B. Adding Data on Request Background and Deliverable
We developed two additional data points to describe the nature of the re-
questor’s question and the Network’s response to that question.
( 1) Background: The aspect of the requestor’s agency duties that his
or her request related to.
( 2) Deliverables: The legal or other service that the requestor asked
that the Network provide and/or that was ultimately provided in
response to the request.
We developed and applied values for these data points to the sample of
requests using qualitative analysis techniques.
15 Through a review of the
sample we coded the requests, or labeled them with one or more text segments summarizing either the request background or deliverable. We developed and applied individual codes (informed in part by background
sources) as we reviewed any and all relevant data recorded for each request,
including, but not limited to, text of the requestor’s question and email exchanges attached to request records.
We generally used two or more pieces of data to decide on code for a request background or deliverable. This was necessitated by the varying
amount of data available for each request; for many requests, the description of the requester’s question did not include enough information to decide on a particular code and so another piece of information, such as an
email exchange, was reviewed.
As new codes were developed, the code list was refined and the data reevaluated. Codes were either removed or added based on their similarity in
scope to other codes. Codes which were narrower in focus than most others
were combined into broader codes. Some codes were removed after the initial review of the sample because they were so broad that they were applied
to almost all requests.
Through this process of refinement, we eventually developed a final list
of codes for both request background and deliverable (see Figure 1) and applied one or more values from both lists to every request which included
enough information to complete an assessment. Deliverable codes were applied to all but two requests in the final sample. Background codes were
14. See infra, figure 1, for a list of codes.
15. See RUSSELL K. SCHUTT, INVESTIGATING THE SOCIAL WORLD: THE PROCESS AND
PRACTICE OF RESEARCH 325 (7th ed. 2011).
16. This fact represents a limitation of the study, since the process of deciding on a code
based on multiple pieces of information required more interpretation than would have been
necessary if coding had been possible based on one consistent piece of data.