Principles of Statistics

Author: Translator:
Publisher:
Publication Date: 1999-09-01
Features: Segment: 2. If the purpose is to understand the general quantitative performance of the overall, you can select medium-level typical units for investigation. 3. If the purpose is to study successful experiences or lessons from failures, you can select advanced typical units and backward typical units, or select typical units from upper, middle, and lower categories for comparison, and then determine several typical units. (IV) Key survey refers to selecting a part of key units from all survey units for investigation. Key units usually should meet the following conditions: 1. The proportion of these units to the total number of units is very small. 2. In the survey indicators, the total value of the indicators of these units should account for a major share. For example, at the national level, investigating only a dozen large iron and steel enterprises such as Baosteel, Shougang, Wugang, and Ansteel can help understand the basic situation of iron and steel production in China. Key surveys can be used for both regular surveys and one-time surveys. When only a basic understanding of the survey subjects is required and there are indeed key units in the overall, conducting a key survey is appropriate. However, since there is a significant difference between key units and general units, the survey data from key units is not suitable for generalizing to the overall population.
Section III: Organization of Statistical Data
I. Audit and Adjustment of Statistical Data
(I) Audit of Primary Data
The primary task in organizing statistical data is to review the accuracy and completeness of the collected data.
1. Accuracy Audit: Mainly reviews the errors in the statistical survey process. Statistical errors are usually divided into two categories: one is registration errors, which are errors that occur during recording, calculation, and summarization, as well as errors caused by falsification such as false reporting or concealment; the other is random errors, which are representative errors produced by sampling inference. Representative errors can be controlled, so the focus of our review is on the first type of error, which is registration errors caused by the quality of statistical personnel. There are two methods for this check: logical check and technical check.
(1) Logical check is a method to examine whether the content of the data is reasonable and whether there are contradictions between relevant items. This requires the auditor to have a high sense of responsibility, a pragmatic work attitude, and a good understanding of the situation and familiarity with the business.
(2) Technical check mainly includes: a. Whether there are omissions or repetitions in the reporting; b. Whether all survey items are filled in, whether the content is compliant, and whether there are any errors in rows or columns; c. Whether the units of measurement are correct; d. Whether the calculations such as totals and products are correct, etc.
2. Completeness Audit: Mainly checks whether there are any omissions in the units to be surveyed, whether the content to be surveyed is complete, and whether there are any data that have not been submitted on time.
(II) Identification and Audit of Secondary Data
The identification and audit of secondary data mainly focus on identifying the authenticity of the data, reviewing the reliability of the data, and can be approached from four aspects:
1. Clarify which unit collected and organized the data for what purpose, thereby determining whether these secondary data are appropriate.
2. Review the methods of investigation and organization to identify the authenticity and quality of the secondary data. For example, generally speaking, data collected using the method of dispatching personnel is more accurate than data collected using the method of correspondence.
3. Evaluate the nature of the data itself, such as direct collected hand data is often more reliable than secondary data, and officially published statistical data (such as statistical bulletins) is more reliable than data from news reports.
4. Check from the relationships between relevant indicators and the dynamic development characteristics. In addition, the scope, time limits, and calculation methods of the indicators must be reviewed.