PART II CASE-STUDY: GUINEA-BISSAU
5. Research methods and organisation of work
5.5 Fieldwork (II): Village-level case-studies
5.5.3 Logistics and pre-testing
In addition to coming to a final decision regarding the survey sites and making arrangements with respect to lodging and transportation, the main preparatory task prior to the actual undertaking of the survey in April-May 2010 consisted of recruiting the interpreters/assistants, which was made through the mobilisation of my local network of acquaintances in Bissau. I sought to recruit two assistants/interpreters (to assist me in each of the two survey sites) who were native speakers of the Manjaco and Fula languages (respectively), spoke fluent Portuguese (or English or French, so that they could communicate easily with me) and had at least completed secondary education.
I was quite fortunate in this respect, given that both Messrs. Duarte Mendes and Amadu Djalo, whom I recruited, not only proved highly skilled translators and assistants, but were also able to provide me with their views on many aspects of the history and culture of their respective regions and ethnic groups. Mr. Duarte Mendes is the External Relations Officer in an association called Uno-Tacal, which seeks to document and preserve the cultural heritage of the people of the Cacheu region, as well as promote development projects in that region. A Manjaco himself, he spends half of his time in Bissau and the other half in his home village of Caiomete. In turn, Mr. Amadu Djalo is a member of the Fula ethnic group who hails from a village in eastern Guinea-Bissau about 50 km away from Braima Sori. He attended university education in Dakar and currently lives and works as a French teacher in Bissau26.
Once the preparatory logistics had been addressed, the questionnaire was then pre-tested (nine interviews) in late March 2010 in a third site: Bugudjan, a predominantly Manjaco village located at a distance of about 30 km from Caiomete. This led to several changes being made to the questionnaire before settling upon the final version, which was then adopted in Caiomete. The version adopted in Braima Sori was identical to the one used in Caiomete, except for a few differences, which are discussed in section 5.5.4 (see also Appendices II and III).
26 See section 5.5.5 (below) for a discussion of issues of positionality in the context of the sources of non-sampling error.
108 A thorny methodological issue that arose after the pre-testing concerned the operationalisation of the “household” concept (of special importance, of course, given that this consisted of a household survey). In the village selected for pre-testing, Bugudjan, people sharing the same dwelling also shared the same ‘domestic economy’, in the sense of pooling their resources in order to produce a food and cash output that is then used to ensure the physical and social reproduction of the household (under the authority of the patriarch). These would typically consist of very large households (compound, joint or extended families), including e.g. the patriarch and his various wives, non-emancipated male offspring and their wives, non-married female offspring and other adult dependents.
In both Caiomete and Braima Sori, by contrast, each residential unit often comprises more than one unit exhibiting relative autonomy in terms of production and consumption. This translates into the existence of more than one stove in the kitchen (usually located in the back of each house: see Figure 5.2). The woman (for it is almost inevitably a woman) in charge of cooking in each stove does so using the stock of food that is obtained either directly (grown, fished, hunted or gathered) or indirectly (bartered or bought) through the work undertaken by the people who ‘belong’ to that stove.
Figure 5.2 A backyard kitchen in Caiomete27
This high degree of ‘inter-stove’ autonomy is not limited to the specific aspect of food (which, it should be noted, accounts for a large share of household consumption): it
27 In this case, comprising four stoves that reflect the existence of four separate households within the same housing unit.
109 reflects a broader economic and financial autonomy, whereby each stove allocates the work of each of its members in the realm of production, as well as the resulting food and cash output in the sphere of consumption - ‘inter-stove’ transfers, which are common in certain circumstances, notwithstanding, insofar as they are relatively limited compared to the overall household output. This methodological problem was not identified until I arrived in Caiomete and it forced me to rethink my survey methods, as I had envisaged regarding each residential unit as the best proxy for a ‘household’. In reality, that was not the case: although the stoves are not fully separated in economic and financial terms, it quickly became apparent that they are indeed the most autonomous resource-allocation units and, accordingly, a decision was made to compile a list of all the ‘stoves’ in the two villages and use it as the sample frame. This option was subsequently validated both by the qualitative and quantitative data obtained during the course of the interviews and by the information that that had also been the methodological option made in the context of the 2009 national-level Population and Housing Census.
Phase 1 of the survey (Caiomete) lasted for three weeks in total, during the course of which I lodged in the home of my assistant and interpreter, Mr. Duarte Mendes. The first three days were dedicated to listing all the stoves in the village: 366 in total, spread out over 123 dwellings. It is a large village for Bissau-Guinean standards, in terms of both its population and the area through which the houses are scattered. Given the need to balance the representativeness of the survey with the time constraints, a decision was made to interview approximately 1/5 of the “household heads” (72 in total). The problems with the very concept of a “head of household” was in this case circumvented easily by the fact that, as of the initial listing of the households, the people in each house would immediately and without hesitation indicate who the person “in charge” of each particular stove was. The actual sampling was then undertaken using a systematic random sampling (“step”) procedure: a number between 1 and 5 was randomly selected and used to indicate the first household in the list to be interviewed. The subsequent primary choices for the interviews corresponded to every 5th household in the list after the first one. The expression “primary choices” is used here because in the case of Caiomete (though not in Braima Sori), even after repeated attempts it was not always possible to interview the people in charge of some households that were part of the sample. There were no instances of refusals to participate in the survey, but on a relatively large number of occasions the people in question could not be found at their homes.
110 According to the explanations that were usually obtained from relatives or neighbours, this was mostly due either to those people being out on the fields working, to temporary migration to work in agriculture elsewhere in the region or to temporary absences related to ceremonial practices and obligations. The latter play a very central role in the social lives of the Manjaco: during the two and a half weeks that I stayed in Caiomete, funeral celebrations lasting for at least two days in each case were held for at least four people from either Caiomete or the surrounding villages, and this translated into relatively prolonged absences of people from their homes. Whenever it was possible to find out when it was that the person in question would return to his/her home, I tried to accommodate this so as not to skip that household. However, whenever this failed to materialise and after at least three attempts at contact at different times of different days, or whenever I was told that the person would not be returning until after my scheduled period of time in Caiomete, a decision was made to skip that household and move on to the subsequent household in the initial list. The final result was that, out of the 72 interviews that were eventually made, 51 corresponded to 1st choices, 15 to 2nd choices, 5 to 3rd choices and 1 was a 4th choice28. The vast majority of the interviews were undertaken in the Portuguese and Manjaco languages, with translation and assistance by Mr. Mendes, although in a couple of cases the interviews were conducted directly in Portuguese given the fact that the respondents were themselves sufficiently proficient in this language.
Upon reaching 50 completed interviews, and given that the logistical arrangements for the second period of fieldwork (in Braima Sori) prevented me from extending my stay in Caiomete, I asked Mr. Mendes to conduct five training interviews on his own (under my supervision), after which he stayed on in Caiomete to conduct the remaining 17 interviews as an enumerator. Given that Mr. Mendes had already actively participated in 50 interviews, which were then followed by five interviews on his own in which I only had to intervene a couple of times, I am quite confident that the recourse to him as an enumerator for the last 17 interviews in Caiomete did not significantly skew the data, nor compromise the reliability of the results29.
In addition to the 72 structured interviews that made up the survey proper, three semi-structured interviews were conducted which provided me with qualitative information on the Manjaco, Caiomete, migration from the region and the village, and the
28 For a discussion of the possible impact of this in terms of non-sampling error, see section 5.5.5, below.
29 See also section 5.5.5, below.
111 village’s socioeconomic organisation: two were made with village elders30, the other with the president of Uno-Tacal31. A three-hour focus group addressing all of these aspects was also conducted with five Portuguese-speaking male residents of Caiomete on one of the last days of my stay32. The reasons for this selection, which constituted an obvious source of bias, consisted of the fact that holding a focus group with participants who could not express themselves in the same language as myself would have required constant interruptions for interpretation purposes and broken the flow of the interaction (which is precisely one of the purposes and strengths of focus groups; Gray 2009:233). Thus, the wealth of qualitative data obtained through the focus group, while enabling a better understanding and interpretation of the results of the survey (as incorporated in Chapter 8, below), had to be taken with special caution.
The first period of fieldwork in Caiomete was then followed by a second period in Braima Sori. In this second fieldwork period, I lodged at a hostel in the nearby city of Gabu and travelled to and from the village every day by bush-taxi. I enlisted the help, as assistant and interpreter, of Mr. Amadu Djalo, with whom I spoke in French, namely in the context of those interviews in which it was necessary to translate the questions into the Fula language.
As had been the case in Caiomete, however, in a couple of instances I was able to conduct the interviews directly in Portuguese.
In the end, the fieldwork period in Braima Sori lasted for only eight days – significantly less than in Caiomete. This was due to a number of reasons. First, Braima Sori is considerably smaller than Caiomete (although still quite large for Bissau-Guinean standards) in terms of its area, population and number of households. Upon compiling a list of all the households (or stoves, for the logic of household organisation is quite similar here to that in Caiomete, but for the fact that the various women in polygamous marriages in this village all belong to the same – their husband’s – household or stove) in the first day of work, a total of 107 households were identified (i.e. less then three times as few as in Caiomete). Second, the actual undertaking of the survey proved much easier than in Caiomete: the respondents could be found at their homes much more easily, which had to do, among other things, with the fact that the Fula dedicate a much lesser portion of their time to ceremonial visits and obligations, as well as to the fact that the cashew harvesting season (March to May) had by then virtually come to an end. A decision was made to
30 Messrs. João Catuluca (13 May 2010) and João Mendes (15 May 2010).
31 Mr. Vitor Caperuto (8 and 12 April, 2010).
32 This focus group took place on 15 May 2010.
112 interview around 1/3 of the “household heads” in the village: 36 in total. No-one refused to participate and, in every case, the interviews were made with the ‘primary choice’
household heads in the sample.
As had been the case in Caiomete, the more structured and quantitative interviews that made up the survey proper were complemented by two longer and less structured qualitative interviews (with the village chief33, on the one hand, and a former migrant and current head of a local farmers’ association34, on the other), in addition to a focus group on migration and the local economy that brought together seven male Portuguese- or French- speaking residents35. Again, the selection of the participants in this focus group was made with a view to ensuring that the conduction of the focus group would not require interpreting. While the focus group provided a significant amount of useful qualitative information, I was and remain aware of the bias thus introduced.