Truths and Lies of Statistics


A completely true fact can also be very deceptive.
How is it possible? As usual, the problem is not in the data but in the people. For a computer, data is not more than that: it is not good or bad, it is not much or little. But people always value the data: we want to know what they mean, whether they are important or not, or whether we should be happy or worried.
At first reaction, we do not process the information: we perceive it . Perception is a quick way to categorize the avalanche of “inputs” we receive, but it is also the equivalent of a security hole: it opens the door to manipulation.
There are many types of manipulation. These are some of the most common:
▪ It can draw attention to debates that “omit” some points of view that they want to hide (as Noam Chomsky tells )
▪ You can play with the vocabulary to give something positive or negative connotations, or to eliminate positive or negative connotations (ex: “collateral damage”)
▪ You can highlight some figures or others: for example, we can compare the last economic data with the best or the worst in history, with the previous month, with the same month of the previous year; or we can compare the accumulated of the year with the accumulated of the previous year. Depending on our interests, we can choose the comparison that most benefits us
▪ Or you can play with the same pair of data and present it in different ways: the impact on our perception can be impressive (via Tim O’Reilly )
The conclusion: the data itself does not deceive, but we must be critical of the selection (what information is highlighted and which is omitted), the presentation and, consequently, our perception.