When faced with a set of data, we can easily find the differences, such as the difference in the daily active users of plan A and plan B, the difference in retention rate, the difference in the number of paying users, and so on.

But these apparent differences may only be due to sampling error this time, not that there is actually a difference between the two. For this reason, there are statistical methods for testing different situations, called hypothesis testing.

Taking Douyin as an example, this article __mobile number list__ briefly introduces three commonly used hypothesis testing methods: analysis of variance, independent sample t-test, significance test of correlation coefficient and their implementation in SPSS. The data is selected from the top 300 Douyin fans displayed by the Douchacha platform as of August 21, excluding celebrities, government affairs and other celebrities (there are extreme values) for analysis.

Hypothesis testing is actually a method of proof by contradiction. The result we want to get is that there is a difference between the two schemes A and B. First, we need to assume that there is no difference between the two, and prove that there is a difference by overturning this assumption.

H0 means that the difference between the two is only sampling error, and H1 means that there is indeed a difference between the two. By rejecting H0 to prove that H1 is correct, the hypothesis test is completed. According to the statistical principle of small probability, when the probability (p) of an event occurring in a test is less than 5%, the event is called a small probability event, and it is considered that it will not occur in this test. At this point, H0 can be rejected and H1 accepted.