CHAPTER 4: DATA AND METHODOLOGY
4.4 P OPULATION AND S AMPLE F RAME
Peters (1996) defines a population as that part of the universe relevant to a specific problem. It is a finite or infinite collection of individual objects and consists of a number of units of enquiry (King’oriah, 2004). A geographical population is therefore a collection of objects with some geographical characteristics in common.
4.4.1 Target population
The target population in this study was drawn from two components of the unit of analysis, namely the geophysical (Nyando Basin) and the social (households).
a) Nyando Drainage Basin and Lower Kano Plains: The river basin has become increasingly accepted as a unit for water resource management (Biswas, 1990;
Hardy, 2005). The rationale stems from the concept of a river as an organic system, where interference with or modification of any part of it will be felt elsewhere in the system. The Lower Kano Plain in Nyando Basin, therefore, is used for the acquisition of economic data, climate and hydrological data (i.e.
rainfall and flow). Also examined are the descriptors of the river morphology and the characterization of river flow within the basin, which Arnaud et al., (2002); Hardy (2005) and Andreo et. al., (2006) have recommended as important for an understanding of the river basin as an open system of geophysical and social interactions.
b) The households: The households formed the main unit of sociological analysis for the acquisition of anecdotal information. Households within the sampled administrative sub-locations were interviewed in order to derive anecdotal information (perceptions of flooding from cotton growers’ point of view). The responses (anecdotal data) from the household with regard to the physical and social characteristics of flooding were supplemented with the results from the physical (hydrologic) analyses.
4.4.2 Sampling and sampling techniques
The importance of sampling and the selection of the sample population in research design are emphasized by Tashakori and Teddlie (1998) and King’oriah (2004). These authors have noted that the whole structure of a sampling survey is, to a considerable extent, determined by the sample where the target population is drawn (households in Lower Kano Plains). Tashakori and Teddlie (1998) also pointed out that there is a need to carefully design a probability sample, where results are representative of the entire population under investigation. In this study the sample populations is represented the households living in the Lower Kano Plain in the lower reaches of the Nyando River. These households have their farms and homesteads located either on the eastern or western banks along the river. These households were treated as a representative sample of the households living in the Kano Plain.
In order to avoid or, at least minimise, the element of bias in the selection procedure, and also to achieve the required precision given the resources available, random sampling were conducted. As noted by the World Agroforesty Centre (2006), increased settlement and crop cultivation takes place in the floodplain and along the banks of the Nyando River. Their report attributes this to an increase in the demand for more land for crop cultivation and food provision. Only the households which are defined by the study area (Figure 4.1) inland of the Nyando River mouth in the Lower Kano Plain were included in the study. Through purposive sampling (non-probability) the interview schedule was mainly focused on the heads of the households (either male of female).
4.4.3 Sampling frame and sample selection
The selection of the households that were to be included in the study began by a delineation of the sub-locations that fell within an area of 48 Km2 inland of the Nyando River mouth (Figure 4.1). The 8 villages identified (within the sub-locations) became the entry points in the selection of households for the acquisition of anecdotal data. Village names were mostly used as boundaries, both on the eastern and western sides of the riverbanks. It was within these blocks that ten administrative sub- locations were chosen and the respective villages were selected for the questionnaire interview schedule.
4.4.4 Sampling size for sociological analysis
Peters (1996) and Neuman (2003) point out that the sample size of the study population must be representative in order to yield reliable results. King’oriah (2004) similarly demonstrates that the choice of sample size depends on the homogeneity or heterogeneity of the population under study with a large sample being preferred for a heterogeneous than homogenous population (King’oriah, 2004). While Peters (1996) and Neuman (2003) propose a sample ratio of 30 % (300 respondents per population of 1000), King’oriah (2004) proposes a sample ratio of 10 % (100 respondents) for a homogenous population.
A more recent study on the impact of land use and land carrying capacity on erosion in Kano Plain was carried out by ICRAF (2001), who estimated between 1000 and 1200 homesteads exist in Lower Kano Plains. The households falling within the sampled sub-locations live in homogenous climatic conditions; share similar socio- economic backgrounds; use similar infrastructures and generally share experiences
with regard to the impact of flood damage on agriculture. Hence, a 10 % sample ratio is considered appropriate for the study. A sample size of 100 respondents, that is 10 % of the population, was considered as adequate under these homogenous conditions. A total of 100 households responded to the questionnaire interview. Figure 4.2 shows the administrative sub-locations where the interview schedule and the acquisition of anecdotal (perceptions) data were performed.
Figure 4.2: The Villages surveyed in Lower Kano Plains (study area)