What Is a Survey?

Abstract

This brief work describes in nontechnical language the operations involved in conducting a sample survey and aspects of assessing the reliability of survey results. It is intended to provide a better appreciation of survey data for nonspecialists who use survey results. Surveys obtain information from samples of limited size in which each member of the population has a known probability of selection in excess of zero so that results can be generalized to the larger population with confidence. Information is obtained through standardized questions to obtain a profile of the population. Respondents are never identified and results are presented in summaries such as tables. The size of the sample depends on the degree of reliability required. Surveys are conducted for a wide variety of purposes when insufficient data exsit to fill a specific information need. The majority of survey results are never exposed to the public. Surveys can be classified according to size target group data gathering methodology or content. They may utilize open or closed questions and may involve more than 1 round of interviewing. The 1st step in conducting a survey is to define objectives as specifically as possible. Planning of survey methodology involves formulating rules for defining and locating persons eligible for inclusion selecting the data gathering methodology designing and testing the questionnaire developing procedures to minimize or control response errors designing and selecting the sample hiring and training interviewers planning management of nonresponses and completing tabulation and analysis. The questionnaire must contain well defined concepts and clear phrasing and should not be overly lengthy. Quality control of fieldwork is achieved in several ways most frequently through observation or a reexamination of a subsample of interviews by supervisors or high level personnel using established procedures to identify obvious errors or omissions in the data. 4 pitfalls that can invalidate survey results and disorient users are failing to follow an appropriate sampling procedure not pretesting field procedures lack of insistence on follow-up of nonresponse cases and inadequate quality control. 2 types of errors may be distinguished sampling errors and nonsampling errors. Calculation of sampling errors is possible if methods based on probability are used in selecting the sample. Information on sampling errors should be made available to all users of the data. Nonsampling errors include effects of refusals and persons not at home of respondent giving incorrect information errors of coding or processing of data and sampling errors committed in the office. A flow diagram of the survey may be helpful in estimating costs.