Estimates of Sampling Error
The accuracy of survey estimates is generally taken to mean the closeness of that estimate to the exact or true value. The true value, which is mostly unknown, is the value that would be obtained, if data could be collected and processed without any errors for all of the units in the population. The error of a particular survey estimate is the difference between that estimate and the true value of the quantity being estimated.
This error can mainly be divided into two groups of errors: Sampling errors and nonsampling errors.
i. Sampling errors
Sampling errors arise from the fact that the observations are confined to a sample of the population rather than the whole population. The sampling errors can be minimised within the given budget if a suitable design is adopted and this error can always be calculated if a probability sample is used.
ii. Non-sampling errors
Apart from the sampling errors, non-sampling errors can arise at every stage during a survey operation. Unlike the sampling errors these non-sampling errors cannot be measured. We can only guess or feel about them. It is vitally important that these errors are controlled. If not properly controlled, the whole survey results may not be of any use. Hence everyone involved in the survey should clearly know what he or she is supposed to do and do that very carefully and correctly.