Do you believe in statistics? They sometimes put "smoke bombs".
Sometimes, there will be some extremely abnormal phenomena in statistical data. Let's look at the following interesting example. Suppose that scientists have developed a new drug to treat diseases. However, the experimental results show that the effect of this new drug is not better than the original drug, as shown in the table:
Simply put, the new drug is only effective for 40% people, while the original drug is effective for 50% people. What's the problem? Is it because this new drug has side effects on some people? Or is there another reason? Therefore, the researchers took gender factors into account and counted men and women separately, as shown in the following table:
We might as well do a practical calculation: for men, the new drug is effective for as many as 70%, while the original drug is only effective for 60%; But for women, the new drug is 30% effective, while the original drug is only 20% effective. Contradictory results have emerged: the new drug is not only more effective for men, but also more effective for women, but it is not as good as the original drug for the whole population! 195 1 year, the British statistician Simpson first discovered this strange phenomenon, so this phenomenon is called Simpson paradox.
Simpson paradox, also known as Simpson effect, is not a paradox.
Its mathematical principle is that at that time, it was not always. If so, there will be Simpson effect. This phenomenon is easy to occur when the data volume and frequency of grouped samples are quite different. For example, in the above example, there are many more women than men participating in the trial of new drugs, while the original drugs are the opposite, and the effect of drugs on men is much better than that of women.
Pie charts are a common way to display statistical results.
This phenomenon often appears in the statistical data of medical and health fields. For example, the "mixed effect" in epidemiology is actually the Simpson effect. Similar things have happened in other areas of human society. The US Department of Labor has released a report showing that during the global financial crisis that broke out in 2009 and affected many years later, the overall unemployment rate in the United States was lower than that during the economic recession in the 1980s. However, after counting the unemployment rate data of college graduates, high school graduates and high school dropouts, we will find that the unemployment rate of these groups during the global financial crisis was higher than that during the economic recession in the 1980s. The reason is that after 2009, the proportion of college graduates in the total population in the United States is much higher than that in the 1980s, and the unemployment rate of college graduates is much lower than that of high school students or high school dropouts.
From 65438 to 0973, the University of California, Berkeley, was sued for gender discrimination, because statistics show that the admission rate of boys is much higher than that of girls. However, the school carefully checked the admission rates of male and female students in various departments of the school and found that this was not the case. In fact, the admission rate of female students in almost all departments is higher. In the end, Berkeley won the lawsuit.