Mitchell Centre Seminar Series

About

Neah Gondal Boston University Mapping Meaning Structures using Directed Network Graphs on Association Rules John Mohr (1998) persuasively argued that formal or structural meaning analysis involves three steps: (1) identifying elements of the cultural system, (2) conceptualizing and measuring relationships among those elements, and (3) representing and interpreting the structure of those relationships based on some type of formal relational analysis. Furthermore, drawing on practice theorists, Mohr also argued that because cultural systems are an embodiment of the everyday social and moral constraints and resources faced by individuals, it is essential to account for the institutional conditions within which meanings occur when making analytical decisions about steps 1-3. Assuming analysts make appropriate choices for cultural elements, my focus in this presentation is on measurement of relationships and their structural representation and analysis. One way to account for the social and moral arrangements within which meanings are structured is to measure relationships between cultural elements based on their co-occurrence within individuals’ consumption, taste, or belief profiles. Consumption of both classical music and visits to art museums showcases a person’s tastes but also reflects their social position (e.g., access to resources, their upbringing, or taste profiles of their social ties). Likewise, belief in fetal personhood and individual choice in childhood vaccination schedules is shaped by social factors such as class, geographical location, and political stance. If such co-occurrences are found among a large number of people in similar social positions, we can infer not only that the relationships between the cultural elements are robust but also that they are meaningfully connected to contextual conditions. Research investigating cultural tastes and preferences has indeed used this approach to interpret co-occurrence patterns as relationships between cultural elements based largely on correspondence analysis, variable correlation networks, and one-mode projections of two-mode networks. In this presentation, I describe and demonstrate the use of another tool to infer relationships between cultural elements, a datamining technique with scant history of use within sociology, called ‘association-rules.’ The key benefit of this technique is that, unlike other methodologies, it generates directed relationships between variables (e.g., preference for opera ? preference for ballet), which has several advantages over existing techniques. For example, association-rules can reveal relationships between more than two variables within a single rule (e.g., A&B?C). I show how such ‘one-sided’ clustering (A goes with B, but B may not go together with A) can be represented and analyzed as network graphs, an approach I call ‘Rulenet.’ I discuss how the proposed technique can provide relatively novel insights into the structural analysis of cultural meaning, less feasible via other techniques, and illustrate the use of Rulenet on survey datasets measuring tastes and beliefs.