A research proposal submitted to the Department of Agricultural Economics in partial fulfillment of the requirements for the award of a Master of Science degree in Agricultural and Applied Economics, University of Nairobi

2017

DECLARATION AND APPROVAL

DECLARATION

This proposal is my original work and has not been presented for examination in this or any other university for the award of a degree

Gichungi Hannah Mumbi

Signature……………………………………………………………………..Date………………………………………..

APPROVAL:

This proposal has been submitted for examination with our approval as supervisors:

Dr. Patrick Irungu (Department of Agricultural Economics , University of Nairobi)

Signature………………………………….. Date………………………………………..

Dr. John Busienei (Department of Agricultural Economics , University of Nairobi)

Signature………………………………….. Date………………………………………..

Dr. Beatrice Muriithi (Social Science and Impact Assessment Unit, ICIPE, Nairobi)

Signature ……………………… Date…………………………….

Don't use plagiarised sources.Get your custom essay just from $11/page

GET CUSTOM PAPER

ABSTRACT

In Kenya, women represent a tremendous productive resource in the agricultural sector, which contributes 26% of the Gross Domestic Product (GDP). More than 80% of farmers in this country are women. Reaching a point of gender equality and empowering women in agriculture is therefore crucial for agricultural development and food security. However, the role of women is often reduced when a new technology is introduced or when a crop enters the market economy. The International Centre of Insect Physiology and Ecology (ICIPE) has developed Integrated Pest Management (IPM) strategy for control of mango fruit flies among smallholder mango farmers, with the aim of increasing mango income, reducing expenditure on pesticides and reducing losses due to fruit fly infestation. Despite the impressive direct impacts of the Icipe’s IPM strategies, no study has been conducted to determine whether adoption of the IPM technology has any influence on gender roles in mango production and marketing decision-making. The study will be conducted in Machakos County to assess whether the IPM strategy influences different roles of women and men in mango production and marketing decision-making, and their share in the benefits. A sample of 300 mango growing households will be randomly selected from the treatment area (Mwala Sub-County) and a similar sample size from the control group (Kangundo Sub-County). Descriptive statistics and fixed effects model will be used to analyze the data.

DECLARATION AND APPROVAL.. 2

ABBREVIATIONS AND ACRONYMS. 6

CHAPTER ONE: INTRODUCTION.. 7

1.1 Background. 7

1.2 Problem statement 11

1.3 Objectives. 12

1.4 Hypothesis. 12

1.5 Justification of the study. 12

CHAPTER TWO: LITERATURE REVIEW… 14

2.1 The role of women in agriculture. 14

2.2 Theoretical review.. 15

2.2.1 Models used to anchor gender roles in decision making. 15

2.2.2 Impact assessment methods of new agricultural technologies. 17

2.3 Empirical review.. 19

2.3.1 Review of past studies on agricultural innovations and gender roles. 19

2.3.2 Determinants of women participation in agriculture. 20

2.3.3 Measurements of women role in agriculture. 21

CHAPTER THREE: RESEARCH METHODOLOGY.. 23

3.1 Conceptual framework. 23

3.2 Analytical techniques. 24

3.2.1 Analyzing changes of gender roles in mango production and marketing decision-making. 24

3.2.2 Impact of IPM on women role in mango production and marketing decision making and determinants of women role in mango production and marketing decision making. 24

3.2.2.1 Decision making index. 24

3.2.2.2 Fixed effects model 26

3.3 Diagnostic tests. 30

3.4 Study area. 31

3.5 Research Design. 32

3.6 Sampling frame. 32

3.7 Data needs and sampling procedure. 32

3.8 Sample size. 33

3.9 Data analysis and techniques. 33

WORK PLAN.. 35

BUDGET. 35

REFERENCES. 37

Appendix 1: Survey questionnaire. 41

LIST OF FIGURES

Figure1. Conceptual framework for the assessment of the impact of IPM technology on women participation in mango production and marketing …………………………………………………..16

LIST OF TABLES

Table 3.2.2: Decisions made in mango production and marketing ………….. ……………………………17

Table 3.2.4: expected explanatory variables………………………………………………………………19

ABBREVIATIONS AND ACRONYMS

DMI Decision making index

FAO Food and Agriculture organization

GDP Gross Domestic Product

GPI Gender Parity Index

ICIPE International Center for Insects Physiology and Ecology

IPM Integrated Pest Management

KSHS Kenyan Shillings

OLS Ordinary least square

UNDP United Nations Development Programme

UNECA united Nations economic commission for Africa

WEAI Women Empowerment in Agriculture Index

CHAPTER ONE: INTRODUCTION

1.1 Background

There is increased interest by researchers and policy makers on gender roles in agriculture (Fuglie and Nin-Pratt, 2012; FAO, 2011). The renewed interest is borne out of recent research that has shown that reaching a point of gender equality and empowering women in agriculture is crucial for both agricultural development and attainment of national food security (Doss, 2014). According to Peterman et al. (2014), the success in agricultural development is greatly influenced by differences in roles between men and women and which, therefore, should be considered. Mainstreaming gender into projects concerned with agricultural progress is important for attaining project effectiveness and poverty reduction (Fort et al., 2001; Fuglie and Nin-Pratt, 2012). Gender inequality, on the other hand, is likely to negatively impact families and communities both economically and socially (Soetan, 2001). Greater gender equality can improve productivity and enhance development outcomes for the next generation (Razavi, 2012).

Introduction of new agricultural technologies is often aimed at raising agricultural productivity. This occurs due to (i) factor substitution, which is the change in a combination of inputs used for production often triggered by changes in inputs’ relative prices, or (ii) technical change in which inputs are reduced in the production of the same level of output or using the same level of input to produce more output (Debertin, 1986). This in turn is expected to raise household income and net social welfare. However, men are likely to take over from women whenever increase in agricultural productivity is associated with better market linkages thus increasing rural gender inequalities (Njuki et al., 2011).

The reduced role of women arises from gender inequalities in adoption of new technologies, often attributed to women lacking access to land, education, extension services, training programs and financial services (Kumar, 1994; Team and Doss, 2011). According to von Braun (1988), investing in ‘women crops” does not necessarily benefit women farmers. Instead, when the new technology increases labor productivity, the activity is frequently taken over by male producers (Fischer and Qaim, 2012). Men also take over from women when agriculture shifts from conventional to more innovative farming methods that increase the income generated from the farm (Njuki et al., 2011).

Farmer groups organized to increase technological innovation and commercialization of agriculture, also often accelerate gender disparities. This happens when men take control over crops that were initially controlled by women when income generated from such crops increase, since they lack access to production resources(Fischer and Qaim, 2012). Whenever a crop begins to appreciate in the market and starts generating higher income, men tend to push their way into the trade (Shiundu and Oniang’o, 2007). Doss (2001) argues that adoption of new technologies affects patterns of labor, land and other resource allocation between men and women. Women face more constraints as they endeavor to engage in market. For instance, lack of secure land tenure, reduces women access to land when its economic importance increases. Men are also likely to take over women activities when crops become more profitable, thus altering labor allocation within a household.

The contribution of new agricultural technologies to the household welfare is often evaluated assuming a unitary household model, where a household behaves as an individual in consumption and production decisions and acts within the principles of rational choice theory (Himmelweit et al., 2013; Agarwal, 1997). This model assumes that all household resources and incomes are pooled, and that decisions are made by an altruistic household head that represents the household’s tastes and preferences. However, evidence has shown that individuals within a household have different preferences and bargaining power to enforce their decisions (Alderman et al., 1995; Agarwal, 1997; Quisumbing and Maluccio, 2000; Himmelweit et al., 2013).

Therefore, based on this argument, an increase in the productivity of a crop resulting from the adoption of a new agricultural technology, is expected to alter the decision making process in a household concerning the crop. Research has shown that men tend to take over from women when a crop becomes more profitable thus reducing women participation in decision making regarding the crop (von Braun 1988; Shiundu and Oniang’o, 2007; Njuki et al., 2011; Fischer and Qaim, 2012). This study is an attempt to understand how the use of integrated pest management (IPM) technology for control of mango fruit fly affects the roles of men and women in decision making concerning mango production and marketing and sharing of benefits.

IPM is an effective and environmentally sensitive approach to pest management. It is developed from traditional pest management approaches and includes a combination of the best mix of chemical and biological controls that are suitable for certain pests, weeds or diseases (Abate et al., 2000(Abate et al., 2000; Ehi-Eromosele et al., 2013). IPM strategies aim at restoring a natural balance between pests and their predators in the ecological systems. They reduce the pest status to tolerable levels while maintaining the quality of the environment (Muchiri, 2012). The goal of IPM is to reduce environmental damage (Isoto et al., 2008), health risks, and lower costs of pesticides use and profit maximization (Kibira et al 2015; Muriithi et al., 2016).

The IPM strategies are designed according the type of pest, the cropping system being used, the crop and ecological zone (Muchiri, 2012). The two main strategies of IPM used for control of mango fruit flies are (i) suppression and (ii) eradication. The suppression strategy controls the population of the pest in order to reduce yield losses while the eradication approach aims to completely eliminate the fruit flies and create ‘fruit fly-free zones’ (Kogan, 1998). Suppression strategy includes baiting application technique (BAT), male annihilation technique (MAT), orchard sanitation, biological control agents and cultural practices. The eradication strategy, on the other hand is expensive and hence not normally used by small scale farmers in developing countries (Kibira, 2015). Examples of IPM suppression strategy are natural enemies such as parasitic wasp from South America used for control of cassava mealybug and the stimulo-deterrent diversionary strategy where stem borers are repelled from maize and simultaneously attracted to a discard crop such as molasses grass (Finlay‐Doney, 2005)(Abate et al., 2000).

The International Centre of Insect and Ecology (Icipe) has overtime developed and implemented an IPM strategy for suppression of mango fruit flies among smallholder mango farmers in sub-Saharan African (SSA) countries. Icipe’s IPM fruit fly control strategy includes parasitoid release, orchard sanitation, male annihilation technique (MAT), food bait and bio-pesticide (Kibira et al, 2015; Muriithi et al., 2016). The IPM strategy is expected to increase household incomes from mangoes through the reduction of expenditure on pesticides and losses due to fruit fly infestation. A study carried out by Kibira et al. (2015) in one of the IPM project action sites in Kenya found that insecticide expenditure reduced by about 46% after application of the IPM strategy. In addition, the amount of mangoes rejected in the market due to fruit fly infestation reduced by 54.5% while net farm income increased by 22.4%. Muriithi et al., (2016) found that applying IPM strategies reduced mango yield losses by an average of 19%. It also reduced expenditure and increased income significantly for IPM users compared to the non-users.

A few studies have assessed the impact of agricultural technology and commercialization on gender. For example, von Braun (1988) assessed the effects of technological change in rice production on gender in Gambia and discovered that women roles are reduced. Fischer and Qaim (2012) investigated the gender implications of farmer groups aimed at commercialization of horticulture in Kenya and found that women tend to lose control over banana production and marketing. However, no study has been explicitly done to assess the impact of IPM technology on gender roles in production and marketing decision making.

In Machakos County, the gender inequality index (GII) stands at 0.62, which is 0.128 above the world average of 0.492 (UNDP, 2014). This indicates a 62% loss in potential human development due to gender inequality. The national average is 0.552 which is 0.06 above the world’s average (UNDP, 2015). GII, which was introduced in 2010 by the United Nations Development Programme (UNDP), measures gender inequalities in 3 important aspects of human development-reproductive health, empowerment and economic status. It indicates the loss of in potential human development due to inequality between men and women. This study therefore seeks to empirically assess whether the adoption of IPM technology for controlling mango fruit fly by smallholder mango farmers in Machakos County has had any influence on gender roles in decision making in mango production and marketing. Consequently, the findings of this study will provide empirical evidence on whether the adoption of IPM technology has escalated gender inequality in Machakos County.

1.2 Problem statement

Despite the contribution of women to agriculture, research has shown that the control of women in agriculture production and revenues tends to be reduced when a new technology is introduced (von Braun, 1988; Hamilton et al., 2001) Hamilton et al., 2001). The reduced role of women arises from gender inequalities in adoption of new technologies, often attributed to women lacking access to land, education, extension services, training programs and financial services (Kumar, 1994; Team and Doss, 2011). Therefore when implementing a new technology aimed at increasing agricultural productivity, the effect on gender roles should be considered.

Icipe have spearheaded development and dissemination of the IPM strategy for suppression of mango fruit flies in Sub-Saharan Africa. The aim of the project is to increase household incomes from mangoes through the reduction of expenditure on pesticides and losses due to fruit fly infestation. The ongoing dissemination and promotional activities of the IPM strategy shows impressive direct impacts as several growers rapidly take up the strategy (Korir et al., 2015). These evaluations of IPM packages for mango fruit fly suppression however have mainly been conducted by assuming a unitary household, where income and resources are pooled and allocated according to a joint utility function. However, there is increasing knowledge that households cannot be treated as if all members have the same preferences in regards to income and resource allocation (Alderman et al., 1995; Quisumbing and Maluccio, 2000) . No study has been conducted to determine whether adoption of the IPM strategy has any influence on gender roles in mango production and marketing decision–making and sharing of benefits. It is therefore difficult to know whether the adoption of IPM technology may escalate gender inequalities that exist in majority of the African rural communities.

This study will use a sample of mango farmers in Machakos County in an attempt to understand the impact of the IPM strategies for mango fruit fly control on gender roles in decision-making related to mango production and marketing, including access to inputs and distribution of market proceeds within the household.

1.3 Objectives

The purpose of this study is to assess the influence of IPM technologies for suppression of mango fruit flies on gender roles in mango production and marketing decision- making among smallholder mango producers in Machakos County. The specific objectives are:

To analyze changes gender roles in mango production and marketing decision-making due to use of IPM strategies for suppression of mango fruit flies among smallholder farmers in Machakos County.
To evaluate the determinants of women role in mango production and marketing decision-making in Machakos County
To assess the impact of the use of IPM technologies on women role in mango production and marketing decision-making in Machakos County

1.4 Hypothesis

Socio-economic, household and institutional factors taken singly have no influence on women’s role in mango production and marketing decision-making in Machakos County.
The use of mango IPM technology has no impact on women role in mango production and marketing decision-making among smallholder farmers in Machakos County.

1.5 Justification of the study

The agricultural sector in Kenya contributes 26% of the gross domestic product (GDP) and about 60% of the total employment. One way of ensuring development in this sector, is expanding the capacity of agricultural technology to increase agricultural productivity and hence farmers’ income (Karugu, 2006). Past studies show that introduction of new technology or improvement of productivity in certain sub-sectors in agriculture such as rice and horticultural production and linking smallholder farmers to markets reduces the role of women in agriculture (Njuki et al., 2011;Von Braun, 1988; Peterman et al., 2014; and Shiundu and Oniang’o, 2007). Therefore, assessing the effect that a new technology has on the different roles of men and women in agricultural production, marketing and decision-making is important to ensure that the new technology does not exacerbate existing rural inequalities that result in the marginalization of particular groups in rural communities.

The results of this study will be useful to researchers, policy makers, non-governmental organizations and other interested parties by contributing to the knowledge of how gender roles in production and marketing and decision making in agricultural enterprises are affected by adoption of a new technology. It will also provide information that will assist the design and implementation of agricultural technologies in such a way that they not only increase the welfare of the entire household, but ensure that gender inequalities are not escalated.

CHAPTER TWO: LITERATURE REVIEW

2.1 The role of women in agriculture

Women play an important role in agriculture globally. In developing countries they contribute 43% of the agricultural labor force (Team and Doss, 2011). According to Quisumbing and Maluccio (2000), women are the engine that drives SSA agriculture and hence the need to focus on the productivity of women farmers. They produce 50% of food globally and produce 60-80% of the staple food (Team and Doss, 2011). Despite this contribution, their role is not often perceived because there is insufficient data showing their contribution to various activities in agriculture. Women are mostly involved in activities that are not considered as economically active employment but are vital to the well-being of rural households (Team and Doss, 2011).

Due to the increased migration of men to urban areas, agriculture is increasingly becoming a predominantly female sector (Luqman et al., 2012). Women are involved in both production and post-harvest activities such as weeding and processing, in most SSA countries. Women make 86% of farmers in Kenya where 44% work on their own right as households heads, while 42% represent their husbands (Karani, 1987). Women tend to produce for more localized spot-markets and in small volumes than men and tend to be more involved in the lower levels of the supply chain and in lower value agricultural products (Tallontire et al., 2005). Moreover, women participation in agriculture does not always lead to increased income for them or increase in the decision-making in sharing and utilization of household income (FAO, 2001).

The role of women in agriculture is often impeded by poor access to inputs, land ownership, information, and credit (Team and Doss, 2011). An understanding of women farmers’ role in agricultural production and marketing decision-making is a prerequisite to devising policies to improve productivity and socio-economic development.

2.2 Theoretical review

2.2.1 Models used to anchor gender roles in decision making

From a gender perspective, it is evident that commercialization of agriculture including adoption of agricultural technologies marginalize women from the control of productive resources such as land, labor and accrued income (von Braun, 1988;Hamilton et al., 2001; Peterman et al., 2014). To understand why gender differences arise when an agricultural technology is introduced or a policy intervention is implemented, various models have been developed to explain household behavior.

The unitary model developed by Becker (1965) assumes that a household behaves as an individual in consumption and production decisions and acts within the principles of rational choice theory (Himmelweit et al., 2013; Agarwal, 1997). It assumes that all household resources and incomes are pooled, and that an altruistic household head that represents the household’s tastes and preferences makes decisions.

According to Quisumbing and Maluccio (2000), this model can be explained by assuming that a household has 2 people male (m) and female (f) that have the same preferences. These household members derive utility from the consumption of a vector of individual commodities x, influenced by a vector of household characteristics g, some of which are unobservable. The household’s utility function is given by U (x; g), which is maximized subject to an income constraint: Y= yj + ym + yf. Where Y is the total household income composed of joint income yj and individual incomes ym and yf. Since the unitary model assumes that preferences of all household members are the same and income is pooled together, resources are allocated to household members according to their ability to translate them into goods from which the household derives utility (Alderman et al., 1995; Quisumbing and Maluccio, 2000).

The unitary model has been rejected to some degree as an explanation of household behavior (Alderman et al., 1995; Agarwal, 1997; Quisumbing and Maluccio, 2000; Himmelweit et al., 2013). This is because evidence has shown that individuals in a household have different preferences and do not always pool their income together. Therefore modeling a household as a single unit, can lead to failure of an intervention (agricultural technology, policy) intended to increase households welfare (Alderman et al., 1995; Quisumbing and Maluccio, 2000).

Collective models were developed to enable the intra-households allocations reflect differences in preferences and bargaining power (Thomas, 1990; Basu, 2006; Xu, 2007). They are based on the cooperative and non-cooperative game theories. The cooperative game theory involves a collaborative decision making process where the results of negotiations among household members results to an equitable benefits for all household members. This model assumes a decision making process in which different household members have different preferences and varying bargaining power to enforce their decisions (Quisumbing and Maluccio, 2000). The bargaining power of an individual is influenced by their human capital or education, access to information, legal rights, bargaining skills and the individual claim to land, labor and income.

The non-cooperative game theory on the other hand is based on the notion that each household member acts in a way that maximizes his or her own utility (Quisumbing and Maluccio, 2000). Personal interests motivate individuals within the household rather than the desire to work in a collaborative manner to maximize the benefits for all household members. The model assumes that individuals in a household cannot enter into binding and enforceable contracts with each other but rather their actions are conditional upon those of others (Alderman et al., 1995).

This study will adopt the cooperative game theory of the household. This is because pooling of income in consumption and production decisions has been rejected in studies such as Alderman et al. (1995) and Quisumbing and Maluccio (2000). Moreover for the analysis of a social institution like the family, the cooperative approach seem more natural and promising than the non-cooperative approach, which leads to outcomes that are generally parento dominated (Fortin and Lacroix, 1997) quoting (Leutold, 1968; Bourguignon, 1984; Ulph, 1988).

2.2.2 Impact assessment methods of new agricultural technologies

Impact evaluation is used to assess the changes attributed to a particular project, program or policy on individuals, households or communities (Baker, 2000) There are 2 designs of impact evaluation namely experimental and non- experimental designs (Khandker et al., 2010). Experimental designs seek to assess an intervention’s effect by identifying a group of subjects sharing similar characteristics and assigning the treatment randomly to a subset of the group. The non-treated subjects then acts as a comparison group to mimic counterfactual outcomes (Khandker et al., 2010) thus selection bias from unobserved characteristics is avoided.

Non-experimental designs are done when it is not possible to carry out experimental designs. In non- experimental designs, a sample is selected from the treated population. Since it is not possible to observe a particular group with and without the intervention in a certain period, a counterfactual group (control group) with similar characteristics as possible to the treatment group is chosen for comparison(Khandker et al., 2010). This is referred to as the ‘with and without’ approach. The ‘before and after’ approach to impact evaluation compares the changes in the key variables during and after the intervention (Wainaina et al., 2012).

This study follows the non-experimental design using both the ‘with and without’ and ‘before and after’ approaches. The statistical and econometric methods used in impact evaluation include ordinary least square (OLS), principal score matching (PSM), difference-in-differences (DiD), instrumental variables (IV), fixed effects and regression discontinuity methods among others(Baker, 2000; Khandker et al., 2010; Wainaina et al., 2012; Muriithi et al., 2016). Ordinary Least Squares (OLS) regression compares treatment and the control groups, controlling for observable characteristics. The estimate of a program impact using OLS is biased if there is omitted variable bias.

Propensity score matching (PSM) involves pairing the treated and control groups which have similar observable characteristics. It is used to correct the estimation of treatment effects controlling for self-selection based on the idea that the bias is reduced when the comparison of outcomes is performed using treated and control subjects who are as similar as possible. However, PSM like OLS cannot match unobservable characteristics hence there can be omitted variable bias(Wainaina et al., 2012).

Regression-Discontinuity (RD) elicits the causal effects of interventions assigning a cut-off threshold above or below which an intervention is assigned. The comparison group is composed of individuals or households that are close to the cut-off point but fall on the wrong side of the cut-off hence they don’t participate in the intervention. RD assumes that after controlling for the criteria used the remaining differences between individuals directly below or above the cut-off score are not statistically significant hence the results will not be biased. However for this to hold, the cut-off ccriteria must be strictly adhered to (Khandker et al., 2010; World Bank, 2015).

Difference-in-differences model (DiD) follow the ‘with and without’ and ‘before and after’ approaches. It estimates the differences between the observed mean outcomes for the treatment and control groups before and after an intervention (Muriithi et al., 2016). DiD model, compares the changes due to an intervention over time and accounts for selection bias due to time invariant and unobservable differences among treatment and control groups. However, it does not control for unobserved time-invariant individual heterogeneity which maybe correlated with both the treatment and other unobserved characteristics.

To control for unobserved time invariant individual heterogeneity, this study will use a fixed effects model. The fixed effects model is a linear regression model that assists in controlling for unobserved characteristics that are time invariant. It assumes that individual specific effects are correlated with the independent variables. However, it does not address the unobserved heterogeneity, self selection and the reverse causality of the outcome (Muriithi, 2013;Muriithi et al., 2016). Therefore a more robust household fixed effects instrumental variable estimation can be used to control for selection bias and reverse causality of the outcome.

2.3 Empirical review

2.3.1 Review of past studies on agricultural innovations and gender roles

A number of studies on the influence of agricultural technology on gender roles have been carried out. Fischer and Qaim (2012) for instance investigated the gender implications of farmer groups aimed at improving access to agricultural technology, training and output market. The study was done among small-scale banana producers in Kenya by comparing the trend in men’s control in production and output of members and non-members over 5 years. The results showed that the groups contributed to increase in male control over income derived from bananas.

Naved (2000) used qualitative techniques to assess the intra-household impact of transfer of modern agricultural technology from a gender perspective. The study assessed how the micro-nutrient rich food intake changed as a result of transfer of a new technology that was distributed within the household and across gender and the impact of implementation of new technology on gender relations. The study found that group-based programs targeting women have a greater potential to address gender relations within the household than programs targeting women as individuals.

Karugu (2006) assessed the effects of technology transfer on gender roles among tea and coffee smallholder farmers in Kiambu County in Kenya. The study assessed the impact of the adoption of technology in tea and coffee production on division of labor along gender lines and also the impact it had on gender roles with respect to post harvest activities. The technology included proper use of improved farming practices such as soil conservation, land preparation, weeding, pruning, quality inputs such as properly constituted fertilizers and pesticides, modern farm equipment, harvesting and postharvest techniques in tea and coffee. Inferential and descriptive statistical methods were used to analyze the data collected. The result showed that there were disparities among the sexes, age groups and type of household. They also showed that female-headed households had less access to resources, education, support services and post-harvest services. However the study did not do a comparison of before and after the technology was adopted by farmers to show whether the disparities among different groups were as a result of the adoption of the technology.

Dolan (2001) examined how contracting French beans has engendered conflicts over rights, obligations and resources in Meru Kenya. The study used gender analysis to examine the changes in division of labor that arose due to contract farming. The results showed that contract farming which requires increased intensification of labor to realize production objectives, has led to a disproportionate increase in women’s work and weakened women control over horticultural production. The study also showed that contract farming increased conflict over male and female property rights and labor contribution to household subsistence.

2.3.2 Determinants of women participation in agriculture

An extensive body of research has demonstrated that households cannot be represented as single or unified units of decision-making (Fortin and Lacroix, 1997; Quisumbing and Maluccio, 2000; Basu, 2006). In the cooperative game theory, an important term used to define individual power is a household member’s “bargaining power” based on their access to and control over resources inside and outside the household (Doss, 1996). Hence, the unit of analysis to evaluate the determinants of participation in production and marketing will be the individual member of a household and not the household as a whole. Various studies have used the cooperative game theory as the theory underlying the determinants of women participation in agriculture

Kiriti et al. (2001) using an ordered probit model evaluated factors that determine female participation in decision making in agricultural households in Nyeri, Kenya. The findings showed that age had a significant influence on the women bargaining power while education, threat of divorce, perceived economic status and employment outside the household, did not significantly influence women participation. Damisa and Yohanna (2007)Enete and Amusa (2010) using a probit model, analyzed women participation in agricultural production in Nigeria. The results showed that women’s level of disposable income, tenure rights and their level of contribution to agriculture had significant impact on women participation in agricultural production.

Oladejo et al. (2011) similarly evaluated women participation in agricultural production in Nigeria using probit analysis. The findings indicated that household size, marital status and local taboos had a significant effect on women participation in agriculture. Women participation was assigned a discrete choice variable (yes or no) where a selected woman was asked to individually indicate whether she participates in agricultural production or not.

Mamun-ur-Rashid and Gao (2012), using multiple regression analysis investigated participation of in fisheries and livestock production activities in Bangladesh. The findings showed that agricultural knowledge and family size had a strong positive correlation with women participation. Education of the female spouse and family income had a negative correlation with women participation in fisheries and livestock activities.

A similar study was conducted by Muntanka (2012) evaluating factors that influence women participation in the Kaduna state agricultural development project in Nigeria using multiple regression. The results showed that age, marital status and education significantly influenced the level of participation while extension contact and access to market were less significant. In this study, the level of women participation was measured according to the number of programmes that the woman was involved in the Women-In -Agriculture (WIA) project.

Woldu et al. (2013)Woldu et al (2013) assessed women participation in Agricultural cooperatives in Ethiopia. Using ordinary least square (OLS) and Tobit models, the study identified which cooperative, household, and individual level characteristics influenced women’s participation in agricultural cooperatives. The findings showed that the major barrier to women’s access are gender biases within households, communities, and cooperatives themselves that favor educated male household heads and land owners over resource-poor women.

2.3.3 Measurements of women role in agriculture

Previous efforts in estimating women’s role in agriculture tend to evaluate their labor contributions, involvement in agricultural production activities and access to production resources (Mamun-ur-Rashid and Gao, 2012, Oladejo et al. 2011, Jaim and Hossain, 2011 , Damisa and Yohanna, 2007). However there has been little household level information regarding their role in decision making (Enete and Amusa, 2010). This study aims to identify women role in mango production and marketing decision-making and to assess the influence of IPM technology on women role in decision making.

Various estimates have been developed to show women roles and participation in decision-making. Fischer and Qaim (2012) used the proportion of banana producing households which men controlled production, output and revenues to analyze changes in control of production and marketing due to collective action. Enete and Amusa (2010) categorized the extent to which women contributed to farm decision making to high=3, medium=2 and low=1.

Decision making index has been used in various studies to estimate women roles in decision making (Baliyan, 2014, Kola’A, 2004). Kola’A (2004) for instance used a decision making index in analyzing the influence of poverty on women decision making within a household. Baliyan (2014), constructed a decision making index to analyze the determinants of women decision-making power in a household. A single indicator is insufficient to measure women role in production and marketing decision-making because it will capture only a particular domain of women empowerment in agriculture (Kishor and Subaiya, 2005). But developing a composite index will include more domains of empowerment. This study will develop a composite decision-making index to estimate the role of women in decision making.

CHAPTER THREE: RESEARCH METHODOLOGY

3.1 Conceptual framework

The role of women in mango production and marketing decision-making is influenced by various factors, which include socio-economic characteristics of women, household characteristics, the policy environment, social norms and gender perceptions as shown by arrows in Figure 3.1. These factors were chosen based on the literature review and the context of the study. Introduction of a new technology or the improvement of productivity in certain sectors in agriculture and linking smallholder farmers to markets often reduces the role of women in agriculture (Njuki et al., 2011, Peterman et al., 2014). Therefore, benefits resulting from the adoption of the IPM are expected to have an impact on the level of women participation in mango production and marketing decision-making.

Figure 3.1: Conceptual framework for the assessment of the impact of IPM technology on women roles in mango production and marketing decision-making

Source: Author’s conceptualization

3.2 Analytical techniques

3.2.1 Analyzing changes of gender roles in mango production and marketing decision-making

The first objective which is analyzing gender roles in mango production and marketing decision making will be achieved through a gender analysis for all activities involved in mango production and marketing decision making to identify roles of men and women. A comparison using mean and mode will be done between the users and non users of IPM. T- Statistics will also be used to compare the changes in gender roles before and after the use of IPM strategies. This data will be collected both at households’ levels using questionnaires and at the community level by conducting focus group discussions (FGDs). To analyze the influence of IPM technology on men role in mango production and marketing decision making, a comparison of the level of men control of mango production and marketing using T-statistics, of the users and non-users, and before and after the use of IPM strategies will be done.

3.2.2 Impact of IPM on women role in mango production and marketing decision making and determinants of women role in mango production and marketing decision making

To achieve the third and second objective which is assessing the impact of IPM on women roles and the determinants of women role in mango production and marketing respectively, a fixed effect model will be estimated. Analyzing impact of IPM on women roles in mango production and marketing will follow the panel nature of the data that will be collected for this study: fixed (within) effect regression estimation. The outcome indicator in this case is the decision making index explained below.

3.2.2.1 Decision making index

Principal component index (PCA) will be used to generate a composite decision making index that will measure the role of women in production and marketing decision making. PCA is a useful statistical technique used to find patterns in data of high dimension and compresses it to fewer dimensions without loss of information (Abdi and Williams, 2010). The PCA technique slices information contained in a set of indicators into several components. Each component is constructed as a unique index based on the values of all the indicators. The main idea is to formulate a new variable, which is the linear combination of the original indicators so that it accounts for the maximum of the total variance in the original indicators (Zeller et al., 2006).

Mburu (2015) used PCA to develop a women empowerment index to evaluate factors affecting women empowerment among beekeepers in Kenya. Krishnan (2010) also used PCA construct social economic index. PCA has also been used in developing poverty index (Zeller et al., 2006) and a neighborhood depravation index (Messer et al., 2006). Muriithi and Matz (2015) also used PCA to develop an asset index. This method is preferred when different indicators have different relative strengths as it helps in identifying and assigning weights to indicators (Zeller et al., 2006). PCA is easy to use and shuns many problems associated with traditional methods such as aggregation, non-linear relationships of variables and standardization (Mburu, 2015 quoting Vyas and Kumaranayake, 2006; and Saltelli et al., 2005).

Fifteen indicators (Table 3.2.2) chosen to indicate the role of women in production decision making are in accordance with 4 key areas of decision making that indicate women empowerment in agriculture (Alkire et al., 2013). Kishor and Subaiya (2005) stated that decision making is a widely accepted measure of women empowerment due the intuitive equality of decision making to power and control. These are decisions on production, productive resources, use of income and leadership.

Table 3.2.2: Decisions made in mango production and marketing

Area	Decision to be made
Land	Buying, selling or hiring land for mango production
Land	What to plant
Labor	How much labor to be hired
Labor	Distribution of labor among different plots
Inputs	Where to acquire inputs
	How much to purchase
	How much inputs to use in a particular mango plot
Training	Who to attend mango production training and other related gatherings
Credit	Where and when to take credit
Credit	What to do with the credit
Group participation	Who will be registered with mango growers group
Group participation	Who should attend growers group meetings
Marketing	Marketing channels to sell produce
Income	Who to receive money from mango sales
Income	How to use money received from mango sales

Source: Author

3.2.2.2 Fixed effects model

The fixed effect model can be expressed as follows:-

………………………………………………………………. (1)

where is the decision making index (DMI) for household in period ;, is the time varying intercept , is the status of using the IPM strategies, shows the impact of using IPM strategies on women role in decision making, is a vector of the observable variables that influence women roles in mango production and marketing decision making. These include the female’s social economic characteristics, household characteristics, policy environment, gender perspectives and social norms.represent the unknown parameters to be measured that will indicate the relationship of women role in decision making and the observable variables. is the unobserved time-invariant household specific heterogeneity that may be correlated with the treatment and other unobserved characteristics. Is the time varying error term. This model assumes that all unobservable characteristics that influence DMI are time invariant and are removed by within transformation.

The following model will be fitted into the data:

(2)

Fixed effects however, does not account for potential self selection bias that may affect the causal relationship between the use of IPM strategies and the outcome (DMI) because of unobservable characteristics such as skills and motivation that influence the use of IPM strategies. This means that there could be a correlation between the use of IPM and unobservable variables that lay within the time varying error term, . As a robust check for fixed effects estimation, a household fixed effects instrumental variables model that will not only account for potential selection bias but also reverse causality of DMI on the use of IPM be estimated.

Table 3.2.3: Description of explanatory variables and their expected signs

VARIABLE	UNIT	EXPECTED SIGN
Female spouse characteristics
Age of female (FAGE)	Years	+
Time spent by female spouse on farming activities (FTIME)	Hours	+
Female access to training on IPM technology (FTRN)	1=yes, 0=otherwise	+
Female marital status (MRT_STATUS)	1=yes, 0=otherwise	–
Husband’s off-farm income (HINC)	1=yes, 0=otherwise	–
Husband’s education (HEDU)	1=yes, 0=otherwise	+
Female off-farm income(FINC)	1=yes, 0=otherwise	+
Female spouse number of years growing mangoes (FEXP)	Years	+
Female access to information (FINF)	1=yes, 0=otherwise	+
Female % investment to mango production (FIVT)	Percentage	+
Female number contacts with extension service providers (FEXT)	Count	+
Female membership to a mango production or marketing group (FGRP)	1=yes, 0=otherwise	+
Female number of years completed in school (FEDU)	Years	+
Female spouse access to credit (FCRT)	1=yes, 0=otherwise	+
Household characteristics
The land owned by household (FRM_SIZE)	Acres	+
The number of male decision makers in household (N MAL)	Count	–
The number of household members (HHS_SIZE)	Count	–
Type of family (TYP_FAM)	1=extended, 0= nuclear	–
Social norms
Acceptance of women to inherit or buy land (LOWN)	1=yes, 0=otherwise	+
Specific activities for women and men in mango production and marketing(ACT)	1=yes, 0=otherwise	–

Education (FEDU) of a female household member is expected to have a positive influence on participation in decision making. It increases the bargaining power of the female spouse since they have outside leverage to enforce their decisions within the household and can make informed decisions. Enete and Amusa (2010) in their study on determinants of women contribution to farming decisions in cocoa based agro forestry in Nigeria found that women’s level of education determined the women involvement in farm decision making.

Age (FAGE) and experience (FEXP) of the female spouse are expected to increase the level of participation of women in decision making because they have more knowledge gained through years of exposure to mango production and marketing. The older the individual is, the more experience they are likely to have in farming hence they tend to be more powerful in decision-making. Kiriti et al. (2001) evaluated factors that determine female participation in decision making in agricultural households in Nyeri, Kenya. The findings showed that age had a significant influence on the women bargaining power.

Group membership (FGRP) is expected to have a positive effect on women involvement in decision making. Involvement of women in a farmers’ group contributes to empowerment and a better position in intra-household bargaining. Fischer and Qaim, (2012) found that group membership had a positive effect on female controlled income share. The study also demonstrated that women use involvement in groups reduced the negative effect on women brought about collective action that encouraged commercialization and adoption of new technologies.

Labor hours (F_LBR) spent in mango production and marketing is expected to positively and significantly influence women participation in decision making. This is because they have invested their time in mangoes hence they are likely to be more involved in the decision making process. Enete and Amusa (2010) found out that number of hours that a woman spends in the farm positively and significantly influenced their level of contribution in decision making.

The higher the percentage that the woman contributes to the investment (FINV) in mango production, the higher will be their bargaining power to enforce their decisions concerning mango production and marketing. Damisa and Yohanna (2007) analyzed women participation in agricultural production and found that the level of investment contribution to agriculture had significant impact on women participation in agricultural production.

Access to credit (FCRT) by women is expected to increase their level of participation in decision making because they are able to invest in production. Women often lack collateral to secure loans to support farm operations hence, they are not able to financially contribute to production (Enete and Amusa, 2010). This leads to low bargaining power in decision making. Social norms are expected to negatively correlate with women participation in mango production and marketing. Culture often places men higher than women and determines that only men will make major decisions and control valued resources (Kiriti et al., 2001).

Access to information (FINFO) is expected to increase women level of participation in decision making. Information on better mango production practices, input prices, the best marketing channels and market prices is likely to have a positive impact on women involvement in decision making. Training increases the skills of women in production and marketing hence it’s expected to have a positive impact on women participation in decision making.

Household size (HHS_SIZE) is expected to have negative correlation with women participation in decision making. Oladejo et al. (2011) found that household size had a significant effect on women participation in agriculture. Farm size is expected to positively influence level of women participation in decision making. Resource requirements including management decisions are expected to increase with farm size hence women are likely to contribute more in decision making with larger farm size (Enete and Amusa, 2010).

Number of male decision makers (N_MAL) is expected to have a negative correlation with women involvement in decision making. The larger the number of adult male in a household, the lower the chance of women making decisions in production and marketing. Luqman et al. (2012) found out that the number of adult male in a household had negative but significant effect on women participation in agriculture.

Women level of participation in decision making in a household is expected to be low if they are married (MRT_STATUS) than if they are single, widowed or divorced because they have less chance in participating. If a woman has off farm income (FINC), she is likely to be more powerful in decision making. Income outside the household may develop greater self-confidence and esteem of the woman which in turn will increase their ability and willingness to influence allocation decisions (Kiriti et al., 2001).

3.3 Diagnostic tests

(a) Multicollinearity

Multicollinearity means that more than two variables are near perfect combinations. When it increases, the regression model estimator of the coefficients becomes unstable and the standard error term of the coefficients can get inflated hence the greater the multicollinearity problem .A multicollinearity test will be conducted using the variance inflator factor (VIF). As the rule of thumb used by many researchers, a VIF greater than 10 indicates that the variable is highly collinear (Gujarati, 2012). 1/ VIF (tolerance) will be used to check the degree of collinearity. If the tolerance value is lower than 0.1, it means that the variable could be considered as a linear combination of the other variables.

(b) Linearity test

Linearity refers to a mathematical relationship or function that can be graphically represented as a straight line, as in two quantities that are directly proportional to each other.A linearity test will also be conducted by plotting the dependent variable against each of the independent variables. If there is a clear non-linear pattern then there is a problem of nonlinearity. When the linearity assumption is violated, the regression analysis will try to fit a straight line to data that does not follow a straight line(Gujarati, 2012).

(c) Heteroscedasticity

Heteroscedastic means that the variance of the error term differs across observations hence consistent but inefficient parameters estimates .A heteroscedasticity test will also be conducted by plotting the residuals versus the fitted values using the rvfplot command in STATA. Heteroscedastic means that the variance of the error term differs across observations hence consistent but inefficient parameters estimates (Gujarati, 2012). If the model is well-fitted, there should be no pattern to the residuals plotted against the fitted values but if the variance of the residuals is non-constant then the residual variance is said to be heteroscedastic.

(d) Autocorrelation

Autocorrelation occurs when members of series of observations ordered in time or space are correlated. It is a violation of the assumption that the size and direction of one error term has no bearing on the size and direction of another. This results to inefficient estimation (Gujarati, 2003). Durbin- Watson d test command in STATA will be used to check autocorrelation. In absence of autocorrelation d statistics is expected to be about 2. The closer d is to 0, the greater the evidence of positive autocorrelation and negative autocorrelation is evident when d gets closer to 4 (Gujarati, 2003).

3.4 Study area

This study will be conducted in two sub-counties of Machakos County (Mwala and Kangundo), which has been selected by the African Fruit Fly Programme as one of their action sites in Kenya. Machakos County stretches from latitudes 0⁰ 45’ south to 1⁰ 31’ south and longitudes 36⁰45’ east to 37⁰ 45 east. It covers an area of 6,281.4 km² most of which is semi-arid. It has an altitude of 700m to 1700m above sea level. It is generally hot and dry and has two rainy seasons. The long rain starts at the end of March and ends in May while the short rains begins end of October and continues till December. The annual rainfall averages between 500mm to 1300mm and is often unreliable (GOK, 2009).

Kangundo and Mwala areas are located in higher altitude hence receive slightly higher rainfall amount. The monthly temperature varies between 18⁰ C and 25⁰C. The hottest months are October and March while the coldest month is July. Mwala has a population of 105,529 people and covers an area of 481.5 KM², while Kangundo has a population of 107,929 people and covers an area of 178.2 KM² .The major crops grown in the two sub-counties are maize, beans, cowpeas, pigeon peas, sorghum, millet, cassava, grafted mangoes and oranges (GOK, 2009).

3.5 Research Design

The research design of this study follows a quasi-experimental approach of with and without and before and after the treatment. The survey will be conducted in Mwala (treatment/with) and Kangundo (control/without) sub-counties. One practical challenge is identification of suitable control sites. In this case, the selected “control” site has the same average climatic potential as the treatment site. In addition, and to minimize any potential interregional spillover effects of the project benefits, the control site will be at least 50km away from the treatment area. Baseline survey will be conducted to assess farmer’s situation before the IPM technology packages are distributed, while the same type of information will be collected after dissemination of the technology.

3.6 Sampling frame

The sampling frame will compose of a census of mango growers in the survey sites compiled by the respective sub-county Agricultural Officers for the two sites targeted for this study. From this list, and using the Cochran sample size formula below, households will be randomly selected from the treatment area (Mwala sub-County) and from the control group (Kangundo sub-County).

3.7 Data needs and sampling procedure

The study will use primary data collected among smallholder mango farmers in Mwala and Kangundo sub-counties with respect to women, household-heads and other household characteristics. A semi-structured questionnaire will be used to elicit data on total mango output, income, labor used, farm size and other input used for production such as fertilizer and herbicides. Random sampling will be used to select mango farmers in the two sub-counties. In addition to household survey, information such as gender perception and social norms will be obtained through informal gender disaggregated focus group discussions with farmers and key informants in the study locations.

3.8 Sample size

In order to ensure that the collected data represent properly the situation of the entire farmers’ community, a process will be followed to establish the minimum sample size which would be representative of the population and achievable with the project budget allocated to the baseline survey. To establish appropriate sample size, we a Cochran sample size formula will be adopted for continuous data given as follows:

where t is the value for selected alpha level of 0.027 in each tail (1.96), s is the estimate of standard deviation in the population, d is the acceptable margin error for mean (Barlett et al., 2001) .

3.9 Data analysis and techniques

Descriptive and econometrics techniques will be used to analyze the data. Descriptive analysis will be conducted to identify the gender roles in mango production and marketing using the Statistical Package for the social sciences (SPSS). Analysis of variance (ANOVA) will also be done to determine the differences in roles between men and women. STATA and Ms Excel will also be used for data analysis.
EXPECTED OUTPUT

Thesis
Journal article

Policy brief

WORK PLAN

2015

2016

Activity

Mar

Apr

May

June

July

Aug

Sept

Oct

Nov

Dec

Jan

Feb

Concept and Proposal development

Proposal submission

Data collection

Data cleaning

Data analysis

Thesis writing

Draft submission to supervisors

Journal writing and submission

corrections

Publication of Journal and submission of final thesis

BUDGET

Activity	Description of Activity	Cost per Unit(Kshs.)	Total Cost (Kshs.)
Enumerators training and hiring	6 enumerators trained and hired for 30 days	3500/enumerator/day	630,000
Facilitation guide and elders	2 field guides/ extension officers/elder to direct the enumerators for 30 days	1000/person/day	60,000
Manuscript preparation and publication in a Journal	Cost for manuscript preparation, submission and publication		20,000
Supervisors per diem during field work	1 supervisor for 5 days	7,500/day	37,500
Student per diem during field work	For 30 days	3,500/day	105,000
Transport cost in the field	Fuel cost	3000 per day	90,000
Communication	Airtime and stationery		5,000
miscellaneous		5% * total expenditure	47375
Total			994875

REFERENCES

Abate, T., Van Huis, A., Ampofo, J., 2000. Pest management strategies in traditional agriculture: an African perspective. Annu. Rev. Entomol. 45, 631–659.

Abdi, H., Williams, L.J., 2010. Principal component analysis. Wiley Interdiscip. Rev. Comput. Stat. 2, 433–459.

Action Aid,2012 public spending in agriculture in Kenya. Is it beneficial to small scale women farmers?

Alderman, H., Chiappori, P.-A., Haddad, L., Hoddinott, J., Kanbur, R., 1995. Unitary versus collective models of the household: is it time to shift the burden of proof? World Bank Res. Obs. 10, 1–19.

Alkire, S., Meinzen-Dick, R., Peterman, A., Quisumbing, A., Seymour, G., Vaz, A., 2013. The women’s empowerment in agriculture index. World Dev. 52, 71–91.

Baker, J.L., 2000. Evaluating the impact of development projects on poverty: A handbook for practitioners. World Bank Publications.

Baliyan, K., 2014. Participation of woman in household decision making: a case study of Muzaffarnagar district, Uttar Pradesh. Bhartiya Krishi Anusandhan Patrika 29, 159–161.

Barlett, J.E., Kotrlik, J.W., Higgins, C.C., 2001. Organizational research: Determining appropriate sample size in survey research. Inf. Technol. Learn. Perform. J. 19, 43.

Basu, K., 2006. Gender and say: a model of household behaviour with endogenously determined balance of power*. Econ. J. 116, 558–580.

Damisa, M., Yohanna, M., 2007. Role of rural women in farm management decision making process: Ordered probit analysis. World J. Agric. Sci. 3, 543–546.

Debertin, D.L., 1986. Agricultural production economics.

Dolan, C., 2001. The’good wife’: struggles over resources in the Kenyan horticultural sector. J. Dev. Stud. 37, 39–70.

Doss, C., 2014. Data needs for gender analysis in agriculture, in: Gender in Agriculture. Springer, pp. 55–68.

Doss, C.R., 2001. Designing agricultural technology for African women farmers: Lessons from 25 years of experience. World Dev. 29, 2075–2092.

Doss, C.R., 1996. Testing among models of intrahousehold resource allocation. World Dev. 24, 1597–1609.

Ehi-Eromosele, C., Nwinyi, O., Ajani, O.O., 2013. Integrated Pest Management.

Enete, A.A., Amusa, T.A., 2010a. Determinants of Women’s contribution to farming decisions in Cocoa based Agroforestry households of Ekiti State, Nigeria. Field Actions Sci. Rep. J. Field Actions 4.

Enete, A.A., Amusa, T.A., 2010b. Determinants of women’s contribution to farming decisions in cocoa based agroforestry households of Ekiti State, Nigeria. Field Actions Sci. Rep. J. Field Actions 4.

Fischer, E., Qaim, M., 2012a. Gender, agricultural commercialization, and collective action in Kenya. Food Secur. 4, 441–453.

Fischer, E., Qaim, M., 2012b. Gender, agricultural commercialization, and collective action in Kenya. Food Secur. 4, 441–453.

Food and Agriculture Organization of the United Nations (FAO), 2001. The state of food insecurity in the world.

Fortin, B., Lacroix, G., 1997. A test of the unitary and collective models of household labour supply*. Econ. J. 107, 933–955.

Fort, L., Martinez, B., Mukhopadhyay, M., 2001. Integrating a gender dimension into monitoring and evaluation of rural development projects. World Bank Wash. DC.

Fuglie, K., Nin-Pratt, A., 2012. A changing global harvest. 2012 Glob. Food Policy Rep.

GOK,2009. Machakos District Action Plan 2009-2013.

Gujarati, D.N., 2012. Basic econometrics. Tata McGraw-Hill Education.

Gujarati, D.N., 2003. Basic Econometrics. 4th.

Hamilton, S., de Barrios, A., Tevalan, B., 2001. Gender and Agricultural Commercialization in Ecuador and Guatemala. Cult. AndAgriculture 23, 1–12.

Himmelweit, S., Santos, C., Sevilla, A., Sofer, C., 2013. Sharing of resources within the family and the economics of household decision making. J. Marriage Fam. 75, 625–639.