• Keine Ergebnisse gefunden

69 c) Job advertisements in June 2008: vacancies announced over the period of 30 days in five national newspapers (print copies) were screened in the archive of the “Daily Times” in Blantyre. Relevant information on health-related vacancies was entered into MS Access, with each advertisement constituting a case (n=53). This did not include positions that required nursing, medicine or biomedical qualifications only. The variables were: date of publication, number of posts, name of employer, location of post, mode of employment, duration, level and type of qualification required. Further analysis was carried out in MS Excel. The resulting table can be found in chapter 4.3.4 under ‘Contemporary data on research and consultancies’.

d) Database for Technical Assistance for the National HIV/AIDS response in Malawi, 2007: This electronic database issued by the National AIDS Commission Malawi was acquired as a CD-ROM. The profiles of n=107 individual consultants are accessible through a fixed user surface. 29 of these individual profiles are attached to profiles of consultancy firms. The electronic format did not lend itself to a transfer into MS Excel or another database.

Moreover, the information contained in the individual profiles was very heterogeneous. Quantitative analysis was therefore restricted to establishing rough categories and making simple frequency counts. The findings are included in chapter 4.3.4 under ‘Contemporary data on research and consultancies’.

Hence, chapter 4 of this study presents the results of the structural data analysis. The focus is on professions or occupations in the first part, while the second part characterises organisations operating in the Malawian health sector and their relationship among each other.

70 they are not analysed as a distinct corpus of data. Most of the text material gathered and produced within the scope of this research, as far as it could be converted into Rich Text Format, was analysed using the software MAXQDA 2010. This predominantly applies to the transcripts of the interviews and the field notes. MAXQDA has been designed for qualitative data analysis and mixed-methods-approaches (quantitative and qualitative). It is offering the possibility to link different elements of data with each other through coding and writing memos, and to apply filter variables for text retrieval.

Expert interviews are not aimed at individual case reconstruction but allow to generate theory about the experts’ larger field of action and its social organisation. Due to the interview guideline, the interview data constituted the most homogeneous section within the corpus of data for this study, which facilitates comparisons between different expert accounts (Meuser and Nagel 2005). The approach, which was taken up by Bogner et al. (2005), is located in the discipline of knowledge sociology, however.

The related constructivist schools of thought and the abstraction process suggested by these authors for the data analysis proved difficult to apply to a case study conducted in a foreign country. Therefore the thematic analysis proposed by Froschauer and Lueger (2003) was chosen as an alternative method.

Thematic analysis is aimed at text reduction and at working out ‘who is saying what in which context’ (Froschauer and Lueger 2003). Considering the different occupations and employers that are investigated in this study, this method allows to work out similarities and differences between certain interviewees and between certain narrative contexts. The term ‘narrative context’ in this case refers to the course of the interview; it is asked in which immediate context within his or her narration the interviewee made a statement. Four different contexts are considered for the purpose of this study, namely whether the interviewee is talking about:

• professions / occupations

• organisations (employers) or contractual work arrangements

• activities and interactions of organisations to provide health services

• opportunities of policy development and international cooperation.

71 Thematic analysis does not require literal transcription. Therefore only the first 6 interviews were literally transcribed, but the recorded material was revisited to write an extensive chronological protocol for each interview. Information was anonymised through masking names and places. MAXQDA allows for the protocols to include hyperlinks to the respective passage in the audio file of the interview. This way, the most relevant sections of the interviews could be easily retrieved later and be transcribed in detail where necessary. Furthermore, relevant variables concerning the circumstances of the interview or attributes of the interviewees can be attached to the files. This feature provides the basis of sorting, filtering and descriptive statistics within MAXQDA. Related field notes and observation protocols can complement the interview data, which is possible through memos that link up different files.

The themes in this study have been deducted from the UNDP literature on capacity development. The UNDP framework provides the four ‘levers of change’, i.e.

knowledge, leadership, accountability and institutional arrangements (UNDP 2010). As special attention was supposed to be given to international mobility and the interaction between Malawian and international actors, I added the themes availability and ownership (UNDP 2003). The interviewees’ statements were collated with these thematic codes, which comprised 21 sub-codes (see Annex 7.6). The UNDP framework is taking an institutionalist approach to capacity development (UNDP 2010), however, it does not provide analytical tools or a research methodology. Therefore, to actually extract relevant findings from the interview data, a combination with the IAD framework was necessary, particularly with the ‘grammar of institutions’ (Ostrom 2005).

Double coding of the interview protocols (thematic codes X narrative contexts) and the use of filter variables (occupation and employer of each interviewee) provided the basis for the identification of relevant interview statements. The results are thus not presented ‘theme by theme’. Rather, the most important themes or statements are presented under the headings referring to occupations, employers or other narrative contexts. Findings from the thematic analysis are presented both in chapter 4 and 5, each referring to different contexts in which statements were made.

72 3.3.3 INSTITUTIONAL ANALYSIS AND DEVELOPMENT FRAMEWORK

The application of the IAD framework in an empirical case study is principally an analysis and synthesis of written or spoken text. Two central elements of the IAD framework appear particularly relevant for the study of HRH in a country like Malawi and shall therefore be introduced here. First, in order to analyse and develop institutional statements, Ostrom (2005) presents a so called ‘grammar of institutions’

to be applied at the linguistic level. The second element is the concept of action arenas (or action situations) as ‘nested holons’. This concept can be used to narrow the focus on the interactions of a particular set of players, while at the same time systematically considering external influences. The integration of different methods and data sources (thematic and structural data analysis) under the IAD framework will subsequently be described in the summary (chapter 3.4).

Grammar of institutions: Ostrom (2005) distinguishes three grades of institutional statements by their binding force, namely shared strategies, norms and rules. A rule is the most obliging institutional statement by this definition, as it contains all five syntax components of the ‘grammar of institutions’. These components are:

• [ATTRIBUTE], identifying the attributes of the actors which are concerned by the statement (e.g. age, citizenship or qualification)

• [DEONTIC], prescribing the mode of action within the scope of the physically possible (i.e. that the actor may, must or must not do something)

• [AIM], indicating at what action or outcome the institutional statement is directed (e.g. setting up a roster, or a certain percentage of skilled birth attendance)

• [CONDITIONS], specifying the circumstances under which the statement applies (e.g. during outreach activities in remote communities, at the quarterly meeting of the management board)

• [OR ELSE], stipulating a sanction which will apply in case of default (e.g.

withdrawal of a license, budget cutbacks)

A norm by this definition includes all syntax elements except for the [OR ELSE], meaning that it still has a prescriptive or moral character. Even if the adherence to the

73 norm is monitored, this is not backed-up by prescribed sanctions. Finally, a shared strategy only comprises [ATTRIBUTE], [AIM] and [CONDITIONS]. It is therefore a non-prescriptive statement taking the form of an advice. Shared strategies rely on prudence and on the expectations that the actors have about each other’s future behaviour.

The present study aims at uncovering strategies, norms and rules from the data, as far as these statements are concerning HRH management and development in conjunction with international aid. Notably, the emphasis in Ostrom’s work lies on

‘rules-in-use’ rather than ‘rules-in-form’ (e.g. laws and written policies), since the actors might not be aware of the latter in their day-to-day decision making. Therefore the institutional statements in this study are largely derived from interview data. This is supplemented by the field notes on the researcher’s interactions with the interviewees and with various other actors involved in international collaborations.

Nested holons: The set of institutional statements concerning HRH in Malawi is principally infinite, since the wider social context also needs to be taken into account.

The field is characterised by a multitude of different kinds of actors operating at the international, national, district and community level, but also transcending these levels. The IAD framework aims at capturing this complexity by focusing on action arenas or action situations. For example, organisations can on the one hand be considered as participants in an action arena interacting with other participants (e.g.

the Ministry of Health interacting with CHAM). On the other hand, they may also host several action arenas within themselves (e.g. the District Health Offices, or CHAM technical colleges). This concept is called ‘nested holons’, with the outcomes of one action arena influencing the other arenas either in a hierarchical or a sequential manner (Ostrom 2005, pp. 11-13). For an empirical case study, it is indispensable to concentrate on circumscribed settings or situations for which the rules and patterns of interaction are to be analysed. While Ostrom (2005) does not provide a clear discriminatory definition, I apply the term ‘action arena’ to the general setting of interaction, and the term ‘action situation’ to a specific constellation of actors within this arena.

74 The focus of this study is on an action arena at the operational level, namely the health district with its planning and coordination activities for the provision of health care.

Particular attention will be paid to the norms, strategies and rules that govern staff deployment and development in this arena. Within one health district, there may be a range of action situations with different actors in specific constellations, who are making operational choices of relevance to health workforce performance. The centre piece of an action situation is its link between actions and outcomes, which can again be captured and refined by institutional statements. The statements can refer to different components of the action situation. On the ‘action side’, they may specify positions, as well as the participants and actions assigned to the positions. On the

‘outcome side’, they may define the envisaged state of the world and the net costs and benefits (pay-offs) assigned to it. Thirdly, institutional statements may relate to the information about and the control over the action-outcome link as such. The components and related types of rules are displayed in figure 2.

Figure 2: Structural components of an action situation Source: Ostrom 2005, p. 189

75 Having established the action arena to be studied also allows the researcher to explicitly turn to its environment (Ostrom 2005, p.15). This refers to exogenous variables on the one hand (in this case the physical, social and cultural attributes of HRH and the larger community in which it is embedded), and to linked action arenas on the other hand (in this case situated at the collective choice and the constitutional level of the Malawian state as a polity).