Torgeir Fyhri 1998

The Gambia: 
The Complexity of Modernising the Agricultural Sector in Africa

Chapter 2: Methodical Considerations


Related items

Pages
The Gambia
The Gambia - Economy
The Gambia - Environment
Economy & Development News 
Environment News 
Economy - Agriculture 

Articles
» Economic history of rural Gambia  
» Main sectors of The Gambian Economy 
» Land Use in The Gambia 
» History: Introduction of the Groundnut to The Gambia 

Library 

Contents

CHAPTER 2: Methodical Considerations
2.1 Data availability
   2.1.1 Questionnaire and qualitative data
   2.1.2 System approach, statistical analysis and quantitative data
2.2 Validity and reliability problems
2.3 A multi-method strategy

OTHER CHAPTERS
Contents

CHAPTER 1: Introduction
CHAPTER 2: Methodical Considerations
CHAPTER 3: Theoretical Framework
CHAPTER 4: Introducing The Gambian Agricultural Sector
CHAPTER5: The Development of the Agriculture Sector 1983-96
CHAPTER 6: Responses to household Constraints and Farm Risk
CHAPTER 7: Testing of hypotheses and Theoretical Discussion
CHAPTER 8: Concluding Remarks
REFERENCES
APPENDIX - Interviews conducted in villages around Farafenni


2 METHODICAL CONSIDERATIONS

Geography is an academic discipline with traditions in studying complex problems often "man-land relationships". This means collecting and using data of different origin and scale, since comparing different geographical and decisionmaking levels is a study object in itself (Holt-Jensen 1991). In addition to comparing the national level with the district level and household level, the relevance of macrotheories in this field are discussed in the light of the findings of the different levels. With this starting point, data collection, reliability and validity problems related to the collected data, and methods selected for this thesis will be presented and discussed below.

2.1 Data availability

"Does anyone have to conduct more studies on Gambian agricultural production?" Babocarr Sarr, the Deputy of UNDP in the Gambia, asked us when visiting his office in Banjul. "This field is probably overstudied".

It is easy to agree with this contention. The Gambia is one of Africa's physically smallest countries with surveyable geographic conditions; it is a typical staple exporting country with a rapid growing population and increasing demand for food. A lot of scientific works have been conducted here on the topic agricultural modernisation. As a result, data availability should be good. The library of the Department of Planning in Banjul was filled up with reports on this topic. Koma (1996) surveys all the reports of the library in a bibliographic paper. However, the scope of most of the reports are mostly connected to gender diversification of labour and irrigated rice projects as Leek (1987), Carney (1986, 88, and 93), Dey (1981), Haswell (1991), Kargbo (1983), Stokke (1988), Johm (1988a, 1988b, 1992, 1996a and 1996b) and commercialisation of rice as Braun (1989), Braun & Puetz (1988), Webb (1989 & 1991).

Another common denominator of these studies are the use of quantitative data as production of a crop in tons, yield per ha and rainfall per year in analyses of national food or commercialisation strategies (USAID 1992, Torrence 1991, Akinboade 1994, Jabara 1990). Less attention, is however, paid to questionnaires of a more qualitative kind and lower geographical level studies where the farmers responses on national agricultural strategies could be understood (Bobb et al. 1994, Taal 1989, Eastman 1987 and Dey 1982). According to DeCosse (1992), additional data of e.g., poverty and off-farm income are of large concern for better understanding of the processes in The Gambia agricultural sector.

2.1.1 Questionnaire and qualitative data

Local agricultural experts through a questionnaire among village participants in the Upper Baddibu district in the North collect the main part of qualitative data collected for this thesis through single interviews Bank Division November and December 1996. A former case study conducted to describe resource management and the tenure system in the same area (Bobb et. al 1994) formed the methodical foundation of the questionnaire. Village participants of both sexes were asked in plenary about perceptions, cope strategies and farming techniques (see Appendix). A Village Extension Worker, which daily advises the farmers in cropping and a monitorer for the Sesamy Growers Association, which is a Gambian NGO, advised and translated during the interview sessions. The questions were answered jointly either by both men and women together or by separate sexes. One interview could last for about two and a half-hours.
The findings of the questionnaire must, however, be considered as preliminary step in research process and not as definitive statements, as only six villages were visited. Therefore data from corresponding studies, as Bobb et al (1994) and Torrence (1991) are added to complete the qualitative data analysed in this thesis.

2.1.2 System approach, statistical analysis and quantitative data

Qualitative analysis and use of statistical agronomic data form the methodological foundation in the more extensive study of processes at a national geographical level. Quantitative analyses include methodical tools stemming from both social and natural scientists. However, these tools are not adapted to analysis including both natural and social statistics. Comparing statistical data with common statistical tools tend to be difficult, as the statistical foundation of this thesis includes that various data as population censuses, rainfall data, agricultural production data, land use data as well as output and input prices. Trolldalen (1984) used a System Approach in the study of the evolution of the environment and agricultural production in The Gambia 1948-83. Data on different variables from different sub-systems as the bio-system, the economic system, the farming system and the demand system where statistically correlated to check the causality relation within the system. According to Holt-Jensen (1991) and Gjessing (1996), the use of a system approach is one way to overcome the barriers between the social and the natural sphere in a statistical analysis. However, when estimates such as land use data differs significantly between different sources as national statistics (NASS-DoP) and FAO statistics (FAO 1995) and scientifically reports such as Ridder (1991) and Dunsmore et al. (1976) the use of both common statistical tools or a statistical system approach might even be misleading.

The analysis of this thesis will not include advanced statistical tools. However, as in Trolldalens System Approach (1984) the empirical findings and the analysis of every influencing element or factor will be presented together for understanding the importance of the entirety. The benefit of this method of presentation is making clear the distinction between the different types of data and their different meaning to the entirety.

Figure 2.1: A model for the evolution of the environment and agricultural production in The Gambia (with correlation coefficients) (Trolldalen, 1984)

The model above, as a tested methodical tool, contains valuable information of the causal relations to agricultural development in The Gambia. However, the model is not significant in this study as the household as an actor and active respondent within the system is omitted. In addition, agricultural policies are not treated explicit in the model, figure 2.1.

Figure 2.2: A model for the evolution of the agricultural production in The Gambia

Figure 2.2, "A model for the evolution of the agricultural production in The Gambia", is basically Trolldalens model (figure 2.1) modified for this thesis. The model becomes significant, adapted to the problem of this thesis, by integrating the household as an actor and agricultural policies as an explicit system. Hence, the variables of Trolldalens model directly influencing agricultural production are in figure 2.2, treated as external conditions in which the active respondent, the farming household respond to. In turn, the action at a farming household level affected by responses, influence agricultural production. Agricultural policies is treated in model, figure 2.2 as concrete efforts aiming at modernising the agricultural sector in The Gambia and thereby on of the external influences in which the farming household respond to.
The model, figure 2.2, is used in this thesis as a visualising of the causal links between the variables and makes thereby a framework for the analysis of quantitative data in chapter 5. However, the model is not used as an independent tool, but one of many tools used in relation to a multi-method strategy. Thus, the analysis of the data related to the model is not a direct succeeding of Trolldalens analysis aiming at testing of System Approach as a methodological tool. Instead, in this thesis the model, figure 2.2, merely is used as a guide to the discussion of the spatial dialectics between the different geographical scales and between social and economical variables, making the agricultural sector developing.

2.2 Validity and reliability problems

Serious validity and reliability problems arise when studying the modernisation process of the Gambian agricultural sector. The most sounding validity problems are related to qualitative data collected from secondary sources. Findings of one study that are seemingly valid in another, such as this, might be invalid because the scientist has a different scientifically starting point.

The validity problems of secondary qualitative sources are very closely related to the reliability problems of both qualitative and quantitative sources. In an interview situation, the respondents might be unwilling to answer questions, or information might be shaded or distorted. According to CSD (1994), farmers are unwilling to provide data on income. Trolldalen (1981) was advised not to ask the farmers about ownership to land during his field study in The Gambia. In the field survey of this study, the farmers tended to avoid direct answers to the question of the quantity of groundnuts sold to Senegal.

In which degree can we then trust national agricultural statistics? Are these statistics reliable? Both the accesses to statistical report and reports analysing these statistics are of high quality compared to other African countries. However, there are large differences in estimates between the different sources. This problem is most visible in the case of land use. According to the official national statistics, National Agricultural Sample Surveys (NASS) (DoP 1996), which most reports are based on, total cultivated area have been fluctuating between 175 000 and 220 000 ha, yearly the last ten to fifteen years. The area lying fallow each year is not measured, but might be one explanation to the yearly fluctuations. FAO (1995) estimates the agricultural area to remain steady around 300 000 ha for the last ten to fifteen years. The problem with this estimate is the lack of definition of agricultural area. Does this term include both actively cultivated and fallow area? Ridder (1991) used remote sensing and GIS methods when estimating cultivate area to be 330 000 ha in 1988 and steadily expanding. This uncertainty about cultivated area creates a problem when measuring yield per ha production of different crops. In addition, due to the limited border control between Senegal and The Gambia and price differences, how much of the groundnut harvest each year is produced in The Gambia and sold to Senegal registered as Sengalese produced? The same problem exists when estimating fertiliser use. According to Akinboade (1994) only between 27 and 36% of fertiliser sold in the first half of the eighties were used in The Gambia. The rest was allegedly exported to Senegal.

On that score, reliability problems also exist when estimating migration, income sources and agricultural versus non-agricultural population due to large differences between different sources.

2.3 A multi-method strategy

The reliability problems of both quantitative and qualitative data culminate in a doubt of conclusions based on analysis were one single method is used. However, the use of multi-method strategy tends to increase the data quality, because multiple sources for evidence might increase the possibility of adapting the findings of an analysis (Dery 1996).

The choice and use of scientific method must, in the tradition of resource geography, be related to the problem examined (Gjessing 1994 and 1996, Gjessing & Trolldalen 1993). Related to the problem of this thesis, an integrated approach, or a multi-method strategy, as promoted by Andersen (1994) and Dery (1996) is chosen. Geography, as agriculture science, implies both the social and the natural sphere; traditionally studied by the two different scientific disciplines. When social science has dealt with agricultural economics, politics and the social meaning of agriculture, natural science mainly has been into agronomic practices, soil and climate science (Richards 1983). The result has often been an artificial division between the different subjects within each discipline as well as between the disciplines. According to scientists as Richards (ibid.) and Grigg (1993), the division is non-exciting in agriculture, which is, as Grigg claims, one discipline of its own. Dealing with the problem of this thesis assumes the necessity of overcoming the barriers between different scientific disciplines and subjects. The risks to which the farmers have to respond are of natural and social, political and economic origin. An integrated analysis, where variables related to both nature and social, economical and political systems are compared, is seemingly a satisfying method coping with the complexity. Integrated analysis implies here, a multi-method strategy, use of different types of data and use of extensive and intensive studies at different geographical scales.

A multi-method strategy will in this thesis mean that the analysis will not rely on single method alone. According to Andersen (1994) and Dery (1996) this strategy has been successful to overcome validity and especially problems in integrated studies.

"The use of different methods has partly been necessary in order to collect and analyse different types of items and variables, and partly been used in combination in order to increase the reliability of the same type of data." (Andersen, 1994 pp.26)

The convergence between different types of data and methods becomes with this strategy an object of analysis. Different types of data are analysed with different types of methods with the aim of control and complete data.

Some of these data, collected in the questionnaire, cannot be ranked and is classified as nominal data. Other data, of both primary and secondary sources can be grouped into classes. Such data, in which their absolute values are unknown, are classified as ordinal data. If the value is known, such as in population figures, the data is classified as interval data. The most exact, data, classified as ratio data, is found where the data can be quantified proportionately in a per cent. Examples of ratio data are production, and land use data (Andersen 1994 and Holme & Solvang 1991). The data collected are of ecological, as well as, social, political, economical and perception related origin.

A fundamental limitation of this strategy is that each single method of both data collection and analysis are not adequate separated from each other. An example very much to the point is the questionnaire conducted in Upper Baddibu district for this thesis. The questionnaire could separately form the basis of an entire analysis if carried trough in a larger format. The questionnaire is instead conducted in a small format to be an adaptation to other sources for data. However, due to the problem selected, a multi method strategy is seemingly satisfying. Few studies are conducted on risk responses related to modernising efforts in the Gambia and Sub-Saharan Africa as a whole. No single adequate method is yet to be constructed for the topic. And as a result, adequate data is lacking. There are very little data on risk cope strategies as sources for off-farm income in rural households. However, analysis of quantitative national production data added with qualitative constraints analysis as conducted by Bobb et al. (1994), Torrence (1991) and Willis (1996) and questionnaires in rural villages are methods that together give a satisfying collection of data related to the problem. In other words, studies at different geographical levels gives meaning in this study as the findings of extensive studies at a national level are added or compared to findings of intensive studies in the villages.


previous     next


© afrol.com. Texts and graphics may be reproduced freely, under the condition that their origin is clearly referred to, see Conditions.
The copyright of this specific text does not belong to afrol.com, but to the author. See special note.

   You can contact us at mail@afrol.com