[1/2] How to Lie with Statistics (How to Lie with Statistics) (Category Management)

This is Darrell Huff's book. 70 years, but it has not lost its relevance. I decided to read it for the rest of the book.Understanding Statistics and Experimental Design. How to Not Lie With Statistics"that I mean." told earlier. And the book did not disappoint my expectations - it is written in plain language, contains no water and talks about various ways to abuse statistics to deceive the audience and manipulate their opinions. The book consists of 10 chapters: 1. The sample is initially biased This manipulation has to do with how we sample. If the sample does not correspond to the population (is not representative)The statistics we compute from this sample can show the numbers we want. But even if we want to make the most honest sample, it is quite difficult to do. For example, the author tells about this in the example of surveys. And here's an example from me, whitepaper.DevEx in ActionAbout developer productivity was described based on surveys of those developers who worked in companies that used the platform. https://getdx.com/ It provides tools for measuring developer productivity. In the end, the survey showed that these tools are useful:) 2. Competently selected average - here we are talking about choosing the average convenient for your particular use case, for example, it can be an average (mean)median (median) and fashion. In general, depending on the type of distribution of your value, these average options can be very different:) 3. Nuances that are modestly silent about It starts with a sample size that may not be mentioned. (With a small size, getting interesting results is much easier.)Unsuccessful results of experiments can also be avoided. (Why talk about uninteresting things?)Plus, you can play with the wording so that it is not clear how the indicator is calculated:) 4. A lot of noise from almost nothing. Here the author tells about the significance and confidence intervals:) And that when specifying specific numbers, it is difficult for us to compare them without knowing the confidence intervals. 5. Graphics are never better. - here we are talking about manipulation of graphs: counting not from the starting point vertically, different scales of the axes, choosing the right time interval to demonstrate the graph of the magnitude on the scale between the beginning and the end of the interval 6. Schematic picture Here the author tells how you can use infographics to deceive people. For example, with a two-fold increase in the monetary index, show twice as much money bag - but we perceive objects as three-dimensional and there is a feeling from this reception that the growth was in the middle of a large scale. 8 (2ˆ3) once 7. A pseudo-justified figure Here the author shows how a random number taken from statistics can be interpreted at your discretion. The main thing is to make a reference to authority and indicate where the number is taken from, and the interpretation is already screwed in:) By the way, this is a frequent manipulative technique. 8. And again, it's "after is after." Here the author says that correlation is not equal to causation. Maybe cause and effect are related to the cycle. (as discussed in the bookThe art of systems thinking"that I mean." told earlier) Or both variables depend on some other third variable, or maybe it's just a coincidence:) 9. How to produce statics (statistical manipulation) Here the author shows examples from the previous chapters and adds games with percentages, recounting the same elements in the calculations, adding together different types of objects and deduces the average. In general, it is very creative:)

Continuation in the next post.

#Math #Statistics #PopularScience #Science #Data