334 ANALYZING DATA
issues, and inconsistent arguments about the impacts of wilderness designation. In contrast, and
as a result of changes made in response to OEPR comments, the final EISs consistently focus on
study areas, evaluate specific scenarios of future development, and clarify the logic of arguing for
or against wilderness designation.
The wilderness review occurred amidst polarized debate about wilderness designation of pub-
lic lands. The debate reflected the fact that the wilderness mandate challenged the BLM’s histori-
cal focus on resource development.^1 Within the agency, this highly contentious debate about
tradeoffs among policy goals was channeled, in part, into debate about technical analyses in the
EISs. Agency staff used technical analyses to generate scenarios, or stories, about future activities
in wilderness study areas and about the impacts of wilderness designation. Through these stories,
they made sense of and framed the issue of wilderness designation. Their analyses reflected nor-
mative positions about what the future ought to bring in terms of resource development and
protection. As a result, technical information in the documents cannot be separated from political
debate about whether public lands should be designated as wilderness. To assess the stories and
arguments in these EISs about designating wilderness areas on public lands, I conducted an inter-
pretive content analysis of them.
DEFINING TERMS: CONTENT ANALYSIS, STORIES,
AND ARGUMENTS
Researchers from varied traditions with quite different research questions look to the content of
written documents as a source of data. As a result, there exist many approaches for analyzing the
content of documents. In political and other social sciences, some researchers have developed
methods for coding documents at the paragraph, sentence, and word levels to characterize con-
tent. Political scientists have used such coding systems to generate numeric data and conduct
statistical analyses to relate political positions to a range of variables, including voting behavior,
post-election policy implementation, and government coalition building (Budge, Robertson, and
Hearl 1987; Gabel and Huber 2000; Ginsberg 1976; Klingemann, Hofferbert, and Budge 1994;
Laver and Budge 1992; Laver, Benoit, and Garry 2003). In contrast, coming from the fields of
discourse studies and communications, Killingsworth and Palmer (1992) used rhetorical methods
to analyze the content of BLM EISs. Retaining data in word form, they characterized and com-
pared style, voice, and language in texts written by agency experts and in the texts of comments
written by citizens in response to draft EISs. Through their analysis, Killingsworth and Palmer
addressed questions about communications between experts and the general public. In all of these
examples, the methodological approach focuses on selected content of documents as data, with
an expectation that systematic analysis of that content-data will provide evidence to answer the
research question at hand.
Early in my research, I posed the question: How did the BLM personnel interpret “wilderness
policy” as they implemented it? In making connections to the academic literature in policy, I used
concepts of narrative and argument articulated by Roe (1994) and Majone (1989). The overall
policy context involved high levels of ambiguity and polarization (arising from multiple and
conflicting interpretations of the value of public lands and wilderness), uncertainty (related to
incomplete information about future conditions for commodity development such as minerals),
and complexity (due to the interconnectedness of issues associated with the multiple uses of
public lands). Roe (1994) argues that narrative policy analysis is useful for understanding and
developing alternative policy approaches in such circumstances. In analyzing the content of the
wilderness EISs, I used definitions of stories and arguments informed by his work. He suggests