Statistical Inference is used to capture Statistical feature of large population by using a small sample of that population. In other words,we statistically infer information of a large system by examining only a proportionally small part of it.
As an example, consider a political election. It is possible to infer the final result of the election just by sampling a small set of data obtained through an exit poll.
Another example, is in the case we want to perform a television audience rating. By sampling a set of houses and which programs they watch, it is possible to infer the real audience rate with high accuracy. In TV and media, you may have often listened that it is announced a cabinet approval rating without calling for an official election. This can be done by sampling population and infer data from that sample.
For cabinet approval rating, can you imagine how many people should be included in the sample?
If we want to know the real cabinet approval rating, we should examine or survey all the population (electors) of a country. Dozens of million people in Japan for example. However, the number of surveyed people is usually within few thousands, which is very small number.
Why can we know the cabinet approval rating of huge population (dozens of millions) by just surveying only one or two thousand people? There is a real result for the cabinet approval rating of huge population and the inferred result. Do you think both results would be very close or very far each other?
The answer to this intriguing questions can be learnt from the Statistical Inference method in Statistics field.
＊We also explain this page in the following video.