Factors influencing the empowerment of women coffee growers in southern Colombia

Abstract This study evaluated the factors influencing the empowerment of women coffee growers in Cauca Department in southern Colombia who are associated with and linked to inclusive global markets. We applied a mixed-methods approach during the research process to adapt the Three Domains of Empowerment sub-index of the Project-level Women’s Empowerment in Agriculture Index with qualitative pre-survey information from participants. After calculating the Three Domains of Empowerment sub-index, we assembled focus groups to expand the analysis and interpretation of the quantitative results. The findings showed that, despite high levels of participation in the productive component of coffee growing and their associativity, women were not completely empowered; 63.3% of participating women reported being disempowered. The three indicators contributing most to their disempowerment were social control, visiting key locations, and work balance. Likewise, through discussions with participating women, we identified sociocultural conditions such as intimate partner violence, household power structures, concerns regarding violent situations related to armed conflict, and institutional weaknesses that limit women coffee growers’ empowerment within both the private and public spheres.


Introduction
Women's participation in agriculture is key to agricultural development in emerging economies.However, women face more obstacles than men when engaged in agricultural production activities due to limited access to raw materials, land, technology, extension services, markets, and credit (Anderson et al., 2021).Over the past two decades, international organisations and national and regional governments have implemented projects and intervention programmes aimed at reducing the gender gap through women's empowerment, based on the association between women's empowerment and reductions in poverty and hunger.According to the review of the literature in Tahir et al. (2018), these programmes have been introduced largely because of pressure from international agencies and donors to achieve the main targets of the UN's Sustainable Development Goals.
In Colombia, unlike other industries, the coffee production industry has made concerted efforts to empower women.Having implemented different programmes and projects since 2006, the National Federation of Coffee Growers (NFCG), a private, non-profit union entity, released the Gender equity policy for women coffee growers report in 2021. 1 The report, which surveyed coffeegrowing households, revealed that gender gaps persist.
According to Johnson et al. (2018), little systematic work has been done to investigate the mechanisms by which interventions can enhance women's empowerment and measure their effects.Intervention programmes of this sort face significant challenges in design, implementation, and evaluation of results and impacts; challenges regarding what and how to measure arise from, among other things, the definition of women's empowerment.
That said, research has not overlooked this challenge.Specifically, according to a review study by Priya et al. (2021) the work of Kabeer (1999) has been enormously influential in the field of women's empowerment research and measurement thereof is the most frequently researched topic in the field.
The use of indices based on indirect measures at an aggregate level has been identified in attempts to gauge empowerment.The main weakness of such indices-despite their facilitation of inter-country comparisons and use in accounting for advances in gender equality (Crookston et al., 2021) -lie in the fact that empowerment is intrinsically experienced by individuals in different domains (Alkire et al., 2013).
In the process of improving these measures, Alkire et al. (2013) introduced the Women Empowerment Agriculture Index (WEAI) to realise the profiles of individual empowerment across multiple domains and utilise indicators relevant to agriculture.The WEAI has been widely used as it allows researchers to assess the state of empowerment and gender parity in agriculture.In addition, to meet the need to measure empowerment at a project level and to monitor and evaluate its impact, Malapit et al. (2019) subsequently developed the Project-level Women Empowerment Agricultural Index (Pro-WEAI).
Like the WEAI, the Pro-WEAI is based on Kabeer's (1999) concept of women's empowerment as a process by which women expand their ability to choose, make decisions, and act to fulfil achievements, thus transforming gender subordination relationships.The ability to make choices incorporates three inter-related dimensions: resources, agency, and achievements.This index considers agency (rather than resources or achievements) to be a more direct measure of empowerment because resources and achievements are present even in situations where women are highly disempowered (Malapit et al., 2019).
The Pro-WEAI comprises two sub-indices: the Three Domains of Empowerment (3DE) and the Gender Parity Index (GPI) and has 12 indicators.The 3DE sub-index assesses empowerment across three domains, namely intrinsic agency, instrumental agency, and collective agency.The GPI measures women's empowerment relative to men's empowerment within their households (Malapit et al., 2019).The Pro-WEAI can be broken down into gender, age, and population subgroups and indicators, allowing women's empowerment-related projects to identify key areas of women and men's disempowerment and design strategies aimed at addressing shortcomings and monitoring outcomes.Thus, it provides key information for the development of more effective participatory action plans focused on strengths and weaknesses of the target population type.
As the Pro-WEAI allows for more efficient interventions to strengthen women's empowerment and reduce gender gaps, it is considered an adequate tool to measure the empowerment of women coffee growers in Colombia.However, the Pro-WEAI has been primarily used in agricultural development projects of large geographical and population scales in Southeast Asia (although not representative at a country level) that are linked to different types of agricultural activities, financed by international cooperation agencies and national governments, and characterised by low group or association membership (Malapit et al., 2019).Further, in some cases, additional measurements (e.g.nutrition) were considered for the purposes of agricultural projects.
Although no research applying the Pro-WEAI has been identified in Latin America, some applications of the abbreviated WEAI (A-WEAI) were found, among which we may highlight the study developed in Guatemala by the International Centre for Tropical Agriculture (Centro Internacional de Agricultura Tropical -CIAT) and the University of Florida.Muriel et al. (2019) analysed a sample of Guatemalan agricultural households where the degree to which women and men have power in their homes and the degree of inequality between women and men within the same household was assessed.
The A-WEAI index is also based on Kabeer's empowerment theory and measures empowerment among five domains (production, resources, income, group membership, and time) using six indicators unlike the Pro-WEAI, which focuses on measuring three agencies (intrinsic, instrumental and collective) and 12 indicators.
In the Productive Support for Agri-food Programme (APAGRO) whose objective was to increase income and food security of small female producers in Nicaragua, a proxy for the A-WEAI index, including only four domains of empowerment (production, resources, income, and leadership) and four indicators to contribute to the empirical evidence on livestock transfer programmes, was used.The analysis shows that improvements on empowerment were mainly driven by participation or membership in community groups or agricultural associations, which could be related to the group training sessions organised by the APAGRO programme to provide technical assistance (Salazar et al., 2018).
Another study, by the Food and Agriculture Organization of the United Nations (2017), identified a series of indicators for measuring empowerment by mapping value chains and prioritising individual-level gender inequality factors to country-wide case studies.These indicators were as follows: Ownership, access to property/lease of property/land; Decision ability (assets, free association, investments)/Empowerment; Access to credit/investments; Access to technical support services/education; Time (use); Access to markets; Political representation (communal and provincial): empowerment; and Access to services.
Previous experiences highlight the influence of Kabeer's empowerment theory and the fact that its measurement does not respond to a single structure of indicators and methodologies, and therefore such measurement must be addressed considering the context and socio-political situation of every region in which women participate in agricultural activities.
We needed to adapt the pro WEAI Three Domains of Empowerment (3DE) sub-index to the context of coffee growers in Colombia.Through this adapted index, we identified factors influencing the empowerment of women coffee growers belonging to the Association of Coffee-growing Women of Cauca (AMUCC) who were participating in inclusive global markets.The results will allow us to identify areas of the participating women's disempowerment, establish a baseline, and create a participatory action plan focused on strengthening their empowerment.The adapted index will also enable researchers and others to evaluate and monitor the impact of different agricultural projects on their empowerment.
Because of their unique characteristics, AMUCC members are of special interest in the measurement of empowerment.For example, they have a wide network of inter-institutional relationships and have participated in different cooperation and development projects intended to strengthen empowerment by improving income and access to resources, and by linking their coffee to inclusive global markets.
The methodological robustness of this research is based on a mixed-method approach, using qualitative and quantitative methodological strategies in all phases of the research.The qualitative strategies facilitated adjusting the concepts and indicators of the 3DE sub-index, redesigning the questionnaire and, finally, broadening the understanding of the quantitative results to respond to the multidimensionality and complexity of measuring empowerment.

Context and study area
Coffee farming is one of the key drivers of the Colombian economy, accounting for 15% of its agricultural gross domestic product (GDP).The Cauca Department comprises the largest number of coffee growers in the country; 99% are small coffee growers (farms smaller than 50,000 m 2 ) in peasant, indigenous, and Afro-descendant communities.Approximately 30% of the Cauca Department's coffee farming is operated by women who simultaneously perform household responsibilities and childcare and take on different roles such as collectors, farm owners, businesswomen, baristas, tasters, and community and union leaders .
The Cauca Department has held the Cauca Coffee Denomination of Origin (DO) 2 since 2011; the certificate is an instrument for the protection of the coffee produced in the region and contributes to a group of coffee families entering the specialty coffee market.
Despite the diversity of natural resources in the Cauca Department and the resulting potential for the economic, social, and cultural development of its territories, it faces challenges such as illegal mining, drug trafficking, inequality, armed conflict, and forced migration.Notably, the Cauca Department territory is of great interest to national and international organisations owing to the large number of municipalities that the Colombian government has prioritised for implementing peace agreements.
A total of 109 women participating in this research reside in this territory; they all belong to the AMUCC, an organisation created in 1999 by women in the municipality of El Tambo who were victims of Colombian armed conflict.Members are residents of nine municipalities, and 96% own a coffee-growing area of less than 30,000 m 2 (AMUCC, 2020).AMUCC members engage in the production of high-quality specialty coffees (Fair Trade Certified in 2010), which they then market to the European gourmet sector.
Because of the influence and importance acquired by the AMUCC, together with other entities, it participated in the creation of TECNICAFE, the first Coffee Innovation Technology Park, in Cajibío in the Cauca Department in 2015, and it holds a position in TECNICAFE's management body.
The AMUCC currently seeks to expand its productive and social bases and reach a total membership of 400 women associated with coffee growing through an inter-institutional public-private alliance by focusing its efforts on empowering its members and the sustainable development of coffee cultivation (AMUCC, 2020).

Sociodemographic characteristics of respondents
The most relevant sociodemographic characteristics of the 109 AMUCC women who participated in the study are presented in Table 1.
The respondents were residents of nine municipalities from the Cauca Department; more than 80% were between 34 and 66 years of age; 66% had an educational level not exceeding basic primary education, and 80% were married or lived in a free union.

Mixed methods
The complex and multidimensional character of empowerment justifies an approach involving the articulated application of quantitative and qualitative instruments.In the social sciences, the mixed-methods approach can investigate social phenomena through coherent designs that integrate quantitative and qualitative methods to expand explanations and understandings of the same object of study (Núñez, 2017).Following Núñez (2017), we prioritised the quantitative aspect of the mixed-methods technique and selected a sequential approach.Thus, the qualitative method (initial interviews) provided the information needed to adjust the 3DE sub-index and the questionnaire.The survey results allowed us to expand our analysis and interpretation of the quantitative results among the focus groups we assembled.

Measuring empowerment through the pro-WEAI 3DE sub-index
A major advantage of the pro-WEAI is that it can be broken up by sub-indices or indicators.Accordingly, we measured AMUCC members' empowerment by calculating the 3DE sub-index, as this allowed us to identify indicators that must be addressed to increase empowerment, design appropriate strategies to overcome the identified deficiencies, and monitor the results of agricultural development projects.
The 3DE sub-index consists of 12 indicators and evaluates empowerment among three types of agency, which reflect the types of generative power.Rowlands (1997) proposed the following: power to (instrumental agency), power with (collective agency) and power from within (intrinsic agency).Power to refers to generative or productive power (sometimes incorporating or manifesting as forms of resistance or manipulation) which creates new possibilities and actions without domination.Power with refers to "a sense of the whole being greater than the sum of the individuals, especially when a group tackles problems together".Power from within is understood as "the spiritual strength and uniqueness that resides in each one of us and makes us truly human.Its basis is self-acceptance and self-respect which extend, in turn, to respect for and acceptance of others as equals."(Rowlands, 1997, p. 13).

Adaptation of 3DE sub-index to the context of Colombian coffee-growing
Despite the Pro-WEAI being a robust and adequate tool to measure women empowerment in agriculture, note that its use has been observed mainly in large-scale projects with different type of crops in Southeast Asia and Africa.Therefore, for its application in the Colombian context, a review and adaptation of the 3DE sub-index was required which involved two types of adjustments: those related to adaptation to the context of coffee-growing women in Colombia and others aimed at incorporating improvements identified by the research team and other authors (Malapit et al., 2019;Narayan-Parker, 2002;O'Hara & Clement, 2018).
Adaptations considering the context of coffee-growing women consisted mainly of (a) modification of the indicator: "participation in groups" (collective agency), (b) the adjustment of adequacy criteria of some indicators such as "participation in groups", "visit to important places" (instrumental agency), and (c) adjustments aimed at increasing the cultural relevance of the questions in the questionnaire.These adaptations are further explained below and more details can be reviewed in the Changes to the Pro-WEAI 3DE sub-index Appendix.
The indicator "membership in groups" required important adjustment for the calculation of the 3DE sub-index as the AMUCC membership criterion invalidated this indicator.During the interviews, women frequently reported that the strength of their associativity was part of their achievements and a key element that allowed them to face difficulties.Accordingly, "membership in groups" was replaced by "participation in groups", and criteria concerning the sufficiency of participants' level of group participation and the influence on its decisions were adjusted.Adequacy was considered to have been achieved if they participated between "a lot" and "totally" and if they could influence the decisions of the group between "much" and "totally".Moreover, after conducting the pre-survey interviews and pilot test, the need to modify the adequacy criteria of some indicators was identified.In the case of "visiting important places", women reported mobilising freely.In this regard, freedom of movement without conditions, such as completing household chores, was a more appropriate criterion for participating AMUCC members.
In response to coffee-growing dynamics in the study area and to increase cultural relevance, agricultural and non-agricultural activities, the list of assets and options on ownership, and use of the land that were required for the indicators of "income autonomy", "control over income", and "ownership of land and other assets" was adjusted.As for the "work balance" indicator, the paid and unpaid work, personal care, and recreational and study activities considered in the questionnaire were adapted after considering the National Survey of Time Use in Colombia (ENUT) as a reference.
Regarding the indicator "contribution in productive decisions", the AMUCC women indicated that participation of all household members in decision-making was important for them, which is why this response option was incorporated.
Due to AMUCC women reported that they assumed household chores to be their responsibility; these activities were included under a single statement when respondents were asked if neglect a household chores could justify "attitudes about intimate partner violence against women".In addition, a question about planning was included to consider sexual and reproductive rights.
By characterising the "relationships between household members" to eliminate biases of what is socially accepted, respect was replaced by "good treatment", therefore expanding the spectrum of possibilities considered by women.This concept is promoted in Colombia through violence prevention initiatives, thus it was considered as a reference that may be closer to the thinking of AMUCC women.
Adaptations aimed at incorporating improvements comprised the introduction of three new indicators for a total of 14 indicators: "critical consciousness", "community participation", and "social control".The indicator "critical consciousness", which investigates participants' recognition of gender rights and stereotypes as structures and processes influencing the unequal position of women in society, was based on the approach by O' Hara and Clement (2018).It underscores the importance of including measures of critical consciousness when measuring agency (Malapit et al., 2019) and responds to changes that the AMUCC members manifested towards their rights, as illustrated by the following quote from an interview: . . . in past times, one lived as a woman according to what the husband or other men said was done, but not now; now as a woman, one has her rights, that is, a woman can do the things she needs to do, and she already has such a right.The indicators "community participation" and "social control" were based on the "individual leadership and influence in the community" domain of the WEAI (Alkire et al., 2013), which captures important aspects of inclusion and participation, accountability, and capacity for local organisation as key elements of empowerment (Narayan-Parker, 2002).Second, one of the lessons researchers learned in the development of the Pro-WEAI was that approaches to measuring collective agency that are independent of group membership should be investigated (Malapit et al., 2019;Yount et al., 2019).
Community participation refers to people's capacity for working together to organise and mobilise resources to solve problems of common interest (Narayan-Parker, 2002).The questions were adapted from the WEAI (Alkire et al., 2013).Membership in influential groups was replaced by "community participation".This indicator assesses women's contributions in their communities.Similarly, it responds to some of the achievements that the participating women declared they wanted to achieve in line with their values.For example, one participant told us: I want my children to know that they have a persevering, fighting mother, who spends her time doing good for others and supporting the whole community. . .it's time to leave them a legacy of solidarity with others, to be there when people need it.
Once the questionnaire was adapted, including incorporation of the questions on respondents' sociodemographic characteristics, a pilot test was carried out to guarantee rigorous validation of the survey structure and the grammatical and semantic coherence of the questions, i.e., validation of content equivalence.The questions were simplified or adapted to the context of the AMUCC members, according to the test results.
We collected data for calculation of the 3DE sub-index in January 2020.Surveys were administered to 109 active AMUCC members in households comprising more than one person.We used a digital questionnaire to monitor the quality of the data in real time.This survey comprised ten modules and collected information on the various indicators associated with the three types of agency: intrinsic, instrumental, and collective.

Calculation of the 3DE sub-index
In calculating the 3DE sub-index, we followed the same steps proposed by Malapit et al. (2019).All 14 3DE sub-index indicators were weighted the same, and each indicator was assigned an adequacy criteria (Appendix), such that a woman who did not meet this cut-off was classified as having reported an inadequacy in that indicator and assigned a value of one and otherwise zero.The number of indicators associated with the three types of agency for which each woman did not reach adequacy was then identified.Women experiencing disempowerment were identified as those who reported inadequacy in three or more indicators (the empowerment threshold) in the weighted calculation of the M_0.
The disempowerment index 〖 (M〗_0) was calculated using the Alkire-Foster method-an axiomatic, count-based approach originally designed to measure multidimensional poverty (Alkire & Foster, 2011).This index registers the percentage of women experiencing disempowerment and the average proportion of inadequacies they experience (Malapit et al., 2019, p. 690).M_0 ranges between 0 and 1, where values close to 1 show a high level of disempowerment and values close to 0 show a high level of empowerment.

Qualitative method
The qualitative work was developed in different phases of the research.Initially, interviews were conducted with 14 AMUCC members to explore their perceptions regarding the concept of empowerment and to validate the types of agency and the 3DE indicators, thus allowing us to adjust the survey prior to administering it.
After conducting the survey, we assembled five focus groups comprising 18 women from six municipalities; we sought representativeness according to age, ethnic group, educational level, and household composition criteria to deepen and clarify some of the survey results on (a) critical consciousness, (b) workload/seasonality, (c) autonomy and contribution in decision-making, and (d) attitudes about intimate partner violence.
We carried out an integrated bi-data analysis, considering the results of the quantitative and qualitative methods and preserving their nature (description, numerical analysis, and textual interpretation), to understand the study objective more deeply.
Finally, a workshop was held with AMUCC members to facilitate their appropriation of and critical reflection on the results of the 3DE sub-index in a participatory manner.Thus, they became methodological tools in an action plan, expanding their critical consciousness.Critical consciousness is understood from the perspective of Schuler (1997) as "the ability to make a critical analysis of the causes and consequences of social relations and, ultimately, to undertake actions aimed at achieving transformation" (p.35).

Ethical considerations
Every participant signed an informed consent form at the beginning of the interaction between researchers and the association's members.

Results
The M_0 score, indicating the percentage of disempowered women and disempowerment intensity was 0.24 (Table 2), while the 3DE score, which represents the achievements of women across the 14 indicators, was 0.76.Approximately 69 of the participating women (63.3%) showed disempowerment.The mean inadequacy score for those women who were not yet empowered was 0.37, i.e., these (disempowered) women reported inadequacy in, on average, 37% of the indicators (approximately five indicators).
Table 3 shows the number and percentage of women who reported inadequacy and disempowerment, the number of respondents who reported inadequacy for each indicator, and the contribution of each indicator to the reported disempowerment.A higher percentage of women reported inadequacy in three indicators: social control (88.1%), visiting important places (79.8%), and work-life balance (71.6%).Notably, the highest percentage of respondents who reported disempowerment and inadequacy was concentrated among these three indicators, which present the largest contributions related to disempowerment at 17.5% (social control), 18.1% (visits to important places), and 14.2% (work balance).

Discussion
The discussion revolves around those indicators contributing the most to women's disempowerment as reported in the previous section and relates to the indicators and elements considered most relevant among the qualitative findings.

Social control in the local and institutional contexts
Social control is the indicator in which most women (88.1%) reported inadequacy.The result of this collective agency indicator contrasted with the other two indicators of group and community participation wherein 15.6% and 11.0% of women reported inadequacy, respectively.
The results show that women's participation occurs at the community level in local spaces.However, in broader institutional spaces where the exercise of citizenship implies accountability, their participation is limited.Three potential reasons for this are the prevalence of armed conflict between various groups, a weakened institutional context, and the absence of training which promotes the exercise of citizenship.
In Colombia, the lack of state presence in different regions and the absence of a state monopoly on force has reduced the nation's capacity to mediate social conflicts.Thus, resolution of these conflicts often occurs outside of public institutions (Orjuela, 1998).Therefore, the dynamics of violence that permeate the Cauca Department inhibit AMUCC members' participation.For example, one respondent told us: When one has resources to give to people, everything is fine, but when one does not have such resources, criticisms, enemies, threats, and things like that come; this is why I didn't like to participate in activities that are community related.
On the other hand, participation in protests drives accusations; thus, there is a widespread fear of participating in this type of activity.In some situations, the capacity for agency is not enough for people to exercise agency, as it is impacted by the structure of the opportunities, they encounter (Narayan, 2005).Thus, distrust in institutions inhibits the exercise of social control as women report not feeling heard in spaces where they have voiced disagreement or claimed a right.Complicating matters, institutions are perceived as inefficient and involving overly long and complex processes.
Although the AMUCC offers members full freedom to participate, we observed that the association does not provide training to guide social mobilisation issues or political capacities.On the contrary, association leaders hold negative perceptions of political exercises and claim to not want any kind of involvement with it, as they mentioned in the focus group meetings.According to Kabeer et al. (2012), access to material resources through financial services alone does not translate into participation in "good governance" at a local level; in addition to better access to material resources, directly strengthening political capacities is required.
In the same vein, Thakur et al. (2022) mention the protection of women farmers'rights as one of the main challenges and obstacles that small landholders in India face to secure production conditions and the prospects for international trade.

Participation in decisions and gender stereotypes in the production and domestic spheres
According to the survey results on participation in decision-making by AMUCC members, 99.1% participated in decisions on coffee cultivation, 97.2% on poultry and other small animal husbandry, and 81.7% on subsistence crops.Ultimately, 65.7% and 61.8% of women made group decisions on the cultivation of coffee and subsistence crops, respectively, while for poultry breeding, 61.3% made decisions alone.As for household purchases, women's participation was very high for routine purchases (93.6%) and very low for large occasional household purchases (9.2%).
These results can be further explained from the contributions of the qualitative results; on the one hand, the women repeatedly stressed the importance of all household members participating in decision-making because they considered it a significant factor in uniting the family and continuing the legacy of coffee cultivation.However, family decision-making can mask the presence of decision-making patterns related to gender-based divisions of labour (Carnegie et al., 2020).Thus, by identifying only who makes the decision when measuring agency, researchers may overlook the actual agency or lack thereof experienced by women in agricultural settings, especially in joint decisions (Garcia et al., 2021).
The AMUCC members had broad technical knowledge of coffee cultivation management; however, they also held gender stereotype-based divisions of labour: "the weeding and stumping works are performed by men because these require the use of force. . .They are more skilled for that job, and we are more skilled at the jobs in the house".The women keep records, feed workers, and handle the coffee drying.They also oversee poultry breeding and gardening, which are activities that demand a lot of time but do not necessarily generate income.As the products yielded are mostly for the family's internal consumption, this work is also linked to caring for the household.The marketing of coffee is the exception, as it is performed exclusively by women through the AMUCC.
This gender-based division of labour implies that women devote an average of eight hours a day to unpaid work activities, such as domestic and childcare tasks and caring for the elderly or people with special needs.They spend an average of two and a half hours per day on coffee-growing activities; this number may be relatively low because the survey was carried out in the second month of the year, prior to the harvest.Indeed, according to the qualitative evidence, women have no free time during the coffee harvest season (three months a year).
Although there are context-related variations in gender-based divisions of labour, social norms, and the production system itself, according to other research, such as undefined study on two coffee-producing organisations in Oaxaca, Mexico, women who grow coffee report not having enough time to complete their tasks, a mix of productive and reproductive work.Disproportionate workloads and time unavailability can limit women's possibility of receiving all the benefits of production and their participation in producer associations (Lyon et al., 2017).In that vein, research on empowerment in abaca, coconut, algae, and pig value chains in the Philippines by Malapit et al. (2020) shows how gender stereotypes influence gender-based division of labour and how women spend their time, which impacts their level of participation and the number of benefits they receive.
In relation to the mobility indicator, Mudege et al. (2015) found that groups of women potato farmers frequently reported a series of social restrictions, such as the performance of domestic tasks and caring for children and the elderly.These duties greatly limited their availability and became a large barrier to their free physical mobility, mainly when social norms designate men as the decision-makers in the household.
In the pilot test and the qualitative research, AMUCC members reported having total freedom to mobilise.However, our findings on the indicator "visit to important places", which evaluates conditions required for women to mobilise, confirmed that the burden of domestic work conditions limits the possibility of becoming involved in activities of interest, such as participation in community groups and training activities.For example, one AMUCC member stated: "[I] [leave] everything ready in the house that needs to be left ready, and then I freely go and participate in my events".
Although AMUCC members recognise that unequal burdens due to gender-based divisions of labour resulting from negative social constructs requiring transformation exist, it remains difficult to carry out reforms because, given their responsibility for domestic work, they would need to first demand assistance and collaboration from men in domestic activities.The verbs "help" and "collaborate" suggest the invisibility of the double workload and the absence of significant male co-responsibility for domestic tasks.These arguments coincide with AMUCC members' responses to the indicators of the "critical consciousness" item, given that approximately 20% reported agreeing with divisions of labour based on gender stereotypes and motherhood mandates; the other members reported a more general rejection.In this regard, although most AMUCC members participate in and contribute to decisions about production issues (99.1%) and actively participate in groups (84.4%) and the community (89%), in many cases, traditional power structures within the household remain intact (Niyonkuru & Barrett, 2021;Tahir et al., 2018).

Male authority and challenges to power relations
To achieve greater power, women coffee growers must constantly face socially constructed and supported male authority figures.This empowerment limitation was observed in their responses to items on all indicators of collective agency, including "visits to important locations" -the indicator of instrumental agency-and some indicators of intrinsic agency, such as "relationships between members of the household", "critical consciousness", and "attitudes of violence".
Women's mobility is conditioned by domestic work and within household power structures.Participating women reported constantly facing situations of confrontation or negotiation with their partners when they attended training or meetings of the groups they participate in: When I started going to training, it hit him hard, and I found him upset and asked him, 'Why are you upset?' I told him, 'You don't have to be upset'.I confronted him and told him he could not be upset. . .'I'm going in order to acquire knowledge'.He understood, and then I taught him [some things].I told him, 'This must be done like this'.
In certain cases, participants' male partners exercised power over them to such an extent that they did not attend training or meetings to avoid conflict with their partners or because their partners directly forbade them from going.This may be related in some way to the high percentage of women (43.4%) who reported feeling uncomfortable expressing disagreement with their partner, although 90.4% of participants consider their relationship with their husbands or partners to be characterised by "good treatment".Confronting and eroding inequalities such as these can lead to discomfort, as can be seen in participants' narratives.For example: "I don't consider myself capable of telling my husband to help with household chores" and "From the start, you can never let the husband take advantage, otherwise you'll never be able to say 'no'".Findings from studies conducted with women in African countries (Fröcklin et al., 2018;Niyonkuru & Barrett, 2021) show more evidence that men can control women's activities and the places they visit.
The exercise of power over another can involve violence (Lukes, 1974).While 96.3% of participants rejected violence in all situations and many held a critical attitude towards violence against women, some reported experiencing psychological, physical, economic, and sexual violence from an intimate partner and being directly or indirectly hurt by this experience.The results of other studies also indicate that violence is a form of domination that continues to pose challenges to the empowerment of women involved in different production activities (Niyonkuru & Barrett, 2021).

Conclusions
Our experiences with AMUCC members indicate that they encounter both empowerment and disempowerment in their social relations during their daily lives.AMUCC members are energised in private and public spheres: as women coffee farmers in the sense of productive activities and as generators and caretakers of life and in the broad sense of reproductive activities.
Their numerous years in high-quality coffee production (along with their households) and agricultural development projects as well as their membership and active participation in the AMUCC have allowed them to improve their technical knowledge through training, equipment access, and credit by implementing good practices for obtaining fair trade certifications and penetrating international markets, as evidenced by adequacy levels above 97% in four of the six instrumental agency indicators.However, an emphasis on productivity-related indicators can bring an economistic perspective that is intended to result in empowerment.Nonetheless, the study results show that, for AMUCC members, their high level of participation in the productive components of coffee growing and their associativity do not allow them to completely overcome disempowerment.Case in point, 63.3% of participating women reported being disempowered and reported the highest inadequacy along the indicators "social control" (88.1%), "visits to important places" (79.8%), and "work-life balance" (71.6%).
The construction of dialogue scenarios, mediated by the preservation of trust and respect for listening, allowed us to investigate together with AMUCC members how the sociocultural conditions they experience in the private and public spheres affect their empowerment.Such conditions include physical, sexual, and psychological violence by their intimate partners within household power structures, fears about violent situations resulting from the Colombian internal armed conflict, and institutional weaknesses that prevent them from more actively exercising citizen leadership.Likewise, the discussions allowed us to identify a gap between participants' "politically correct" attitudes and the rights violations that are typical in their daily realities.Collectively recognising this gap with AMUCC women in a participatory manner facilitated their acceptance of the importance of generating association-led pedagogical processes aimed at rendering these violations visible, strengthening dialogue with state institutions responsible for the restoration of rights, and energising institutional and community support networks to prevent and address rights violations within families and communities.
It should be noted that, although the AMUCC can promote its members' empowerment, social and cultural gender roles being reproduced in society entail relations involving subordination that sink their structural roots into the subjectivities of men and women, thereby rendering their transformation of such roles difficult.This limits the potential of the AMUCC in facilitating the women's empowerment process, as reflected in most indicators.
Finally, it is important to emphasise that relationships between types of agency must be conceived in a systemic way, noting their potentiating and synergetic-not mechanical-character between them.Expanding "power to", "power with", and "power within" is possible for AMUCC members if the actions they implement are based on strategic interest: i.e., with the political intentionality of transforming subordination relationships (Young, 1997).Concretising transformative political intentionality also requires examining disempowerment.Here the factors influencing empowerment must be read from a relational perspective because overcoming economic dependence or improving income alone guarantee neither the expansion of critical consciousness nor the transformation of subordination relationships.

Limitations
The overall size of the population we targeted does not allow us to deepen the results by expanding the sample along categories such as educational level, ethnicity, gender, and region.Additionally, our contextual adaptations to the 3DE sub-index and failure to calculate the GPI do not allow for comparisons with other international works that have applied Pro-WEAI to measure the empowerment of women in agricultural development projects.These limitations should be considered when reviewing the scope of the research.

Self-efficacy
No change.

Attitudes about intimate partner violence against women
Of the five situations proposed, two were changed: (B) 'neglecting children' and 'burns the food' was changed to 'neglecting household chores' and (E) 'makes a decision or refuses to use contraception methods' was included.
It was identified that AMUCC women are usually in charge of household chores, including food preparation and childcare, and assume that such chores are their own responsibility; in that regard, those items are both included under the broader category of 'household chores'; the planning decision item was added to consider sexual and reproductive rights.

Relationships between household members
Respect for good treatment was changed [1] when relationships between household members were characterised.Thus, the name of the indicator was adjusted.
The questions about whether they respect another person in their household and whether they feel respected were consolidated into a single question: 'Do you consider the relationship with [MEMBER OF THE HOUSEHOLD] to be characterised by good treatment?'When respondents were directly asked whether they respect other family answers, their answers implied a value judgment about themselves that could be mediated by what is socially accepted.In contrast, in the pre-survey interviews, participants showed that they considered treating each other with respect (although these responses could be biased by what is socially accepted) as positive.Therefore, we replaced 'respect' with 'good treatment' to expand the spectrum of possibilities considered by the women.In Colombia, this concept is promoted in violence prevention programmes, which is why it was considered a reference that may be closer to the thinking of AMUCC members.
In the pilot study, the question 'Do you trust your [RELATIONSHIP] to do what is best for you?' was changed to 'Do you believe that the actions carried out by [PERSON] are in accordance with your best interests?'However, the question generated confusion and gave rise to multiple interpretations, so we decided to eliminate it.

Critical consciousness*
The indicator and the conditions that must be met for adequacy were created.
This indicator was created considering the work by O' Hara and Clement (2018) to investigate the recognition of gender rights and stereotypes, such as structures and processes influencing the unequal position of women in society. (Continued) Yoshioka Vargas et al., Cogent Social Sciences (2023) The answer option 'household' was added because women repeatedly highlighted in the pre-survey interviews that it was important for them to involve all household members in decision-making.This is related to challenges and objectives involved in their intention to pass on the legacy of coffee cultivation (given the achievements they have made through their work).They believe that participating in decision-making is important for uniting the family and motivating sons and daughters to continue working in the field to develop better opportunities and prevent them from migrating to cities.

Ownership of land and other assets
The number of questions on land ownership was reduced from five to three.In asset ownership, the descriptions of some articles were adapted to the AMUCC member context.They differentiated urban houses and rural houses as two types of 'great assets'.
To ease survey administration, it was considered important to simplify the questions on land ownership and whether the land was cultivated.In some cases, AMUCC members cultivate land that can be leased or that is owned by a family member.Accordingly the following three questions were established for land ownership: G3.01.'Do you own any of the land your household cultivates?' G3.02.'Who owns the land cultivated in your household?' G3.03.'Do you cultivate land alone or do you cultivate it together with another party?'

Access to and decisions on financial services
No change.

Control over use of income
In addition to questions about input in decisions on how to use income, the following question was added: 'Who owns the income from [ACTIVITY]?' The question 'Who owns the income from [ACTIVITY]?' was added because participants indicated in the pre-survey interviews that they had income from non-agricultural economic activities that they considered exclusively their own (e.g.work outside the farm, product sales).Therefore, we considered it important to identify whether they generated income they considered exclusively their own from certain activities.This question did not affect the calculation of the index.

Work balance
Participants were asked to record the amount of time spent on given activities on a typical day by 30minute time intervals rather than 15minute time intervals.
We referred to both the original survey activities in the Pro-WEAI and the National Survey of Time Use in Colombia when considering which activities to include.Given the complexity of locating codes for each activity during administration of the pilot test and the detailed monitoring of the hours spent on each activity, we decided to group the activities under the following five codes: 1) personal, recreational, and studyrelated care; 2) unpaid work: housework, community activities and paperwork; 3) paid work; 4) coffee cultivation; and 5) caring for children, adults (sick, elderly), or people with special needs.In the survey, each of the above-referenced groups of activities was considered.We coded coffee cultivation and caring for people separately because we considered it important to identify in detail how much time women spent on each of these activities.
In the pilot test, we noticed that women described their daily activities sequentially rather than according to time intervals.
When we asked the women to estimate how much time they spent on the given activities, 15-minute time intervals were difficult for the women to reference.Thus, 30-minute time intervals were used. (Continued)

Indicator
Changes to the Pro-WEAI 3DE sub-index

Visits to important locations
Questions about the different locations that could be visited were adapted to the context of the AMUCC member.
In the pre-survey interviews, the women did not report major limitations to their mobility.Further, during the pilot test, when asked who usually decides which locations they visit and whether they visit the given places, they indicated that it was taken for granted that they decide these things themselves.Thus, instead of asking who usually decides whether they can go to a [location] and if anyone would object, we asked whether women felt they must meet any conditions to visit the given places.
Beyond the frequency with which participants visit the given places, not needing to meet conditions to visit the given places was considered a cut-off for their having achieved adequacy.The given conditions were adapted to the context of the participating women and according to the experiences referred to in previous qualitative studies.

Participation in groups*
The indicator was created to enforce stricter conditions for having achieved adequacy because all respondents are AMUCC members.
Instead of considering whether the respondent was an active member of at least one group as a cut-off for having achieved adequacy, we asked the degree to which they perceived they could participate and influence the group.Adequacy was considered to have been achieved if they participated between 'a lot' and 'totally' and if they could influence the decisions of the group between 'much' and 'totally'.

Community participation*
This indicator and conditions for adequacy were created.
In the pre-survey, the women reported active participation in different groups that influence the community.They referred to directing actions from their positions for the benefit of this participation and recognised themselves as leaders.Their leadership was focused on identifying the needs of the community and helping solve different problems.Accordingly, some questions about the leadership domain in the WEAI were referred to in developing this indicator of public participation and having the agency to exercise voice through participation in collective action (Alkire et al., 2013).

Social control*
This indicator and conditions for adequacy were created.
In view of women's active participation in their communities, we investigated their participation in group activities aimed at social control.As for the previous indicator, we referred to some questions on the leadership domain from the WEAI (Alkire et al., 2013).

Note: * New indicators
[1] Good treatment refers to practices and relationships of love and respect that one has with oneself and others.The keys to good treatment are offering security, using both simple and accessible verbal and body language, knowing how to listen, being respectful and kind, and acting in a courteous and empathetic manner.

Adequacy criteria
Agency type Indicator Adequacy criteria

Intrinsic Autonomy in income
A Relative Autonomy Index (RAI) was used to calculate this type of motivation as well as the answers to four statements about motivation regarding how to use the income generated in agricultural and non-agricultural activities.Regarding the question "Are you like this person?","Yes" answers were assigned a value of one, and no answers were assigned a value of zero.Statements were weighted using the following scheme: 0 for Choice 1 (no alternative), −2 for Choice 2 (external motivation), −1 for Choice 3 (introjected motivation), and+3 for Choice 4 (autonomous motivation).A reported RAI of greater than or equal to one indicates adequacy in this indicator.

Self-efficacy
Responses to the eight statements on the self-efficacy scale must be either "I agree" or "I totally agree".The score must be greater than or equal to 32.

Attitudes about intimate partner violence against women VCA**
The respondent must point out that the husband or partner has NO justification for physical or psychological abuse of his wife or partner among the five proposed situations.

Relationships between household members**
The respondent must point out that the relationships with her husband or partner and the other members of the household (over 13 years old) are characterised by good treatment "most of the time" and that she feels comfortable not agreeing with the husband or partner and other members of the household "most of the time." 3

Critical consciousness*
The respondent must respond to all statements about gender stereotypes with "strongly disagree" or "disagree". (Continued) Yoshioka Vargas et al., Cogent Social Sciences (2023) The respondent must meet at least one of the following conditions for all agricultural activities in which she participates: 1) makes decisions alone, 2) makes decisions jointly with at least some contribution to the decision, or 3) feels like she could make a decision if she wanted to (at least to a medium degree).

Ownership of land and other assets**
The respondent must report owning individually or jointly at least one of the following assets and/or items: 1) two large assets or 2) land.

Access to and decisions about financial services
The respondent must meet at least one of the following conditions: 1) belongs to a household that used a source of credit in the past year and participated in at least one single or joint decision about this; 2) belongs to a household that did not use credit last year, but could have borrowed from at least one source; or 3) has access, individually or jointly, to a bank account or other financial institution.

Control of use of income
The respondent must participate in decisions regarding how to use both income and products from all agricultural activities in which she participates unless no decision has been made.

Work balance**
A workload indicator (CT) was constructed to capture the time spent on each activity.The hours dedicated to primary activity (SAP) and half of the time dedicated to secondary activities (SAS) are added, with CT being: CT = SAP + (0.5 × SAS) The respondent shows adequacy in this indicator if they have a workload of less than 10.5 hours a day.

Visits to important locations*
The respondent must meet at least one of the following two conditions: 1) visits at least two locations (the urban centre, family, friends/neighbours) once a week and are not required to fulfil any conditions for mobility or 2) visits at least one location (health, community meeting, trainings, places outside the community or village) at least once a month and does not require any conditions for mobility. (Continued)

(Continued) Collective Participation in groups*
The respondent must meet one of the following two conditions in at least one of the groups in which she participates: 1) she participates in the group between "much" and "totally" or 2) she believes she could influence group decisions if she wished to do so between "much" and "totally".

Community participation*
The respondent must meet at least one of the following three conditions: 1) she has contributed money or time to village or community development projects, 2) she has contributed money or time to build or maintain religious sites, or 3) she has contributed to collaborative practices in her community. 4

Social control*
The respondent must meet at least one of the following three conditions: she has participated in oversight of 1) the proper use of public resources, 2) the misconduct of authorities or other officials, or 3) conflict resolution in the community.
Note: * New indicators ** Indicators with changes to the adequacy criteria.

doi.org/10.1080/23311886.2023.2204633 Table 3. Percentage of women who reported inadequacies by indicator and relative contribution to disempowerment Respondents who reported inadequacy* Respondents who reported disempowerment and inadequacy**
The column on respondents' reported inadequacy shows the proportion of respondents who reported inadequacy in a given indicator regardless of their empowerment status.**The column on respondents who reported disempowerment and inadequacy shows the proportion of respondents who reported disempowerment and inadequacy in the given indicator.