What is statistics? What is the function of statistics, which is very important for writing a statistical paper, so how to write this statistical paper well?

Statistics is a comprehensive science that infers the nature of the measured object and even predicts the future of the object by searching, sorting and analyzing the data. It uses a lot of professional knowledge in mathematics and other disciplines, and its application scope covers almost all fields of social science and natural science.

First, the definition and theoretical basis of statistical data accuracy

How to define statistical data is accurate, of course, it can't be absolutely accurate, but it should reflect the development trend of things within a reasonable range. In this sense, it has some connection with the law of large numbers. Of course, the accuracy of statistical data is relative in different environments and occasions, which may be correct in one case and inaccurate in another. That is, the accuracy of statistical data changes with the change of time, place and conditions. Therefore, in specific circumstances, it is particularly important to define the standards of statistical data. We are not sure about the accuracy of the statistics. We should pay attention to key statistics to ensure their accuracy. For example, statistical data related to the national economy and people's livelihood and social and economic implementation must be as accurate as possible. At the same time, we should not only pursue the accuracy of statistical data, but also consider the economy and timeliness of statistical survey data, so as to grasp the accuracy of statistical data.

Based on the principle of large numbers and central limit theorem in probability theory and mathematical statistics, in the process of statistical estimation, it is necessary to scientifically select samples, sample corresponding estimation methods for estimation, and conduct serious analysis and research to keep the error of statistical data within an appropriate range and ensure the accuracy of statistical data.

Second, the evaluation criteria and analysis methods of statistical data accuracy

The evaluation of the accuracy of statistical data is mainly to ensure the high accuracy of statistical data, and the changing trend of statistical data conforms to the logic and actual situation. So as to correct errors and correct data in time.

1. qualitative analysis method

Qualitative analysis is used to verify and evaluate the changing trend and law of statistical data. That is, to observe whether it conforms to the development law under objective circumstances, or to compare it with relevant statistical data. For example, it can be compared vertically, that is, compared with the statistical data of the same period last year or previous years. Or horizontal comparison, that is, comparative analysis with the statistical data of similar units from multiple angles and levels. Check whether it conforms to the changing trend and development law of the objective situation.

2. Quantitative analysis method

The evaluation index of statistical data is established through quantitative analysis, and the accuracy of statistical data is quantitatively analyzed. That is, the error of statistical data must be within a certain range. The established evaluation index must be clear in the index meaning of statistical data, simple in calculation and easy to obtain data. At the same time, the established statistical indicators should have practical significance. Statistical data of different departments, industries and regions can be compared horizontally, and statistical data of different periods can also be compared vertically.

Third, how to evaluate the accuracy of statistical data?

How to evaluate the accuracy of statistical data is to verify, analyze and evaluate the obtained statistical data. By evaluating the statistical data, we can get the data error, and then we can get more accurate statistical data by adjusting and analyzing the error.

1. Basic evaluation of statistical data

It mainly evaluates the standardization of statistical work, the reliability of statistical basis, the accuracy of statistical statements and the legitimacy of statistical data. Specifically, it includes 1, normative evaluation of statistical process. That is, to evaluate every process and link of collecting and sorting out statistical data. 2. Reliability evaluation of statistical raw data. That is, it mainly depends on whether the original data recorded are standardized and sound. 3. Accuracy evaluation of statistical statements. That is, to evaluate whether the data error rate, qualified rate, report type and report index are complete. 4. Validity evaluation of statistical data. That is, whether the generation process of statistical data conforms to the relevant laws and regulations of statistical law, and whether statistical data has been tampered with, falsely reported, concealed or forged.

2. Logical evaluation of statistical data

It is logical and feasible to evaluate the changing trend of statistical data. All kinds of statistical data can be analyzed from the perspective of rationality, reliability and logic. 1, evaluated by the balance method. Comprehensively observe the data of various increase and decrease relationships, and observe the difference of operation results. 2. Use the method of general equality to evaluate. Comprehensive observation of the same indicator, whether the data results appearing in different tables are equal. 3. Use the method of correlation balance to evaluate. That is, there must be some logical relationship between multiple indicators. If the data is abnormal, it needs to be carefully evaluated. 4. Evaluate by the correlation between actual use and data.

3. Use the method of mathematical statistics to evaluate.

Use the method of mathematical statistics to evaluate. That is, the data obtained from the investigation are brought into the formula for calculation and compared with the prescribed evaluation criteria. 1, sampling survey. On-site verification of statistical data of sampling survey, and calculation of reporting error rate or survey error rate to evaluate the accuracy of data. 2. Regression analysis method. Find a mathematical equation by statistical method, establish the relationship between independent variable and dependent variable, and use the given value of independent variable to calculate and estimate the value of dependent variable. At the same time, consider practical factors. 3. Trend prediction method. According to various statistical data, the changing trend is analyzed by drawing. 4. Analysis of variance. According to the random variables and their distribution laws in probability theory and mathematical statistics, the random events and laws are tested to find out the reasons. That is, the obtained statistical data are randomly grouped, and variance analysis is carried out on this basis. If there are obvious differences between the variances of each group, it can be determined whether there is statistical error in this group of statistical data. Details are as follows:

1. Divide samples with sample size of ab into one group.

2. Calculate the sum of squares m between groups and the sum of squares n within groups. The sum of squares between groups is the sum of squares of deviation between group average and total average; The sum of squares within a group is the sum of squares of the deviations between the observed values of each group and the group average.

3. Find the variance between groups and the variance within groups. Divide the sum of squares between groups and the sum of squares within groups by their degrees of freedom respectively. You can get the variance between groups and the variance within groups.

4. Find the value of f and judge whether there is an error.

If so, it means that there are differences and errors in the mean values of each group.

4. Use relevant methods to evaluate

Because of the internal relationship between statistics reflecting a phenomenon, if one statistic changes, another statistic will also change. We can test another indicator with higher reliability. That is, 1, the relative comparison method. We can compare and analyze the related data of indicators with internal relations and the accuracy of the obtained statistical data. 2. Group evaluation method. Observe whether the grouping of a general index according to the structure conforms to the objective law, and judge the accuracy of statistical data accordingly. 3, the average and relative number method. For various indicators that are not directly related, average index and relative indicators can be compared to find out the differences. 4. Logical reasoning. For the statistical data based on contact, analogy is made to verify the accuracy of the statistical data. For example, a time series average difference detection method is used. This method is used when individual factors have different effects on the accuracy of statistical data or the statistical data is inaccurate. That is, expand the scope of investigation, investigate the average situation of change in a certain period, and find out the average level of change and development because this long-term trend is not affected by individual factors.

Note that the number of chronological items in this model depends on the specific situation. If the scale of historical data is large, the number of slow chronological items can be larger. If the stability of historical data is good, the number of quick chronological items can be smaller. The value of significance level depends on historical data.