A census is a complete count of a given entity, whether population, housing, agricultural holdings, businesses, or other, in a given geographic area at a given time. Because of financial and time constraints, the information collected in a census is usually basic and limited. A survey is admin istered to a subset of a population to obtain detailed information about certain groups, such as school-age children or mothers, or to obtain other information, on a sample basis, about the entity in question (households, businesses, agricultural establishments, etc.).
A survey can provide valuable infor ma tion about the population being studied, but because of the limitations of samples, the results can be generalized only for relatively high-level geographic areas, such as the country as a whole, or (depending on the way in which the sample was selected) for specific regions or areas of the country (e.g., urban vs. rural). A census, however, attempts to cover the entire geographic area of the subject population, so it can provide reliable information at very low levels of geography. In addition to counts (e.g., number of people, number of housing units, number of farms, number of businesses, etc.), a census usually provides a profile of additional related characteristics such as fertility, housing quality, acreage, number of employees, and so on.
Census and survey data are used to plan for education, health facilities, administration, and other needs. In order to implement programs and activities, statistics are needed by government administrators and by private users, including businesses, industries, research organizations, and the general public. These statistics also may serve as measures of existing conditions for small areas, providing a basis for planning development programs, and perhaps establishing a basis for action.
In order to obtain an accurate census or survey, the data must be as free as possible from errors
and inconsistencies. Statistics derived from 'dirty data' (that is, data which still contain errors) may produce an inaccurate profile of the country or geographic area. Therefore, before any tabulation programs are run, the data should be checked for errors and changed so that important data items are valid and consistent. This is not to say that correction of data
after they are collected can compensate for poorly collected data. It is not practical (if not impossible) to produce a data file which is 100 percent error free. Every effort at accuracy should be made in all stages of the census or survey.