Child Labour on Ghanaian cocoa farms: sweet chocolate with a
bitter aftertaste
Rense Hermans Bachelor Thesis University of Amsterdam 17-06-2019 Source: Pixabay Source: PixabaseChild Labour on Ghanaian cocoa farms: Sweet chocolate with a bitter aftertaste Course: Bachelor Thesis Project
Course code: 734302042Y
Bachelor Human Geography, Urban Planning and International Development studies (SGPL) College Social Sciences
University of Amsterdam
Assignment date of final version: June 17th, 2019
Supervisor: Dhr. dr. A.F.M. (Fred) Zaal Author: Rense Hermans (11279907) Adres: Bethanienstraat 18,
1012 CA, Amsterdam Email: rensehermans@live.nl
Table of Contents
Preface ... 4
Abstract ... 5
Chapter 1: Introduction ... 6
Chapter 2: Theoretical Framework ... 9
2.1 Introduction ... 9 2.2 Child Labour ... 9 2.3 Child Work ... 9 2.4 Quality of education ... 10 2.5 Local perceptions ... 12 2.6 The market ... 13 Chapter 3: Methodology ... 16 3.1 Research questions ... 16 3.2 Research description ... 17 3.2.1 Research type ... 17 3.2.2 Research method ... 17 3.2.3 Research design ... 18
3.2.4 Research analysis and data collection ... 18
3.3 Description of the case ... 19
3.4 Hypothesis ... 20
3.5 Graphical conceptual scheme ... 20
3.6 Operationalization... 22
Chapter 4: Analysis ... 23
4.1 Introduction ... 23
4.2 Presentation of the data 4.2.1 Ghanaian cocoa households ... 24
4.2.2 Quality of education ... 26
4.2.3 The Market ... 28
4.2.4 Local perceptions ... 32
4.3 Data analysis: Logistic Regression Analysis ... 35
4.3.1 Introduction ... 35
4.3.2 Land Preparation by household labour ... 36
4.3.3 Weeding by household labour ... 37
4.3.4 Pesticides by household labour ... 37
4.4 Data Analysis: Likelihood of Child Labour on Ghanaian cocoa farms ... 39
4.4.1 Introduction ... 39
4.4.2 Results ... 39
Chapter 5: Conclusion and discussion ... 41
Chapter 6: Recommendations ... 43
Chapter 7: Literature list ... 44
Preface
The thesis that you are about to read is the result of research on child labour on Ghanaian cocoa farms. The thesis means the end of my bachelor degree Human Geography, Urban Planning and International development studies that I followed the last three years. The reason why I chose this study had mainly to do with the fact that I enjoyed geography the most at high school. But probably it has also to do with my broad interest in the world, learning about other countries and cultures. Furthermore, the variety of courses and subjects that the study offers attracted my interest.
I have never been in Ghana nor did I learn anything in particular about this country during my bachelor-degree. However, I followed a course called Children and International Development. This course discussed issues on child labour, quality education for children and different
perceptions of childhood. The course was interesting and I appreciated what I learned about the relevance of the topics. The thesis gave me the opportunity to use the knowledge that I gained during the course and apply it to one specific case, namely child labour on Ghanaian cocoa farms. I want to make a special thanks to my supervisor Fred Zaal for all his assistance during the project. He always made time available to discuss the thesis, provided excellent and useful feedback and gave me new insights. I also want to thank Sjoerd de Vos, who helped me when I was struggling with the statistical program SPSS and Anna Laven, for her help and making it possible to do an internship at the Royal Tropical Institute (KIT).
Abstract
In Ghana cocoa is the most important cash crop (International Cocoa Initiative, 2016). Ghana is with Ivory Coast responsible for 70% of the world wide cocoa production (Food Empowerment Project, 2009). The production of cocoa is labour intensive and according to the study of Tulane University (2015) it is estimated that 2 million children engaged in hazardous work in Ivory Coast’s and Ghana’s cocoa industries. Hazardous work under children is defined as child labour. A root cause of child labour that is often cited has to do with poverty. However, child labour has still some grey areas that involves more than poverty. The thesis investigates child labour on Ghanaian cocoa farms. It has its focus on the relationship between the use of child labour, quality of education, local perceptions and the market. The thesis attempted to delve deeper into the topic and go further on the notion that child labour is mainly related to poverty. To investigate the topic, multiple literature sources are used and furthermore a quantitative data-base constructed by the Royal Tropical Institute (KIT).
Chapter 1: Introduction
For most people, chocolate is a delight. In the Netherlands for example, a person consumes annually 4.6 kilograms on average (Lambert, 2016). Cocoa is not only used for the manufacture of chocolate products but as well for beauty products such as perfume or lip moisture made out of cocoa butter (Di Boscio, n.d.). The latter is even of far greater economic importance.
Ghana and Ivory Coast are responsible for 70% of the total amount of produced cocoa in the world (Food Empowerment Project, 2019). In Ghana, cocoa is the most important cash crop for farmers (International Cocoa Initiative, 2016). However, even though the taste of chocolate is sweet, the aftertaste can be bitter. According to the National Plan of Action for the elimination of worst forms of child labour in Ghana (2017), 21.8% of children are working on the cocoa farms. A study conducted by Tulane University (2015) estimated even that two million children are engaged in hazardous work in Ivory Coast’s and Ghana’s cocoa industries. Child labour prevents children from acquiring quality education and it is harmful to their health.
The production of cocoa is labour intensive. Much of the production relies on smallholder farming. Those small-scale farms are mostly not bigger than four hectares (Roldan et al, 2014) According to Owusu-Amankwah (2015) it is estimated that a cocoa agro-forestry system needs around 206 person-days per hectare during the establishment phase and 37 during the
operational phase. Labour intensive activities that cocoa farming includes are for example the land preparation. Trees have to be cut down, all the vegetation has to be slashed and the entire bush has to be burned. Weeding takes place around three times a year. When they use of machetes, people have to climb into trees higher than three meters to cut the mistletoe. The mistletoe is a woody parasitical shrub that has to be cut down to prevent the destruction of the fruits (Ogunmefun et al, 2013). Cocoa producers are making use of a combination of family, hired and communal labour. However, the main source of smallholder cocoa farmers is traditionally household labour. This includes both adults and children (International Cocoa Initiative, 2016).
The cocoa industry continues today to invest more in the sector to increase the production of cocoa. Even though Ghana has progressive child laws and they ratified major conventions that are related to labour rights, child labour (i.e. inappropriate work, see below) still exists on cocoa farms. The greater investment in the cocoa sector creates a growing concern that smallholders are going to use more family labour, which may include children as well (International Cocoa Initiative, 2016).
Children are protected against child labour under a number of human rights treaties. For example article 32.1 of the CRC (Convention on the Rights of the Child) provides: ‘’States Parties recognize the right of the child to be protected from economic exploitation and from performing any work that is likely to be hazardous or to interfere with the child’s education, or to be
harmful to the child’s health or physical, mental, spiritual, moral or social development’’ (Child Rights International Network, 2018). Roggero et al (2007) state that there is a significant correlation between child labour and youth mortality. Furthermore, according to the National Plan of Action (2017), nine in every ten children that were involved in child labour has suffered from some forms of abuse. Thus, there is a need for more in-depth research to understand the underlying factors and - more importantly - to find a way to end the continuation of child labour in Ghana. If not, many children’s lives will continue to be deprived and the ‘’bitter’’ aftertaste of chocolate stays.
The thesis tries to have a closer look at what influences the use of child labour in cocoa production in Ghana. The research question of the thesis is as follows:
‘’Which factors affect the use of child labour by Ghanaian cocoa households?’’
Poverty among cacao farmers is often perceived as the major determinant of child labour. However, there are many other factors that influence child labour (Zdunnek et al, 2008). A study of the Tulane University and Walk Free Foundation (2018) state that poverty is the root cause of child labour. Households do not earn enough money and due to this they cannot hire adult workers with the result that children are send to work. Poverty may not always be the direct and only reason. The thesis tries to look behind the general assumption of poverty. The thesis attempts to research three other factors that may be of influence and may even interact with each other. These factors are the local quality of education, the local cultural mind-set, and the market.
The thesis starts with the theoretical framework. It explains the main concepts that are used and investigated in the thesis. It continues with the methodology. It supplies an explanation about the type of research that is conducted, which methods of research are used and how the data has been collected. Subsequently, it provides an analysis of the data. It attempts to clarify the data and it applies statistical methods to find significant results. In the end, the thesis gives a small
summary to give an answer to the research data and it discusses the results and furthermore it provides recommendations.
Chapter 2: Theoretical Framework
2.1 Introduction
The theoretical framework discusses the variables that are used in the thesis. The dependent variable is the use of child labour by Ghanaian cocoa households. To explain this variable, child labour and child work are discussed. Afterward, the theoretical framework studies the independent variables quality of education, local perceptions and the market.
2.2 Child Labour
Child labour is defined in different ways by different organizations and institutions. The
International Labour Organization (ILO) uses a very clear definition. According to the ILO, child labour is inappropriate work, which is work that is mentally, physically, socially or morally dangerous and harmful to children and interferes with their schooling (Ansell, 2017). In other words, child labour abuses children, deprives them from being a child and puts them in (physical and psychological) danger. For the ILO it is a priority to eliminate the worst forms of child labour such as hazardous employment (ILO, n.d.). The worst forms are defined by the ILO as all forms of slavery or practices that are similar to slavery. This includes the sale and trafficking of children, debt bondage and serfdom, and furthermore forced or compulsory labour. The last includes as well compulsory recruitment of children for use in armed conflict. On cocoa farms these definitions could be applied on activities such as the carrying of heavy loads, use of sharp tools, exposure to agricultural chemicals such as pesticides or exposure to long working hours. During these activities, most injuries are experienced by children. In Ghana hazardous work is prohibited for all children under 18. The minimum age for a child be allowed to normal work is 15 (Tulane University, 2015).
2.3 Child Work
Not all the work that is delivered by children can be perceived as child labour. There is a distinction between exploitative child labour and child work that is often defined as light work. According to Berlan (2013) it is difficult to draw the line between ‘’acceptable’’ and
‘’unacceptable’’ work for children. This is especially the case in rural agricultural situations when certain kinds of work form part of socio-cultural traditions. Child work can be defined as the notion that a child helps on the farm with non-hazardous activities. The work should not interfere with their schooling or their possibility of being a child (Cacao Barometer, 2018). The activities in which the children are involved contribute to their own personal development and
the welfare of their family. It helps them to learn skills and achieve experience to prepare them to become productive members of society during their adult lives. Examples of activities include helping their parents around the home, assisting in a family business or earning pocket money outside school hours during school holidays. Examples in a cocoa household are gathering of cocoa pods or removing the beans from the cocoa pods under the supervision of adults and not for too many hours. The legal age that a child can conduct light work is 13 according to the Ghana’ Children’s Act, however, the types of light work should be carried out under certain conditions (Republic of Ghana, 2008).
2.4 Quality of education
The enrolment of children into school is seen as a victory for children rights. It is perceived as a key to end child labour (Berlan, 2004). Compulsory education can be a method to ensure that children cannot enter the (full-time) workforce (Ansell, 2017). In Ghana, although the school enrolment is relatively high, this is also true for the number of working children (Berlan, 2004). According to the research rapport of Tulane University (2015), 96% of all children working on cocoa farms are attending school. The majority of children in Ghana do attend public school at the secondary level and according to the Ministry of Education, the net enrolment is high for kindergarten and primary school (91% for boys and girls) (Cocoa Barometer, 2018).
Most of the cocoa farmers no longer see cocoa farming as a lucrative activity. Receiving education is believed by the parents to be the way out of cocoa production for their children. Nonetheless, the dropout rates are high especially after the examinations (Berlan, 2004).
Furthermore, children who are combining work with the school are in a disadvantaged position. Children do not make sufficient time for school and therefore have poor learning outcomes (National Plan of Action, 2017). On school they are suffering from physical pains and exhaustion. Due to this, there is as well a clear difference between the results of children who are working and non-working children (2004).
A feature that impedes the educational future of children is the quality of education that the school provides. The public education system faces several problems. There is a lack of financial and material resources, an inadequate school infrastructure and particularly a shortage of qualified teachers in rural and remote areas (Cocoa Barometer, 2018). Furthermore,
paradoxically, the school environment is not always free from child labour. Often children are sent out to work for school during the school hours. The students have to clear land owned by
the school. Clearing the land is arduous manual work. With machetes the children have to cut the thick vegetation down and afterward they have to gather the weeds that have been cut. The land that has been cleared by children is used by the school as farmland, to produce food that subsequently serves for school lunches for which the school is responsible. This lowers the costs for the school (Berlan, 2013).
Even though education has been made free through the implementation of the Free Universal Basic Education Program, still parents have to pay the school. Formally primary and junior secondary schools do not have the legal right to charge school fees from parents. However, most of the schools are asking fees for school meals or other activities to earn some money because they are struggling with tight budgets (Bøås & Huser, 2006). In September 2017, the Ghanaian government made secondary education free as well. The president, Nana Akufo-Addo promised that there would no fees added for example for school meals, textbooks or admission. However, despite the good intentions, the implementation was not well executed Money did not arrive always in the remote and poor communities and due to this, schools are still charging fees with forced school dropout rates to go up as result (Mitchell, 2018).
Berlan (2004) states a dilemma that parents are facing: will education offer their children better opportunities after they leave school? Or is it perhaps better that their children acquire
experience that could be useful for later by working on the farm instead of learning a little and taking most of the family income away? In other words, is the local school worth the financial sacrifices that a family has to make, what are the educational outcomes for a child? Furthermore, many third world economies are struggling with high rates of unemployment. Education does indeed not always mean that someone will have access to a job.
The National Plan of Action (2017) states that there is the presence of a perpetual poverty cycle, ‘’Today’s child labourers become tomorrow’s poor adults, with families they cannot cater for.’’ Child labour causes poverty. Children do not get developed enough to have competent
knowledge and skills to acquire decent jobs when having legally employable age that would bring a better income. However, on the other side, Berlan (2004) states that poor quality in schools may actually be driving children to work.
2.5 Local perceptions
The western conception of childhood means that it is a time free from work in contrast to adulthood. However, in many other societies, children are expected to work. It is their duty to help their families and by doing so develop skills that are deemed necessary for adulthood. Working is an essential part of childhood education that produces more resilient adults. In Ghana, it is believed that it is important that children socialize in this manner. Due to this, they grow up to become a useful and responsible member of society. From an early age on, their duties and hard work and having respect towards their parents’ authority are emphasized. Children cannot contradict or stand up to what is asked by an adult (Berlan, 2004). Furthermore, children who do express their views are perceived as social deviants. It is unrespectful and should, therefore, be punished. They are bringing shame on their parents by showing that they are not raised properly and not aware of important cultural values such as respect and
obedience (Twum-Danso, 2010). With reference to this cultural mind-set, cocoa producers justify the presence of children that are working on the farms. It is part of Ghanaian culture to work hard and help your parents (Berlan, 2004).
Even though Ghana was the first country that ratified the Convention of the Rights of the Child and incorporated certain articles in their country’s children act, the cultural mind-set has implications for the implementation of that act. Especially regarding Article 12, (‘Children have the right to be actors in their own lives and they are not only dependent on the decisions that are made by adults which makes them passive recipients’), stakeholders are struggling with implementation due to the influence of cultural values. One of the main reasons is the
contradictory character of the Ghanaian conception of childhood. As stated, childhood is seen in Ghana as a period of obedience and respect towards parents. Because children are fed by their parents, they are under parental control and do not have the right to speak, but only to listen. The result is that a large number of children is struggling to make their own views known. Children find it even difficult to talk and formulate their own views if not being in company with their parents (Twum-Danso, 2010). Furthermore, children have limited knowledge of their own rights of being protected from exploitation and labour issues (National Action Plan, 2017). Ten years after ratifying the CRC in 1990, the Ghanaian government acknowledged that the article had been the most problematic one to enforce (Twum-Danso, 2010). This notion can be as well reflected in the report of National Action Plan (2017). It states that one of the major issues is that there is a low public awareness on child labour. Large sections of the population have minimal knowledge of the concept and consequences of child labour. Furthermore, the report
states that the Ghanaian population has a poor disposition towards the rights and wellbeing of children.
Another feature that is interesting to add, is how children perceive their own situation. Between 2001 and 2003, Berlan (2009) conducted fieldwork in Ghana and she followed children on cocoa farms. According to Berlan, the children showed a lot of resilience and ingenuity while trying to cope with their situation. However, they did not see themselves as passive victims who are physically and psychologically damaged by their work. They saw them selves as social actors who are trying to make the best of the oppressive and difficult circumstances in which they are situated. Children were saying that working on the farm could be hard but they did not describe it as a bad thing. This perception is as well in line with the broader pattern of child socialization. In the Ghanaian society, ‘’laziness’’ is widely detested. As stated before, parents believe that children should be involved in their work to become productive and hard-working individuals. Although children do not always describe labour as a bad thing, this does not necessarily mean that the work is not child labour according to the definition used by the International labour Organization (ILO). For example, it is deemed inappropriate when it interferes with their schooling. Applying the standards of child labour in the literature, child labour is distinguished from child work, which may be acceptable. However, in the real world it is difficult to determine the difference between child work and child labour. The typical child worker, as said above, can be defined as a child that assists its family with farming while attending school (Ansell, 2017). However, according to Berlan (2004), children in Ghana who are attending school, but also have to work, have lower school results than children who do not work. Possible reasons she gives are not having time for homework, exhaustion and physical pains. Another feature that is perceived is that many children drop out of school temporarily during the harvest season (Tulane University, 2015). So, in short, the cultural mind-set of local communities in Ghana may clash with the practices and ideas from western societies and national legal frameworks, and could cause child work, which is acceptable, to become child labour, which is not acceptable.
2.6 The market
The Ghanaian government has a complete monopoly over the cocoa industry value chain. They indirectly purchase the beans from the farmers and sell them to foreign buyers. Farmers sell their cocoa beans to a local cocoa-buying company. This company sells it to the government cocoa board that sells it subsequently on the world market to chocolate manufacturers such as Nestle and Cadbury (Berlan, 2004).
The government cocoa board is called COCOBOD. Since 1984, COCOBOD is in control of the production and marketing of cocoa, coffee and shea nuts. Due to COCOBOD, cocoa from Ghana has achieved a reputation for high quality beans in international markets. COCOBOD and its subdivisions have the responsibility to provide inputs and deliver products to farmers, traders and buyers. However, even though COCOBOD tries to keep up premium prices, there is a low producer price in cocoa farming. This issue causes concern among many organizations around the world. On the one hand they try to solve problems such as poverty and hunger in rural areas and on the other they want to preserve the cocoa supply in the long run (Roldan et al, 2013). All companies, even Fair Trade organizations (although they claim to be buying directly from the farmers) have to deal with the channel described above. Due to this, chocolate manufacturers have only limited power to monitor labour conditions on individual farms, for example, in respect to the use of children. Manufacturers are often restricted to trade directly with
producers (Berlan, 2004). Roldan et al (2013) states that ‘’Small scale farmers have little or no contact with stakeholders especially with chocolate manufacturers’’. The contact that cocoa farmers have is with agents of COCOBOD when they are providing training and extension services. However, that is only around twice a year. Due to the fact that COCOBOD has almost complete control over the Ghanaian cocoa production, it puts the farmers in isolation from external organization that are willing to provide help and improve the production (Roldan et al, 2013).
Another issue that occurs in the relationship between farmers and stakeholders, is the fact that there is a lack of written agreements ("farming contracts"). Furthermore, if there is a contract, there is little or no information about the contracting terms and conditions for the farmers. Additionally, according to Roldan et al (2013), awareness among farmers of sustainable production or certifications to motivate better producer price is lacking as well.
Another important characteristic of the market is the fact that the government determines the price they pay the farmers for cocoa. Annually they set the buying price of cocoa, also called the Farm Gate Price. This fixed price, according to COCOBOD, is 70% of the world market price. However, in reality it is significantly lower (Cocoa Barometer, 2018). Farmers are complaining that the purchasing clerks, who buy cocoa from the farmers, are paying less than the fixed price (Roldan et al, 2013). Due to the fixed price, the price is independent of the price fluctuations on the world market. Because of this monopoly, the government can make considerable profits.
These profits are mostly invested in urban areas instead of the poorer rural areas. This causes resentment among rural dwellers (Berlan, 2004). According to Roldan et al (2013) the Ghanaian government should invest for example more in the infrastructure such as the construction of better roads. Due to the poor quality of the infrastructure, many smallholders who are living in remote areas are struggling to develop. They have less access to reach training centers and furthermore to transport their beans from villages to the ports and they can participate less in the value chain of cocoa.
The result of all this is that the poverty gap between rural and urban is increasing and little progress has been made to counter rural poverty (National Action Plan, 2017).
Chapter 3: Methodology
3.1 Research questions
The main research question that the thesis attempts to answer is:
‘’Which factors affect the use of child labour by Ghanaian cocoa households?’’
As said in the introduction, poverty is mostly seen as the most important determinant of child labour. The National Plan of Action (2017) states that child labour causes poverty and that there is question of a poverty cycle. The thesis tries to look further than the notion of poverty as the main cause and looks at other possible factors account. The factors that the thesis studies are the quality of education, the market and the local perspectives. First the situation will be described summarily to provide some context. Thereafter, characteristics of the unit of research and of Ghanaian cocoa households will be outlined in order to understand how they are situated in the case. How many members does a household contain? What are the most important resources and what is the highest level of education that they achieve? Subsequently, the factors on which the thesis focuses will be studied. There will be an attempt to find a relationship between these factors and of course with the unit of research, Ghanaian cocoa households, to give an answer to the research question.
The main research question can be specified in the following sub-questions:
1. What is the role of the quality of education in use of child labour in Ghanaian cocoa households?
2. What is the role of the market in the use of child labour in Ghanaian cocoa households?
3. What is the role of local perspectives on childhood in the use of child labour in Ghanaian cocoa
3.2 Research description
3.2.1 Research type
The relationship between theory and research knows two approaches, namely inductive and deductive. By the use of an inductive approach, the theory is the outcome of the research, or in other words, while conducting research the investigator attempts to develop a new theory. The research contains generalizable inferences out of observations and other findings. One of the advantages of an inductive approach is that it gives the researcher the opportunity to make new findings without being limited within a certain theory (Bryman, 2016).
In contrast to an inductive approach, a deductive approach goes the other way around. Bryman (2016) states that the researcher draws on the knowledge and theoretical ideas about a
particular domain in order to deduce a hypothesis and translate it into operational terms. After making a hypothesis within a certain theory, the researcher collects data to construct his findings. The hypothesis will be tested by these findings to confirm or reject the hypothesis subsequently. Afterward, the theory can be adapted. So, in short, by the use of a deductive approach, findings are made from the point of view of a certain theory that is eventually tested. One of the advantages of a deductive approach is that the theory provides certain guidelines for the researcher (Bryman, 2016).
The thesis maintains an analytical inductive approach. The thesis attempts in a certain way to develop a new theory or view on child labour on Ghanaian cocoa plants. It does not restrict itself with the theory that child labour is mainly caused by poverty. It makes use of other concepts that could be seen as the tools to develop a new perspective on child labour. Furthermore, the thesis contains a set of hypothesizes that is tested by analyzing the collected data. When the new findings do not correspond to the hypothesis, the hypothesis can be adjusted and new analyzes can be made to test again the hypothesis. In short, the nature of the thesis can be perceived as iterative. The hypothesis is during the process of analyzing the data reformulated until the moment that no falsification can be found.
3.2.2 Research method
The research method used and the main analysis applied in the thesis is quantitative. For the analysis use is made mainly of an existing data-set with numerical data. An attempt is made to explain the data. In this thesis the quantitative data of the Ghanaian cocoa households are
analyzed. The different variables are compared and their frequencies are analyzed. A logistic regression analysis has been carried out using the statistical program SPSS. Logistic regression is a technique for estimating models with a dichotomous dependent variable (Treiman, 2009). Additionally, a risk indicator/index has been made with the help of the numerical data and SPSS.
3.2.3 Research design
The thesis makes use of a single case study design. A case study is defined by Bryman (2016) as a research that is concerned with the complexity and particular nature of the case in question. Yin (2014) states that a case study investigates a contemporary phenomenon (the ‘’case’’) in depth and within its real-world context. The use of a case study design is a logical choice for this study. By choosing a case study design, Ghanaian cocoa households can be analyzed intensively and in detail. Ghanaian cocoa households are the only unit of research that the thesis attempts to outline in depth. It investigates the case in an explorative way. By the use of a case study the unique characteristics of cocoa households can be revealed that may potentially be connected to child labour. The results are explained and described by the use of written literature.
3.2.4 Research analysis and data collection
The thesis makes use of several documents. To get a better understanding of the topic, a number of academic articles and books were studied. The academic articles used are mostly about child labour in the Ghanaian cocoa sector, on different definitions of child labour and possible causes. Furthermore, the literature studied provided information about the rights of children and the cultural perceptions in respect thereof. Bryman (2016) states that linking the existing literature with your own research questions, findings and discussion is an important way of
demonstrating the credibility and contribution of your research.
Policy documents provide essential quantitative demographic information such as school enrolment, employment rates and market values of cocoa during the years. also, how child labour is defined in these documents and with which causes it is related according to these documents, how these causes are perceived and their possible solutions, are all derived from these documents, mostly sourced from the Government of Ghana.
An important source of information that is used for writing this thesis is a data-set. By the use of this data-set, a quantitative analysis has been made. The data-set used is made by the Royal Tropical Institute (KIT). KIT is an independent centre of expertise and education for sustainable
development. In their work they are assisting governments, NGO’S and private corporations around the world. They are providing help to build inclusive and sustainable societies and furthermore they are measuring their impact. Themes on which their work is focused are health care, gender, economic development and intercultural cooperation while guided by the
Sustainable Development Goals of the United Nations (KIT, n.d.).
The collected quantitative data is analysed with the help of the statistical program SPSS. The thesis presents the data by constructing tables and afterward it compares those tables by putting two different variables together in a contingency table.
3.3 Description of the case
The unit of research of the thesis are the cocoa-producing households in Ghana. Approximately 800.000 farm families are active in the cocoa industry (COCOBOD, 2019). cocoa cultivation is divided over six regions: Eastern Ghana, Ashanti, Brong-Ahafo, Central Ghana and the Western region (see appendix for map). However, the production of cocoa has shifted mostly westward due to fluctuating rainfall and decreasing fertility of other regions (Bray, 2014). As said in the introduction, cocoa is the country’s most important cash crop of Ghana on the foreign exchange (Vigneri et al, 2016). As already was discussed in the introduction, Ghana produces together with Ivory Coast 70% of the total amount of cocoa in the world (Food Empowerment Project, 2019). It is estimated that they produce an estimated 675.000 tons respectively (Roldan et al, 2013). In Ghana, the cocoa crop generates about $2 billion in foreign exchange. Furthermore, the cocoa sector is a major contributor to Government Revenue (Ghana Cocoa Board, 2019).
The data used for this thesis were collected by the KIT research team between November 2016 and January 2017. Partners while conducting the research were Lindt Cocoa Foundation, IDH the Sustainable Trade Initiative, Jacobs Foundation, Rainforest Alliance, UTZ program and German Initiative on Sustainable Cocoa (GISC). The research conducted covered a broad range of subjects, for example, the household demographics and crop diversification. The study surveyed 1560 cocoa households in Ghana and 37 group discussions were conducted. KIT collected data in Ghana as well as in Ivory Coast. However, since the thesis focuses on Ghana only, the data regarding Ivory Coast will not be used. The large sample size was important because it makes it possible to perform analyses of statistical significance. Furthermore, the data may potentially be generalized at a higher level. According to the data-base around 40% of the households have been questioned in Western Ghana, 20% in Ashanti, 19% in Eastern Ghana, 16 in Brong Ahafo
and 5% in Central Ghana. The main aim of the study was to investigate why households grow cocoa and they analyzed the major aspects of cocoa of production and marketing.
The fact that this database was not explicitly focusing on child labour limits its usefulness somewhat, but it remains the most relevant database available for our study. The data-base does, however, provides insight on which data is lacking to investigate child labour further. It is useful to test the data-base to make subsequently recommendations for following research.
3.4 Hypothesis
It can be fairly assumed that the reason that the biggest income of the households is cultivating cocoa the fact that it is the most valued crop to produce in the region. A hypothesis could be that a household with a sufficient income is willing and able to send their children to school instead of subjecting them to child labour. If this is true, then the number of hours of child labour should become less if they are send to school. Instead the amount of hired labour should high and the amount of household labour should be low. However, household labour does not give any information on the number of hours spent by children. Furthermore, according to the literature, there are cases that children are still active on the farm although they also attend school. For example, it could be the case that children are going to school during the week, but are working for their parents in the weekends.
Another hypothesis that can be made, is that when a household is more modernized, the local perceptions are modern as well in the sense that in the household no child labour is used. This could be demonstrated if households have any certifications by official programs such as Rainforest Alliance or Fairtrade. Furthermore, if a household followed training regarding child labour, the presence of child labour should be low.
3.5 Graphical conceptual scheme
The thesis delves deeper into the concepts that are explained in the theoretical framework. It investigates the concepts are related to the presence of child labour on Ghanaian cocoa households. The concepts of the theoretical framework could be seen as the independent variables that influence the dependent variable, namely the use of child labour on Ghanaian cocoa households. However, the independent variables could have a relation between each other
as well. As stated in the theoretical framework, the investment in rural areas is less than in urban areas. This may have an influence on the quality of education. The figure beneath is the graphical conceptual scheme that presents the relation between the variables.
Local
perceptions
The use of child labour by Ghanaian cocoa households
Quality of Education
Figure 3.1: Graphical conceptual scheme
The Market Local
3.6 Operationalization
The operationalization presents how the concepts are defined. It is divided in dimensions, indicators and sources from where the information has been received during the thesis.
Concept Dimension Indicator Source
Education Quality of education Children enrolment hExamination results Dropout rates
Value of certificate and the price of education (Governmental) reports Data-set Academic Articles Cultural mind-set/
Local Perspectives Cultural perceptions of childhood Certifications
Awareness of child labour Certified by official organizations (Governmental) reports Academic Articles Data-set
The market Fixed Price
Household benefits
Price level
Earned money of every household member Distribution of household expenses (Governmental) reports Academic Articles Data-set
Child Labour Type of work Age of children Working time of children on the cocoa farm
Presence of working children on the cocoa farm
Hazardous or not hazardous activities Age of children expressed in years
How many hours children are working daily/weekly on the farm
Number of children that are working on the farm
(Governmental) reports
Academic Articles Data-set
Chapter 4: Analysis
4.1 Introduction
The analysis makes use of variables of the data-set and tries to interlink them with the discussed variables of the thesis. However, the study of the KIT does not completely match the research question of the thesis. Due to this, the collected data do not always fit exactly with the variables of the thesis. KIT (2018) states in the methodology under the section of limitations: ‘’Finally, the research did not cover some important topics, such as child labour. This would require a specific methodological focus to do justice to the issue’’. Nevertheless, the data-set is still useful for the thesis since it contains variables that are related with for example education, the used type of labour or the sources of loans and on what the loans are spent.
Even though the data-base is limited for this research, by using sources from literature, some statements can be made. However, because of these limitations a definitive answer to the research question cannot yet be given and the results are limited to likely assumptions. The collection of the data was conducted by the research team of the KIT and not by myself. Another consequence is that the chosen variables from the data-set, are not completely in accordance with the presented operationalization. This has to do with the fact that the exact variables are not available. However, again, it was possible to add these variables using literature. Though, further research is needed to construct a data-base that does contain more specific variables to investigate child labour more in depth.
First, the Ghanaian cocoa households are analyzed. Different variables of the data-set are presented that can be perceived as important and give chosen variables of the thesis more understanding. Afterward, the analysis does the same with the concepts of quality of education, market and local perceptions.
Most of the data is presented in three categories. The first row displays to what subject it is related. The second provides the frequency and the third the percentage. Furthermore, each table also specifies ‘’missing’’. This means that a respondent could not give an answer to the question. For this several reasons are possible. It could be related to the period of data collection.
4.2 Presentation of the data
4.2.1 Ghanaian cocoa households
The research unit of the data-set, as said before, is Ghanaian cocoa households. The thesis investigates the use of child labour in these households. This set comprises 1318 households. The first variable presented is of how many members a household consists. According to the data, the number of household members varies from 1 to 16. The biggest group of households (218) has 5 family members. Afterward, the second biggest group of households (205) has 6 family members and the third biggest group (182) has 4 family members. See in the table below for the other household numbers:
Number of household members Frequency Percent
1 33 2,5 2 80 6,1 3 125 9,5 4 182 13,8 5 218 16,5 6 205 15,6 7 168 12,7 8 106 8,0 9 64 4,9 10 61 4,6 11 23 1,7 12 22 1,7 13 8 0,6 14 9 0,7 15 3 0,2 16 3 0,2 Total 1310 99,4 Missing System 8 0,6 Total 1318 100,0
Table 4.1: Number of household members (Source: KIT Amsterdam, Tyszler et al, 2018)
The average number of household members is 5,85. According to the United Nations (2018) the average number of members in a household in Ghana is 3,5. Therefore on average the Ghanaian cocoa households have 2,35 more members. General statements that can be made to explain this number, is that household sizes in the rural areas are bigger than the urban areas.
Other variables that can be presented with the data-set are the ages of household members divided in different age-categories. Furthermore, the data provides this information separately in sexes (boy or girl, male or female). The separate tables on the age categories can be found in
the index. However, all the information presented in those tables can be summarized in the table that follows. It presents the total amount of household members of the different age categories in sexes in their frequencies and percentages. Furthermore it contains the averages per
household and the percentages of the sexes together. Age
category Girl/female Girl/female average Boy/male Boy/male average Total girl/female & boy/male Girl/f emal e perce ntage Boy/ma le percent age Percentag e total girl/fema le & boy/male Younger than 4 338 0,28 398 0,3 736 9 10 10 Between 5 and 14 1019 0,78 1179 0,9 2198 27 30 29 Between 15 and 65 2179 1,66 2133 1,63 4312 59 54 56 Over 65 173 0,13 217 0,17 390 5 6 5 Total 3709 (…) 3927 (…) 7636 100 100 100
Table 4.2: Age categories of the households (Source: KIT Amsterdam, Tyszler et al, 2018)
The age structure on a national level in 2018 does not differ much from the age structure in the households (the results of the table and of the national age structure are rounded off to whole numbers). The age category on the national level between 0 and 14 is 39%, which is the same for the households. Between 15 and 65 it has a percentage of 58%, thus 2% more than the households. Lastly, the age category over 65 has a percentage of 4, which means 1% less than the households (CIA World Factbook, n.d.).
Based on the information contained in the table, the following assumptions may be made. Children younger than 4 (736 household members, 10%) are presumably not active during farm activities because of their very young age. This is to be expected as they are not able to conduct serious work on the farm. Without being repetitive, in the theoretical framework it was
discussed that the legal age for a child to conduct light work is 13. However, if the work contains also hazardous activities such as transporting heavy loads or applying pesticides, it is perceived as child labour. Thus, the second assumption that may be made is that 29% of the household members could be under the risk of conducting child labour. Furthermore, in the theoretical framework it was discussed that the minimum age for a child in Ghana to conduct normal work, thus not hazardous, is 15. The final assumption that may be made is that the percentage of family members that can be legally active on the farm is 56%. However, the data-base does not contain any information about which activities are performed under which age. This means that
children between 15 and 18 are as well under the risk of conducting hazardous activities and thus child labour.
4.2.2 Quality of education
The first independent variable that is analyzed is the quality of education. The data-set contains a couple of variables that can be related with education. First of all, the data-set includes the highest achieved education level of each respondent. Respondents could choose one of seven possible answers: (1) they did not complete a formal education, (2) they finished primary school, (3) junior high school/ middle school, (4) Senior high school, (5) University, (6) Technical college or (7) other.
Furthermore, the respondents indicated their ability to read in English or French. The
respondents could answer the question with a ‘’yes’’ or ‘’no’’, or in other words, the results are presented as a binary variable.
The data-set provides as well data that is related to the education of their children. Data that could be perceived as useful for the thesis should be as well related to children. One of the questions that the respondents were asked was if all their boys or girls with the age between 5 and 14 were attending school. The results are shown for boys and girls separately. The answer could be ‘’yes’’ or ‘’no’’. By the help of the statistical program SPSS, a new variable has been created that shows whether all the children (boys and girls) of one household go to school. The discussed questions are presented in the following tables:
Education level Frequency Percent
no formal education completed 315 23,9
Primary school 167 12,7
Junior high school (JHS) / middle school 617 46,8
Senior high school (SHS) A/O level 131 9,9
University 44 3,3
Technical college / vocational 30 2,3
Other 7 0,5
Total 1311 99,5
Missing System 7 0,5
Total 1318 100,0
In general, the education system in Ghana operates on a 6-3-4-4 system. Primary school is 6 years, Junior High School 3 years, Senior High School 4 years and afterward 4 years of University to achieve a bachelor degree (U.S. Embassy in Ghana, n.d).
The biggest group (617) has achieved Junior High School. The second biggest group (315) has not completed a formal education and afterward the third biggest group (13) has completed Senior High School.
The question on whether the respondent is able to read English or French, results in the following table:
Ability to read in English or French Frequency Percent
no 739 56,1
yes 579 43,9
Total 1318 100,0
Table 4.4: Ability to read in English or French of respondent (Source: KIT Amsterdam, Tyszler et al, 2018)
More than the half of all respondents is not able to read in English or French. This result seems to contradict the results regarding highest achieved education. Of all respondents, 23,9 % did not completed a formal education and still only 43,9% can read. An assumption that could be made is that most of the respondents followed education in a Ghanaian language and due to this did not learn to read in English or French. However, according to the U.S. Embassy in Ghana (n.d.) the official language of instruction is English. The first three years students may study in any of the eleven local languages and afterwards they will study in English. Furthermore, according to UNICEF (2013) the literacy rate of adults in Ghana is 71%. Presumably this has to do with the low quality of education. UNICEF (n.d.) states that many students in Ghana do not benefit from good education. This has for example to do with the overcrowded classes and not sufficient trained teachers and schoolbooks.
The tables related to (1) boys and girls between 5 to 14 that are currently attending school and (2) if all children of a household are attending school, are as follows:
All household girls ages 5 to 14
currently attending school Frequency Percent
no 14 1,1
yes 633 48,0
Total 647 49,1
Missing 671 50,9
Total 1318 100,0
All household boys ages 5 to 14
currently attending school Frequency Percent
no 16 1,2
yes 718 54,5
Total 734 55,7
Missing 584 44,3
Total 1318 100,0
Table 4.6: School attendance of all household boys ages 5 to 14 (Source: KIT Amsterdam, Tyszler et al, 2018)
All household children currently
attending school Frequency Percent
Valid no 28 2,1
yes 878 66,6
Total 906 68,7
Missing System 412 31,3
Total 1318 100,0
Table 4.7: School attandence of all household children ages 5 to 14 (Source: KIT Amsterdam, Tyszler et al, 2018)
The first impression that these results give is that there is a high percentage of children that are attending school. Out of 647 girls, only 14 are not attending school and out of 734 boys, only 16 are not attending school. The missing cases could be related to the fact that a respondent does not have a boy or a girl. The table related to all household children also shows a high percentage of children that are attending school. Out of the 906 children, 28 children are not attending school. According to the UNESCO Institute of Statistics (2019) is the national net enrolment rate in primary and secondary education of 2018 is 83,59 and 59,01. Primary education is for the ages between 6 and 11 and secondary from 12 till 17. Furthermore, education in Ghana is compulsory from the age of 4 to 14.
4.2.3 The Market
The operationalization approached the variable market as the income sources and expenses of a household. How much money does a household earn from producing and selling cocoa? And furthermore, on what do they spent their money, what has importance for them. In other words, it presents in a way the economy of the household. The first variable presented is the main income sources of the last year. However, selling crops are not their only sources of income. According to the data-base there is as well data available about the loans of cocoa households. It tells if a household has borrowed money, the source of that money, on what they used the loan and if the household has repaid borrowed money.
The discussed variables give the following tables:
Households biggest source of income last year Frequency Percent
None / not applicable 2 0,2
Sale of cocoa 1116 84,7
Sale of other crops 102 7,7
Sale of livestock or livestock products 1 0,1
Sale of fish 1 0,1
Sale of bush products (bush meat,
charcoal, wood etc) 1 0,1
Labouring for other people on their
farms 2 0,2
Own small business or trading 50 3,8
Salary employment with a company 2 0,2
Salary employment in government job
(teacher, nurse, police, agric officer etc) 21 1,6 Remittances - money given by friends
and family who are living away from the household 10 0,8 Other 9 0,7 Total 1317 99,9 Missing System 1 0,1 Total 1318 100,0
Table 4.8: Households biggest source of income last year (Source: KIT Amsterdam, Tyszler et al, 2018)
The biggest source, as could be expected, is the sale of cocoa with 84,7%. The Cocoa households are the target of the research so it would be strange if another crop or activity would be their biggest source of income. Furthermore, according to the literature, the cocoa crop has the highest value on the market. The KIT (2018) states that the most frequently cited reason why households grow cocoa is the fact that it provides for a higher income than any other crop option. According to the participants of the research, the income was ‘good’, ‘high’ or higher than other crops. Furthermore, they added that the price was increasing the last years.
According to the data-set there are 10 possible sources to borrow money. Only the three biggest sources are presented in the tables. Those sources are relatively much higher than the other sources. The source from which most of the money is borrowed are lbc’s or pc’s (licensed buying company or purchasing clerk). Afterwards, family and friends are the biggest group and at last a bank. This gives the following tables:
Loan source: lbc or pc Frequency Percent
no 221 16,8
yes 132 10,0
Total 353 26,8
Missing System 965 73,2
Total 1318 100,0
Table 4,10: family and friends as source of loan (Source: KIT Amsterdam, Tyszler et al, 2018)
Table 4.11: Bank as source of loan (Source: KIT Amsterdam, Tyszler et al, 2018)
Other sources from which money has been borrowed are village moneylenders (25), microfinance institutions (9), saving and credit cooperatives/ credit unions (7), traders (3), companies (3), vsla/islc (informal savings and loans groups) (1) or another source (5).
According to the table, there are 965 households missing. This is logical because only 353 of the 1318 households are borrowing money. This is as well presented in the following table if a household has borrowed money last year:
Loan source: family and friends Frequency Percent
no 248 18,8
yes 105 8,0
Total 353 26,8
Missing System 965 73,2
Total 1318 100,0
Loan source: bank Frequency Percent
no 285 21,6
yes 68 5,2
Total 353 26,8
Missing System 965 73,2
Total 1318 100,0
Money borrowed last
year Frequency Percent
Valid no 965 73,2
yes 353 26,8
Total 1318 100,0
In the following table presents the information if a household has repaid their loan:
Loan repaid Frequency Percent
Valid no 34 2,6 yes 236 17,9 repayment period is ongoing 83 6,3 Total 353 26,8 Missing System 965 73,2 Total 1318 100,0
Table 4.13: Has the household repaid their loan? (Source: KIT Amsterdam, Tyszler et al, 2018)
As can be perceived, the majority of households (236) have already repaid the loan they made last year. Only 34 households have not paid back and for 83 households is the repayment period ongoing.
The data-set shows the commodities on which the households spent their loans. Remarkable is that most of the households spent their loan on school fees. As said in the theoretical framework, the education in Ghana is for free. However, schools are still requiring payment of extra costs. Maybe due to this households have to borrow money. In addition, a considerable part of the loan is spent on chemical inputs (fertilizer, herbicides and pesticides) and on hired labour. The graph below presents the loan use by the households.
Graph 4.1: on what does the household spend their loan? (Source: KIT Amsterdam, Tyszler et al, 2018) 154 121 68 54 46 33 32 25 22 16 7 5 4 3 2 1 1 0 20 40 60 80 100 120 140 160 180
Loan use by households
The last presented variable of the Ghanaian cocoa households, is their division in wealth
quintiles according to the Wealth Index Construction. The Wealth Index Construction contains 5 groups that go from poor to rich. The index is based on data that is related to household’s ownership. These owning features may describe their wealth. Examples of how the index is constructed are the materials that are used for the house, types of water access, the availability of electricity or internet (DHS Program, n.d.). The data-base does not include a variable with the exact income and expenses of a household. The variable of the Wealth Index may be perceived as the closest variable to income. Ownership can be perceived as a dependent factor of income. The Wealth Index of the households gives the following table:
Wealth Quintiles according to DHS
Wealth Index Frequency Percent
Valid 1st quintile 283 21,5 2nd quintile 592 44,9 3rd quintile 240 18,2 4th quintile 26 2,0 Total 1141 86,6 Missing System 177 13,4 Total 1318 100,0
Table 4.14: DHS Wealth Index (Source: KIT Amsterdam, Tyszler et al, 2018)
4.2.4 Local perceptions
The final independent variable could be seen as the most difficult one to define by the use of the data-set. This has to do with the fact that the data-set does not contain any variables related to local perceptions. However, some variables may be used to make a number of assumptions. The first variable is if households are trained on child labour. Respondents could answer ‘’yes’’ or ‘’no’’. This gives the following table:
Training on child
labour Frequency Percent
no 304 23,1
yes 339 25,7
Total 643 48,8
Missing 675 51,2
Total 1318 100,0
Table 4.15: Has the household reveiced training on child labour? (Source: KIT Amsterdam, Tyszler et al, 2018)
Another variable that can possibly be related to local perspectives is the certification that a household has. It may be said that a household perceives a certification as important for their farm and that they acknowledge the demands to achieve a certain certificate. This could be for
example a certification by Rainforest Alliance (RA), Fairtrade (FT), UTZ Certified or Organic (Org). On this question, the respondents could answer ‘’yes’’, ‘’no’’ or ‘’don’t know’’. This produces the following table:
Certified by RA, FT,
UTZ or Org Frequency Percent
no 313 23,7
yes 455 34,5
don't
know 550 41,7
Total 1318 100,0
Table 4,16: certification of the household (Source: KIT Amsterdam, Tyszler et al, 2018)
As can be perceived, the number of households that are certified is higher. However, it is significant that the biggest group has no idea if they are certified.
The reason to include the variable if a household is certified has to do with the fact that those certifications are related to child labour. For example, the child labour and forced labour guidelines of Fairtrade (2015) state that any of these forms of labour are strictly prohibited in Fairtrade operations. Furthermore, they are trying to play a positive role to eliminate
exploitative labour practices and adopting rights-based approaches in the global supply chains. However, it is still difficult to say that a certified product is free from child labour. Fairtrade America (2018) states that no person or product certification system provides a 100% guarantee that children are not included in the working process. UTZ, which is part of the
Rainforest Alliance, endorses this notion as well. Even though they are claiming that child labour is prohibited on UTZ certified farms, they state that it is impossible to guarantee that child labour does not occur on their farms ‘’No system can monitor every farm 24 hours a day, 365 days a year’’ (UTZ Certified, 2019).
Lastly, the variables of which type of labour are used for different activities are presented. Categories that the data-base for example contains are household labour and hired labour. Household labour could mean that children are active as well during farm activities, however the information does not make a division of which family member is active. It does not immediately mean that children are used. If hired labour is active, it is possible that a household makes use of such labour in order not to have to send their children to work.
By combining the data of different household labour and hired labour activities, the following table were made. Respondents could give the answer ‘’yes’’ or ‘’no’’ to all the different activities and it is displayed in percentages of the number of households.
Participation of Household labour Yes in % No in % Of N
Land clearing 39,84 60,56 635 Land preparation 79,55 20,45 494 Planting 91,94 8,06 682 Granular fertilizer 84,59 15,41 519 Liquid fertilizer 55,46 44,54 696 Manure 78,08 21,92 73 Herbicides 74,59 25,41 673 Pesticides 55,46 44,54 1154 Funguicides 62,2 37,8 971 Weeding 55,4 44,6 1249 Pruning 67,63 32,37 1078 Harvest 80,88 19,12 1313 Pod breaking 81,62 18,38 1311 Fermenting 92 8 1313 Transporting 66,02 33,98 1080 Drying 95,5 4,5 1312
Table 4.17: Participation of household labour on the farm (Source: KIT Amsterdam, Tyszler et al, 2018)
Participation of hired labour Yes in % No in % Of N
Land clearing 71,5 28,5 635 Land preparation 36,44 63,56 494 Planting 24,34 75,66 682 Granular fertilizer 33,53 66,47 519 Liquid fertilizer 53,59 46,41 696 Manure 39,73 60,27 73 Herbicides 38,48 61,52 673 Pesticides 56,85 43,15 1154 Funguicides 40,27 59,73 971 Weeding 71,02 28,98 1249 Pruning 38,59 61,41 1078 Harvest 43,03 56,97 1313 Pod breaking 15,33 84,67 1311 Fermenting 8,45 91,55 1313 Transporting 44,26 55,74 1080 Drying 6,78 93,22 1312
Table 4.18: Participation of hired labour on the farm (Source: KIT Amsterdam, Tyszler et al, 2018)
The first impression that both tables give when comparing them with each other is that the participation of household labour is higher than hired labour. The highest percentage of household labour is for the activity planting. Labour intensive activities for children could for example be land clearing, land preparation, weeding, transporting and pruning. Furthermore, activities as working with pesticides and herbicides could be harmful because according to the literature most of the times there is no protection clothing available.
4.3 Data analysis: Logistic Regression Analysis
4.3.1 Introduction
To compare the variables with each other and to investigate if there is a relationship between the dependent variable and independent variables, a logistic regression has been conducted. A logistic regression analysis is a predictive analysis that describes the data and explains the relationship between one dependent binary variable and the other independent variables. Furthermore, it analyses also the relation between the independent variables (Statistics solutions, n.d.).
After running a logistic regression analysis with the help of SPSS, several tables are displayed. These tables provide informative characteristics showing whether there is a significant relation between the variables. One of these characteristics estimates the coefficient of determination. SPSS provides two of these coefficients, Cox & Snell R Square and Nagelkerke R Square. The rate goes from 0 till 1, or in words, from weak to strong. Furthermore, another coefficient that
provides information about the direction of the relationship is B. This coefficient can be negative or positive. A negative relation means that how higher the independent variable is, how lower the dependent variable is and the other way around when the coefficient is positive. At last, the significance (Sig.) provides shows whether there is any correlation between the variables. This characteristic also provides information on which independent variable can be considered the most important variable. Or in other words, the variable that has the biggest influence on the dependent variable. In general, the significance has to be less than 0,05 in order to say that there is any statistically significant meaning of relationship between the variables. Another coefficient that indicates the importance of a variable is the coefficient Wald (Bryman, 2016).
The data-set of the KIT does not provide directly a dependent variable that is related to child labour. However, the assumption that can possibly be made is that the variables about
household labour include child labour. Based on this assumption three variables of household labour are analyzed. Moreover the chosen household labour activities are assumed as labour intensive and harmful to children. The analysis includes the following household activities as dependent variables: land preparation, weeding and pesticides. The logistic regression analysis includes the independent variables if a household had training on child labour, if they borrowed money, if they are certified by an official organization, if all their children between 5 and 14 are currently attending school and if there is the presence of hired labour.
4.3.2 Land Preparation by household labour
During the analysis of the variable land preparation by household labour, 223 households were included. For the other household no data were received. Due to this these households are not included in the result.
The Logistic Regression provides the following results:
Coefficient of determination (Model Summary)
-2 Log
likelihood Snell R Cox & Square
Nagelkerke R Square 164.718a 0,060 0,108
Table 4.19: Coefficient of determination (Source: KIT Amsterdam, Tyszler et al, 2018)
Variables in the
Equation B S.E. Wald df Sig.
Training on child labour 0,118 0,453 0,068 1 0,794 childschool -19,783 40192,933 0,000 1 1,000 Money borrowed last year -0,366 0,422 0,753 1 0,386 Certified by utz, ra, fairtrade, organic or others -0,275 0,473 0,338 1 0,561 Land clearing done by hired labour -2,511 1,034 5,897 1 0,015 Constant 24,145 40192,933 0,000 1 1,000
Table 4,20: Logistic regression analysis of land preperation by household labour (Source: KIT Amsterdam, Tyszler et al, 2018)
The coefficient of determination is 0,060 (Cox & Snell R Square) and 0,108 (Nagelkerke R Square). These coefficients can be considered as weak. These coefficients also may explain that in general the independent variables in the other table have a weak B coefficient. However, the B coefficient has a negative direction. This means for example that if a household borrows money, the household labour during land preparation is a bit less. The only variable that has any form of significance is the variable land clearing done by hired labour (0,015 < 0.05). This was expected because when a household is able to hire labour, they are needed less for work on the farm. An aspect that may be surprising is the fact that training with regard to child labour does not seem to have effect on the amount of household labour. It has a weak positive B coefficient. However, it cannot be perceived as significant.
4.3.3 Weeding by household labour
The following tables represent the relationship between the independent variables and the dependent variable weeding by household labour. The logistic regression analysis included 436 of the 1318 households. The analysis gave the following results:
The coefficients of determination of this relationship are 0,314 and 0,421. These coefficients can be considered low. In contrary to land preparation by household labour, training with regard to child labour has a negative direction. The B coefficient is -0,669 and furthermore it is significant (0,005). According to the results of the logistic regression analysis, it can not be proven that household labour decreases when a household borrows money or is certified by an official organization or when all children between 5 and 14 go to school. The last independent variable, weeding done by hired labour, shows a strong negative direction related to household labour. Or in other words, when the hired labour is high, the household labour is low.
4.3.4 Pesticides by household labour
The total number of households with data relevant for the use of pesticides for which the logistic regression analysis could be applied with household labour as dependent variable, amounts to
Coefficient of determination (Model Summary)
-2 Log
likelihood Snell R Cox & Square
Nagelkerke R Square
434.062a 0,314 0,421
Table 4.21: Coefficient of determination (Source: KIT Amsterdam, Tyszler et al, 2018)
Variables in the equation B S.E. Wald df Sig. Training on child labour -0,669 0,238 7,887 1 0,005
childschool 1,282 0,955 1,801 1 0,180
Money borrowed last year 0,004 0,258 0,000 1 0,987
Certified by utz, ra, fairtrade, organic or others 0,329 0,199 2,738 1 0,098
Weeding done by hired labour -4,251 0,603 49,782 1 0,000
Constant 2,513 1,119 5,045 1 0,025
405. The logistic regression analysis shows the following results: Coefficient of determinatio n (Model Summary) -2 Log likelihoo d Cox & Snell R Square Nagelkerke R Square 427.176a 0,230 0,314
Table 4.23: Coefficient of determination (Source: KIT Amsterdam, Tyszler et al, 2018)
B S.E. Wald df Sig.
Variables in the equation Training on child labour 0,115 0,241 0,229 1 0,633 childschool 0,616 0,916 0,451 1 0,502 Money borrowed last year -0,237 0,267 0,789 1 0,375 Certified by utz, ra, fairtrade, organic or others -0,078 0,193 0,165 1 0,685 Pesticide application done by hired labour -2,374 0,265 80,492 1 0,000 Constant 1,497 0,961 2,426 1 0,119
Table 4.24: Logistic regression analysis of pesticides by household labour (Source: KIT Amsterdam, Tyszler et al, 2018)
The last analyzed relationship between the independent variables and the dependent variable also shows weak coefficients of determination. Cox & Snell R Square gives 0,230 and Nagelkerke R Square gives 0,314. The variable training on child labour results in a weak positive B direction instead. Furthermore it seems that if the children are attending school, this means that it lowers the amount of household labour. However, this variable cannot be perceived as significant and the B coefficient is not strong. According to the results, if a household is certified, the application of pesticides by household labour decreases. However, this relationship has also a weak B coefficient and cannot be perceived as significant.
