Sometimes when communicating with technology or business, if the other side pops out this effect from time to time, the law, and happened to have not heard, at this time can only pretend to understand.

In fact, these concepts are not mysterious, today we sorted out a number of areas of common effects and laws, with easy-to-understand examples to help explain, so that we quickly understand grasp!

Simpson’s Paradox

Simpson’s paradox refers to the fact that when the data of the two groups are compared and divided into multiple dimensions, group A performs better than B in each dimension, but group A as A whole does not necessarily perform better than B.

Don’t understand? Let me give you an example

Recently, various universities in the UK have sent out offers. As a result, two colleges of a certain university, law school and college of Arts, are suspected of gender discrimination in admissions. Let’s take a look at the admissions situation and analyze it.

Law School Enrollment

College of Arts enrollment situation

According to the admission ratio data in the table, the admission ratio of female students is 33.6%>15.1% and 91.1%>80.1% higher than that of male students

But when you put together the data from the two schools

Law school, college of Arts data collection

It turns out that the percentage of women admitted is lower than that of men,

This is the classic Simpson paradox, in which two sets of data under certain conditions, when discussed separately, satisfy certain properties, but when combined, can lead to opposite results.

In fact, “the countryside surrounded the city, armed seizure of power” also has a similar idea.

2. Matthew Effect

The Matthew Effect comes from the Biblical parable: “To everyone who has, give twice as much, so that he may have more; And if he hath not, take away from him all that he hath, and leave him none. This is colloquially explained as “the stronger the strong, the weaker the weak”.

The Matthew effect is very common in business. For example, in the recommendation algorithm, users judged to be of better quality will get more resources, which will also form feedback. The more resources obtained, the more users will be judged to be of better quality, thus exacerbating this effect (similar to short videos with more likes, more exposure, more likes).

Benford’s Law

Benford’s law, which states that in a pile of real life data, the occurrence rate of the number with 1 as the first digit is about 30%, has not been rigorously proved.

It acts as a monitoring indicator, and when a set of data does not conform to Benford’s law, there is reason to suspect fraud. So this law is often used to detect whether the financial statements of listed companies are fraudulent and whether there is fraud in elections.

It should be noted that it can be used to check whether various data are falsified, but pay attention to the conditions of use: 1. At least 3000 pieces of data; 2. No manipulation.

4. Survivor bias

The survivor bias is a result of natural selection: the survivors have lost their voices.

People only see the results of certain kinds of screening, and they don’t realize that they ignore the people who are being screened out.

For example, you must have heard this sentence: “What’s the use of studying? My classmate in primary school, his grades were in a mess. He quit school before finishing junior middle school, and now his business can be big.

However, the actual situation is that the group of children who can read in a class also have a good business in the future, some rent a house to eat instant noodles, and some stay at home to eat the elderly, but the average living standard is higher than those who do not read. But some of the kids who don’t go to school are probably in debt and in hiding, unemployed and out of work, and you don’t see those people. You see the survivors and the business is big.

Pareto’s Law

The name may not be familiar to you, but you must have heard of the 80/20 rule. Pareto, a management scientist, found that 20% of the people in the society hold 80% of the wealth through studying a lot of facts.

For example, only 20% of active users are paying users, 20% of paying users contribute 80% of revenue, etc. Of course, 20% and 80% are only statistical data, and their essence is that “there is a general imbalance between cause and effect, effort and harvest”, that is, the certainty and predictability of the existence of the imbalance relationship.

The 80/20 rule tells us to focus on more essential things. Doing things without planning is likely to waste 80% of your energy producing 20% of what you want.

In the counting work, some students may often have this feeling when doing analysis. They ran n sheets of data, but only used four or five data when writing the report.

Therefore, the habit of analysis is to first think about the causes of problems, assign corresponding weights to each possible cause, and then verify each cause in the simplest and most convenient way to quickly eliminate the wrong direction, rather than giving detailed explanations on each cause.