Addressing the Problem of Non-response and Response Bias

Most of the statistics needed for national planning are derived from large scale sample surveys with households as reporting units both in the developed and developing countries. Examples of these surveys are Income and Expenditure, Employment, Food consumption and Nutrition, Agriculture, Health, Education, Establishment, etc. The required information on these topics is obtained from the selected households or firms in an Establishment survey. The households to be canvassed are usually selected by probability sampling. As it happens in all surveys some selected units may not be contacted, or fail to respond to the interview or when they do provide distorted or inaccurate responses. These introduce some element of bias in the estimates.