This paper mainly studies how to sort out a large number of data obtained from educational science experiments, investigations and tests, describe the whole picture of the data, and thus express the characteristics and relations of a certain aspect of things or phenomena. Its specific contents include: how to group and classify data, how to describe the distribution of a group of data with statistical tables and charts, and how to simplify data by calculating the number of features such as concentration and difference, so as to further describe the characteristics and panorama of a group of data. The method of descriptive statistics is not only suitable for a set of data that only represents a local situation (sample), but also suitable for data that represents the whole situation (population) obtained from comprehensive surveys, experiments or tests. It can simplify and summarize a large number of scattered and chaotic digital data, and better show some characteristics of things. It is helpful to study and explain the essence of the problem.

Second, infer statistical data.

Inferential statistic is developed on the basis of descriptive statistics, and mainly studies how to infer the relevant characteristics of the crowd from the information provided by local data. In scientific research, it is usually impossible or unnecessary to observe and study all the problems studied one by one, but to extract some individuals from the whole as samples for analysis and research. This includes how to estimate and infer the overall situation from local statistical data, how to test hypotheses, how to analyze the factors that affect the change of things, how to compare the differences between two things and more than two things, and how to make decisions on uncertain things and phenomena. These are the problems that inferential statistic studied and solved. The specific contents of inferential statistics include: estimation methods of population parameters, various hypothesis testing methods, hypothesis testing methods of counting data, and various nonparametric statistical methods.

Third, experimental design.

Experimental design is a research scheme or scheme made by researchers to reveal the relationship between independent variables and dependent variables in the experiment before the experiment. Its contents include: how to extract experimental samples, and how large samples to extract to achieve the expected accuracy; What assumptions are put forward and which independent variables (factors) should be manipulated to cause the changes of dependent variables; Which irrelevant variables are controlled to reduce the experimental error; How to arrange the experimental steps; How to observe the experimental process; What statistical methods are used to process and analyze the experimental results, etc.