The word religion is widely used to describe a set of social practices, a taxon that includes such paradigmatic examples as Christianity, Islam, Judaism, and Hinduism. But the word can also be applied to things that are not explicitly religious, such as Shinto or hockey, or to a set of practices specific to a place or group, such as hunting or farming. The variety of definitions suggests that there is no single, unproblematic meaning for the term religion.
Over the past forty years or so, scholars have pulled back from a stance that treated the notion of religion as something that exists outside the study of human culture. Instead, they have viewed the concept of religion as a social construction. Several different approaches to this topic have been developed, including those of Clifford Geertz and Talal Asad. These scholars take a Foucauldian approach to understanding the way that concepts of religion are shaped by power relations, and they try to show that what people think of as religion is a matter of choice, not nature.
The idea that religion is a social construction has implications for the question of whether it is possible to study it objectively. For example, some scholars try to avoid the problem of stipulative definitions by using polythetic categories, a method that involves sorting social kinds according to their properties. The key, however, is that the polythetic system must be able to handle cases of overlapping or contradictory property sets.