8 Metodología
8.7 Descripción de las coberturas identificadas
8.7.12 Mosaico de cultivos, pastos y espacios naturales
Validity means truth interpreted as the extent to which an account accurately represents the social event to which it refers (Hammerssley 1990). The extent to which the collected data may be relied upon was also thought of in advance. This research was designed to use the case study strategy and a variety of data collection instruments. By using this methodology, it would be possible to triangulate the collected information at all stages. Thus, if the item is unreliable, then it must also lack validity since validity tells us whether the item measures or describes what it was supposed to measure. However, a reliable item is not necessarily also valid (Bell 1983). The main challenge in this research was to design the research such that if another researcher used the same instruments under the same conditions, he/she would be likely to get the same results.
Since internal validity is a concern of casual case studies, that is, to determine whether “x” leads to “y”, the problem can be dealt with by pattern matching of the individual case results so that one is sure that “x” is the only one that leads to “y” and no other possibility of “x” leading to, say, “z”. In order to address internal validity, explanation building of one item after another and time series analysis (population changes, land prices) were employed, and the combined effect of the issues were considered. A convergence of such results would confirm and strengthen the internal validity (Silverman 1993).
The multiple data collection process was one of the ways used to check validity. Silverman (ibid.) defines participant observation as a field strategy that simultaneously combines document analysis, respondent and informant interviewing, direct participation and observation. Such a method makes a great deal of sense, for it goes beyond the recognition of the partiality of the data to a more general practice of “method triangulation” which serves to avoid partial views and presents something like a complete picture.
Reliability and generalisation
Reliability refers to the degree of consistency with which instances are assigned to the same category by different observers or by the same observer on different occasions (Hammerssley 1992a). Kirk and Miller (1986) identify three types of reliability, namely quixotic reliability i.e. the circumstances in which a single method of observation continually yields an unvarying measurement: diachronic reliability, i.e. the stability of an observation through time (ibid.); and synchronic reliability, i.e. the similarity of observations within the same time frame. A standard way of assessment is through triangulation of methods comprising interviews, key informants, observations and reports. This is the type of reliability used in this research.
In order to improve reliability in observation, Silverman (1993) suggests that observers need to keep three sets of separate notes, namely: Short notes made at the time of observation; expanded notes made as soon as possible after each field session and a field work journal to record problems and ideas that arise during each stage of the fieldwork and a provisional running record of analysis and interpretation. The reliability of interview schedules is achieved if the respondents are made to understand the questions in the same way. This may be achieved through pre-setting of questionnaires, training of interviewers and use of ABC answers as much as possible.
As regards generalisation from cases to populations, a case study research does not follow a purely statistical logic. The issues should be understood in terms of generalisability of cases to theoretical propositions (using the identified variables), rather than in terms of populations or the universe (Yin 1994; Silverman 1993). The two cases were studied to enable replication of the findings and draw conclusions across them. The case studies were embedded such that apart from examining the global nature of the activities taking place in the case, attention was given to certain sub-units using a specific tool (commodity chain analysis) in order to understand and get clear measures of the facts. Although each case had a specific purpose, each of them was examined separately using the same procedures and methods in order to preserve context.
Thus, the three propositions formulated earlier guided the analytical generalisations. Using the propositions that were broken down into measurable sub-units, the findings of the two case studies were contrasted and compared. The outcome of the synthesis was used to assess whether the results of the two cases replicated each other or were divergent. If they replicated each other, then the issue of external validity would have been addressed and the results could be accepted for a much larger number of cases. From these propositions, generalisations could be made about the findings of similar case studies using the same unit of analysis.
A portrait of respondents
There are two reasons that make it important to formulate a portrait of respondents. First, such a devise assists in the justification of the validity and reliability of the results of the research. The results of this research need to preserve Mlandizi as Mlandizi and Masaki as Masaki as explained by resident respondents. Secondly, it enables the reader to understand the types of problems faced by farmers whilst providing pointers to the success or failure of policies pertaining to rural and urban linkages. A portrait would also enable us to understand the
ethnic composition of the residents, which has a very strong relationship to patriarchal norms. This is important, as it would inform us about their culture in relation to the type of agriculture practised.
Of the 24 respondents in Mlandizi, 8 household respondents originated from Kilwa and had migrated to the area more than 20 years ago, whilst 4 respondents were born in Kibaha (Ruvu) and in Bagamoyo Districts. Four respondents came from Morogoro, three from Liwale, two from Tanga, two from Dar es Salaam and one from Kilimanjaro Region. Most of these respondents had shifted to the village more than 10 years ago.
In Masaki 28 respondents were interviewed, of whom 20 were Zaramos claiming to be born in the village, and four were in-migrants from other regions, for instance, Kilwa and Lindi; 4 respondents lived and worked in Dar es Salaam, but they owned land in Masaki. Masaki seems to be a typical Swahili village with very little influence from outside. One reason for having few in-migrants could be the poor access road to the village.
The composition confirms that there is substantial in-migration originating from Dar es Salaam and other regions into the villages, especially those villages with good access such as Mlandizi. Indeed, most of the people are not indigenous, (not Zaramos and the Kweres), because during the villagisation programme, there was no selection of who belongs to which tribe.