This paper reflects on the relationship between economic (quantitative) and anthropological (qualitative) approaches to the analysis of poverty in developing countries. Drawing on detailed evidence from Mozambique, we argue that different research approaches do not merely see the same poverty from different angles; rather, they tend to conceptualize poverty differently. These different views can be traced to alternative philosophical positions along three axes - the ontological character of poverty; its generative mechanisms; and epistemological priorities. The quantitative analysis provides an indispensable numerical snapshot of trends in consumption and its broad correlates over time (e.g. via poverty profiles). In contrast, anthropological work focuses on lived experiences of poverty, which is rooted in a view of poverty as a process of social marginalization. While the policy implications of an economic approach focus on overall economic development and enhancing inputs at the household level, a relational view of poverty suggests the need to address counter-productive relations of power and to carefully target interventions at the poorest. Clarifying the specific philosophical commitments invoked in different forms of poverty research can shed light on the different kinds of research questions they are able to address.