Unraveling the connection between coffee farmers’ value chain challenges and experiential knowledge: the role of farm family resources

ABSTRACT

Purpose

Multiple value chain challenges confront smallholder farmers, which necessitate context-specific solutions. Family resources, such as information and production inputs, are valuable assets for farmers. When properly used, farmers’ family resources can help them in learning how to address value chain challenges. Yet, the learning in rural agricultural value chain literature still does not inform how family resources influence farmers’ learning.

Design/methodology/approach

Face-to-face interviews with 214 coffee farmers were used to investigate how family resources shape farmers’ experiential learning process. The data was analyzed using PLS-SEM.

Findings

Results show that family resources play a crucial role in farmers’ experiential learning process, particular in reflecting on and addressing value chain challenges they are confronted with.

Practical implications

Smallholder farms, as a collective and farmer-centered experiential learning context, can serve as a source of inspiration for extension agents bringing the paradigm shift from technology transfer to participatory advisory services to reality.

Theoretical implications

The study contributes to experiential theory in the context of agriculture by advancing a model on how rural family support can function as a resource to change the mechanisms underlying farmers’ experiential learning.

Originality/value

The smallholder farm is a node in larger social learning networks (e.g. Innovation platform), where resources such as information, labor, emotional support, and production inputs, circulate.

KEYWORDS:

• Agriculture
• coffee value chains
• social learning
• experiential learning process
• family resources
• smallholder farmers

1. Introduction

Approximately 90% of farms around the globe are operated by families (Graeub et al. 2016; Lowder, Sánchez, and Bertini 2021). They provide 80% of the world’s food, jobs, and 2.2 trillion dollars in income (Bosc et al. 2013; Graeub et al. 2016). With a value of 19 billion USD, or 70% of total coffee exports, coffee is the most important crop enterprise for over 50 low-income countries in terms of export earnings (Kuma et al. 2019). Coffee contributes 20% of Uganda’s total exports and provides a significant source of income for 1.7 million smallholder coffee farmers (UCDA 2020). However, Uganda’s coffee exports are low when compared to African counterparts such as Ethiopia, Kenya, and Rwanda, despite its potential as Africa’s second-largest Arabica coffee exporter after Ethiopia (ICO 2020). This is mostly due to the sector’s reliance on smallholder farmers,¹ who face several challenges in their farming process i.e. production, harvest, postharvest handling, and marketing. At production, for example, insect pests and diseases (Liebig et al. 2016), cause up to 57% coffee yield loss (Cerda et al. 2017), as well as low quality (Velmourougane, Bhat, and Gopinandhan 2010; Pimenta, Angélico, and Chalfoun 2018; Walker et al. 2019) which inturn leads to low and fluctuating coffee market prices (Abrar, Solomon, and Ali 2014; Kidist, Zerihun, and Biniam 2019). The latter is a typical example of a complex coffee farming challenge that necessitates multiple solutions. Complex farming challenges have several dimensions (Schut et al. 2015), are embedded in interactions across different organizational and social settings (Giller et al. 2008) and involve multiple actors (Hemmati 2012). As a basis, a range of actors (e.g. researchers, donors, policymakers, and practitioners) have embraced the coffee value chain approach as a solution to farmers’ challenges (Kaplinsky, Terheggen, and Tijaja 2011; Collins, Dent, and Bonney 2016; Ponte et al. 2014).

Innovation platforms (IPs) are the most common operationalization of coffee value chains in low-income countries (Pali and Swaans 2013; Camacho-Villa et al. 2016; Kilelu, Klerkx, and Leeuwis 2014). IPs are structured interfaces among farmers where they tap into the capacities of diverse actors (e.g. processors, traders, transporters, input suppliers, output handlers, policymakers, extension agents, and researchers) to learn how to address their farming challenges (Tui et al. 2013). For instance, Ochago et al. (2021) found that when challenged with pest and disease infestation, coffee farmers joined together, deliberated, shared their experiences, and purchased certified coffee inputs as a group. In this case, farmers rely particularly on experience-based knowledge as it has practical, personal, and local relevance and is accumulated over long periods by doing, experimenting, and watching (reflective observation) (Šūmane et al. 2018). Consequently, farmers who learn to address their challenges are more able to generate context-specific solutions (Janssen and Swinnen 2019).

Experiential learning (EL) is a learning approach that involves addressing challenges in the farming process (Percy 2005; Pincus et al. 2018). Individuals learn to overcome challenges through reflecting on prior challenges, sharing practical ideas with others, and working together to solve challenges (Laforge and McLachlan 2018; Oreszczyn, Lane, and Carr 2010; Milestad et al. 2010; Lubell, Niles, and Hoffman 2014; Okumah et al. 2021). The existing research indicates that farmers’ EL processes are reliant on resources obtained through family relationships, among other factors. When faced with challenges along the coffee value chain, for example, Ochago et al. (2021) found that farmers rely on information, labor, emotional support, coffee production inputs, linkages to training avenues, and supportive actors by family members. These resources help to engage in learning activities that lead to the development of challenges-solving knowledge. Farmers learned how to properly space coffee plants, apply fertilizers, and spray against pests and diseases through observing and interacting with fellow family members during regular farming tasks (e.g. planting, pest and disease scouting and control). In agreement are studies by (Hoang, Dufhues, and Buchenrieder 2016; Fisher 2013; Sutherland and Burton 2011; Hoang, Castella, and Novosad 2006; Pratiwi and Suzuki 2017; Danielsen et al. 2020) that found that family relationships increased farmers’ experiential knowledge. For example, farmers acquire information about pest and disease management through their spouses, friends, and neighbors, according to Danielsen et al. (2020). A few studies, on the other hand, argue that family interactions create homogeneous and redundant knowledge within the family (Fisher 2013), inhibiting the acquisition of new knowledge outside the family (Smith, Anderson, and Moore 2012; MacGillivray 2018). Family relationships, from this perspective, may isolate farmers from knowledge brokers (such as advisors and extension staff), reducing their ability to carry out farm tasks, let alone develop new knowledge about farm techniques (Fisher 2013; Tregear and Cooper 2016). Nevertheless, a variety of research has now demonstrated a favorable association between family resources and farmer learning, i.e. the relationship between challenges in the farming process and the level of knowledge gained through learning activities. However, the extant studies on family resources and farmer learning are primarily qualitative, descriptive in character, and focused on the relationship between family resources and knowledge – one component of the experiential learning process. It is still unknown how family resources influence the experiential learning process. As a starting point, the goal of this study is to understand how the farmer’s EL process is influenced by the resources he or she has access to through family ties.

2. Theoretical foundation

Many scholars agree that EL is an important component of learning methodologies for farmers in rural areas who want to improve their ability to cope effectively with various complex farming challenges (Pincus et al. 2018; Roberts 2006; Ochago et al. 2021). Experiential learning has been used as a foundation for extension interventions in the context of adult learning (Roberts 2006) including interventions at family farms (Abbey, Dowsett, and Sullivan 2017). Kolb’s experiential learning theory is commonly employed in existing research to explain how learning unfolds (Matsuo and Nagata 2020; Morris 2020). Experiential learning, according to Kolb’s model, is a context-dependent process in which experiences are transformed into experiential knowledge (Kolb 2015; Kolb and Kolb 2009). Kolb’s definition of experiential learning indicates four interlinked concepts: (1) the experiences, (2) the knowledge created, (3) the transformation of the experiences, and (4) the context, for example, farmers’ families and their resources.

2.1. The experiential learning process

2.1.1. Experiences

According to Kolb (1984), learners must have tangible experiences to learn. Existing research on experiential learning describes experiences as challenges (Ochago et al. 2021; Morris 2020). The EL process comprises solving unique, context-specific, and ill-structured challenges (Blair 2016; Asfeldt and Beames 2017). In this paper, the value chain challenges faced by smallholder coffee farmers are highlighted. Smallholder farmers produce the majority of coffee, but they face a variety of challenges throughout the coffee value chain. Next to pests and diseases (Liebig et al. 2016) harvesting and postharvest practices account for more than 60% of a coffee bean’s overall quality loss during drying and hulling (Hameed et al. 2018). Finally, coffee quality, which is a product of both pre- and post-harvest operations, is the cause of low and fluctuating coffee market prices (Abrar, Solomon, and Ali 2014; Kidist, Zerihun, and Biniam 2019). Although these challenges are well-known, there is little research in the agricultural value chains and learning literature on how challenges kick-start farmers’ EL (Schut et al. 2019; Probst et al. 2019). For example, Ochago et al. (2021) in their research of farmers’ experiential learning in coffee value chains found that challenges like pests and diseases, poor quality and quantity of coffee, and low and variable coffee prices stimulated farmers’ EL. In line with a coffee value chain perspective, this study uses these interrelated elements to identify farmers’ challenges: challenges during production, harvesting, postharvest handling, and marketing.

2.1.2. Experiential knowledge

Knowledge is knowing something and knowing how to do something (Sternberg 2002). According to Johanson and Vahlne (1977), experiential knowledge is information learned solely from personal experience. Experiential knowledge is created when a farmer generates, finds, and captures solutions to challenges in its value chain (Newman and Conrad 2000; Martín-de Castro et al. 2011). Accordingly, experiential knowledge refers to a farmer’s ability to align information with one’s own or with knowledge from other farmers and apply it to challenges-solving activities. Coffee IP farmers, for example, gain knowledge on new farming methods such as proper plant spacing, line planting, composting, fertilizer application, spray against pests and diseases, selective picking of red ripe cherries, etc. (Tahir et al. 2020; Iorlamen et al. 2021; Chichaybelu et al. 2021). Farmers get to know about value chain actors (e.g. fellow farmers, processors, traders, transporters, input suppliers, extension agents, researchers, governmental, and non-governmental organizations), and farming techniques through their IPs (Ochago et al. 2021; Lamers et al. 2017). Following extant research, this study uses two interconnected parts: knowing new value chain actors, and farming methods to define farmers’ experiential knowledge.

2.1.3. The transformation of farmer challenges into knowledge

In the Experiential Learning theory model, two ways of transforming experience are reflective observation and active experimentation (Kolb, Boyatzis, and Mainemelis 2001). According to Di Stefano, Pisano, and Staats (2015); Beard and Wilson (2013), reflective observation requires seeing, hearing, and discussing the experience – what happened, how it happened, and why it happened. Schön (1987)’s reflection theory breaks down reflection into two parts: reflection in action (Cajiao and Burke 2016) and reflection on action (Ajjawi and Boud 2018). Decisions made while practicing or ‘how teachers think on their feet,’ are referred to as ‘reflection in action’, p. 12 (Farrell 2012). Reflection-in-action is nearly entirely concerned with the challenges-solving process. People claim they are reflecting, for example, when they are deeply thinking about how to address complex challenges, according to Moon (2013). Reflection-in-action entails using observational analysis, listening, and/or touch or ‘feel’ to solve challenges. Because challenges in farming are multi-dimensional, these necessitate complex solutions. This often involves challenges solving-knowledge acquisition through the adaptive process of experimentation (Cajiao and Burke 2016; Di Stefano, Pisano, and Staats 2015). Reflection on action, on the other hand, takes place after the activity has been completed (Schön 1987). In other words, reflection-on-action is the act of looking back to evaluate what happened (Ajjawi and Boud 2018). So, the reflection includes identifying challenges, determining root causes, and considering viable remedies (Miller and Maellaro 2016). When faced with coffee value chain challenges (challenges at production, harvest, postharvest handling, and marketing), farmers, according to Ochago et al. (2021), reflect on their current knowledge and interactions with other value chain actors such as fellow farmers, processors, traders, transporters, input suppliers, extension agents, and researchers. Farmers’ level of experiential knowledge (knowing new value chain actors, and farming methods) increased when they reflected on their current knowledge. Hence, the following hypotheses were tested:

H1a. Farmers reflect on their current knowledge and interactions with other value chain actors when confronted with challenges.

H1b. Farmers who reflect on their current knowledge and interactions with other value chain actors increase their knowledge of farming methods and value chain actors.

Alternatively, as a direct response to challenges, farmers can act immediately through active experimentation (Kayes, Kayes, and Kolb 2005). Farmers are known to experiment (Leitgeb et al. 2014; Meynard, Dedieu, and Bos 2012). They experiment with new seed varieties and alternative production processes, and look for new ways to promote their products through their social networks. Farmers are, in fact, part of a larger social context, emphasizing the importance of networks. Skaalsveen, Ingram, and Urquhart (2020) found that the level of knowledge and experience amongst farmers was largely affected by the experimentation of farmers on their farms while exploring new ideas and techniques, and communicating this experiential knowledge through informal learning networks. When farmers used their current knowledge to solve challenges and interact with other value chain actors, their level of experiential knowledge increased (Ochago et al. 2021). Therefore, active experimentation occurs when farmers use their existing coffee value chain challenges solving-knowledge and interact with other value chain actors to increase their level of experiential knowledge. The following hypotheses were tested:

H1c. Farmers who face coffee value chain challenges use their current knowledge and interact with other value chain actors.

H1d. Farmers who use their current knowledge and interact with other value chain actors increase their knowledge of farming methods and value chain actors.

2.2. The moderated effect of farmer family resources

A family is a social construct that includes grandparents, parents, siblings, spouses, and eventually children and grandchildren (Pylyser, Buysse, and Loeys 2018; Finch 2007). Interactions among family members enable farmers to learn to overcome challenges through sharing resources including information, knowledge, labor, emotional support, coffee production inputs, linkages to training avenues, and supportive actors (Ochago et al. 2021). In their study, Ochago et al. (2021) found that when farmers faced challenges in their farming activities, they utilized their family resources to perform two main learning activities i.e. reflection and experimentation. Farmers sought advice from family members on good agronomic practices such as seed selection, land preparation, pest and disease management, pruning, mulching, postharvest handling techniques such as proper harvesting techniques, sorting, and drying well, storage, and marketing aspects such as collective marketing and alternative buyer sourcing. The challenge of inadequate labor was overcome by enlisting the help of family members to dig and maintain the coffee plantations, harvest coffee, and aid in the supervision of other hired pickers, guard coffee against thieves, and transport products to the sale point, to name a few responsibilities. While this study found that family resources regulate the association between challenges and learning activities in rural value chain contexts, it does leave a vacuum that has to be filled, in particular, to test the effect of family resources on the relationship between challenges and learning activities. This study hypothesizes (H2) that;

Hypothesis 2a: Farmers’ access to family resources positively moderates the relationship between their coffee value chain challenges and reflection on current knowledge and interactions with other value chain actors.

Hypothesis 2b: Farmers’ access to family resources positively moderates the relationship between their reflection on current knowledge and interactions with other value chain actors and their knowledge of farming methods and value chain actors.

Farmers’ ways of transforming experience (reflection and experimentation) have been related to the acquisition of experiential knowledge through the use of the farmer’s family resources. Farmers’ family resources enable farmers to reflect on and experiment with their existing coffee value chain knowledge, as well as interact with other value chain actors to increase their experiential knowledge, according to (Ochago et al. 2021). Even though this research suggests that family resources influence the association between farmers’ ways of transforming experience and their experiential knowledge in coffee value chain contexts, there is no research that associates individual farmer’s family resources with experiential knowledge. This study hypothesizes that;

Hypothesis 2c: Farmers’ access to family resources positively moderates the relationship between their coffee value chain challenges and their use of current knowledge as well as interaction with other value chain actors.

Hypothesis 2d: Farmers’ access to family resources positively moderates the relationship between their use of current knowledge as well as interaction with other value chain actors, and their knowledge of farming methods and value chain actors.

Integrating the previous sections suggests that the indirect relationships between farmers’ value chain challenges and their experiential knowledge via reflection and experimentation may be conditional on-farm family resources. Figure 1 depicts this dual-stage moderated mediation model. According to the model used in this study, the relationship between farmers’ value chain challenges and their reflection, as well as the relationship between their reflection and their level of experiential knowledge, will vary significantly depending on the level of family resources they have access to. Then, the relationship between farmers’ value chain challenges and their experimentation, as well as the relationship between their experimentation and their level of experiential knowledge, will vary significantly depending on the level of family resources they have access to.

3. Methods

3.1. Study location

The study took place in the districts of Kapchorwa, Manafwa, and Namisindwa in the Bugisu Sub Region of Uganda’s Eastern region. The district of Kapchorwa is divided into seven sub-counties. Manafwa is made up of ten sub-counties, whereas Namisindwa is made up of seven. Kapchorwa and Manafwa districts’ coordinates are 1.3350° N, 34.3976° E, and 0.9064° N, 34.2866° E, respectively (Google Earth, 2022). Kapchorwa, Manafwa, and Namisindwa have population estimates of 113,500, 157,900, and 220,000 people, respectively, according to the Uganda Bureau of Statistics (UBOS 2015).

Figure 1. Dual stage moderated mediation model (research framework).

Agriculture is the principal economic activity in the area, which is divided into three zones: highland, midland, and lowland. These terrain zones determine the types of farming activities that farmers engage in, as well as the crops that are grown. The highlands and midlands are dominated by coffee and bananas, while the lowlands are dominated by maize and bananas. Coffee is mostly grown by smallholder farmers on farms that are less than one acre in size, often intercropped with bananas (Jassogne, Lderach, and Van Asten 2013). Coffee yields in Kapchorwa range from 1556 kg/ha to 1776 kg/ha in Manafwa/Namisindwa. Under good management methods, the average yields for Arabica coffee in both districts are below the national average of 2000 kg/ha. The high occurrence of diseases and pests is mostly to blame for the low output potential.

3.2. Target population, sampling, and data collection

Coffee IP farmers in Uganda’s main coffee-growing regions of Kapchorwa, Manafwa, and Namisindwa were studied. IPs represent dynamic learning environments that support the adoption of innovations and where farmers interact. At the same time, there is a lot of heterogeneity among IPs in Uganda, in supporting services as well as in structure and membership. This is more advantageous since it gives a more level playing field for evaluating farmer learning than selecting individual farm households. Finally, due to their horizontal and vertical connections, the innovation platforms are currently the most popular farmer grouping. A total of 214 respondents (Appendix A) were interviewed for an average of 1 h and 15 min each using a standardized survey questionnaire. A stratified random sampling procedure was applied to recruit participants for the survey. As a sampling frame, a list of 450 current coffee IP farmers in the study site was used. The main author obtained a list from the research assistants at Makerere University’s Value chain Innovation platforms for food security (VIP4FS) project coordination office, which was validated by the district IP coordination team (IP facilitators/coordinators/chairpersons) during a one-day meeting with the main author. Because coffee is a male-dominated enterprise, the main author stratified the names obtained by gender. After that, he sorted the names and used Excel’s RAND function to select every second name on the sheet. Pretesting with a comparable group who did not participate in the study was used to assess the applicability of the structured interview instrument. Face-to-face interviews were conducted with 22 respondents (twenty by research assistants and two by the main author) in two central locations: Tegeres Sub County, Kapchorwa district, and Butta Sub County, Manafwa district. The pre-test helped to ensure interview time, question clarity, and a common understanding of the interviewing code words in the local languages. The completed questionnaires were used by the main author to create data templates and analysis of emerging results. The preliminary data analysis resulted in the refinement of the survey tool for the actual data collection. Appendix B contains the items for the variables that were constructed using the existing literature. Likert scale items were used to investigate all study components. Respondents can express their real feelings using Likert-type scales. Factors like reliability influence the number of response categories on a scale (Bendig 1954; Dawes 2008; Preston and Colman 2000; Krosnick 2018). Leung (2011) observed no differences in reliability, mean, or standard deviation for 4, 5, 6, and 1-point Likert-type scales. For both the research attributes and the responders’ group in this study, a five-point Likert scale seemed appropriate. The responses were graded, with options ranging from strongly agree (5) to strongly disagree (1). During the data collection stage, each research assistant conducted a face-to-face interview with a respondent at their home. All interviews were completed for one district before moving on to the next, and the interview results were recorded during the interviews on hard copy questionnaires. The main author interviewed one respondent on the first day and one respondent halfway through the interviews for each district as a quality measure and to formalize how he would later analyze this data. He ensured data quality through thorough training of research assistants and using research assistants who are proficient in the local dialects. He held three separate training sessions for the research assistants. Also, he held team debriefs every day after the data collection exercises to share lessons and challenges to ensure a uniform interpretation of the survey questions.

3.3. Data analysis

The partial least squares structural equation modeling (PLS-SEM) approach (Hair et al. 2019) with the support of statistical software SmartPLS 3 was used to obtain the PLS-SEM results (Henseler, Ringle, and Sarstedt 2015). In a range of areas, including agricultural science and psychology, partial least squares structural equation modeling (PLS-SEM) is a frequently used method for analyzing complex inter-relationships between observable and latent variables (Willaby et al. 2015). PLS-SEM has advantages when working with complex models, non-normal data, and small samples (see Hair et al. 2019 for more information), and it is ideal for models with higher-order constructs (Hair Jr et al. 2017), like in this study. Almost all PLS-SEM studies frame their approach in a confirmatory sense, that is, a literature review is followed by the development of formal hypotheses, and finally the model estimation (Henseler 2018). In the current study, which is interdisciplinary and which addresses a new field of research on experiential learning, the multi-variate statistics are used more in exploratory than confirmatory ways (Henseler 2018). Specifically, this study bridges the gap between formal and informal education by integrating together educational psychology, experiential learning, and agricultural systems, i.e. an innovation platform/agriculture value chain as organizational learning settings/community programs. PLS-SEM analysis is divided into two parts: the measurement model and the structural model (Hair et al. 2019). On one hand, the measurement model uses quality attributes such as outer loadings, Cronbach alpha value, composite reliability, and average variance extracted. The structural model, on the other hand, uses coefficients, P-values, and Confidence Intervals.

4. Results

4.1. Assessment of the measurement models

Before evaluating the structural model linkages, PLS-SEM provides routines to test for measurement reliability and validity. Hair et al. (2019) have well-documented corresponding guidelines, which include: the evaluation of the loadings, Cronbach’s alpha, ρA, composite reliability, the average variance extracted, and discriminant analysis for reflective constructs (Tables 1 and 2).

Table 1. CA, ρA, CR, and AVE.

Constructs	CA	ρA	CR	AVE
Challenges (CE)	.711	.715	.811	.465
Experiential knowledge (EK)	.772	.785	.843	.519
Reflective (RA)	.728	.728	.817	.472
Active experimentation (AE)	.836	.844	.879	.548
Farm family resources (FMRa)	.784	.804	.847	.481
Farm family resources (FMRb)	.696	.712	.802	.451

Notes: Significant at ***p < 0.01; **p < 0.05 level (2-tailed), CA = Cronbach’ alphas, ρA = rho_A, CR = composite reliability,and AVE = average variance extracted.

Table 2. Discriminant validity.

	AE	CE	EK	FMR	FMR2
CE	.352[.233; .454]
EK	.305[.180; 410]	.214[.115; 301]
FMRa	.393[.267; .507]	.161[.110; .162]	.569[.437; .697]
FMRb	.487[.345; .617]	.135[.106; .117]	.332[.185; .451]	.675[.543; .787]
RA	.364[.256; .485]	.278[.144; .348]	.41[.280; .558]	.445[.323; .599]	.312[.172; .365]

Notes: [] 95% Confidence Intervals Bias Corrected (BCa CI).

Outer loadings, reliability, and validity measures are used to select items to include in the model. The first PLS algorithm run revealed that some items had low outer loadings (see appendix B). The results were satisfactory after removing the items with low loading and rerunning the PLS algorithm. All Cronbach alpha values and rho A (ρA) values in Table 1 were greater than 0.7, indicating internal consistency and reliability (Hair Jr et al. 2017). In appendix B, the majority of loadings were satisfactory and extremely significant (p < 0.01). While certain indicator loadings were less than 0.7, they were kept since the composite reliabilities of the constructs were more than the acceptability criterion of 0.70 (Hair, Ringle, and Sarstedt 2011). This result showed that the indication was reliable enough (Hair Jr et al. 2017). Furthermore, all AVE values were significant within the 0.5 thresholds, indicating good convergent validity. The bootstrapping procedure with 5000 samples was used for discriminant validity with the no sign changes option, bias-corrected and accelerated (BCa) bootstrap confidence interval, and two-tailed testing at the 0.05 level (Aguirre-Urreta and Rönkkö 2018; Cheah et al. 2019). Results in Table 2 revealed that the heterotrait-monotrait (HTMT) values were lower than the 0.85 conservative thresholds (Henseler, Ringle, and Sarstedt 2015). These findings demonstrated discriminant validity (Hair Jr et al. 2017).

4.2. Assessment of the structural model

4.2.1. Mediation analysis

Mediation analysis measures the degree to which a variable contributes to the transmission of change from a cause to its effect. Table 3 shows a considerable beneficial correlation between challenges and reflection. The bootstrap (.023; .306) and statistics (β =  1.94) values for the variable suggest substantial effects. For this reason, H1a which states that the farmers reflect on their current knowledge and interactions with other value chain actors when confronted with challenges is endorsed. The findings strongly indicate H1b because its statistics (β = .027) is a substantial path. This variable’s coefficients and bootstrap results are both highly significant at p < 0.01. Hereafter, hypothesis 1b, which posited that farmers who reflect on their current knowledge and interactions with other value chain actors increase their knowledge of farming methods and value chain actors, was accepted. Furthermore, the results of the H1c&d tests (β =  .233 & .160) were identical to those of the H1a-b tests. Henceforth, H1d was approved.

Table 3. Model relationships between challenges, reflection and experimentation, and experiential knowledge.

Hypothesis	Relationships	Coefficients	95% Confidence Intervals Bias Corrected(BCa CI)
Hypothesis 1a Farmers reflect on their current knowledge and interactions with other value chain actors when confronted with challenges.	Challenges → Reflection (H1a)	1.94**	[.023; .306]
Hypothesis 1b Farmers who reflect on their current knowledge and interactions with other value chain actors increase their knowledge of farming methods and value chain actors.	Reflection → Experiential knowledge (H1b)	.027***	[.135; .371]
Hypothesis 1c Farmers who face coffee value chain challenges use their current knowledge and interact with other value chain actors.	Challenges → Active experimentation (H1c)	.233***	[.097; .332]
Hypothesis 1d Farmers who use their current knowledge and interact with other value chain actors increase their knowledge of farming methods and value chain actors.	Active experimentation → Experiential knowledge (H1d)	.160**	[.006; .288]

Notes: Significant at ***p < 0.01; **p < 0.05 level (2-tailed).

Both reflection and active experimentation were used to buffer the relationship between challenges and experiential knowledge (Table 4).

Table 4. Mediation effects of the farmer’s reflection and active experimentation on the relationship between challenges and experiential knowledge.

Relationships	Coefficients	95% Confidence Intervals Bias Corrected (BCa CI)
Challenges → Reflection → Experiential knowledge	.052**	[.001; .084]
Challenges → Active experimentation → Experiential knowledge	.037*	[.009; .094]

Notes: Significant at **p < 0.05, * 0.1 level (2-tailed); Reflection → Active experimentation (β = .284***, .157; .384).

4.2.2. Moderation analysis

The next step analyzed the moderating role of farmer family resources on reflection and active experimentation as mediators of farmers’ experiential learning (H2a-d). In a moderated mediation model, the moderating variable strengthens or weakens the relationship between the independent and mediator variables, as well as the mediator and the outcome (dependent) variable; thus, mediating effects shift as the moderating variable changes. Table 5 shows the positive and significant regression coefficient of the interaction effect between challenges and farmers’ family resources on reflection (β = .112, p < 0.1). The interaction effect of reflection and family resources on experiential knowledge had a positive and significant regression coefficient (β = .131, p < 0.05). This provides preliminary support for a conditional indirect effect. The moderation effect was further verified by the bootstrapping test, with a 95% BCCI of [.008; 0.227] for the link between challenges and reflection as well as [.023; .238] for the link between reflection and experiential knowledge (Table 5). Because farm family resources have a positive and moderating effect on the relationship between challenges-reflection, farmers’ ability to think about their current knowledge and interactions with other value chain actors is enhanced when they attempt to address their coffee value chain challenges. Furthermore, family resources positively moderate the relationship between farmers’ reflection and their knowledge of farming methods and value chain actors, implying that if farmers have access to farm family resources, their knowledge of farming methods and value chain actors increases after thinking about their current knowledge and interactions with other value chain actors. Thus, H2a&c is supported.

Table 5. Moderation effects of farmer family resources on the mediated relationship between challenges and experiential knowledge.

Hypothesis	Relationships	Coefficients	95% Confidence Intervals Bias Corrected
Hypothesis 2a: Farmers’ access to family resources positively moderates the relationship between their coffee value chain challenges and reflection on current knowledge and interactions with other value chain actors.	Challenges × Farm family resources → Reflection (H2a)	0.112*	[.008; .227]
Hypothesis 2b: Farmers’ access to family resources positively moderates the relationship between their reflection on current knowledge and interactions with other value chain actors and their knowledge of farming methods and value chain actors.	Reflection × Farm family resources → Experiential knowledge (H2b)	0.131**	[.023; .238]
Hypothesis 2c: Farmers’ access to family resources positively moderates the relationship between their coffee value chain challenges and the use of current knowledge and interaction with other value chain actors.	Challenges × Farm family resources → Active experimentation (H2c)	−0.120**	[−.235; −.015]
Hypothesis 2d: Farmers’ access to family resources positively moderates the relationship between their use of current knowledge as well as interaction with other value chain actors, and their knowledge of farming methods and value chain actors.	Active experimentation × Farm family resources → Experiential knowledge (H2d)	−0.140**	[−.283; −.009]

Notes: Significant at **p < 0.05, * 0.1 level (2-tailed).

Using the same method, the moderating effects of family resources on the connection between challenges and experiential knowledge via experimentation were investigated. The interaction effect of challenges and farm family resources on active experimenting yielded a negative and significant regression coefficient (β = −.120), as shown in Table 5. Table 5 also reveals that the interaction effect of active experimentation and farm family resources on experiential knowledge had a negative and significant regression coefficient (β = .140). This lends preliminary credence to the idea of a conditional indirect impact. The bootstrapping test confirmed the moderation effect, with a 95% BCCI of [.008; .227] for the link between challenges and active experimentation and [−.235; −.015] for the link between active experimentation and experiential knowledge (Table 5). The farmer’s access to family resources has a negative moderating influence on the relationships: Farmer challenges – farmer active experimentation and farmer experiential knowledge imply that if farmers have access to farm family resources, their ability (capacity) to use their current knowledge and interactions with other value chain actors when attempting to address their coffee value chain challenges is diminished. Again, there is no knowledge of farming methods and value chain actors acquired as a result of their present knowledge and interactions with other value chain actors. Consequently, H2b&d is not supported.

5. Discussion

In prior agricultural extension studies on social networks and farmer learning in rural areas, family relationships were found to be crucial in farmer access to knowledge-learning outcomes e.g. (Fisher 2013; Tregear and Cooper 2016; Pratiwi and Suzuki 2017). Farmers learn from their past experiences and through interactions with other family members, according to this strand of literature (Chantre, Cerf, and Le Bail 2015; Burton et al. 2020; Dolinska and d'Aquino 2016). Indeed, family interactions encourage trust-based peer-to-peer learning through the exchange of experiences, challenges, and hands-on learning (Berkvens 2012; Kroma 2006; Abbey, Dowsett, and Sullivan 2017). However, the literature does not indicate how, and under what conditions, farmer experiential learning takes place. The goal of this study was to determine how the farmer’s access to family resources influenced their experiential learning process. The role of challenges in experiential knowledge is discussed from the perspective of farm family resource access in this study, which is in line with Kolb’s EL theory (Kolb 2015). The farmer’s access to farm family resources affected their experiential learning process in several ways, according to this study. The interaction between challenges and farm family resources, in particular, has a positive and negative impact on farmer learning activities. The farm family’s resources are more important to the relationships: challenges-reflection and reflection-acquisition of experiential knowledge rather than challenges-experimentation and experimentation-acquisition of experience knowledge.

5.1. The moderating effect of farmer’s access to farm family resources on the relationship between challenges and reflection, as well as relationship between reflection and experiential knowledge

First of all, having access to farm family resources allows farmers to reflect on previous solutions to value chain challenges to gain new knowledge for solving future value chain challenges. To put it another way, farmer family members’ involvement in their farming decision-making, advice, and encouragement helped farmers to reflect on their current knowledge and interactions with other value chain actors. This finding is congruent with the findings of (Ochago et al. 2021), who found that when confronted with coffee value chain challenges, farmers reflect on their present knowledge and interact with other value chain actors such as fellow family farmers. Then, by reflecting on their current knowledge and interactions with other value chain actors, farmer family members’ involvement in their farming decision making, advice, and encouragement improved their knowledge of new networks and farming practices. These findings add to (Hoang, Dufhues, and Buchenrieder 2016; Fisher 2013; Sutherland and Burton 2011; Hoang, Castella, and Novosad 2006; Pratiwi and Suzuki 2017; Danielsen et al. 2020; Ingram 2010; Samiee and Rezaei-Moghaddam 2017)’s studies of social networks in learning, which found that family ties increase the acquisition of experiential knowledge. This study findings, in particular, add to this earlier research by systematically relating farmer’s family resources to the experiential learning process, rather than just isolated parts of learning i.e. the experiential knowledge outcome of this process. This research reveals that specific family resources (for example, farmer family members’ involvement in farming decision making, advice, and encouragement) have a positive effect on the acquisition of new knowledge through reflection when faced with challenges.

5.2. The moderating effect of farmer’s access to farm family resources on the relationship between challenges and active experimentation, as well as the relationship between active experimentation and experiential knowledge

In contrast, access to farm family resources hinders active experimentation. Family resources, in particular, have a negative effect on active experimentation and the acquisition of new knowledge through active experimentation. These findings agree with studies on social networks and learning such as those (Fisher 2013; Tregear and Cooper 2016), who found that strong bonding network ties such as those of the family have a negative influence on farmer learning. Differently from their study, though, this study reveals indirect conditional effects of family resources that lower farmer active experimentation. Particularly, farmers’ ability for active experimentation was negatively impacted by family emotional support, trust, and engagement in coffee marketing decisions, restricting their ability to build knowledge about new networks and farming methods. These findings are most likely explained by the nature of the active experimentation. Unlike reflection, which was mostly an individual activity, experimentation was largely collaborative. Farmers require significantly more resources to experiment than those provided by the family due to the nature of the rural coffee value chain setting. For example, the land at the study site is small and already allotted, tests on phytosanitary measures and spraying to control coffee pests will have to be conducted on rented/purchased land. At this point, farmer experimentation is backed by collaboratively mobilized resources such as land, labor, seeds, and so on, via networks other than the family (Schut 2017; Kusters et al. 2018). Farmers then develop knowledge through exchange visits, look and learn (observation), and so on (Vellema et al. 2013). In this respect, the existing knowledge within the family network can explain the negative outcome. Experimentation requires existing information inside a specific network. Family interactions are closed networks that generate homogeneous and redundant knowledge within the network (Fisher 2013), preventing the acquisition of new knowledge outside the family (Smith, Anderson, and Moore 2012; MacGillivray 2018). Because family members rely on other family members for resources such as advice, they have been removed from knowledge brokers (such as extension personnel) over the years, resulting in a limited ability to explore and gain new knowledge through experimentation.

5.3. Implications

These findings have significant implications for family learning, in terms of theory, management, and policy. From a theoretical perspective, learning models based on social interactions, such as those found in a farm family, can stimulate as well as hinder higher-order learning through challenges. In terms of reflection, this study fills in the empirical gaps in Kolb’s experiential learning model by demonstrating, through the integration of family embeddeness-based and experiential learning theories, that the availability of family resource support can potentially increase experiential learning. Regarding active experimentation, this study fills in the empirical gaps in Kolb’s experiential learning model by demonstrating, equally supported by the integrative approach of family embeddeness-based and experiential learning theories, that the availability of family resource support can potentially decrease experiential learning (Bergsteiner, Avery, and Neumann 2010; Jarvis 2012).

Farm family resources are often the most beneficial resource for coffee producers who engage in transformative learning. Farmers can participate in transformative learning activities targeted at addressing their challenges in a context-specific and socially interactive way, challenging interventions to change (Leeuwis 2004). Smallholder farms, as a collective and farmer-centered experiential learning context, can serve as a source of inspiration for extension agents bringing the paradigm change from technology transfer to participatory advisory services to fruition. The use of tools like Participatory Rural Appraisal (Mwongera et al. 2017), which allows for a more extensive gradual, and iterative definition of challenges and solutions in direct exchanges with key stakeholders – farmers – could provide useful insights for possible adjustments in agricultural extension research and development. First, given that farm families rely on guidance on how to carry out their value chain activities, a viable method in the family farm form of agriculture would be to target influential household members for challenges-based learning actions. This entails experienced mentors providing individual home coaching, with an emphasis on things such as family assets, value chains, and people. This approach leads to more adaptable transformative and social learning arrangements in which farmers can openly share their previous challenges, knowledge of potential solutions produced and implemented, and other resources with EL. This is especially advantageous in developing countries, where rural extension and agricultural information services are still in short supply. Second, farm households should raise awareness about the necessity of sharing experiences while also providing each individual with access to useful information for reflection. Consequently, extension agents and policymakers should identify key decision-makers in farm households (mainly household heads) as a starting point for encouraging farmer reflection when confronted with challenges and the acquisition of new knowledge through reflection. This is because, there are distinct preferences among household members in a household farm system, and these preferences can influence their learning activities and outcomes. For example, it is generally known that women and children in coffee farm households are involved in coffee production through harvesting rather than marketing or allocating coffee sales revenues. In Appendix C, all additional family resources that facilitate farmers’ experiential learning process are related to decision-making. Because reflection necessitates seeing, hearing, and talking about the experience (Di Stefano, Pisano, and Staats 2015; Beard and Wilson 2013), sensitizing decision-makers on the importance of equity in coffee-growing activities would improve reflection and knowledge development through reflection.

Family resources have a detrimental impact on farmers’ active experimentation when faced with challenges, as well as knowledge acquisition through active experimentation. According to the findings, coffee farmers that actively experiment rely on new knowledge and external networks to expand their learning. As a result of these findings, extension agents and policymakers should continue to develop learning interventions, such as cooperative experiments involving various farm household members, when faced with challenges. This will coincide with the emphasis on agricultural knowledge production, which corresponds to a broader interest in multi-actor learning networks involving various stakeholders and bringing together and capitalizing on the diverse forms of knowledge possessed by those (Ingram et al. 2018; Moschitz et al. 2015).

Also, because reflection as a learning activity must be elicited consciously by learning actions (Ajjawi and Boud 2018), policy-makers can use the family as a unit to identify practical interventions to local challenges and improve targeted rural agriculture value chains by connecting different stakeholders to farm households at the community level. The family farm is frequently a node in larger learning networks (e.g. Innovation platforms) where new ideas, techniques, seeds, and other items circulate. Learning activities can help farmers to identify practical solutions by having discussions with peers and experts, comparing practices in similar contexts to their own, and participating in hands-on activities (Adamsone-Fiskovica and Grivins 2022; Ingram et al. 2018; Chancellor, Priebe, and Mkenda 2019) throughout the learning process. Concrete experiences, for example, can be aided by visualizing a farmer’s challenges, whereas reflection can be aided by facilitated discussions. Planned joint experimentation activities beyond farm families can aid experimentation. Subsequently, policymakers will be able to incorporate the role of farming households into rural agriculture research and development strategies, acknowledging them as crucial actors in agricultural knowledge production and dissemination (Dabire et al. 2017; Téno and Cadilhon 2017; Vissoh et al. 2017; Ingram et al. 2020; Moschitz et al. 2015; Tisenkopfs et al. 2015). This also entails a greater appreciation of local, indigenous, technical, and informal knowledge, as well as individual farmers’ innovative potential (Šūmane et al. 2018).

6. Conclusion

Consistent with the idea of social embeddedness (Granovetter 1973, 1985), resources accessed through ongoing personal relations (i.e. embedded) may moderate the mediating effect of learning activities on the challenging experiences to experiential knowledge relationship. Hence, more challenged farmers demand more family resources to engage in a variety of learning activities that result in high levels of experiential knowledge. Thus, the goal of this research was to find out how the farmer’s access to family resources influenced their experiential learning process. This study dissects the experiential learning process as a whole and then shows how different farmer’s family resources influence the acquisition of new knowledge through reflection and active experimentation.

Farmers’ family resources, according to the findings, have both positive and negative effects on their experiential learning processes. The evidence in this study has numerous implications for theory, practice, and policy. The results demonstrate how the availability of family resource support can potentially increase or decrease experiential learning by integrating the family embeddedness perspective – a nuanced lens of the social embeddedness perspective (Granovetter 1973, 1985; Uzzi 1997) that focuses on embeddedness within the specific context of family ties and experiential learning theorization (Kolb 2015). This study’s findings, in particular, add to previous research e.g. Danielsen et al. (2020) by systematically relating farmer’s family resources to the experiential learning process, rather than just isolated Specific family resources (for example, farmer family members’ involvement in farming decision making, advice, and encouragement) have a positive effect on the acquisition of new knowledge through reflection when confronted with challenges according to this study. Furthermore, these findings are consistent with previous research on social networks and learning (Fisher 2013; Tregear and Cooper 2016), which found that strong bonding network ties, such as those of the family, had a negative impact on farmer learning. However, unlike their work, this analysis indicates indirect conditional effects of family resources on farmer active experimentation.

7. Areas for further research

While the current study focuses on the moderating effect of farmers’ farm family resources on their experiential learning process in the IP environment, other studies in non-IP settings may be undertaken. Moreover, the current study adapts four items used to measure reflection by Kember et al. (2000). Farmers’ social networks, as guided by qualitative findings (Ochago et al. 2021), are included in addition to the four items. None of these items have previously been theorized, grouped, or used in the way that the current study does. Other studies that take such item combinations into account may be conducted.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The authors acknowledge generous funding received by the Australian Centre for International Agricultural Research (ACIAR) to fund a Ph.D. project as part of the program “Value chain Innovation Platforms for Food Security” (VIP4FS) partnering the World Agroforestry Centre (ICRAF), University of Adelaide (Australia) and Wageningen University, The Netherlands. Supervision was provided also thanks to the Entrepreneurial Learning in Inclusive Agribusiness (ELIA) project, as part of the Senior Expert Program (SEP) funded by the Dutch Scientific Council (NWO) in collaboration with the CGIAR’s Climate Change, Agriculture and Food Security (CCAFS) Program.

Notes on contributors

Robert Ochago

Robert Ochago is a Ph.D. fellow at the Business Management & Organization, Wageningen University & Research.

Domenico Dentoni

Domenico Dentoni, Ph.D., is a Full Professor at Montpellier Business School, Montpellier Research in Management, University of Montpellier, France.

Jacques Trienekens

Jacques Trienekens, Ph.D., is a Professor at Business Management & Organization, Wageningen University & Research.

Notes

1 Smallholders are farmers who own small pieces of land and rely almost completely on family labor to raise subsistence crops and one or two cash crops. They are defined by their restricted resource endowment. Because of smallholder farmer’s restricted resource endowment, the terms ‘family farm’ and ‘smallholder farm’ are frequently interchanged. See (Kostov, Davidova, and Bailey 2019; Garner and de la O Campos 2014; Lowder, Skoet, and Raney 2016).

Appendices

Appendix A. Respondents interviewed

Table

District	Sub county	Number of respondents
Kapchorwa	Municipality-Western Division	8
	Tegeres	1
	Kabeywa	12
	Municipality-East and Central Divisions	12
	Chema	24
	Kapteret	6
	Chebonet	5
	Kapchesombe	3
	Sipi	3
Manafwa	Butta	24
	Bukhofu	17
	Nalondo
	Khabutoola	18
	Luwa Town Board-Mukoto	8
	Bukhofu-Bukusu	4
Namisindwa	Bukhoho	16
	Bumbo	43
	Namweru	1
	Bunamulingi	7
	Mukhuyu	2
3 Districts	22 Sub Counties	214

Appendix B. Variable measurement

Table

Variable	Operationalization	Question asked	Questionnaire items	Outer loading at first PLS Algorithm rum	Outer loading at PLS Algorithm re-run
Challenges	Drawing on Kolb’s experiential theory, Matsuo and Nagata (2020) described experiences as both expected and unexpected as depicted in the following questions: What was your course of action? What happened? What were the expected or planned results? What were the unexpected or unplanned results?. Additionally, Morris (2020) defined experiences as context-specific challenges. Based on the above authors, Ochago et al. (2021) used the existing coffee value chain literature e.g. (Cerda et al. 2017; Pimenta, Angélico, and Chalfoun 2018; Hameed et al. 2018; Kidist, Zerihun, and Biniam 2019) to craft a definition of coffee farmers’ experiences as challenges faced by farmers at production, harvest, postharvest handling, and processing and storage and Marketing. Therefore, this study uses three interrelated elements to identify farmers’ challenges: challenges during production, harvesting, postharvest handling, and marketing.	Please indicate by ticking the appropriate box how often you faced challenges in the last 5 years (2015-now) in Challenges at production, harvest, postharvest handling, and processing and storage and marketing (on a 5 point scale where 1 = ‘never’ and 5 = ‘always’)	Pest and disease infestation	.297	low loading
			Production inputs (e.g. pesticides, fertilizers, prunning saws, labour etc.)	.261	low loading
			Extreme weather changes (e.g. sometimes too hot or too much rain)	.271	low loading
			Knowledge about coffee agronomic practices (e.g. weeding, spraying against pests and diseases, fertilizer application etc.)	.369	low loading
			Harvesting skilled coffee picking labour	.393	low loading
			Thieves	.481	low loading
			Knowledge about coffee picking (CE1)	.584	.536***
			post-harvest handling and processing knowledge about coffee sorting, washing, fermenting, drying, storage etc.	.444	low loading
			post-harvest handling and processing coffee pulping labor	.424	low loading
			post-harvest handling and processing equipment (e.g. pulping machines, drying materials etc.)	.492	low loading
			post-harvest handling and processing too much rain slowing down the coffee drying process	−.189	low loading
			Post-harvest handling and processing storage (e.g. space, materials-bags)(CE2)	.630	.649***
			post-harvest handling and processing knowledge of proper coffee storage	.479	low loading
			Market coffee buyers (e.g. few, unreliable, untrustworthy)	.326	low loading
			Market coffee prices (CE3)	.504	.716***
			Market fluctuating coffee prices (CE4)	.600	.767***
			Market transport challenges (CE5)	.486	.576***
			Market finances to run the coffee business	.449	low loading
			Market coffee prices information (CE6)	.678	.695***
Experiential knowledge	Drawing on Kolb’s experiential theory, Matsuo and Nagata (2020) described experiential knowledge as a learning outcome. Based on these authors, Ochago et al. (2021) used agricultural value chain literature e.g. (Kabambe et al. 2012; Thiele et al. 2011) define experiential knowledge as knowing about new value chain networks, farming methods, and technologies, as well as self-personal strengths and weaknesses (Ochago et al. 2021). Following extant research, this study uses two interconnected parts: knowing new value chain actors, and farming methods to define farmers’ experiential knowledge.	Please indicate how much knowledge do you have – compare to other IP members – in the following domains: Know about new value chain networks, and farming methods (on a 5 points scale where (1) ‘strongly disagree’ to (5) ‘strongly agree’)	Beneficial network relationships for coffee production (EK1)	.705	.747***
			Beneficial network relationships for coffee harvesting (EK2)	.707	.707***
			Beneficial network relationships for coffee post-harvest handling and processing (EK3)	.691	.671***
			Beneficial network relationships for coffee marketing	.419	low loading
			Knowledge about methods to increase my coffee production (yield)(EK4)	.740	.784***
			Knowledge about methods to improve my coffee harvesting (EK5)	.713	.680***
			Knowledge about methods to improve on post-harvest handling and processing methods	.468	low loading
			Knowledge about methods to market coffee	.293	low loading
Reflection	According to Matsuo and Nagata (2020), Kolb’s experiential learning cycle ought to the term reflective observation and instead adopt reflective analysis. This is because reflection entails identifying challenges, determining root causes, and considering viable remedies. The authors proposed the following questions: Were there any successes and failures with the experience? What were the factors causing success and failure?. Moreover, problem-solving involves reflecting on past problem-solving strategies as well as through sharing practical ideas with others and joint problem solving making social interactions critical. Thus, using value chain literature (Kabambe et al. 2012; Thiele et al. 2011), Ochago et al. (2021) defined the first component of reflection as a farmers’ reflection on their interactions with other value chain actors	How often – in the last 5 years – have you reflected on your interactions with existing relationships as a way to tackle post-harvest and marketing challenges, compared to other IP members? I reflect on interactions with … , (a) Operation Wealth Creation (OWC)/NAADS, (b) National Agriculture Research Organization-Buginyanya (c) Uganda Coffee Development Authority (UCDA) (d) NUCAFE (e) Makerere University (f) etc. (on a 5 point scale where 1 = ‘never’ and 5 = ‘always’)	National Agricultural Research Organization (NARO)	.206	low loading
			Uganda Coffee Development Authority (UCDA)	.176	low loading
			Makerere University	.407	low loading
			Kyagalanyi Coffee Limited (RA1)	.646	.683***
			Kawacom Uganda Limited (RA2)	.549	.599***
			Great lakes Ltd	.339	low loading
			Coffee A Cup	.273	low loading
			Gumutindo	.297	low loading
			Farmer groups	−.162	low loading
			Innovation Platforms	−.059	low loading
			National Union of Coffee Agribusinesses and Farm Enterprises Ltd (NUCAFE)	.202	low loading
			Cooperative union (e.g. Kabeywa, Bukusu,Arokwo etc.)	−.072	low loading
			Contact/model/influential farmers	−.213	low loading
			Politicians, gate keepers	.052	low loading
	Kember et al. (2000) measured reflection using four items. These items include: (1) I occasionally question how others do something and try to come up with a better way, (2) I enjoy thinking about what I've been doing and considering alternative solutions, (3) I frequently reflect on my actions to see if I could have done better, and (3) I frequently re-appraise my experience to learn from it and improve for my next performance. This study uses the above literature to define the second component of reflection as a farmer reflecting on their current knowledge to solve challenges.	Compared to other IP members, I … .. (i) question the way other coffee farmers production methods and try to think of a better way (on a 5 point scale where 1 = ‘never’ and 5 = ‘always’)	question the way other coffee farmers production methods and try to think of a better way (RA3)	.640	.685***
		(ii) like to think over my coffee harvesting methods and consider alternative ways of doing it.	like to think over my coffee harvesting methods and consider alternative ways of doing it (RA4)	.633	.679***
		(iii) re-appraise my coffee post-harvest handling and processing so I can learn from it and improve for my next performance	re-appraise my coffee post-harvest handling and processing so I can learn from it and improve for my next performance (RA5)	.686	.724***
		(iv) reflect on my coffee marketing sales to see whether I could have improved on what I did.	reflect on my coffee marketing sales to see whether I could have improved on what I did (RA6)	.612	.607***
Active experimentation	Generally, active experimentation is defined as doing or putting to use the existing knowledge (Matsuo and Nagata 2020; Wang et al. 2020). With the aspect of learning in the rural value chain context being socially interactive, (Ochago et al. 2021) used the above research to define active experimentation using two sub-components as follows: (i) having used existing networks to solve value chain challenges; (ii) using existing coffee farming knowledge to solve coffee value chain challenges. Therefore, active experimentation occurs when a farmer uses their existing coffee value chain challenges solving knowledge and interact with other value chain actors to increase their level of experiential knowledge.	How often – in the last 5 years – have you used the knowledge obtained through relationships to tackle production and harvesting challenges compared to other IP members? I used knowledge from … , (a) Operation Wealth Creation (OWC)/NAADS, (b) National Agriculture Research Organization-Buginyanya (c) Uganda Coffee Development Authority (UCDA) (d) NUCAFE (e) Makerere University (f) etc.(on a 5 point scale where 1 = ‘never’ and 5 = ‘always’)	Operation Wealth Creation (OWC)/NAADS	−.035	low loading
			National Agricultural Research Organization (NARO) (AE1)	.716	.782***
			Uganda Coffee Development Authority (UCDA)(AE2)	.570	.621***
			Makerere University	.126	low loading
			Kyagalanyi Coffee Limited	.225	low loading
			Kawacom Uganda Limited (AE3)	.610	.706***
			Great lakes Ltd	.080	low loading
			Coffee A Cup	.300	low loading
			Gumutindo	.262	low loading
			Farmer groups	.339	low loading
			Innovation Platforms	.061	low loading
			National Union of Coffee Agribusinesses and Farm Enterprises Ltd (NUCAFE)	.189	low loading
			Cooperative union (e.g. Kabeywa, Bukusu, Arokwo etc.)	−.005	low loading
			Contact/model/influential farmers	.018	low loading
			Politicians, gate keepers	.003	low loading
		Compared to other IP members, I use my knowledge about coffee: (on a 5 point scale where 1 = ‘never’ and 5 = ‘always’) (i) Production	use my coffee production knowledge (AE4)	.722	.727***
		(ii) harvesting	use my harvesting knowledge (AE5)	.722	.748***
		(iii)Post-harvest handling and processing	use my post-harvest handling and processing knowledge (AE6)	.717	.703***
		(iv)Marketing	use my marketing knowledge (AE7)	.703	.729***
Farmer’s family resources (FMR). Note: The 11 items in the paper are further divided into two groups based on loading, reliability, and validity issues, as follows: FMR1 to FMR5 loaded first and thus acted as FMRa, while the remainder acted as FMRb.	- advice, information, or knowledge (e.g. on the coffee quality demanded by coffee buyers)- labor (production, harvesting, processing, and marketing) - emotional support i.e. encouragement, inspiration, motivation to start or continue coffee farming, and attending training - Time/grant a spouse permission to attend training meetings, family members connect other family members to training avenues, financing stakeholders, etc. - farmers borrow or share agro-inputs e.g. pulping machines- Financing coffee farming activities e.g. providing funds, and collateral for accessing finances as a group are a few of the resources accessed by farmers through their families across the coffee value chain (Ochago et al., 2021)	Please indicate whether the support from your family members helped you solve challenges at coffee production, harvesting, post-harvest handling, and marketing – in the following domains: production, harvest, postharvest handling, processing and storage, and marketing (on a 5 point scale where 1 = ‘never’ and 5 = ‘always’)	I obtain advice from my family members on coffee production (FMR1)	.642	.758***
			I obtain advice from my family members on harvesting	.061	low loading
			I obtain advice from my family members on post-harvest handling and processing	−.026	low loading
			I obtain advice from my family members on marketing	−.203	low loading
			My relatives encouraged me to e.g. start coffee farming (FMR2)	.731	.823***
			My family members mentor me in coffee farming activities (FMR3)	.346	.452***
			My family offers me emotional support during coffee production activities (FMR4)	.576	.665***
			My family members are knowledgeable about coffee, I have learned a lot from them (FMR5)	.602	.679***
			I have no restriction from my family on whether or not to attend trainings on coffee farming	.358	low loading
			My family members are a source of inspiration for my coffee farming activities (FMR6)	.509	.490***
			My family paid for my school fees until my current education level	.074	low loading
			My family members are involved in decision making at the production stage (FMR7)	.726	.803***
			My family members are involved in decision making at harvesting (FMR8)	.562	.704***
			My family members are involved in decision making at post-harvest handling and processing (FMR9)	.599	.722***
			My family members are involved in decision making at marketing	.143	low loading
			The trust my family members have in me motivated me to start up coffee farming (FMR10)	.565	.612***
			My family members offer coffee production labor (FMR11)	.543	.568***
			My family members offer coffee harvesting labor	.048	low loading
			My family members offer coffee post-harvest handling and processing labor	.040	low loading
			My family members offer coffee marketing support-transporting, negotiating coffee prices etc.	.083	low loading
			My family members offer me production inputs-land (FMR10)	.333	low loading
			My family members offer me production inputs-pruning saws	.151	low loading
			My family members offer me post-harvest handling and processing activities-pulping machines	−.017	low loading
			I have access to business contacts through my family members	.246	low loading
			My family members do finance my coffee farming operations	.148	low loading
			My family members act as collateral for me to access finances for my coffee farming activities	.217	low loading

Appendix C. Pearson correlations between decision making and farmer’s family resources

Table

	My family members are involved in decision making at:
	Production	Harvesting	Post-harvest handling and processing	Marketing
I obtain advice from my family members on coffee production methods	.389**	.230**	.249**	.151*
My family members are knowledgeable about coffee, I have learned a lot from them	.400**	.253**	.245**	.157*
My relatives encouraged me to e.g. start coffee farming	.414**	.334**	.374**	.296**
My family offers me emotional support during coffee production activities	.327**	.159*	.201**	.343**
The trust my family members have in me motivated me to start up coffee farming	.293**	.277**	.350**	.448**
My family members offer coffee production labor	.238**	.236**	.297**	.311**
My family members offer me production inputs-land	.191**	–	–	.183**

Notes: significant at ***p < 0.01; **p < 0.05 level (2-tailed).