There has been considerable controversy and debate over recent years about the most appropriate method of measuring poverty. This debate has included, among other issues, the questions of absolute versus relative poverty, the merits of money income as a measure of the standard of living and the associated selection of poverty lines and equivalence scales, and the selection of alternative indices of poverty. A review of the literature indicates that the choice of differing approaches to poverty measurement can lead to differing estimates of poverty. In the face of such results an evaluation of the impact upon poverty estimates of alternative measurement methodology is appropriate. This thesis assesses the impact upon the estimated level of poverty of variations in some of the key poverty measurement parameters. The expenditure data derived from the 1975-76, 1984 and 1993-94 Household Expenditure Surveys have been analysed to assess the sensitivity of poverty estimates, derived from a range of poverty indices, to variations in the generosity of the equivalence scales, the level of the poverty line, and the choice of the indicator of the level of resources used. The sensitivity of each poverty index to variations in these parameters is assessed at both the aggregated level and for the specified household types, while those population subgroups particularly susceptible to poverty are also identified. The poverty distributions derived for each of the survey years are compared to evaluate the impact upon changes in the level of poverty over time of variations in the underlying parameters. The thesis concludes that both poverty estimates at a point in time, and poverty trends over time are sensitive to variations in the equivalence scales, in the level of the poverty line, in the selection of the indicator of the level of resources, and in the choice of poverty index itself. In light of these results, a review of recent Australian poverty research concludes that insufficient attention has been paid to the sensitivity issues associated with the measurement of poverty.

Non-monetary deprivation indicators are now widely used for studying and measuring poverty in Europe. However, despite their prevalence, the empirical performance of existing deprivation scales has rarely been examined. This thesis i) identifies possible conceptual problems of existing deprivation scales such as indexing, missing dimensions and threshold; ii) empirically assesses the extent of possible error in measurement related to these conceptual problems; and iii) offer an alternative way for constructing deprivation measures to mitigate the identified conceptual problems. The thesis consists of four stand-alone papers, accompanied by an overarching introduction and conclusion. The first three papers provide empirical evidence on the empirical consequences of the missing dimensions and threshold problems for the measurement and analysis of poverty, while the fourth paper exemplifies a concept-led multidimensional design that can reduce the error introduced by these conceptual problems. The analysis is generally held for 25 EU countries using European Survey of Income and Living Conditions (EU-SILC); only in the second paper, the analysis is done for the UK using British Household Panel Survey (BHPS).