Giving people analysis software and expecting them to excel in data science is like giving them a stethoscope and expecting them to excel in medicine. Bob Hayes
Statistics are numbers
One thing to remember: statistics are simply numbers. And because they are numbers, they look like facts. Even more so, natural, untouchable facts.
How we choose to interpret the numbers depends on us and a key mental tool: critical thinking. It's time to think about how we think.
We will not stray too far into the field of statistics, mathematics, psychology or philosophy. But the more mental models we integrate, the better we can interpret the information before us.
Statistics are not facts. They're interpretations. They're not complete natural phenomena. It is people who choose what data to collect, how to calculate it, which results they will share and how they will describe them.
An example of relevance and narrative
When consulting a graph, we have to ask ourselves if it is relevant to the question we want to answer.
The cumulative sales graph is a perfect solution to improve the narrative, when sales are not promising. Unlike a graph, which represents sales per quarter, this one will always be growing, as it accumulates sales. If we want to know if sales have improved, increased or decreased, this graph will not give us relevant information.
10 Logical fallacies
What is a logical fallacy?
A logical fallacy is a flaw in reasoning. Logical fallacies are like tricks or illusions in our thinking.
Aristotle is the first known philosopher to establish a list of logical fallacies. These fallacies are common mistakes made when discussing and thinking. Being aware of these errors is extremely useful in sharpening our analytical skills.
Developing thought awareness is known as metacognition and is a key component of critical thinking.
When it comes to analyzing data or evaluating a conclusion from the data or models presented, it is worth considering the reliability of what is presented.
Among the most common errors are data quality, problems in model development and interpretation of results.
1. The fallacy of incomplete evidence
It is the selection of a cluster of data to fit our argument, or finding a pattern that fits an assumption.
It is used to prove an argument, or to use confirmation bias and reasoned reasoning instead of deductive reasoning in its analysis. It may involve searching for patterns, but ignoring contradictions.
2. The player's fallacy or Monte Carlo fallacy
Believing that past events affect futures in terms of random activities, as in many games of chance, such as roulette spins.
So, while there may be a small chance of heads coming off 20 times in a row if you flip a coin, the chances of heads coming off on each individual toss are still 50/50, and are not influenced by what happened before.
The hot hand fallacy is the player's reverse fallacy. We think that an unlikely chain of "luck" will remain in place.
3. Black or White or "The False Dilemma"
The false dichotomy or assumption/establishment of a binary state when there is none.
Present two alternative states as the only possibilities, when in fact there are more possibilities.
Also known as the false dilemma, this insidious tactic has the appearance of forming a logical argument, but under more detailed scrutiny it becomes clear that there are more possibilities than the option presented. Binary thinking, in black and white, does not allow for the different variables, conditions and contexts in which there would be more than the two possibilities presented. It frames the argument deceptively and obscures rational and honest debate.
4. The fallacy of authority
If an authority thinks something, it must be true.
It is important to note that this fallacy should not be used to dismiss expert claims or scientific consensus. Appeals to authority are not valid arguments, but neither is it reasonable to ignore the claims of experts who have demonstrated in-depth knowledge unless one has a similar level of understanding and access to empirical evidence.
However, it is entirely possible that the opinion of a person or institution of authority is wrong. Therefore, the authority held by such a person or institution has no intrinsic relationship to whether his or her claims are true or not.
5. Get in the Car
Appealing to popularity or the fact that many people do something as an attempt validation.
The flaw in this argument is that the popularity of an idea has absolutely nothing to do with its validity.
If it did, then the Earth would have stayed flat for most of history to accommodate this popular belief.
6. Personal Incredulity
Having difficulty understanding or not knowing how something works and concluding that it is probably not true
Complex issues such as biological evolution through natural selection require some understanding before an informed judgment can be made on the subject at hand; this fallacy is generally used in place of that understanding.
7. The straw man
The act of twisting someone's argument to make it easier to attack.
By exaggerating, misrepresenting, or simply completely making up someone's argument, it is much easier to present your own position as reasonable, but this kind of dishonesty serves to undermine honest rational debate.
8. Ad hominem (from the Latin 'against man')
Attacking your opponent's character or personal traits in an attempt to undermine his argument.
Ad hominem attacks can take the form of openly attacking someone or more subtly questioning their character or personal attributes as a way of discrediting their argument. The result of an ad hominem attack can be to undermine someone's case without having to commit to it.
9. Correlation and causality
Presuming that a real or perceived relationship between things means that one is the cause of the other.
Correlation does not imply causality. Similarities between two statistics or trends do not imply that one caused the other.
Many people confuse correlation (things that happen together or in sequence) with causality (that one thing actually causes the other to happen). Sometimes correlation is a coincidence, or it may be attributable to a common cause.
10. The anecdote
The use of personal experience or an isolated example rather than a strong argument or convincing evidence.
It is often much easier for people to believe someone's testimony than to understand complex data and variations on a continuum. Quantitative scientific measurements are almost always more accurate than personal perceptions and experiences, but our inclination is to believe what is tangible to us and the word of someone we trust about a more "abstract" statistical reality.
Next resources for Segmentation
Data Studio allows you to connect multiple tools to analyze the data of your e-commerce, allowing you to design a ...
Conoce los errores más comunes en el análisis de datos para que no te suceda a la hora de afrontar un proyecto de estas ...
Mediante la metodología MDM podemos asegurar la consistencia de nuestros datos maestros para evitar errores mayores en n...