Information Technology Reference
In-Depth Information
question remains: How can we know that we have identified the correct causal-
ity in a single case or small N study? Or in other words (those of Kelly, 2004),
how can we distinguish between the
contingent
and the
necessary
? Ideally, ques-
tions such as this could be answered at the level of the method, and would not
have to be achieved on the level of a specific study, i.e., for each application of the
method. What would need to be established for each application of the method is
that it was realized with an acceptable quality standard. We can, hence, distinguish
between the
argumentative grammar
(Kelly, 2004) of a method, the logic by which
evidence is linked to claims (conclusions, interpretations, and recommendations),
and the
provenance
of the data and claims (the manner they were brought about
in a specific study). Together, these two aspects contribute to developing
trust
in
the outcomes of a study: via the argumentative grammar, it is established that find-
ings resulting from the application of the method are
in principle
warranted, and
the provenance information can be used to establish that the method was applied
correctly in a specific study. Since the argumentative grammar of the various meth-
ods used in learning research have been and continue to be extensively discussed
(approaches and technological applications discussed in the section above have also
focused on this level), we will focus here on the provenance aspect that underlies
many methodologies and data, but has received little attention in learning research
so far. It is also the aspect that could be addressed by information technology.
Provenance of Data and Findings
The term provenance is mostly used in Arts, where it is used to describe the history
of ownership of a work of art. In computer science, Groth et al. (2006) define the
provenance of a piece of data as “
the process that led to that piece of data”; as
such, provenance is synonymous with data lineage, pedigree, or history and applies
to quantitative as well as qualitative data. Reviews of related lines of research
in computer and information science are provided by Bose and Frew (2005) and
Simmhan, Plale, and Gannon (2005).
Provenance can relate to the use of data both within a study (e.g., How was the
data created? How transformed? How did it enter into the conclusions?) as well as
across studies (e.g., Which other studies refer to this data?). In this way, provenance
can be used to assess data quality and reliability, to establish ownership of data,
to establish an interpretation context, or to replicate a study (Chorley, Edwards,
Preece, & Farrington, 2007).
The general architecture of a provenance system (Groth et al., 2006) relates
provenance-processing services (such as for data audits) to the provenance records
gained from an application environment (such as a number of qualitative and quan-
titative data transformation and analysis services) as shown in Fig. 12.6. In the
general framework described by Groth et al., provenance information is represented
as assertions (so-called p-assertions) made by an actor pertaining to a process. The
...
