When we make a complete enumeration of all the data points in a population, we would expect that the data obtained will be completely free from errors. But, in practical situations collection of this is never the case because it is hard to avoid errors of observation. Also in the process of collecting data, some values might be wrongly tabulated which will lead to errors in our data.
Since the data is obtained via complete enumeration, it does not have any error due to sampling. The errors arising in this manner which are caused by factors other than sampling of data are called non sampling errors.
Non sampling errors can arise in the following ways:
- Use of inappropriate statistical unit.
- Inaccurate methos of survey, observation, measurements, etc.
- Errors in data processing operations such as coding, verfication, etc.
- Erros when printing the data in tabular form.
Examples of non sampling error:
- Suppose that an employer is transcribing the attendance record of the employees into the register. The employer can make a simple typing mistake which can cause non sampling error.
- Suppose that a reseacher is dealing with a large grouo of data. Since the data size is very large collecting all the data will be vary tediuos and the reseacher might make mistakes when tabulating the data.
Sampling error vs Non sampling error:
In some situations, the non sampling errors may be large and deserve greater attention than sampling errors. As a general rule sampling errors decrease as the size of the sample increases whereas the non sampling errors increase as the size of data increases. We should choose our sample size in such a way that both sampling and non sampling errors are minimised and the error does not adversely affect the final result.