After selecting the three clusters, the identification of the population size and actors to be surveyed and interviewed relied on the criteria defined by network analysis and
‘snowballing techniques’. The ‘snowballing technique’ is about asking participants and interviewees to identify possible actors that fit the selection criteria and can participate in the research (Ritchie et al., 2003). Actors were selected on the basis of their participation within the CFA and their contribution to the technology transfer in each nucleus.
According to Wellman (1997), network analysts:
‘[…] must define the boundaries of a population, compile a list of all the members of this population, collect a list of all the direct ties (of the sort the analyst interested in) between the members of this population, and employ a variety of statistical and mathematical techniques to tease out some underlying structural properties of the social systems’ (p.26).
Thus, the researcher sets a limit on the network actors and relationships. By following this criterion, the indefinite collection of data is avoided, as well as the necessity of manipulating large amounts of information.
In this research, the limits of the population were defined by what this thesis calls the
‘third level network boundary’. As illustrated in Figure 4.2, network boundaries were defined by three levels of actors and relationships: the first level refers to participants of an agreement, in this case, all smallholders and leading change agents belonging to a
nucleus, and the connections between them; the second level includes the contacts to which the actors in the first level were connected, and the connections between them;
and the third level refers to the contacts to which the actors in the second level were connected. It was required that the total population of smallholders, leading change agents within each nucleus (first level), and other actors who demonstrated a pivotal role in diffusion, participated in the surveys.
Figure 4.2: Network levels for data collection
Source: Own elaboration from survey data and interviews
The preliminary analysis of surveys, together with the snowballing technique, were used to complete the list of actors to be interviewed and surveyed. A diverse universe of actors who were able to influence technology diffusion included: 1) small-scale growers40; 2) medium and large-scale producers41; 3) anchor companies; 4) farmer organisations and alliances between agribusiness companies; 5) academic and research entities; 6) government institutions; and 7) social organisations. The diversity of participants helped to achieve a range of responses on the interview questions and to triangulate the information provided by interviewees and survey respondents. Table 4.2 shows the distribution of the total number of surveys and interviews amongst the participants.
40 See page 81 for the definition of small-scale producers that was used in this thesis.
41 See page 81 for the definitions of medium and large-scale producers that were used in this thesis.
Table 4.2: Number and distribution of surveys and interviews conducted by type of actor
Type of actor No. Survey No. Interviews
Case study: Nucleus A
Source: Own elaboration from survey data and interviews
As can be seen in Table 4.2, 120 small-scale producers were surveyed (out of 2,331 smallholders that were located in the Central zone) and, from this population, 15 were interviewed. This selection of survey participants was guided by the type of nucleus and small-scale producers required in this research, namely clusters in which agro-industry companies had technical relationships with their fruit providers and in which producers were willing to adopt technical practices addressed to deal with the epidemic. It was previously explained that the epidemic had a significant impact on small-scale producers of the Central zone who, in several cases, stopped adopting crop and PC-related practices
and seeking technical assistance (p.113-114). This was mostly the case of producers belonging to the towns where ICA announced the official declaration of a plant disease epidemic (see p.9). Contrary to that case, we selected 120 small-scale producers that: 1) were located in towns of the Central zone with a latent risk of epidemic, 2) were still able to treat and control the disease in their plantations, and 3) were receiving regular and significant support from agro-industry firms to deal with the crisis, such as the assistance provided by Anchor Companies A, B and C.
Based on the preliminary SNA of the survey feedback and the interview responses given by medium and large-scale producers, representatives of anchor companies, and leaders of farmer organisations, it was possible to select a group of 15 smallholders for the interview stage. Of these, 3 were producers with a high number of network links, 3 were producers with a low number of links and 9 were producers that were identified to be playing a crucial role in the decision-making processes of technology diffusion.
Interviewed smallholders and technical assistants of agro-industrial companies were the main source of information and data. Amongst other interviewees were researchers and staff of Fedepalma and Cenipalma, which lead the development and the transfer of agricultural practices for PC management in Colombia. We also interviewed the head of what this thesis called the ‘UG alliance’42, which was the regional strategy developed by some anchor companies to deal with PC disease in the Central zone and included the anchor companies leading the technology transfer in nuclei A, B and C.
This study employed primary and secondary sources of information, as well as qualitative and quantitative data. The primary source data were gathered from interviews and surveys that were undertook in two stages of fieldwork, from March to July 2015. Most information was collected from primary sources such as surveys and semi-structured interviews based on specific oil palm clusters, and, to a lesser extent, observations from
42 The real name of the alliance is not disclosed in order to comply with confidentiality agreements that were signed by the author with the companies of this study.
meetings. This information was complemented with secondary resources such as articles, academic documents, handbooks, sectoral and governmental documents, online news, reports, and statistics of the agribusiness that were mostly collected from public websites of sectoral organisations.
An additional reason that justifies the combination of methods in this work was the need for verification or triangulation. According to Creswell (2013), Mabry (2008) and Yin (2013), triangulation enables a researcher to enhance validity and check the accuracy of the data collected through different methods (triangulation by method) and sources of evidence (triangulation by data source). In short, this research ensured the reliability of the data collected by using qualitative and quantitative methods. Quantitative data, for example, was cross-checked during interviews and meetings with other actors.
The strategy to reach actors mainly involved the support of Anchor Companies A, B and C and agro-industrial organisations. Cenipalma acted as the intermediary between the researcher and the three anchor companies. These companies then helped to reach the majority of smallholders and provided logistical support to access the farms of producers efficiently. The strategy to reach the rest of the interviewees mostly involved telephone and email contacts.