(A) the scientific design of the survey program
The sampling investigation process is a systematic project, which requires careful design and scientific investigation scheme in advance to reduce the non-sampling error in the investigation process. The scientific design of investigation scheme should focus on two aspects: sampling framework and questionnaire design. In the aspect of sampling frame, designers are required to have a certain understanding of the distribution structure of the survey population before compiling the sampling frame. Before sampling, they should check the sampling frame, identify and deal with possible problems, and take certain measures to remedy them. In the aspect of questionnaire design, designers are required to ensure that the questions are in line with the survey theme, and to see whether the respondents can fully understand the survey intention and are willing to cooperate to make correct answers. Specifically, we should pay attention to the following points: (1) The questions of questionnaire design should focus on the purpose of the survey, and it is forbidden to ask some irrelevant questions. (2) The questions raised should be clear in meaning, and the words used should be appropriate, concise, logical and polite. (3) The words used in the questionnaire should be objective and should not be biased or induced. (4) Be sure to consider the actual ability and conditions of the respondents, try to avoid questions that the respondents are difficult to understand or can't answer, and at the same time use less technical terms. (5) For some sensitive issues, we should try to downplay their sensitivity. (6) The content of questions should be single, and several questions should not be put together. (7) The content of the questionnaire should not be too much, so as not to take up too much time of the respondents and affect their work and rest.
(2) Strengthen training and improve the overall quality of the sampling survey team.
Sampling survey requires high quality of sampling survey team, which requires statisticians not only to have high professional quality, high sense of responsibility, good professional ethics and high professionalism, but also to have strong practical work ability. The quality and ability that a questionnaire designer should possess is to master a lot of statistical inference knowledge, understand the relevant knowledge of the survey content, have certain psychological knowledge and know something about computer data processing. The qualities and abilities that investigators should possess are as follows: first, they should have a realistic working attitude and sense of responsibility, and be good at contacting the masses; The second is to be familiar with the purpose and content of the investigation; Third, be able to master and skillfully use various scientific investigation methods, means and techniques, and reasonably select and use investigation methods according to the characteristics of collected data. The quality and ability of a summary analyst should be able to accurately process data and information, skillfully apply statistical analysis methods and skillfully write investigation reports. To improve the overall quality of the statistical team in an all-round way, we should adopt various ways and methods to cultivate a high-quality statistical team by holding short-term training courses, regular and irregular theoretical seminars, exchanging experiences, and going to colleges and universities for further study.
(3) Pay attention to the selection and management of investigators.
In the sample survey, investigators directly deal with all kinds of people in society. If the investigators do not pay attention to interview skills and the art of wording, and are not good at dealing with different types of people, it is impossible to get the effective cooperation of the respondents. If the investigators lack a realistic working attitude and sense of responsibility, and have no hard-working spirit, it is impossible to obtain accurate statistical data. Therefore, the selection and management of investigators is an important task and the key to reduce investigation errors. First of all, we must choose the right investigators. Investigators should be selected from the aspects of theoretical professional knowledge, moral quality and adaptability, and those with good ideological and professional quality and strong working ability should be selected. We should strengthen the management of investigators and establish an effective restraint mechanism to prevent investigators from practicing fraud.
(D) Use various ways to reduce the mistakes of respondents.
Sampling survey requires respondents to provide accurate and complete information. However, in the actual investigation, due to various reasons, respondents often fail to answer or deliberately make mistakes, which leads to the occurrence of respondents' mistakes. In view of this situation, on the one hand, we should take appropriate measures to educate and influence the respondents, so that they have a correct understanding of the investigation and are willing to cooperate effectively with the investigators; On the other hand, effective measures should be taken to reduce the occurrence of errors under investigation. For the investigation of some questions involving personal privacy and business secrets, randomized answer technology can be used to keep the information provided by the respondents confidential and improve the answer rate. For the errors caused by the biased data provided by the respondents intentionally or unintentionally, the survey data can be corrected by checking relevant records, logical checking or re-investigation. Those who deliberately distort the facts, make false reports or conceal them, and deliberately practice fraud for some benefit, thus causing serious consequences, shall be investigated for their legal responsibilities.
(five) do a good job in the summary of survey data to prevent summary errors.
For the errors in the process of investigation data collection and data transmission, two measures should be taken from the causes.
Surface control. First of all, we should strictly review the survey data and make a scientific arrangement plan. The second is to adopt advanced computer finishing technology.