Errors in statistics or any statistical investigation can be broadly classified in two types:

a)      Sampling errors and b) non sampling errors

a)      Sampling errors: Even after taking care in selecting sample, there may be chances that true value is not equal to the observed value because estimation is based upon a part of the population not on the whole. Hence sampling give rise to certain errors known as sampling errors.

Sampling errors are present only in sample survey. These are due to faulty selection of sample, biasness of the investigator and may be due to when investigator select convenient members of the population for sample.

Sampling errors are of 2 types:

i)                    Biased sampling errors and ii) unbiased sampling errors

Biased sampling errors arise due to biasness on the part of the investigator, biasness due to non response, biasness in the technique of the approximation,  biasness in the measuring instrument.

Unbiased sampling errors or compensatory errors are the errors in which the ultimate result would be neutralized. If the observations are large in number then these unbiased errors will not effect the final result. For example: The chance of making an overestimate is almost same as the chance of making an underestimate. Like the values 275, 325, 345 are rounded to nearest number 300. In this case 325 and 345 are overestimated and 275 is underestimated. Therefore these unbiased sampling errors are also known as compensatory errors.

This is about the sampling errors in statistics and its types.

Also read: Error in statistics and its reasons

b)     Non sampling errors:

These are the errors which are not in the human control. These errors can be traced at any stage of inquiry. They are also present in sampling as well as census methods. They increase with the increase in number of units to be examined. They arise due to following reasons:

i)                    Lack of trained and qualified investigators

ii)                  Due to wrong answers to the questions

iii)                Due to incomplete coverage

iv)                Biasness of the investigators

v)                  Vague questionnaire

vi)                Faulty list of population

vii)              Wrong method of asking questions

viii)            Wrong calculations while processing the data

ix)                Failure of respondent memory to recall past events

x)                  Error while printing the results

These are the reasons of non sampling errors.

The above are the various types of errors in statistics.