• Keine Ergebnisse gefunden

3. METHODS AND MATERIAL

3.1. METHODS

3.1.2. CONTENT ANALYSIS

A quantitative-qualitative content analysis has been chosen as the adequate method for carrying out the discourse analysis. Commonly, content analysis has been used in research of political communication as well as in mass media analysis since it can reliably and validly interpret texts, pictures, movies, and music pieces (Friedrichs, 1973: pp.317-18).

Mass media studies, regarding public spheres, applying content analysis include Ferree et al., 2002) and applied to forest issues: Krumland, 2003), and Hütte (1999). In Krumland (2003), content analysis was carried out on texts from daily newspapers in order to examine the political effect media reportages have on the positions of the forest and nature conservation sectors. In Hütte (1999), content analysis was carried out to make a European comparison of the technical discourse for both forest and nature conservation sectors on the concept of sustainability.

Content analysis is a method which systematically and objectively captures characteristics of processes of social communication (Friedrichs, 1973: p.315). Social communication will be here understood as: a process in which -within a social environment- an actor (sender) „says‟

something (message) in a certain medium (communication channel) which is to be received by another actor (receiver; ibid.). Or as Laswell has so famously summarized: “who says what to whom, why

50

and with what effect?” (Laswell, 1948). Content analysis can be carried out to reveal each and all of these questions. It can focus on the senders, the message, the receivers, and/or the intentions and the consequences of the communication process. In this study the focus, within the communication process of forest science, is the questions of who communicates (sender) and on what is being said (message), being the chosen channels newspaper articles and articles published in scientific peer-reviewed journals (see below).

Definitions of content analysis are many in number. Berelson has defined content analysis as the “research technique for the objective, systematic and quantitative description of the manifest content of communication” (Berelson, 1952: p.18). Content analysis is objective in the sense that on hand certain clearly specified rules, material is assigned to specified categories. The existence of these clearly specified rules of categorization decrease the possibility that the personal bias of the coders influence the process (Bryman, 2001: p.178). Being systematic means that the application of the specified rules is done in a consistent manner in order to again reduce the possible bias (ibid.). Berelson‟s definition mentions quantitative descriptions referring to the quantitative accounts of the material analyzed through categories. However, qualitative aspects may as well be incorporated20, which may lead to a complementary way of doing content analysis, namely a quantitative-qualitative content analysis. Berelson finally mentions manifest content of communication as the research objective, in which he is referring to the actual message being delivered in other words, what is being said.

A complementing definition of content analysis, which is not as restrictive as the previous definitions, is delivered by Früh (2001). He defines content analysis as “an empirical method to describe systematically and intersubjective-comprehensibly the content and formal characteristics of messages”21 (Früh, 2001: p.25). This definition intentionally leaves out characteristics such as „quantitative‟ and „manifest‟ of the previous definition which, according to Früh, have been the cause for more confusion than clarification. Früh states that a main qualitative criterion of content analysis is that the results must be independent of individuals (must be objective) in order for conclusions to be valid; it must allow results to be repeatable, communicable, and criticizable (p.37).

There are many advantages of content analysis (see Bryman, 2001: pp.189-90; and Früh, 2001: p.39): it is a transparent research method; it allows longitudinal analysis; it is an unobtrusive method, because it considers, for example, written texts as newspaper articles or scientific journal articles, being thus not dependent on the cooperation of test persons;

it is a flexible method: researcher is not tied to certain deadlines regarding the data collection; no changes in the research object are present throughout the research; and it allows information to be generalized about social groups that are difficult to gain access to. All these qualities make content analysis the adequate method for carrying out the research here undergone.

Considerations must be given to the disadvantages of the method such as (Berelson, 1952: pp.191-92): the documents up for analysis must be authentic, credible, and representative of all possible documents, if they do not fulfill this criteria the conclusion of the analysis must be taken with consideration; some interpretation on the part of

20 For example: incorporating „frames‟, through which the points of views of the different speakers are collected.

21 Own translation.

Methods and Materials

51

coders is always possible and so it is impossible to construct a category system that leaves out some interpretation; problems arise when latent content (the hidden meaning of a message) instead of manifest content (apparent content of the message) are examined; an answer to the question of why a content occurs is mostly based on inferences.

The analysis of content is carried out, once hypotheses have been erected, through a category system, which allows the coding of the units relevant to the problem dimensions, which in turn reflect the hypotheses. In the category system the various dimensions of the hypotheses are operationalized. The category system is the most important part of the method because on it depends the quality of the research, being then content analysis only as good as the categories which make up the substance of the research (Berelson, 1952). Here, the research categories research categories employed all have a theoretical background. The objective of the construction of categories is to be able to make statements about the structure of the researched material through frequency or multidimensional tables (Friedrichs, 1973: p.322).

For this work, two units of analysis are considered. On the one hand the entire article (be this a mass media or scientific article) and on the other, the speaker and her/his statement.

Regarding the entire article, formal categories comprising physical units (physically delineated aspects; Krippendorff, 1980: p.61) are recorded such as journal source, article date, number of authors, and type of article (such as articles, review, or editorial, which have been proven to have on average a substantial citation impact Moed and Leeuwen, 1995: p.461). On the other hand, the primary source of structure and meaning of a text is the speaker; the speaker decides what to say, how to say it, and what others should take it to mean. Not one speaker speaks in a text but many can do so. Thus, the statements of the different speakers carrying out either directly or indirectly speech acts in the articles (both media and science) comprise the second unit of analysis; these statements are the propositional units of analysis (by which the objects –statement of speakers- and their attributes –such as interest position assigned to others- are recognized; Krippendorff, 1980: p.62). By statement is meant all those speech acts incurred on by one speaker relevant to the topics being examined (Ferree et al., 2002: p.50). Speakers and statements are coded when they have direct relevance to forest and one of the three issues selected.

3.1.1.1. CATEGORY SYSTEM

The category system, contained in a coding book, is crucial for carrying out content analysis. Since here the objective is to examine both media and science discourse on forest two category systems were built. In order to be able to compare relevant categories no great changes were carried out, however certain categories were only applicable to media discourse and others only to science discourse. Both these category systems were based on the theoretical considerations presented in the previous chapter and can be seen in annex I.

The category system must comply with five important points (see Bryman, 2001: pp.188-89; and Holsti, 1969: p.95). It must have discrete dimensions (dimensions are totally

52

separated, no overlap between them); it must have mutually exclusive categories (no overlap in the categories supplied for each dimension); it must be exhaustive (for each dimension, all possible categories should be available); it must give clear instructions:

categories must be well defined (interpretation about each dimensions and what factors to take into account for each category must be clear); it must be clear about the unit of analysis (such as newspaper article or statements on defined topics); and the classification must be consistent to allow comparisons.

The category systems reflect both a formal-descriptive approach relating to the entire article as unit of analysis and a diagnostic approach where answers to questions such as what the speaker means with their message and which characteristics it possesses (Früh, 2001: pp.41-2).

The theoretical discussion, resulting in the five hypotheses tested required specific categories to be constructed in order to obtain the data for proving them. Two main groups of categories were built according to the hypotheses constructed. The first group of categories refers to those which needed to prove the existence of a deliberative or empowered discourse, as well as determine whether collaboration is a form of a deliberative discourse. The second group refers to those variables that help to verify or falsify the existence of medialization of science. Table 3.1 gives an overview of the different categories that were constructed based on the previous theoretical discussion.

Details of the categories may be found in annex I.

Table 3.1. Overview of categories providing evidence of deliberative science discourse or medialization of science (Source: own construction)

Assessment event / statement of speakers - Communication tools

History, technology, metaphors, symbols, descriptions, others

- Politization

Interest positions: speaker / causer

For both systems, categories regarding the entire article were coded. Such as date, source, authors (number of authors and country and institution affiliation of authors for the scientific articles), place of event (country and level of event), assessment of event were coded.

Speakers and their statements were coded within the propositional unit of analysis (represent standing). On this level, categories such as assessment of statement, the interest position speakers place others in, the existence of risks, the solutions mentioned and if solutions were mentioned, what type of political instruments was used when giving solutions, and the communication strategies they used were coded. For the media category system, the country of speakers was also coded. For the science category system only the affiliation country of authors writing the article could be recorded.

Methods and Materials

53

The speakers were coded according to their main occupational sector. Speakers can be scientists, members of the government, politicians, enterprise representatives, non-governmental organization representatives, single persons or communities amongst others (full list and definition of actors and their occupational status is found in the annex I).

Journalists as well as other members of the mass media were as well coded. Especially journalists in the mass media discourse have an important role when speaking, as these are the actors that write introductions to topics and in doing so bring their own interpretation patterns in the discussion. In the scientific discourse, actors were coded by looking to the reference section of the articles. If the reference indicated a contribution to a scientific journal, or a scientific book, then these actors were coded as scientists.

The statement of speakers was coded regarding the interest position (in accordance with von Prittwitz, 1990) they assigned to other actors involved in the issue discussed. The framing of the issues was then done considering the actor-group acting as speaker and the party they named as responsible (causer) for the problems faced in the issue. The same categories as when coding the occupational sector of the speaker were used adding society and nature as actors that can be the causers, victims or addressees of the issues at hand (these cannot be direct speakers).

The assessment of the event as well as the speaker‟s statements, were coded as positive, negative, ambivalent or neutral assessment.

Regarding the communication tools, metaphors, descriptions, historical references, and the use of symbols were coded as strategies that speakers make use of when delivering their statements and which can later serve as evidence for the medialization degree of the discourse.

3.1.1.2. CODING PROCESS

After the category systems were built, the training of coders was carried out. In this phase coders were familiarized with the category systems, sample articles were jointly coded at first and then separately and results afterwards compared. If differences in categorizing arose, further clarification and training was undergone. The coding process implied the use of a software-based data mask (in SPSS).

Coding must be done in a consistent manner (Bryman, 2001: p.189). There must be inter-coder (coding must be consistent between inter-coders) and intra-inter-coder (each inter-coder must be consistent over time) reliability in order for the category system to be valid. Inter-coder reliability was assured in that the same selection of articles, coded by different persons, was controlled and compared for their results. Minor differences were revealed and afterwards corrected; reliability between coders was high. The same procedure was repeated after all articles had been coded. The intra-coder reliability was examined through coding a selection of articles again after some time had elapsed since their original coding. Results were consistent.

54