Comparative poverty research flourishes, especially since comparable income data are easily available through the Luxembourg Income Study. However, a number of methodological pitfalls in comparative poverty research are often overlooked. There is a vast amount of literature on sensitivity of measured results to the choice of income definitions, poverty lines, and equivalence scales, but other effects have been rather neglected in comparative poverty research. How does the underlying survey design affect results and cross-national comparability? Are low-income strata adequately represented in those surveys, is there a systematic bias of response rates among those groups, and how does it vary across countries? In addition, some types of income - such as means-tested benefits, being particularly relevant for poverty research - tend to be under-reported in some surveys. This paper uses the data available in the Luxembourg Income Study for three countries - Germany, Sweden and the United Kingdom - to exemplify the limited comparability of widely-used income data used in poverty research. In a first step, the paper summarizes the available evidence on methodological problems caused by differing data sources and survey designs, household definitions, and flawed reporting of some income components. Especially means-tested benefits tend to be under-reported in income surveys; so income for poverty-prone groups of the population may be underestimated, and, by this token, income poverty may possibly be overestimated. In a second step, this issue is illustrated by a simple simulation exercise: Entitlements to means-tested benefits are imputed for each household in the sample, based on the institutional regulations in each country. Compared to actual poverty rates in the original sample, imputed poverty rates are markedly smaller, if not reduced to zero. Even if one accounts for an incomplete take-up of benefits, a large gap between actual and simulated poverty rates still remains, largely caused by problems in survey design. The paper concludes with a number of recommendations for improving income surveys from the perspective of comparative poverty research.
