ACCESSING AND GENERATING DATA 123standings (themselves, as discussed in chapter 1 of this volume, dependent on ontological and
epistemological presuppositions) and the questions thus framed, in turn, define what “counts” as
evidence. As Kuhn (1970) noted, observation is theory laden. That does not mean, however, that
observation is “theory determined,” a point made by I. Shapiro (2002), for otherwise, human
beings could never learn. Empirical evidence can still be unexpected and can startle the researcher
and upset accepted theories and habitual understandings. It is precisely that potential for the
proverbial upsetting of the apple cart that renders it so important to broaden the scholarly imagi-
nation about what “counts”—not only as social science evidence, but as legitimate social science
research questions and topics.^5
Considering the range of possibilities for evidentiary sources in the context of a hermeneutic
philosophy of social science highlights the extent to which the repertoire of human meanings is
expressed in various genres, such as the range of artifacts considered by Weldes. The principle of
“intertextuality” holds, then, not only across different (literal) texts but across data genres. A
researcher unread in the Bible or the Qur’an, for instance, would miss out on the resonances to
their audiences of U.S. president George W. Bush’s, Israeli prime minister Ariel Sharon’s, or
Saudi Arabian crown prince Abdullah’s speeches. But that researcher would also miss something
of significance (or that was significant at the time) in agency and polity meaning in an analysis of
the National Aeronautics and Space Administration if he did not know that it named its “Enter-
prise” spaceships, at many Americans’ requests, after the fleet in the science fiction series Star
Trek (Weldes 2003b).
Admitting of a broader range of evidentiary sources is not simply a matter of imagination; it is
also a matter of philosophical presupposition. For a science that focuses on numerical “facts,”
admitting less bounded, more ambiguously interpreted materials into the realm of “data” is, or
can be, a stretch. This predisposition is evident in the common distinction between “hard” and
“soft” data, reflecting the perceived centrality of numerical (“hard”) data to the enterprise of
“true” (i.e., methodologically positivist conceptions of) science.
One of the temptations—we think it an especially American one—in dealing with governmen-
tal documents, numerical reports in particular, is to treat them as inherently “honest” or “truth-
ful.” Perhaps this is a hallmark of positivist-influenced or ontologically realist research: We think
that interpretive researchers, especially those influenced by feminist methods, with their empha-
sis on standpoint epistemologies (see, e.g., Hartsock 1987), and critical theory, which questions the
ties between knowledge and power, are less likely to take government pronouncements on faith and
more likely to “lift the rug” and “look beneath the surface.” The reflexivity embedded in these
approaches leads researchers to question the production of knowledge (including of their own role
in that). This questioning may lead to exploring silences in discourse, what is not being said, such as
the “hidden transcripts” (J.C. Scott 1990) underlying official pronouncements.
Julia Paley (2001), for example, describes the ways in which residents of a barrio in Chile
resisted the efforts of the census taker and administrators of other survey questionnaires. In one
case, the respondent registered the fact that his choice of television station depended on context:
who else was watching, what was on, and so forth. Not satisfied with the response, the questioner
tried again. He met again with a reply that did not satisfy him. At that point, and despite all
training to the contrary, the surveyor “fed” the respondent the answers on the sheet. Such situa-
tions are familiar to survey researchers, and tremendous efforts are made to train questioners not
to yield to such counterpressure. But a second case describes resistance of a different sort. On
being asked what sort of electric appliances she had in her house, the respondent told the census
taker that she had none, despite the fact that several were visible and even running during the
“interview.” The respondent later explained to Paley: “Why should I say I have these things when