Bachmann, Lorenz B. R.: Review of the Agricultural Knowledge System in Fiji - Opportunities and Limitations of Participatory Methods and Platforms to promote Innovation Development -


Chapter 3. Methodological framework

The framework conditions for the field study were very favourable as the researcher lived and worked in Fiji for more than two years before the research phase of this study began. This good framework comprised of:

A participatory action research methodology was used for this study. The concept of this approach is reviewed in the first section. The role of the researcher and that of other actors in the research process is examined in the second section. The implementation of the fieldwork with all research activities and important aspects of the research process are discussed in the third section. A summary of the tools applied in the research process is provided in the last section.

3.1 Participatory Action Research (PAR)

Participatory action research is an explorative qualitative research method. In the past the scientific acceptability of qualitative methods was often questioned<23>. Today, this discussion has become less controversial and the use of qualitative methods has been more frequently accepted. Wolcott (1990, 26) refers to the discussion and states “There is no longer a call for each researcher to discover and defend them anew.“ For this reason a detailed justification of the method will be omitted here. Instead, a brief explanation of the method and its adaptation in the context of this study will be given.

The origins of PAR can be traced back to Lewin (1946). His ideas can be seen as an attempt to bridge the gap between science and practice. He complained “research which produces nothing but books is not sufficient“ Lewin (1953, 280). This statement expressed very well the researcher‘s own feelings during the review phase of the PRAP programme. It was a regional research programme that had practically no impact on farmers as the ultimate beneficiaries of development efforts. Thus, looking for ways to bring together science and practice became a key issue of this study.

Lewin (1946) described action research as advancing in a spiral of steps, each of which comprises planning, action, and evaluation of that action. In this step-by-step approach, insights into complex situations are obtained. With each step insights gradually build up. The research process is determined by stages of information gathering, planning of actions, evaluation of those actions and re-planning for a new cycle in the light of the insights that were gained in the previous cycle of the spiral. After re-planning the process


continues in a new cycle with modified actions and evaluations. The concept is illustrated in Figure 4.

Figure 4: The action research spiral

Source: modified from Kemmis and Mc Taggert 1988,11.

The action research cycle is in fact a learning cycle. In this context learning occurs at different levels. BAWDEN (1988, 6) calls this “double loop learning: learning about learning about things“. A researcher tries to find out something about certain events or things and may use a certain ’window‘ for ’finding out‘. Reflecting on the way of finding out represents the second level of learning.

Hagman et al. (1998, 228) also takes up the idea of learning in cycles and even speaks of triple loop learning. Within a learning cycle several sub-loops can be drawn. Starting at field level, findings will help to revise existing concepts (conceptual level) and these in turn may have an influence on the organisational set-up of an institution (institutional level). Different levels of learning were relevant in the analysis of the Fiji AKIS. In this line of reasoning, the researcher suggests to speak of multiple-loop learning. In addition to the three levels mentioned above, the managerial/organisa-tional level was added. New concepts have an immediate influence also on management and the way action is organised. Taking an example, the loops could be compared to a spider web. If one pulls at any point in the web, changes will be felt all over the net. The same applies to the different loops. In the case of the study, the PRA findings in the field had an influence on the concepts on how field work should be done. It also had implications for management on the question of how things should be organised and finally it also had an influence on the institutional set-up of the organisation. The model is depicted in Figure 5. Depending on the subject researched, multi-loop learning implies that different other levels (or dimensions) of a given study case may be investigated in a single research cycle. In this respect the model helped to draw connections between important research variables.

Besides the analysis of the AKIS, this type of research or multi-loop learning mode was relevant during the phase of the PRA training course. While the practical subject of the PRAs was to investigate farmers‘ problems and their interaction with the Ministry, the second level of learning investigated the best ways of teaching PRA tools.


Figure 5: Multiple-loop learning in action research

Source: modified design after Hagmann et al. (1998, 228).

An important feature of participatory action research is its dialogical nature, which enhances participation. Several actors may be involved actively in the research process. Ehret (1997, 35) explains what creates participation: “Discussion with the clientele, iteration, triangulation, verification - all are essential parts in PAR for obtaining information. Thus, participation is not an add-on; it is the basis of research. The style of co-operation and the attitude of partners determine their interaction and the joint search for improvements. There is no ’subject‘ which has to be researched, but a partner who is accompanied in a situation-improving process". As one focus of this study was directed on improving research and extension services, the main research partner was the Ministry. But also other actors played particular roles in gaining information and discussing findings and implications. In this respect, particular farmers and project experts should be mentioned.

Another important aspect of the methodology is that it is capable of recording change. This is of particular importance for the analysis of a dynamic subject such as an agricultural knowledge system. Using learning cycles in timed intervals permits the investigation of changes in this dynamic system. In order to better understand the reasons for the current situation and also to identify potentials for future development, such an approach requires an analysis of the past. In other words a static ’photographic shot‘ of the system can only catch a certain moment in time. To better understand the direction of progress of the system, several such shots are necessary to gain a reasonable insight and make a prognosis on how the system will or should evolve. This was attempted in this study in several ways. The development of agricultural production and exports were used as an indicator for the efficiency of the Ministry to develop new commodities and increase production. Developments were reviewed in the period 1976 to 1998. The development of staff and resources were reviewed in the period 1984 to 1998. This allowed the investigation into which extent staff and resources were adequate in enabling


good services to the Ministry. Changes in the working approaches and methodologies used by research and extension were recorded at the beginning of the field phase in autumn 1994 and again at the end of research phase in 1998. This long period of observation enabled changes to be traced, which are not visible in short term observations. Thus, it allowed accurate judgement on the impact of participatory methodologies that were promoted by PRAP and other donor projects<24>.

3.2 The role of the researcher

In PAR, the research process is influenced by the relationship to the clientele and the tasks at hand. In this respect the researcher‘s own personality has an influence, too. The researcher is required to keep close contact with his partners in research. Therefore, it is almost impossible to eliminate subjectivity. The researcher is part of the process and thus his personality influences the research process and its outcome.

Before, during and after the field phase, the researcher held different positions that were linked to his professional career and employment status. A description of the roles and their influence on the research process is given here.

Before the field phase the researcher worked for two years in the EU Delegation. His post of donor representative enabled him to meet with Government representatives and the top-level management staff of other institutions. In this period, first contacts with management at MAFF were made, which facilitated later work in the Ministry. Through this development a good overview about the country and its institutions was gained. The research design for study was developed in this period.

Next, the researcher held the position of interim PRAP Programme Co-ordinator (11.94 -4.95). This post required, in particular, leadership and negotiation skills. Leadership was necessary to run the programme with its eleven components and 14 experts. While, negotiation skills were needed to merge the interests of the donor with those of the various national and regional partner organisations. In this period, the research design was finalised, a first workshop was conducted, and the terms for the PRA training course were negotiated with the Ministry.

During the core field phase (5.95 - 12.95), the researcher worked as consultant attached to PRAP P11. Now, full attention was given to the organisation and implementation of the research activities. The core task was the PRA training course. The task required the researcher to perform several roles or functions: trainer and teacher, resource person, facilitator and moderator and, of course, researcher. The trainer/teacher role was evident for the PRA training course. The role as resource persons related to specific areas of competence and experience. These were useful for the PRA team members, but also in particular during the discussions with farmers. The differences in experience and knowledge between the researcher and the main partners (farmers, researchers, extensionists) were an asset for discussions. As information was shared, discussions became more meaningful and practice oriented. The role of facilitator and moderator was


required in particular during the PRA fieldwork. ’Handing over the stick‘ to the farmers required staying behind and enabling them to express their own views. Also during the preparation and analysis stages, the researcher tried to facilitate rather than to teach in order to enable more learning by self-discovery of the team members. Lastly, the actual role of ’the researcher‘ was to keep an overview, plan the activities, and to observe and analyse the on going processes. An important aspect of the research role was the systematic documentation of the process and its findings.

In the period after the field phase, (1.96 - 7.2000) the researcher‘s role can be best characterised as that of an external observer and analyst. The activities in Fiji were followed via project progress reports, internet publications<25> and email correspondence with former colleagues. These activities highlight the role of the observer. The aspect of analyst refers to the intensive literature studies and the analysis of all the data collected in the field which in turn provided new insight and room for reflection.

Working in these different roles and positions made it easier to understand other actors and to slip into other roles. It increased the awareness that the position affects the rational, worldview, context and behaviour of any actor in a knowledge system. Changing roles facilitated the understanding that ones current own perception is only just a tiny part of reality. This again helped to become more aware of one‘s own subjectivity, which in turn increased the tolerance and interest for other views and ideas. But the different roles held, also created some irritations. In particular at the beginning of the field phase, partners were not always clear about the role of the reseacher. It was surprising for them that the ’donor representative‘ or ’program co-ordinator‘ suddenly worked on the field level. Overall, the different roles held permitted a deeper penetration into the subject and enabled a more comprehensive understanding.

3.3 Chronology of research activities

The methodological aspects of the research activities are discussed in two steps. At first, the main research activities are discussed in chronological order. All activities are explained briefly, and reasons for a better understanding of the sequence of the research activities are given. In a second step key methodological aspects of selected research activities are conferred in detail. Particular reference is made to the following points: workshop design and composition of participants in the workshop, a training course in participatory methods, questionnaire design with survey units and sample size.

The research design and problem definition for this study were formulated earlier in 1994. The practical field implementation started with a workshop to examine linkages, and technology development and diffusion at MAFF in November 1994. The workshop result indicated a clear lack of participatory methods at MAFF. For this reason, a training programme for participatory methods was set up. The training programme started in May 1995 and ended in December 1995. This constituted the main field phase. The training


course was devided in three cycles starting with an RRA and followed by two PRAs. Each RRA/PRA was divided into the phases, preparation, field work, separate analysis, joint analysis and write-up of report. During the training course, in September 1995, a second workshop was held. This workshop focused on an analysis of the information flow at MAFF. A training report and detailed reports on all RRAs/RPAs were published shortly after the researcher left Fiji in 1996. Then, the analysis of the field phase started. In a detailed literature review the researcher became aware of the latest AKIS concepts suggested by Röling 1995 and Engel 1995. This led to the formulation of the platform model as presented earlier in chapter 2.2.4. A new research topic (RT 4) on platforms was added to the original research design. Necessary questions were included in a follow-up questionnaire that was distributed in Fiji in November 1997. Besides testing the platform model, the purpose of this follow-up questionnaire was to assess changes in the working methods used at research and extension and assess the impact of the training course of participatory methods. The follow-up questionnaire was analysed in 1998 and the present study was compiled in the following two years. The chronology of research events is summarised in Figure 6.

Figure 6: Time chart of research activities

In the following paragraphs important methodological aspects of the main research activities are conferred.


Investigation of the Fiji AKIS

In chapter 2.2.4, the procedure to analyse the AKIS in Fiji has already been outlined in broad lines. Some additional information shall be provided here to specify the different sources of information used. For the examination of the core actors in the Ministry Research and Extension Division, project reports and in particular the Ministries annual reports and other publications were used. These findings were complemented by own observations and informal interviews with Ministry staff and other project experts during the field phase. The review of the role and problems of farmers in the AKIS relied mainly on the results of the RRA and PRAs. Workshops on linkages and information flow investigated the other actors who played significant roles.

Table 5: Composition of
workshop participants

Group of participants

Workshop I

Workshop II

Top management












Other divisions



Project experts



Observers, others






Linkage and information flow workshops

The first main field activity focused on the analysis of existing linkages and information flow in the Ministry. Two workshops were convened to gain insights into this subject. The first workshop on linkages was held in Suva in November 1994 and the second workshop was held at Koronivia research station a year later, in September 1995. One day was allocated for the first, and two days for the second workshop. The composition of participants in the two workshops is shown in Table 5. In total, 35 respectively 39 participants attended the workshops. Attention was placed on a good representation of all MAFF divisions and management levels. The first workshop had a stronger emphasis on the core divisions research and extension, while the second workshop had a higher representation of top management and the section for Economic Planning and Statistics (EP&S). In addition, 6 observers mainly from regional and educational institutions were invited to broaden the range of actors.

In relation to the total staff numbers in the various divisions, about 15 % of the permanent staff were present. Top management was represented by about 1/3 of all members in the second workshop. The six project experts in the workshop corresponded to about half of the expert community working with MAFF at the given time<26>. Thus, the composition of both workshops can be rated as a good cross section of the different actors in the AKIS. Working in small groups of 4 to 6 participants combined with plenary discussions made it possible to analyse linkage and information flow problems in depths. The group results were documented on pin boards and the researcher noted, every evening, observations and main contributions during the discussions in a mind protocol<27>.


Training course in participatory methods

The training course had tree main objectives:

Reaching these objectives required the preparation of a training programme, the moderation of the training sessions, the observation of the training progress and the actual implementation of the field work. Dealing with these multiple tasks simultaneously required quick decision making and a frequent re-planning of activities. This was facilitated by a constant monitoring of the progress of the team on a joint plan of operation that was visualised on a wall in the office. This also helped to keep other partners up to date with our programme.

Table 6: RRA and PRA
cycle length in weeks









Field phase




Analysis 1




Analysis 2




Total weeks




The RRA/PRA training course was designed as a ’train the trainer‘ course. The idea was that the researcher would train a number of 6-8 MAFF staff that would become the key resource persons for PRA and train other MAFF staff in the future. As a training method, it was considered best to have a practical ’on the job training‘ concept. This meant that participatory methods should be taught by repeating and applying the tools several times. To improve linkages between the divisions, it was planned that staff from all major divisions of the Ministry should be included in the course (Research (RD), Extension (ED), Economic Planning and Statistics (EP&S) and Animal Health and Production (AH&P)). The selection of suitable candidates was left in the responsibility of the Ministry. This, however, created a few initial problems. Although the training course was agreed on by MAFF four months before the start, the Ministry had difficulties in selecting the relatively high number of staff needed<28>. Finally, the Ministry could send only four staff members: extension (2), research (1) and EP&S (1). These trainees will be further referred to as the PRA team. For the entire training period the PRA team was located at the main research station in Koronivia.

The training course was split in 3 rounds, a first RRA and two following PRAs. Each round constituted a separate learning cycle that permitted recommendations for the following round. The length of the three cycles is depicted in Table 6. The first RRA had a medium length with 10 weeks, the second cycle was the most detailed one with 17 weeks and the last PRA had an emphasis on ’rapidity‘ and was completed in 3.5 weeks. While all cycles had a similar field phase with 0.75 to 1.5 weeks, preparation and analysis times differed strongly. The first and the last cycle had short preparation periods


(1,5 weeks), while PRA 2 was prepared in more detail (4 weeks). Analysis for the first two cycles required 7 to 12 weeks. It was grouped in two almost equally long phases of separate and joint analysis. The rationale of this division was to enable a better judgement on the abilities of the PRA team members and in particular to test their ability to work independently. In practise, analysis consisted of a short 1-2 day joint discussion of the main field findings and an agreement on how to analyse the information was made. Then, the researcher left to Germany and the PRA team worked independently for a period of 4-5 weeks. Upon the researchers return, the results of this separate analysis were then compared and a joint report was prepared. The reports were presented in half-day workshops to MAFF staff and other interested experts.

The preparation and implementation of the RRA and PRAs followed a structure of 16 steps proposed by Nagel et al. (1989, 7). The steps cover various aspects starting from problem definition, across research design, implementation and analysis, to utilisation of results. The sequence of steps was originally designed to implement formal surveys with the help of RRA. As a working hypothesis, it was assumed that the sequence might be considered as universal to implement any type of data collection process such as RRA or even PRA. During PRA 2, the sequence was tested completely. In the first and the last cycle, some steps were omitted to save time.

To obtain feedback on the training measures, an evaluation questionnaire was filled out by all PRA team members after each step and at the end of each training cycle. In addition, the researcher kept a daily diary to record the activities carried out and observations that emerged in the process.

During the field work of the three RRA/PRAs, an estimated total of 148 farmers in 20 locations participated in the exercises. In the first RRA, 33 farmers were consulted. In the following two PRAs another estimated 64 and 48 farmers consecutively participated in the PRAs. All PRA activities were facilitated as group exercises. To investigate differences among farmers, three types of groups, average, advanced and women groups, were formed. Group sizes varied from location to location consisting of between 3 to 8 farmers. Sometimes farmers joined the group spontaneously or left prior to the full completion of exercises. For this reason the total number of farmers could only be estimated<29>. This large number of farmers combined with the grouping, permitted a good insight into the diversity of farms in Fiji. At the same the time, the field phase required intensive co-ordination with the field extension offices of the Ministry. This, together with many informal discussions with the PRA team members, permitted the study of the interaction farmers on the one side and the Ministry on the other.


The above experiences during the PRAs, together with a study of existing documents, formed the basis for the individual analysis of the research and extension divisions of the Ministry. These results were compared with the findings of a formal questionnaire distributed at the end of the research period, three years after the first workshop. The


purpose of this follow-up questionnaire was to gain information for two lines of inquiry. First, its function was to trace the changes in the AKIS due to the various project interventions and in particular the training course on participatory methods. Second, the model of platforms for innovation development was presented with a short paper and a discussion of the concept was stimulated with a range of open questions<30>. The main purpose here was to evaluate, to which extent, this model might be suitable to reduce existing linkage problems in the Ministry. The questionnaire was distributed to five groups: research, extension, top management, project experts and the trainees in the training course on participatory methods. To account for the different working approaches used in the different divisions, the questions were formulated specifically for each group. The questions regarding the interaction and linkages between the divisions were the same for all groups. The set of questions on platforms was equally identical for all groups. The length of the questionnaire was adjusted so that it could be filled within a 1 to 1½ hour time period.

Table 7: Questionnaire sample size







in %

Top management












PRA team




Project experts








The questionnaire was sent via email to Fiji and copied locally by PRAP P11 and then distributed within the Ministry in December 1997. To secure the anonymity of MAFF staff, the questionnaires distributed did not ask for the name of respondents. Only the division, position and function were recorded. The questionnaires were recollected by P11 and returned to the researcher in February 1998. In total, 60 questionnaires were distributed of which 41 were returned. Table 7 highlights the sample size and distribution of respondents. The average rate of returned questionnaires with 68 % may be considered as good. All members of the PRA training course replied and also extension staff had a very high rate of replies with 88 %. Two thirds of experts and half of the researchers replied. The number of replies from top management was disappointingly low with only 2 questionnaires returning. Despite several reminders, no further questionnaires were returned<31>.

The information was analysed in a spreadsheet program using descriptive statistical methods (means, averages, and percentages).

It may be asked why the various divisions were not interviewed with a questionnaire at the beginning of the field phase. This point was considered, but at that time another project (SCEP) had already started a questionnaire survey to review in particular the research and extension division<32>. In order to avoid duplication of efforts and not to hinder


the few Ministry staff with yet another survey, the researcher decided to refrain from distributing a parallel questionnaire.

Case studies

Four case studies were selected to illustrate the work of the Ministry with the help of a few practical project examples. The cases are examples of projects that were implemented by MAFF in co-operation with different donors. Each case examines the following elements: brief history, project goals, research and extension approaches, farmers‘ participation, configuration of actors involved, and a comparison of the main strengths and weaknesses.

The first case study is an example of the Ministries approach during the 1980s. It shows a large scale project with a straight “transfer of technology“ approach. The other three cases describe projects that applied participatory methods for technology development and diffusion. As the first case study is an older MAFF project, the assessment was based on discussions with former staff, visits to the project area and available reports. The three other case studies were all EU funded projects. The researcher was familiar with these projects through several field visits during implementation in the role of donor representative and later as program co-ordinator. Internal project reports and discussions with the project experts and other local project staff provided complementary information.

The case studies were also used to validate the platform model. The assessment focused on the question to which extent the projects, seen as platforms, contributed towards innovation development and diffusion. In this respect the study cases complement the findings on platforms derived from the follow-up questionnaire.

3.4 Tools and data quality

Given the complexity of the subject studied, it is evident that a proper analysis could not be achieved with a single or standardised method only. Studying the Fiji AKIS and all the interaction of several actors required a comprehensive approach with a systematic combination of various tools and methods.

To explore something unknown requires an open mind, flexibility in the concept and step-by-step progressing. Qualitative research tools, such as PRA tools and in particular open approaches to questioning, are ideal for such explorative research. It is not the aim of this study to describe these tools here in detail. Descriptions on the tools and manuals on PRA are already published in considerable numbers<33>.

Qualitative research tools can produce a wealth of information. At the first stage much of this information is descriptive. Therefore, the information may be difficult to analyse and many researchers of the positivist fraction consider it as less reliable, only case specific and generally subjective. To improve the reliability of qualitative information, it is


necessary to cross check the information gained with one tool or method with other tools or methods. Triangulation is such a systematic method of cross checking. During the training course a graphic tool to display triangulation was developed. Figure 7 illustrates the method at the example of this study.

Figure 7: Triangulation of tools and research topics

The research questions or variables on one side were faced with the different tools used for the investigation on the other side. Then all variables are linked with arrows to the respective tools. The chart permitted to control that all research questions (variables) were investigated via several (at least 2) tools or means of investigation. This chart depicts very well the triangular (or even multi-angular) links of research questions and tools. It creates a visual possibility to verify whether there is a good balance between research questions and tools used. Two different types of triangulation were used; methodological and data source triangulation. The first refers to the different methods or tools used. The second checks if information remains the same at other times, other locations or with other


informants. For example in this study farmers‘ problems and interactions with the Ministry were investigated with several tools such as seasonal calendars, maps, life profiles, gross margin exercises or field walks, just to name a few. Data source triangulation included the selection of several locations and of different respondents (average and advanced farmer groups, women groups).

The above approach was one aspect to ensure a good quality of information. Another has to do with the method of participatory action research by itself. The nature of advancing the research process in cycles produced the chance to verify findings from one cycle to the next. An intention was not to extract findings, but to discuss them, derive conclusions and feed these back into discussions again in order to allow sharing and validating of conclusions and emerging new ideas. This approach was crucial for the analysis of the AKIS. Regarding the different viewpoints and tools used, it cannot be concluded that any one perspective is right or wrong, better or worse than another. The many viewpoints and methods create an overall richer picture of the AKIS, and help to gain a better understanding. This richness of picture may be seen as an indirect quality indicator that the recommendations derived are well founded and constitute a good starting point for a reorientation of the Fiji AKIS.



Equivalent to the debate on hard and soft systems thinking discussed earlier in chapter 2.2.2 .


An overview on the support measures of the main donors is provided in chapter 6.1.1.


The internet developed fast in Fiji. Government, the local journals, magazines and the private sector started publishing information in the net. This was a good opportunity to stay in touch with the ’other side of the world‘.


Due to fluctuating numbers, a precise calculation was difficult. For Ministry staff, more details are given in chapters 5.3.3 and 5.4.2.


Own observations and records were complemented by the project reports of the funding projects.


Several aspects added to the problem. The number of staff available at the Ministry was rather low and it was therefore difficult to attribute staff for new tasks. Another problem had to do with the qualifications of staff. As staff with good qualifications were scare, there was a certain concurrence among donor funded project to compete for these counterparts.


The estimate is based on 4 farmers per group, a conservative assumption. The total number of farmers (including part time participation) is therefore even higher. In the first RRA, 12 locations were visited. In the following 2 cycles, 4 more each were visited. Among the 20 locations visited, 12 locations were settlements and 8 villages.


The questionnaire was distributed by email to most experts. The simple reply function of these email programmes allowed, in several cases, for discussion as the questionnaires went back and forth several times.


Work overload of management may be the main reason for not replying.


The results of the report on the Research Division were also used as additional reference in this study (cf. Dowling and Kumar 1995). The report on the Extension Division was not yet completed when the researcher left Fiji. Later the report was classified as confidential so that it became impossible to obtain a copy.


A selection of manuals and trainer guides have been published in the meantime (e.g. Theis and Grady 1991, Leurs 1993, Schönhut and Kievelitz 1994; Pretty et al. 1995).

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.

DiML DTD Version 2.0
Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML - Version erstellt am:
Thu Mar 29 16:23:51 2001