scientists\u201d. Personally I find it a little <\/span>cringe.<\/span><\/i><\/p>\n

Or the empirical evidence of certain statistical results is denied.<\/span><\/p>\n

anchoring bias<\/b><\/h3>\n

the owner sets a turnover target x for the current year. As anchor we have this. The statistician points out that this is unrealistic, for example looking at the average of aggregate sales per month, taking into account<\/span> uncertainty<\/span><\/a> (variance) or in some cases objective limits such as business and\/or market bottlenecks (demand).\u00a0<\/span><\/p>\n

Technological anchoring, linked to other biases that follow: companies become fossilized \/ anchored with Excel despite having few specific reasons to prefer it to Google spreadsheets.<\/span><\/p>\n

availability bias<\/b><\/h3>\n

of owners and managers rely on intuition because a statistical analysis is less available in terms of accessibility of vocabulary, analysis time, costs linked to the analysis, etc.<\/span> Why complicate the bread?<\/span><\/i>An old Italian singer would say.<\/span><\/p>\n

Intuition can also be replaced with familiarity with a certain decision-making process, analysis software, etc.<\/span><\/p>\n

sunk cost<\/b><\/h3>\n

the owner spends a 5-figure amount for a certain sales strategy in a more competitive market but with potentially greater results. The statistician, or data scientist, says that there are safer markets. The owner,<\/span> to protect<\/b> their own integrity, they do not listen to the statistical point of view despite <\/span>they have no results<\/b> in that market. It would cost them too much in terms of self-esteem to admit the mistake.<\/span><\/p>\n

Another technological example: the company does not want to change the management system despite causing technical debt. For example, recovering data from that management system, instead of having a CRM (a kind of database for potential customers, customers and more), costs much more because it requires a person, as data cannot be extracted programmatically and automatically. If you want, you can automate the person's work, but it still costs more than changing the approach (software).<\/span><\/p>\n

status quo or inertia<\/b><\/h3>\n

one of the reasons why we still need to talk about digital transformation in SMEs. Fortunately, large consultancy firms do this type of work.\u00a0<\/span><\/p>\n

This limit acts as a substrate for the previous cases.<\/span><\/p>\n

Types of entrepreneurs and managers most prone to bias<\/b><\/h2>\n

And here we get into psychometrics, which I also mentioned<\/span> in an (Italian) episode of the podcast<\/span><\/a> and in a<\/span> reel in Italian<\/span><\/a>. If we take the most accepted quantification of personality in the literature, \"The 5 personality traits\", also called OCEAN, we cannot easily find what we want for the categories of people mentioned.<\/span><\/p>\n

However, on a general level:<\/span><\/p>\n