Data Analysis as Process

52

53

After going through a thousand pages of text, more than 400 (substantive) categories had emerged. And while talking with different groups of actors about the same issues – for each group emerged a slightly different set of categories that showed each groups’

different perspective on the same issue. In this case, the water utility managers had categories for decision-making like ‘stressing long-term benefits’, ‘visionary’, and

‘making compromises’. On the other hand, the municipal managers had decision-making categories such as ‘collaborating’, ‘calculating’, ‘individual interests’, and

‘compromising as last resort’.

Below I provide an example of what these categories looked like in NVIVO in practice (Table 3-1). In the table, I categorized interview text according to ‘collaboration’ and for example I found an interviewee talking about working together by saying: “we had to collaborate with the neighbors because we do not have any groundwater ourselves”

which I then coded and categorized as collaborating ‘based on constraints’.

Table 3-1 Practical Example of NVIVO Tree Nodes

After the first round of coding (open coding), the processual aspect of the data was extracted by categorizing what the interviewees said about the past, present, and future.

From this data, it emerged that a managerialization is taking place but in order to verify this data, I used the content analysis of the three journals I reported in the previous text.

My ‘hunch’ was that if the interview data showed that management issues became more

54

important over the past decade, this would also be visible in the articles published during that time. Hence, I undertook the content analysis of the key publications of each sector.

Each journal was analyzed according to the main focus of its articles it had published during a certain period of years. The data included the two periodicals Kuntatekniikka (1998 – 2006) and Kauppalehti Newspaper (1995 – 2006). The content analysis for these publications was performed by using Microsoft Excel although NVIVO could have been used for that analysis as well⁵. For Kauppalehti, all articles related to water services from 1995 to present were screened according to their full text. For Kuntatekniikka, first and second level coding according to the titles and abstracts already available English language summary available for each issue from 1998 to present, from the publisher’s web site.

The categories emerging from the coding related to the focus of the article, which were technical, managerial, citizen, society, customer, regulation, policy, and general (see Table 3-2). After that round of coding, I decided to present the technical, managerial, and customer categories in a longitudinal chart that is presented in the chapter about municipal actors. I focused on these three categories because they had verified the impression that I had from the interview data, that technical issues had become less important over time, and that managerial and customer issues became more important.

This was the moment when I first had the idea that there was something like a rising municipal entrepreneurialism.

Table 3-2 Example of First-Level Content Analysis of Kuntatekniikka Journal Kuntatekniikka Issue of 06/2005

Technology Managerial Citizen Society Customer Regulation Policy General

Environmental art project with company 1 1 1

Lightning facilitates safety 1 1

3D scanning advantages 1

Infraguide 1

Canadian forum for municipal technology 1 1

New law on railroads 1 1

Changes in law on maintenance 1 1 1

GPS technology and virtual networks 1

Data management systems 1 1

Precipitation reservoirs 1

New techniques for WSS rehabilitation 1 1

Everyday life at Vihti water works 1

Traffic patterns and mobility management 1 Increased car ownership as challenge 1

Road as bottleneck 1 1

5 I found MS Excel more helpful for drawing tables and charts because I have been fairly familiar with it already, and it was possible to perform the coding also with that particular program although it probably required more time.

55

As a result of these journals’ longitudinal content analysis, I turned once more to my qualitative data and conducted a round of coding, where ‘water commercialization’

emerged from the previous categories as the main category. Since it appeared as the main category and the most interesting category, I selected it as the main category. The following stage meant to apply the selective coding technique (integrating and refining the theory) that helps to identity those categories, which have a relationship with the main category, making them sub-categories. At the stage of selective coding theoretical saturation is reached, which is the point in category development at which no new properties, dimensions, or relationships emerge during analysis (Strauss and Corbin, 1998).

Based on Glaser’s (1998) distinction between substantive and theoretical codes, more than 400 categories emerged that represented substantive codes (any empirically emerging phenomena that is interesting to the researcher and that is categorized) but during the stage of selective coding, theoretical coding had to be used to drop those categories that were not theoretically and conceptually relevant to explain the mechanisms of the main category of commercialization. Theoretical coding means to sensitively go through the categories and detect those which contain theoretical concepts and families of concepts, such as decision-making, that included ‘rationality’,

‘collectivism’, or ‘individualism’. As Glaser (1998) writes, the grounded theorist needs to be aware of theoretical families of codes because they make the researcher aware of their occurrence. However, they cannot be forced into the theory but rather have to emerge from the data.

After dropping most of the categories and only keeping those emerged as being connected to the phenomena of commercialization, the kept categories were analyzed once more and a model was built that explained the mechanisms. The theoretical categories that had emerged from the data, such as ‘rationality’ or ‘individualism’ were used to create the processual model of commercialization mechanisms. The categories that were finally left over from the originally more than 400 categories represent the headlines in the empirical chapter that explain the institutional logic grouped into actor’s domain, its principles of organizing, and its criteria for effectiveness especially related to its own organization and to water services.

56

Table 3-3How Grounded Theory Categories transformed into Chapters

In each chapter that analyses the logics of a particular actor group, it was reported on

‘how the water sector should be organized’ and the nature of decision-making from this actor’s point of view. The table above depicts the municipal actor’s logics, which comprise thinking collectivistic towards the inside of the organization (‘internal collectivism’) while individualistic towards the organizations’ environment (‘external individualism’) and applying managerial concepts grounded in ‘rationality’ to its decision-making.