Comparative Research Design - The Politics of Water Institutional Reform in Neopatrimonial States - page 49

the interviewees provide more information when confronted with the concrete situation com
pared to when the interview is conducted in a room (Messer, Townsley 2003: 30f; De Negri et
al 1998: 57 60).
Interviews were conducted with representatives of the WUA and other local organiza
tions as well as with villagers who were selected randomly. Several interview manuals were
developed for different target groups. These interviews in the respective village were endorsed
by interviews with representatives of responsible district agencies and the donor project offic
es. All manuals as well as the lists of interviews can be found in the appendix. Group discus
sions were conducted with members of the WUA council and randomly selected groups of
villagers. However, they were difficult to organize and only one group discussion with WUA
council members was conducted in each case study. Group discussions with the villagers they
occurred rather spontaneously. Transect walks (sometimes car rides) were conducted along the
channels with the respective water managers. While local officials are usually able to speak
Russian fairly well, the population is not necessarily able to do so, especially the women.
Therefore, the local case studies were conducted together with one or two local field assistants
who translated when interviewees preferred to talk in Kyrgyz, Uzbek, or Tajik rather than
Russian.
Qualitative Data Analysis
For the analysis of the data, qualitative content analysis was employed. Qualitative content
analysis aims at identifying the main contents of the material with a step by step reduction and
dissection of the textual material. It has to be conducted systematically in order to ensure that
it is intersubjectively comprehensible (Mayring 1991, 1992: 22 41; Lamnek 1995: 207 218).
As far as possible, all expert interviews were recorded and afterwards transcribed, the
Russian ones by a Russian native speaker. However, interviewees sometimes reacted reluctantly
to the wish of recording the interview, especially in the lower levels of the administration. In
these cases, recording had to be refrained from in order to get any information at all. Open
interviews were sometimes recorded, depending on the concrete situation. Interviews at the
local level were usually not recorded for the above mentioned reasons. Since these were mostly
conducted together with a local field assistant, it was possible to draw up detailed protocols as
the functions could be shared between an interviewer and a minute taker.
The obtained data were analyzed with MAXqda software. Beside the interview transcripts,
also protocols of informal conversations and observations were typed into MAXqda. The
process of qualitative content analysis was based on the model developed by Mayring (1991;
1992). The first step of data analysis was the coding all of the texts (Kuckartz 1999: 75 100).
This allows the systematization, decomposition, comparison of the data. The code system was
developed deductively based on the interview manual, followed by two inductive revisions.
Then, the relevant codes were paraphrased. The codes that referred to purely technical or
procedural knowledge (like data on water use, laws, structures) had not to be further reduced.
Codes referring to interpretation patterns and assessments (such as characteristics of the politi
cal process or the administration) were individually reduced further, explicated and structured.
The code system is attached in the appendix.