Field Work in Sample Surveys of Household Income and Expenditure

Abstract

The last few years in this country have seen a considerable increase in the use of sample surveys for the collection of data on personal incomes, saving, and expenditure. There have recently been three major inquiries: the Ministry of Labour household budget inquiry; the Oxford Institute of Statistics Savings Surveys; and the project of the Department of Applied Economics, Cambridge, the Social Accounts of Cambridgeshire. (For a general description of the methods and objects of these three inquiries see references. 1,2) But while there has been considerable discussion about the design for such surveys, relatively little has been said about two other important aspects of the direct collection of income and expenditure information, namely the reliability of the material collected and the cost of collection. The existence of non-sampling error, while recognised, has not received very much attention, probably because of the difficulty of measuring it. In the course of the Cambridgeshire survey, however, it has been possible to form some views about the ways in which certain kinds of non-sampling error could be reduced, as well as to build up some data on costs. I shall be concerned with three questions. Firstly, what can be done to increase response in surveys of human populations (particularly financial surveys), since incomplete response may always be a source of error. Secondly, what can be done to reduce recording errors arising from the mistakes of both the interviewer and the respondent. Thirdly, what, if any, will be the effect of the suggested steps upon the cost of field work for such financial surveys. The material drawn upon is, in the main, that obtained in a large-scale budget inquiry conducted by the Department of Applied Economics, Cambridge, as part of a research project, the Social Accounts of Cambridgeshire, designed to develop methods for the direct collection of national income and social accounting information. The field work was carried out during the period April 1953 to March 1954.