Como Mentir Com Estatistica -
The most fundamental trick in the statistical liar’s toolkit is the biased sample. Huff famously illustrates this with a survey showing that Yale graduates earn a high average salary. The unspoken catch? The survey only contacted successful alumni whose addresses were on file, ignoring those who had moved away or fallen into obscurity. In a modern Brazilian context, Como Mentir com Estatística would warn against a poll claiming “90% of São Paulo residents support a new policy” when the poll was conducted only in a wealthy, gated community. The lie is not in the arithmetic (90% is mathematically correct), but in the hidden assumption that this tiny, unrepresentative group speaks for the whole.
Perhaps the most pervasive form of statistical lying, however, is the confusion between correlation and causation. Huff provides a classic example: there is a strong correlation between the number of firemen sent to a fire and the damage caused. A lazy or dishonest analyst might conclude that “more firemen cause more damage.” The truth, of course, is the reverse: bigger fires require more firemen and cause more damage. In the age of big data, this fallacy is everywhere. A study might show that children who read more books have higher test scores. Does reading cause intelligence, or do intelligent parents provide both books and good genes? Como Mentir com Estatística teaches the reader to always ask: “What else could explain this?” Como Mentir Com Estatistica
Finally, Huff addresses the deceitful graph. By truncating the y-axis (starting a bar chart at 50 instead of zero), a minor 10% increase can be made to look like a spectacular, vertical explosion of growth. Similarly, a pictogram—a row of dollar bills or bags of coffee—can be distorted if the illustrator scales both the height and width of the image, making a doubling of data look like a quadrupling of size. The most fundamental trick in the statistical liar’s
Beyond sampling, the book exposes the seductive power of the “average.” Huff famously distinguishes between the mean, the median, and the mode. A developer wanting to boast about high salaries in a new office might use the mean if a few executives earn millions, making the average look impressive. A union leader wanting to show that workers are underpaid might use the median , which is unaffected by the executives’ fortunes. Without specifying which average is being used, a statistician can paint wildly different pictures from the same set of numbers. As Huff wryly notes, “The average you get depends entirely on what you choose to average.” The survey only contacted successful alumni whose addresses
In 1954, Darrell Huff published a slim, illustrated volume that became an unlikely phenomenon. Titled How to Lie with Statistics , it was not a manual for criminals, but a survival guide for citizens. Decades later, its Portuguese translation, Como Mentir com Estatística , carries the same provocative charge. The book’s central thesis is as unsettling as it is simple: numbers, often revered as the language of objective truth, are remarkably easy to manipulate. Huff’s work is not an indictment of statistics as a field, but a warning against the misuse of statistical reasoning by advertisers, politicians, and the media. Ultimately, the book teaches that the greatest lie is not a false number, but a misleading context.
In conclusion, How to Lie with Statistics is less about lying and more about seeing. Huff’s genius was to realize that the most dangerous lies are not bold fabrications, but subtle distortions of truth—a biased sample, a convenient average, a false cause. In an era of algorithmic feeds, political spin, and corporate “data-driven” claims, the lessons of Como Mentir com Estatística are more urgent than ever. The book does not ask us to distrust all numbers, but to become critical readers of them. After all, as Huff famously quipped, many people use statistics the way a drunk uses a lamppost: for support, not for illumination.
