Five rules, teach you to see the truth behind the data

In the era of chaotic information, people seem to have lost a cognition and discerning ability, and passively accept information about various data information "attacks". Although statistical data is easy to lie, there is no statistical data, and lies are easier.

So how to dig valuable data from the world full of false information, adverse research and poor motivation?

We have taken the five rules from the book "Subscribe Truth" and taught everyone to use statistics correctly to make "communication with data" more credible.

Not chaotic, not trapped in love

The first step of learning can start with learning to control emotions.

When you see the data results, pay attention to your reaction. If you see the statistical results, whether you feel angry or happy, or if you can't believe it, you have to stop and reflect on it.

When we encounter statistics from a certain world problem, and think about whether to praise on social media, or fiercely refute, stop, first ask ourselves a question: "Why is my emotions so excited?"

We do this not only for ourselves, but also a social responsibility. We have seen how the social pressure has the impact of our ideas and ways of thinking. We must set a conclusion slowly, learn to control our emotions and put aside their positions, and only pay attention to the facts itself.

In this way, we can not only think more soberly, but also provide the correct thinking mode for others, that is, we do not stand by the members of a political faction, but we are thinking and reasoning with individual identity and reasoning with an unspoken attitude. of.

Emotions can control people's thinking. Therefore, when interpreting statistics, professional knowledge and technology are important, but if you do not put the reins on the mood of emotions, and let it be suspicious of our time, we will eventually lose our forefoot.

Inserting angle and bird's eye view of the worm

Try to learn the problem from two angles: the perspective of the worm and the bird's eye view.

It is easy for people to understand what they see from their own perspectives as the whole picture of things. Psychologists call this "naive realism", that is, they think that they are seeing the truth without any deviation. This innocence of this one -leaf obstacle will seriously mislead us.

Natural realism can make people understand many things. For example, the Morrian public opinion survey agency investigated nearly 30,000 people in 38 countries on a series of social issues. It was found that these people -can represent most of us -the understanding of things is seriously inconsistent with the reliable statistics. The following is an example.

Our understanding of the murderous crime rate is wrong. We think that the murder case has been rising since 2000, but this proportion has been declining in most countries that have been investigated.

We think that the number of terrorism in the past 15 years is higher than 15 years ago, and the number of people has declined.

We think 34%of people have diabetes, and the real number is 8%.

We thought 75%of people used Facebook. During the 2017 survey, this number was 46%.

The news reported incidents are also data to some extent. Although they are not representative data, they really affect our views on the world. In Kanmann's words, they are "fast numbers" -the number of people can get the number of conclusions at once.

The number provided by the bird's eye view is boring and rigorous, but it is comprehensive and profound. The data seen in the perspective of the worm is lively, but it is more one -sided. It is not easy to balance the two perspectives.

We need to remind ourselves often that we may ignore other things while understanding these things. Statistical, like other disciplines, must complement each other with rigorous logic and personal experience. It is the ideal way to combine the two to organically combine the two.

See the definition of data

When we want to understand any statistical results, we must first think about what the result is actually. New crown pneumonia's outbreak caused similar problems.

On April 9, 2020, media reports said that in the past 24 hours, 887 people in the UK died in the new crown, but I happened to know that this number was wrong. Scottish Statisticalist Hillary Berd made a thorough investigation and he told me that the real figures were likely to reach about 1,500 people.

Why is the number difference?

Part of the reason is that some people die at home, and the official only counted those who died in the hospital, but mainly because those hospitals that expanded the new crown were too late to update the number of deaths, and they often lag a few days.

The death data announced on Thursday may be the number of deaths on Sunday or Monday. Due to the surge in the number of deaths in these days, the data of our data three days ago was likely to underestimate the severe nature of the current situation.

Many problems are because people go wrong from the beginning.

They are obsessed with statistical technical issues, such as asking about sampling errors and errors, the number of debate numbers rising or decreased, believing, doubt, analysis, and analysis of various numbers. Question: What are the statistical objects? What standards do you use?

We must first figure out what the statistical object of the data is, and the second is mathematics calculation.

Learn to watch data under the macro situation

Pulling away from seeing the problem can give you a macro feeling. Every time you see a statistics, you can think about it, is this a big number?

We take former US President Trump to build a wall on the US -Mexico border as an example to talk about what the macro feeling is. Building a wall will cost $ 25 billion. Is this number big? This sounds a bit big, but to truly understand this number, you need something as a reference. For example, the US defense budget is nearly 700 billion US dollars per year, that is, $ 2 billion per day. Therefore, the cost of building a wall is equivalent to the two weeks of military expenditures of the US military.

Or, there are about 325 million people in the United States, except for 325 million US dollars, the cost of this wall is about $ 75 per person. This number is large or small, you can judge yourself, but I guess these comparisons, your judgment will be more reasonable. Pulling away from seeing the problem can give you a macro feeling.

If you can remember some numbers with a ruler, they can bring you a lot of convenience. You can use numbers to compare (a 10,000 -word report seems to be long, but a normal novel is 10 times), or the average is also considered (US defense budget is more than 2,000 US dollars per person per year).

These ruler numbers, whether you have remembered in your brain or you checked, can be used for calculations. This is a simple matter, but it is very inspiring.

Understand whether the statistical sample covers comprehensive coverage

In fact, we must often ask: Who is the data less? What did you miss?

In terms of data, scale does not mean everything. Two problems should be dealt with: sample error and sampling deviation.

The sample error reflects such a situation: sometimes it is purely accidental that people who are randomly sample do not reflect the real opinions of the people. "Error range" means this risk, but the larger the sample, the smaller the error will be. A random interview with 1,000 people is a big sample for public opinion tests for any purpose.

There is also a larger trap to avoid public opinion surveys, that is, sampling deviation. The sampling error means that the sample that is randomly selected does not reflect the real situation, and the sampling deviation is that the sampling does not cover the full sample type.

How many bold racial and gender discriminations in the society, you just know it around. But in general, what we count, or who missed it, is caused by the carelessness, not serious prejudice and some unintentional life when we choose.

Unless we collect data by yourself, we will do limited statistics omissions. But at least, when others provide data for us, we can and should remember to ask who or which content may be missing.

Big data looks very comprehensive and may be widely used, but "a lot of one" is an illusion that is easy to make people: everything is in my own control. In fact, we must often ask: "Who is there in the data? What did you miss?"

This is just one of the reasons for us to treat big data cautious. Big data represents a huge change in data collection and statistical methods. The impact of this change needs to be examined.

- END -