Rituals of identification: A critical discourse analysis of public opinion polls about welfare

Public opinion polls do more than elicit "opinions" from anonymous respondents. They stipulate the shape, scope, and very terms of those opinions, and moreover, presuppose a host of ideological assumptions underlying those opinions. Based on a critical discourse analysis of polls about welfare from 1992-1994, this dissertation argues that polls perform an important symbolic and instrumental role in the social construction of political reality. I argue that: (1) As a form of discourse, polls elicit symbols and information; (2) As discourse, polls convey symbols and information; (3) As a form of political ritual, polls construct the idea of public.; Chapter One introduces the dissertation. Chapter Two presents a history of various conceptions of public opinion in the twentieth century and outlines three perspectives on public opinion polling: populist, critical, and constructionist. The chapter also reviews scholarship about the relationship of polls to language, the press, and policy.; Chapter Three analyzes polling as a form of communicative interaction and proposes a functionalist discourse model of polling that takes role constraints as well as the syntactic and semantic features of poll questions into account.; Chapter Four examines, from critical and feminist perspectives, polling as a form of persuasive communication that invokes ideologically conservative presuppositions about welfare, work, and dependency. The chapter also briefly describes welfare and its history in the United States.; Chapter Five explores polling as a form of political ritual and explores how the term public is itself far more than a statistical abstraction--it is a powerful symbol that evokes a cultural order rife with contending identities, hierarchies, and boundaries.; Chapter Six statistically investigates the relationship of structural and content features of poll questions about welfare to the strength and ideological direction of response. The results suggest that structural and content features account for at least 38% of the variance in aggregate conservative response to polls conducted in 1994.; The dissertation concludes by arguing that the presumption of self-evident truths, values, and meanings embedded in polling discourse about welfare creates an ideologically saturated but non-negotiable field in which the production of consent is virtually guaranteed.