37 | P a g e analysis). This was then followed by three case studies (exploratory analysis), a survey data in Zimbabwe. Lastly model specification (and verification of the specified model using collected data). This study followed this order in order to firstly establish a learning curve from Japan’s experience before understanding the situation prevailing in Zimbabwe and estimating a new cooperative model. The research also utilised triangulation methods in the data collection process. “Triangulation is the application and combination of several research methodologies in the study of the same phenomenon” (Denzin, 1978, p. 261). By combining multiple observations, theories, methods, and empirical materials, this can help to overcome the weakness or intrinsic biases and the problems that come from a single method, single-observer, and single-theory studies. Thus, the purpose of triangulation is to obtain confirmation of findings through the convergence of different perspectives.
38 | P a g e From the available seven districts in Mashonaland East Province, the study purposefully selected Goromonzi district. The sampling unit is an entirety of similar items or a group of persons that have one or more characteristics in stock that are of interest to the researcher. It is one into which an aggregate is divided for sampling purposes (OECD, 2001). The sampling unit was taken to be any household that was using the services of a cooperative, i.e., registered or unregistered active members. Many farmers belong to different types of cooperatives in Goromonzi with some of them in informal marketing and supply groups (sometimes of four or five households). These several individuals made up a complete collection of research units or objects that collectively formed the research domain for generalisation of my findings.
In stage two, to determine the statistically significant sample size; a 95% confidence interval, a 10% margin of error and a household target population of 5034 CA and 430 A1 were keyed into the Qualtrics online tool (Qualtrics, 2018). The research uses a margin of error equal to 10% instead of 5% to reduce the time and financial costs of data collection, as mentioned. At this margin of error, the online tool recommended that the survey should select a minimum of 95 CA households and 79 A1 settlement model households as a representative sample of the population. From a list of farmers provided by the Goromonzi extension officers within the selected CA and A1 farms, the study used non-probability sampling9 to select 100 farmers in each farming/settlement model who were members of a cooperative for questionnaire interviews (the study dropped eight households from the A1 because of incomplete data). I used non-probability sampling method in this stage of the multi-stage cluster sampling to minimise monetary and time expenses. It enabled us to identify (from the extension worker provided list) farmers who conformed to my desired criterion. Non-probability sampling may be the only feasible alternative, in particular, if the relevant population remains vague and hard to define (Blumberg, Cooper, & Schindler, 2008; Cooper & Schindler, 2013, pp. 358-361). The implications on research bias of utilising non-probability sampling came in the form of reduced dependence on the probability theory and hence reduced overall inferences about the entire population (Guo & Hussey, 2004, pp. 10-13). To strengthen the data for triangulation, I additionally interviewed eight key informants, including ministry officials and leaders in the national cooperative movement. These cooperatives were producing a wide range of products
9 A sampling method in which sample elements are chosen not based on predetermined probability but the research purpose and resources available (Guo & Hussey, 2004, p. 2)
39 | P a g e from eggs, horticultural crops (tomatoes, vegetables and potatoes) to dairy and poultry products while some were in input or equipment sharing cooperatives.
2.3.2 Sampling in Japan
Japanese data collection presented challenges of its own, given the fact that Japanese society is a ‘closed society’. For farmers to open up to a foreigner and answer questions about such personal things as income and socio-political issues that they go through proved to be complicated. The researcher had to activate various connections, with the help of the academic supervisor in order to find cooperatives that would agree to discuss their cooperative activities.
The researcher had to build an outstanding amount of rapport before being allowed to carry out research. Therefore, the sampling design for Japan involved purposive sampling mixed with snowball sampling to choose the cooperatives to study as well as the respondents to interview within the cooperative movement.
Given the fact that the research sought to develop a learning curve from the study of the Japanese cooperative movement, the selected three different types of cooperatives. Learning from the extensive literature review carried out, the sample had to include i) a larger-scale cooperative whose contacts with the JA-Zenchu10 (state) cooperative movement was healthy and active, whose ideology was in line with that of the JA-Zenchu cooperative movement; ii) one small-scale cooperative movement that was independent from the JA and whose ideological standing was different from that of the JA-Zenchu, and iii) one small-scale cooperative that was relatively new and had higher levels of independence. These specifications represent structural differences in cooperative organisation. From the purposive sampling method employed, I sampled 1) Sanbu Yasai Network, 2) JA Green Ohmi cooperative and 3) Nose-Yotsuba cooperatives. The research site of Sanbu Yasai Network was the furthest away from the researcher. The initial plan had been to study a cooperative in the rice-producing region of Niigata prefecture, but the study could not secure good rapport in that region, and no permission was secured from the cooperative leaders on time. However, the JA Green Ohmi became an excellent substitute for JAs in Niigata.
10 Central Union of Agricultural Co-operatives (JA-ZENCHU) is the apex body or national level association that represents agricultural cooperatives across Japan in matters of extension, improved living conditions, auditing, policy representation and public relations (Esham, Kobayashi, Matsumura,
& Alam, 2012, p. 944)
40 | P a g e 2.4 Data collection: Instruments and techniques
A data collection instrument refers to devices used to collect data (Seaman, 1991, p. 42). In this research, data was collected using questionnaires and interviews. Primary and secondary data collection was done in both Japan and Zimbabwe. The type of data that has not unprocessed or unanalysed is called original research primary data and is extremely rich with information for researchers. Data from this method is original as the research instruments are designed for that particular research and administered directly to the respondents making it more authentic. Secondary data, in contrast, is information or data that already collected and recorded by someone else, usually for other purposes (Cooper & Schindler, 2013, p. 130).
Secondary data saves time and money since it is readily available in libraries and online (sometimes for a fee). Secondary sources of data used included textbooks, publications, journals, government/private sector databases and reports in both Japan and Zimbabwe.
Secondary data collection was mostly utilised in the case of Japan.
2.4.1 Questionnaires
Primary data was collected from Goromonzi using a comprehensive research questionnaire.
The questionnaire contained a list of open-ended and fixed response questions in order to save on time and resources. The administered study questionnaires focused on establishment, management, production, profitability, economic viability, governance, policy environment, access to market and most importantly, independence and sustainability of cooperatives.
Structured questions are also useful in avoiding irrelevant answers, and they make analysis easier. The questionnaire was designed based on a tried and tested questionnaire from the SMAIAS. Given the limited amount of resources, the study had to employ extension workers as research assistants. This approach cut down on expenses such as accommodation and transport usually paid to research assistants that are not based on the research sites.
Furthermore, extension workers had motorbikes and could collect my information while carrying out their extension duties at the same time. The instrument had a section that collected phone numbers to enable us to make follow-ups and phone calls to clarify some sections during the questionnaire checking and data cleaning processes.
2.4.2 Interviews
While the study mostly utilised interviews in the Japanese data collection, the study employed them in Zimbabwe as well. A structured (closed style) or unstructured (open style) interview
41 | P a g e involves dialogue between two people, where the researcher asks questions, and the respondent provides the answers (Frey & Oishi, 1995, pp. 1-2). Interviews obtained detailed information on the extent to which the land reform and other socio-economic and political policies in Japan were implemented, under which conditions, reasons for its success, significant challenges encountered and how these challenges were overcome to achieve set objectives. Both open and closed-ended interviews were carried out in Zimbabwe and Japan.
In Zimbabwe, structured and unstructured interviews were carried out with stakeholders in the agricultural policy field as well as with key informants. The research utilised both structured, and unstructured interview methods and a total of 10 interviews were carried out as shown in Table 2.1. The interview with key informant 2, who is an expert scholar was carried out via skype video call while the rest were one on one interviews.
Table 2.1: List of interviewees and their affiliation Interview
type
Position or role in the cooperative movement Number
Structured interviews
Current cooperative registrar, 1
Assistant cooperative registrars 2
Former cooperative registrar 1
Chairperson of Central Association of Agricultural Co-operatives (CACU11)
1 Head of cooperatives in the ministry responsible for cooperatives 1 Key informant 2- Expert scholar on cooperatives 1 Unstructured
interviews
Leader from two different cooperatives 2
Key informant 1- Independent co-operator 1
Source: Own study (2018)
The interview with key informant 2, who is an expert scholar was carried out via skype video call while the rest were one on one interviews.
Choosing a data collection technique and tools for Japan was quite complicated. Given the limited Japanese language skills that the researcher possessed, it made more sense to utilise a structured questionnaire to collect data. However, most of the questions that are easily obtainable through a questionnaire are readily available (though in Japanese language) on several other online and physical platforms. Thus, to get more in-depth information about the
11 The Central Association of Cooperative Union (CACU) is the Apex body responsible for representing all cooperatives at the national and regional levels. It has its headquarters in Msasa in Harare and is currently chaired by Mrs Maziva. It is the Zimbabwean equivalent of the JA-Zennoh in Japan.
42 | P a g e current and trajectory of the Japanese movement, the study had to employ interview techniques of collecting data. Many of the interviews carried out with the actual farmers or members of three different cooperatives had to use structured interview guides to direct the flow of the interviews (Picture 2.1).
Picture 2.1: Field interviews with a small-scale cooperative member, Chiba, Japan, 2019
Source: Field research photos taken by author, 2019
The technique was more flexible than the questionnaire, which needs clarity if respondents are unclear about specific terms. With the help of a translator in some cases to avoid losing information or meaning lost in translation, the researcher carried out 24 in-depth interviews in Japan with stakeholders knowledgeable in the field and history of Japanese land reform and agricultural cooperative movement. These included i) 3 leaders from each of the three different cooperatives, ii) an expert and scholar of cooperatives, and iii) 20 interviews with farmers or members of the three different cooperatives.
2.4.3 Participatory and observatory data collection
In the initial stages of the data collection process in all three research sites in Japan, the researcher had to undertake participatory research through observation. In Nose farm, the researcher did an internship at the farm, and the data collection process included going through the everyday lives of the cooperative members and the cooperative workers. The reason to employ this method was partly because of low Japanese language skills of the researcher as well as limited resources to hire a language interpreter. After three weeks of participatory research, the study then managed to secure structured interviews. The same situation was
43 | P a g e obtaining in Chiba prefecture. The researcher utilised the host family system, and the family spoke excellent English. Thus, the researcher agreed to provide his labour services to the host family in exchange for accommodation, food and research interpretation services from the host family.
2.4.4 Reliability and validity of research instruments
I have clearly defined and described my methodology to improve the reliability and validity aspects of my research design and technics. In research, reliability and validity of instruments refers to how accurate an instrument is in estimating a variable (Polit & Hungler, 1995, p. 296).
Given the fact that the research sought to develop a cooperative model for Zimbabwe’s agrarian cooperatives, I received various forms of assistance from experts in research such as from the SMAIAS which gave much input especially with the designing of the research questionnaire.
After developing the questionnaire, it went for pre-test using a total of eight farmers.
Information from the pre-test was brought back to the desk, analysed and used to improve the questionnaire before final administration was carried out. Although this method increased the cost of research, it was necessary to improve the validity and reliability of the questionnaire.
The same procedure (formulation, pre-test, improving and then administration) applied for the structured key informant interviews, which had a standardised interview guide for the three different cooperative movements in Japan. Study result reliability is important because it ensures that the same results would be obtained if the study were to be replicated by other researchers using the methods.
2.4.5 Research unit of analysis
In Zimbabwe, several reproduction, economic production and accumulation processes occur at the farming household level (Scoones, Marongwe, Mavedzenge, Murimbarimba, & Sukume, 2010, p. 57). The household has decision-making powers, and its miscellany in gender, age, socio-economic interests, political influence and economic status (Matondi, 2001, pp. 67-68) makes it ideal for our study. The same is somewhat true for Japan where most small family-run farms (commercial and non-commercial) make decisions at the household level.
Additionally (most importantly for the case of Japan), most social organisations in the rural areas admit members as a household and not necessarily as individuals. Thus, a household (father, mother and children) will enjoy benefits from a cooperative joined by the father (or mother) as the head of the family. This factors enabled the study to set the research unit of
44 | P a g e analysis as the household engaged in agricultural activities. This made the household as the most appropriate unit of analysis for my study.