Sampling error is related solely to the sample taken for conducting a census of the population (or data).
There can be errors where the sample is too small or it does not represent the characteristics of the whole population (or data type).
Such errors can be reduced by taking random samples where each unit of the selected data and the change is measured.
Non-sampling errors can be random or systematic.
Random means unpredictable which leads to wrong estimations and systems tend to accumulate over the entire data.
It can result in bias in the results.
It is caused by other factors like coverage error where the unit in the sample is incorrectly excluded or included or duplicated.
Non-response refers to the failure of getting an accurate response from the sample due to refusal, non – contact, absence or other factors.
It can be complete in case the respondent is unavailable or temporarily absent or he refuses to participate in the survey.
It can be partial when the people participating in the survey are unable to clearly understand the questions.
Processing error can happen when there is an error in the way the data is entered, edited or coded in the system and this can lead to inaccurate outputs.
Response error can happen in conditions when the respondents give erroneous response intentionally or accidentally.
It can happen that the questions or instructions are not clear, or the respondents are not able to recall the exact information or sometimes, the queries are just answered in a socially desirable manner.
The interviewers can influence the way the respondent answers as if they are too friendly or aloof, it can change the answer as per the response of the interviewer. Hence, they should remain neutral and allow the respondent to give real answers.
Measurement errors can happen in certain samples due to confusion created by the words used in the queries or it can happen because low –quality data is used on multi-item scales for market value survey or for global real estate investment or to get the answers to changing patterns of commodity production and consumption like how many kilometers you traveled in the last year, or how many glasses of water you drink in a week, or what is your daily calorie consumption in breakfast? Such queries can lead to proxy response or it can provide irregular or general answers - which can increase the change of getting an error.