INTRODUCTION xixartifacts are physically gathered up and brought back to the lab for analysis; whereas in the
nonexperimental, field-based studies^16 conducted in the human sciences, the primary “data” and
their sources are left in their locations of origin (or should be). What are brought back are the
researcher’s copious interview and/or observational notes and/or notes on documents, although
copies of documents, interview tapes, and the like may be brought out of the archive or the field.^17
This formulation makes clear that the data of such studies are not the people themselves, or the
events and conversations and settings and acts, or even the documents, but rather the researcher’s
views of these, as encapsulated in her notes. “Data,” in this approach, are not things given (datum,
data, from the Latin “to give”), but things observed and made sense of, interpreted. What is ac-
cessed are sources of data; the data themselves are generated, whether by the researcher interacting
with visual/tactile/spatial sources or coproduced in conversational and/or participatory interactions.
This understanding of “data” as constituted by human researchers’ observations renders problem-
atic the creation of databases of interpretive data for other researchers to use. So-called raw data
may be the “least interpreted” form (in contrast to succeeding stages in the research process), but the
“interpretive moment” cannot be escaped: It colors all stages of the research process, such that
human science data are never really “raw” and “unprocessed.” Other researchers would be getting
processed, not “raw,” data—“cooked” and filtered through someone else’s interpretive schema.^18
VARIETIES OF INTERPRETIVE ANALYTIC METHODS
At this point (if not before; see below), “data” in hand, methodologies part company, largely
around the question of the legitimacy of “word data.” At the epistemological level, methodologi-
cal positivists posit the superiority of quantitative data over word data. From this perspective,
words are best translated into numbers for purposes of statistical analysis.^19 Interpretive research-
ers reject the assumption of the superiority of quantitative data over other forms of data (e.g.,
sound, visual imagery, built space). They do not reject quantitative data per se. Instead, they take
an interpretive perspective on numbers: That communities choose to count particular phenomena
reveals much about what communities value and the problems that are, or are not, recognized as
central to their identities and concerns (see, for example, Czarniawska-Joerges 1992 for an inter-
pretive analysis of budgets).^20 Interpretive researchers, then, respect the form or genre of the data,
and word data are retained in their original form for purposes of interpretive analysis.
It is common knowledge that there is a wide range of “advanced” methods of statistical analy-
sis: Markov chain Monte Carlo ideal point estimation, multinomial logit analysis and multino-
mial probit analysis, ARIMA (autoregressive integrated moving average), MANCOVA
(multivariate analysis of covariance), and MANOVA (multivariate analysis of variance), cluster
analysis, factor analysis, principal components analysis, and more.^21 It is less widely known out-
side of the interpretive research community that there is a broad range of methods for the interpre-
tive analysis of word and other data genres, some of them (e.g., semiotic squares) no less “advanced”
or complicated than nonparametric random-effects analysis or Poisson regression. Several of
these are explored in part III of the book. To display interpretive methods in all their “infinite”
variety, Table I.1 lists here many more than are addressed there; we also note that this table is
suggestive and by no means complete.^22
This variety of analytic methods is suggestive of the fact that all interpretive researchers do not
speak with one voice on some of the central philosophical and procedural issues. Interpretive phi-
losophies have only been available beyond the German- (and, to a lesser extent, French) reading
world since the mid-twentieth century or so. Their explicit, conscious, and intentional extension into