1 Authors: Patrick Mvuyibwami1, Desire M. Kagabo1, Livingstone Byandaga1, Anton Eitzinger1, David Karanja2, Annuarite Uwera3, Wilber Ssekandi5, Stanley Nkalubo5, Julius Mbiu4, Eric Dushimirimana1, Mireille U. Muhigirwa1, Eileen Nchanji1, Denyse Uwera1, Mathieu Ouedraogo1, Chris N. Mwangi1 and Joseline Kiogora1 1Alliance of Bioversity International and CIAT 2Kenya Agricultural and Livestock Research Organization (KARLO) 3Rwanda Agriculture and Animal Resources Development Board (RAB) 4Tanzania Agricultural Research Institute (TARI) 5Uganda National Agricultural Research Organisation (NARO) Citation: Patrick Mvuyibwami, Desire M. Kagabo, Livingstone Byandaga, Anton Eitzinger, David Karanja, Annuarite Uwera, Wilber Ssekandi, Stanley Nkalubo, Julius Mbiu, Eric Dushimirimana, Mireille U. Muhigirwa, Eileen Nchanji, Denyse Uwera and Mathieu Ouedraogo (2024). Assessing the Current Conditions of Climate Information Services in East Africa. The Alliance of Bioversity International and CIAT. Assessing the Current Conditions of Weather and Climate Information Services in East Africa Household survey Enhancing Climate Resilience in East Africa project (ECREA) Baseline report December 2024 2 Contents Acknowledgement ............................................................................................................ 5 Executive Summary .......................................................................................................... 6 Chapter 1. Introduction ..................................................................................................... 8 Chapter 2: Research methodology .................................................................................. 10 2.2. Description of the study areas ......................................................................................... 10 1. Rwanda .................................................................................................................................................... 11 2. Uganda .................................................................................................................................................... 12 3. Kenya ....................................................................................................................................................... 12 4. Tanzania .................................................................................................................................................. 13 Climate risks and agriculture (Beans) .......................................................................................................... 13 2.3. Research design .............................................................................................................. 14 2.3.1. Sampling Techniques and Sampling Frame ...................................................................................... 15 2.4. Collection of data ............................................................................................................ 17 2.5. Data analysis techniques ................................................................................................. 18 2.6. Research limitations ........................................................................................................ 19 Chapter 3: Results ........................................................................................................... 20 3.1. The socio-economic characteristics of surveyed farmers ................................................... 20 3.2. Accessibility of Weather and Climate Information (WCI) and Impact-Based Early Warnings (IBEWs) among the study farmers .......................................................................................... 26 3.2.1. Delivery of WCI and IBEWs through available communication channels ......................................... 26 3.2.2. Preferences and barriers/challenges in accessing WCI and IBEWs ................................................... 33 3.3. Use of Weather and Climate Information (WCI) and Impact-Based Early Warnings (IBEWs) among the study farmers ....................................................................................................... 39 3.3.1. Utilization, Usefulness, and Accuracy of WCI and IBEWs among the study Farmers ........ 39 3.3.2. Preferences and Barriers/Challenges in Using Information on W&C and IBEWs.............................. 43 3.4. The Impact of WCI and IBEWs on Climate Change Resilience among the Study Farmers ..... 45 3.4.1. Farmers' Adaptive Capacity to Climate Risks using WCI and IBEWs ................................................. 45 3.4.2. Impact of WCI-Informed Decisions on Crop Yield, Household Income, avoided losses, and Damage Reduction .................................................................................................................................................... 53 Chapter 4: Discussion of the results ................................................................................. 61 4.1. Social Economic characteristics and vulnerabilities of the surveyed farmers ...................... 61 4.2. Access and Barriers to Receiving WCI and IBEWs Among Study Farmers ............................ 67 4.3. Use of WCI and IBEWs in Climate Preparedness and Adaptive Actions Among the surveyed Farmers ................................................................................................................................. 73 4.4. Outcomes of Using WCI and IBEWs: Crop Yield, Household Income, and Avoided losses .... 77 Chapter 5: Conclusions and Recommendations for ECREA Project Implementation ......... 81 5.1. Conclusions ..................................................................................................................... 81 5.2. Recommendations for ECREA Project Implementation ..................................................... 82 REFERENCES ................................................................................................................... 85 3 Appendices: .................................................................................................................... 93 List of figures Figure 1 ECREA baseline intervention area .................................................................................................... 11 Figure 2: The land allocation decision-making scheme and determinants of the decision makers ................. 62 Figure 3: Country comparison on gender, age and education ........................................................................ 62 Figure 4: Farmer’s main roles in the bean value chain and livelihood activities ............................................. 63 Figure 5: Country comparison on household land size and land allocation to bean production ..................... 64 Figure 6: Farmer groups to which household heads belong ........................................................................... 65 Figure 7: Farmer groups to which spouses of the surveyed farmers belong ................................................... 66 Figure 8: Climate shocks and their prevalence in the study area .................................................................... 67 Figure 9: The access to WCI ............................................................................................................................ 68 Figure 10: The most likely reasons why the farmers do not receive WCI and whether they would like to receive the information ........................................................................................................................ 69 Figure 11: The most likely reasons why the farmers do not receive IBEWs and whether they would like to receive the information ........................................................................................................................ 69 Figure 12: The most likely reasons why the farmers do not like to receive WCI ............................................. 69 Figure 13: The sources(channels) through which farmers receive WCI ........................................................... 70 Figure 14: The sources(channels) through which farmers receive IBEWs ....................................................... 70 Figure 15: The level to which WCI is shared within the farming communities ................................................ 71 Figure 16: The level to which IBEWs is shared within farming communities .................................................. 71 Figure 17: The types of WCI received by the study farmers ............................................................................ 72 Figure 18: The types of IBEWs received by the study farmers ........................................................................ 72 Figure 19: The usefulness of WCI for helping farmers anticipate and prepare for climate events and disasters ............................................................................................................................................................. 74 Figure 20: The usefulness of WCI for helping farmers anticipate and devise a preparedness plan for making timely informed decisions .................................................................................................................... 74 Figure 21: The accuracy of WCI for helping the farmers predict the actual weather events and hazards ....... 75 Figure 22: The most likely reasons why the farmers do not use the received WCI ......................................... 75 Figure 23: The most likely reasons why the farmers do not use the received IBEWs ...................................... 75 Figure 24: Actions taken by Farmers Based on WCI........................................................................................ 76 Figure 25: The Impact of the decision taken based on WCI ............................................................................ 78 Figure 26: The most likely attainable increase in income as a result of using WCI .......................................... 79 Figure 27: The most likely attainable bean yield increase as a result of using WCI ......................................... 79 Figure 28: The impact of action taken based on IBEWs .................................................................................. 80 Figure 29: The most likely attainable avoided losses as a result of using IBEWs ............................................. 80 Figure 30: The most likely losses as a consequence of not using IBEWs ......................................................... 80 4 List of Tables Table 1: Demographic information for the farmers who participated in the baseline household survey ....... 21 Table 2: Land ownership and decision-making factors in bean production .................................................... 23 Table 3: Participation of the surveyed farmers in farmer associations/cooperatives ..................................... 25 Table 4: Access to WCIS and IBEWs among the surveyed farmers .................................................................. 28 Table 5: Types of WCI and IBEWs received by surveyed farmers .................................................................... 28 Table 6: Sources of information for WCI and IBEWs among the surveyed farmers ......................................... 29 Table 7: WC& IBEWs information sharing among the surveyed farmers ........................................................ 29 Table 8: Effect of education on regularity of access to weather and climate information .............................. 30 Table 9: Effect of education on types of WCI received by surveyed farmers ................................................. 31 Table 10: Effect of household land size on the regularity of access to IBEWs and the types of weather and climate information received by surveyed farmers ............................................................................... 31 Table 11: Effect of farmer associations membership status on the regularity of access to weather and climate information and the types of the information received by surveyed farmers ....................................... 32 Table 12: Gender and farmer groups membership status and types of Impact-Based Early Warnings received ............................................................................................................................................................. 33 Table 13: Preferences and barriers for farmers who do not receive WCI ....................................................... 35 Table 14: Preferences and barriers for farmers who do not receive IBEWs .................................................... 36 Table 15: Effect of education levels on farmers’ barriers to IBEWs access ..................................................... 37 Table 16: Gender and farmer association membership status and barriers to accessing IBEWs ..................... 38 Table 17: Concerns and barriers for farmers who would not like to receive the WCIS and IBEWs .................. 38 Table 18: Perception of the surveyed farmers on use, usefulness and accuracy of WCI ................................. 40 Table 19: Perception of the surveyed farmers on use, usefulness, and accuracy of IBEWs ............................. 41 Table 20: Effect of land allocated to beans on perceived usefulness of IBEWs for making preparedness plans ............................................................................................................................................................. 41 Table 21: Effects of gender and land-size allocated to beans on perceived predictive accuracy of IBEWs and gender on perceived suitability of WCI for bean production ................................................................. 42 Table 22: Effect of farmer group membership status of spouses and household heads on perceived anticipatory accuracy of IBEWs and types of IBEWs received by surveyed farmers .............................. 43 Table 23: Reasons for non-utilization of received WCI & IBEWs ..................................................................... 44 Table 24: Types of informed decisions made as informed by Weather and Climate Information ................... 47 Table 25: Shock or stress that occurred between 2022 to 2023 in the study area .......................................... 48 Table 26: Effect of education on the prevalence of weather and climate events among surveyed farmers ... 49 Table 27: Effect of land allocated to beans on climate-related shocks experienced by surveyed farmers ...... 50 Table 28: Effect of gender on types of unfavorable weather and climate events experienced by surveyed farmers ................................................................................................................................................. 51 Table 29: Actions taken in response to Impact-Based Early Warning (IBEWs) information ............................ 52 Table 30: Impact of informed decision made based on W&C information ..................................................... 54 Table 31: Estimated increase in yield, income, and the monetary value of avoided damages as result of using the received WCI .................................................................................................................................. 55 Table 32: Outcomes of actions taken based on the received Early warnings (IBEWs)..................................... 55 Table 33: Estimated monetary value of avoided losses of crops, livestock, and assets as a result of using IBEWs ................................................................................................................................................... 56 Table 34: Estimated monetary value of crop losses, livestock and assets as a consequence of not using IBEWs ................................................................................................................................................... 56 Table 35: Effect of education on the outcomes of farmers' WCI-informed actions ......................................... 57 Table 36: Effect of education on crop yield outcomes of farmers' WCI-informed actions .............................. 57 Table 37: Effect of education on farmers estimated monetary value of avoided losses because of their WCI- informed decisions ............................................................................................................................... 58 Table 38: Effect of education on farmers’ household income increases because of WCI-informed responses 59 Table 39: Effect of landholdings on the outcomes of IBEWs-informed actions ............................................... 60 Table 40: Effect of gender and farmer group membership status on the outcomes of farmers' IBEWs- informed actions .................................................................................................................................. 60 5 Acknowledgement Gratitude is extended to all individuals and organizations involved for the successful completion of the baseline study for the ECREA project in Kenya, Rwanda, Uganda, and Tanzania. Special appreciation is directed to the farmers, PICSA workshop participants, and survey respondents whose insights and experiences were vital in shaping this research and enhancing understanding of the agricultural landscape in the region. Acknowledgement is also due to the ECREA project implementers and partners for their efforts in facilitating the survey process, providing logistical support, and ensuring local community voices were included. The dedication of the ECREA project team to data collection, analysis, and interpretation is particularly noteworthy. This report is a key component of the East Africa Climate Resilience Enhancement (ECREA) initiative, led by the Alliance of Bioversity International and CIAT, with funding from the Foreign, Commonwealth, and Development Office (FCDO) through the UK Met Office. The project is carried out in collaboration with the Intergovernmental Authority on Development Climate Prediction and Applications Centre (ICPAC), National Agricultural Research Institutions (NARS), and National Meteorological and Hydrological Services (NMHS) of Kenya, Rwanda, Tanzania, and Uganda. Additional partners such as the Rwanda Water Board, Shamba Shape Up, ishamba, and Radio Huguka will also make a significant contribution. ECREA aims to strengthen the provision and utilization of Weather and Climate Information Services (CIS) in the selected East African nations while evaluating existing tools and methodologies to identify effective strategies, best practices, and necessary resources for enhancement. This initiative fosters collaboration with national and local entities, as well as farmers’ cooperatives in the targeted countries. 6 Executive Summary The Enhancing Climate Resilience in East Africa (ECREA) project aims to strengthen climate resilience in East Africa by improving access to timely, context-specific Weather and Climate Information Services (WCIS) and Impact-Based Early Warnings (IBEWs). It supports tailored agro-climate advisory services for farming communities. A baseline household survey was conducted in Uganda, Rwanda, Kenya, and Tanzania, targeting over 800 participants from the common bean value chain. The survey used a mixed-methods approach, involving Interactive Voice Response (IVR) and paper-based surveys, descriptive statistics analysis, and binary and multinomial logistic regression modeling to assess adaptive capacity, resilience, and access to and use of WCIS and IBEWs. The survey revealed gender disparities in agricultural participation, with male representation from 51% in Kenya to 75% in Rwanda. Women were more represented in Uganda and Kenya, indicating an opportunity to increase female engagement in resilience initiatives. The farming population was predominantly young-to-middle-aged (30%-53%), with Tanzania having a higher proportion of younger farmers (43%). Education levels varied, with Rwanda and Tanzania showing higher primary education levels (59%), while Uganda had 24% of respondents with university education. These findings highlight the need for technically and linguistically accessible and gender-responsive capacity-building strategies. Agricultural livelihoods are highly vulnerable to climate shocks, with over 70% of respondents engaged in crop farming. Limited diversification into livestock (Kenya 22.6%) and small businesses (Kenya 14%, Tanzania 11%, and Uganda 6%) indicates the need for more climate- resilient livelihoods. Bean cultivation was largely subsistence-based, except in Uganda (54.3%) and Tanzania (52%), where market-oriented production was more common. Land ownership varied significantly, with 56% of Rwandan respondents owning less than 0.5 hectares, while larger plots (above 5 hectares) were more common in Tanzania (16%). Land allocated to bean cultivation was highest in Tanzania (24%) and lowest in Rwanda (13%), emphasizing the need for land-efficient agro-climatic practices and targeted WCIS and IBEWs. Gender dynamics in decision-making varied across regions. Joint decision-making on land and crop selection was most common in Rwanda (55%), while men dominated in Tanzania (39%). Women played a major role in crop selection in Tanzania (50%). However, men held the dominant roles in crop variety, land allocation, and the use of WCIS and IBEWs. Promoting gender inclusivity is crucial for improving collective ownership of adaptation efforts. Membership in cooperatives and advisory groups was high, especially in Tanzania (84%) and 7 Rwanda (79%), but membership in marketing and financial groups remained low. Expanding access to these groups could strengthen farmers' market positions and, as a result, financial resilience. Access to WCIS and IBEWs varied significantly. Regular access to WCIS was highest in Kenya (44.9%) and Tanzania (44.4%), while Rwanda (29.5%) and Uganda (31.8%) had lower access. Low-education-level farmers, smallholder farmers, female farmers, and farmers disconnected from farmer associations were less likely than their counterparts to report receiving WCI and IBEWs regularly. Access gaps were notable, particularly in Rwanda (34.8%) and Uganda (39.5%), where many farmers, especially female, low-education-level, and smallholder farmers, and those disconnected from farmer associations reported lacking access to the information. For IBEWs, Tanzania (42%) and Uganda (40.7%) had the highest access levels. Radio was the primary dissemination channel, reaching 64% of farmers in Rwanda, with peer sharing prevalent (over 84% in Rwanda and 92% in Tanzania). However, knowledge gaps, timing issues, and infrastructure challenges hindered access. Despite these challenges, farmers expressed strong demand for WCIS and IBEWs, with 91%-98% of respondents showing interest. Farmers who used WCIS reported benefits such as better crop yields (45.32% in Rwanda, 47.90% in Uganda), and yield increases of less than 0.1 tons per hectare (48.60% in Rwanda, 45.37% in Kenya). In Tanzania, 40.32% reported reduced crop damage due to adverse weather. IBEWs helped prevent losses, with 49.0% of Rwandan and 32.3% of Ugandan farmers avoiding losses by acting on early warnings. The seasonal value of avoided losses was under $30 USD for 45% of Rwandan and 30.6% of Ugandan farmers. Without IBEWs, potential losses of up to $100 USD were observed in Uganda and Kenya. Logistic regression analysis revealed that most farmers, particularly, female, low-education-level, and subsistence farmers, and those disconnected from farmer associations were less likely to report having achieved increased crop yield, household-income, and avoided-loss outcomes because of taking informed actions. Current services have modest impacts on crop yield, damage reduction, and loss avoidance, highlighting the need for improved precision and agro-climatic relevance. The survey highlights limited access, usage, and inclusivity of weather and climate information services, particularly for female farmers, low-education-level and smallholder farmers, and farmers who do not belong to farmer associations. This lack of inclusiveness hampers equitable access to critical climate adaptation resources. Furthermore, regional differences in service utilization, trust issues, and delivery timelines underscore the need for tailored, context-specific intervention. 8 Chapter 1. Introduction Climate change and variability are significant threats to global agricultural productivity, undermining food systems and exacerbating food insecurity, which currently affects approximately one billion people worldwide. This number is expected to rise if action is not taken (Mc Carthy et al., 2018; Zhao et al., 2017). The impacts of climate change, including flooding, unpredictable rainfall patterns, rising sea levels, droughts, soil erosion, and extreme weather events, pose significant risks to financial stability and food security (Coulibaly et al., 2020; Dube et al., 2016; Gezimua & Rahut, 2021). These challenges are particularly severe in sub-Saharan Africa, where agriculture relies heavily on rainfall as the primary water source (Förch et al., 2011; Kristjanson et al., 2012). Research has shown that climate shifts, particularly changes in rainfall and temperature patterns, negatively impact physical, biological, and socio-economic systems, making populations more vulnerable to various risks and disasters (Amwata, 2013; IPCC, 2012; Tasokwa, 2011). This vulnerability is further exacerbated by the occurrence of extreme weather events, such as floods and droughts, which continue to undermine rural livelihoods, especially in regions like Kenya's arid and semi-arid zones. Here, climate-related risks significantly affect livestock and small-scale farming, the primary sources of income for many (JARVIS et al., 2011). Additionally, increasingly erratic rainfall patterns disrupt growing seasons and affect the distribution of rainfall, presenting considerable challenges for smallholder farmers in sub-Saharan Africa (Benjamin & Pierre, 2014; Camberlin & Okoola, 2003; HANSEN et al., 2011). The growing uncertainty surrounding water availability and rainfall patterns necessitates urgent action to strengthen climate resilience (Ongoma & Chen, 2017). The consequences of these climatic changes extend beyond agricultural disruptions; they can also substantially decrease crop yields, threatening food security and potentially causing severe shortages (Brown et al., 2023; Kogo et al., 2021). Such disruptions in food production have serious long-term implications, as malnutrition can hinder cognitive development and stifle economic progress (Alam et al., 2020; Khandelwal et al., 2020). Beans, a vital commercial and nutrient-dense crop in East Africa, have the potential to improve household food and financial security. However, like many crops, they are highly vulnerable to erratic weather conditions (Blackwell, 2012). In response to these challenges, it is crucial for East African bean growers to have access to accurate weather and climate information (WCIS) and impact-based early warnings (IBEWs) to safeguard their livelihoods against climate-related risks. Although weather and climate information services and early warnings have the potential to enhance resilience, their awareness, accessibility, and use 9 remain limited in the region. In light of this, the Enhancing Climate Resilience in East Africa (ECREA) project seeks to address these gaps by promoting climate resilience and improving the quality of life for East African communities. The project focuses on empowering local populations to adapt to climate change impacts while ensuring equitable access to WCIS and IBEWs. The study was conducted in Uganda, Rwanda, Kenya, and Tanzania to assess the climate- change-and-variability adaptive capacities and vulnerabilities of farming communities. The study involved diverse groups, including women, men, youth, people with disabilities, bean value chain actors, and agricultural practitioners. The specific objectives of the study were: (1) To analyze socio-economic and demographic factors affecting participation in climate initiatives, identify key vulnerabilities, and propose interventions to enhance climate resilience; (2) To evaluate the awareness, accessibility, usability, and impact of WCIS and IBEWs for bean value chain actors and agricultural practitioners in the ECREA project intervention areas; (3) To establish inclusive WCIS and IBEWs systems based on Gender Equality and Social Inclusion (GESI). The findings from this study will inform policymakers, practitioners, and stakeholders about the current state of climate resilience in East Africa. By addressing existing gaps in knowledge, accessibility, and use of WCIS and IBEWs, the study aims to guide the development of strategies to strengthen resilience, foster inclusivity, and promote sustainable agricultural practices in the region. 10 Chapter 2: Research methodology This section outlines the study area, research design, types of data, as well as the methods for data collection, and the corresponding analytical approaches that were undertaken to achieve the study's objectives. 2.2. Description of the study areas ECREA project is covering four targeted countries from East Africa (Rwanda, Kenya, Uganda and Tanzania) eight districts in Rwanda, nine districts in Uganda, four counties in Kenya and five districts in Tanzania where common bean crop (Phaseolus vulgaris) is commonly grown and all of which are highly vulnerable to climate change and its associated risks. (see Figure 1). These nations are in the East African Rift Valley, a region marked by diverse geographical features, varying climate zones, and ecological systems that significantly influence agricultural practices. These countries share a common dependence on rain-fed agriculture, which makes them particularly susceptible to climate variability and extremes (Palmer et al., 2023). Beans (Phaseolus vulgaris), a staple crop in these regions, are particularly vulnerable to climate changes due to their sensitivity to fluctuations in temperature, precipitation, and extreme weather events such as droughts, floods, and pests (Asfaw et al., 2009). Almost all beans are produced by small rural farmers, traditionally for subsistence consumption. Demand from urban areas is always increasing as rural-urban migration is increasing and according to Deborah Potts (Potts, 2013), consumption in African urban areas is growing fast. East Africa is one of the fastest urbanizing areas in the whole continent exceeds 4.5%; Uganda is the fastest with an urban population growth rate of 6%. Even the middle classes of these countries are consuming beans rather than meat (Statista, 2024). Beans are also important as a cash crop for many farmers and for cross-border trade in bean deficit areas depending on climate stresses and market volatility. Poverty in this region is largely rural phenomenon (R.A. Kirkby, 1992). Moving forward, coordinated efforts are needed to further expand the role of legumes in food systems globally. 11 Figure 1 ECREA baseline intervention area 1. Rwanda Rwanda is characterized by its hilly and mountainous terrain, which creates both opportunities and challenges for agricultural production. The country is highly dependent on its agricultural sector. Agricultural sector constitutes 90% employment opportunities in the economy and 70% export revenue in the country (Aubert, 2018). On the side of provision of food, 91% of domestic food is generated by the agricultural sector, with beans being one of the primary crops grown for both subsistence and market purposes (Rwirahira, 2009). Rwanda faces climate risks such as irregular rainfall patterns, droughts, and soil erosion, which impact crop yields. Beans are particularly sensitive to shifts in seasonal rainfall, and erratic weather patterns have increasingly reduced productivity, threatening food security (Austin et al., 2020). In the highland areas, increased rainfall intensity leads to soil erosion, affecting the viability of beans and other crops. In contrast, lowland areas are more prone to drought, which can severely limit bean growth and reduce yields. Rwanda’s farmers often face challenges in accessing timely weather and climate information, which further complicates their ability to adapt to these shifting patterns (Nsengiyumva et al., 2022). The gender dynamics in agricultural decision-making also play a significant role, with women, who make up a significant proportion of the farming workforce, having limited access to critical climate information. 12 2. Uganda Uganda is a geographically diverse country, with regions ranging from lakeshores to dry savannahs. Agriculture remains a critical part of the economy, and beans are a staple crop grown in many areas, particularly in the central, western, and southwestern regions. The country faces significant climate risks, including increased rainfall variability, prolonged droughts, and flooding (Kogo et al., 2021; Mubiru et al., 2018). These changing patterns are having an adverse effect on bean production, as erratic rainfall makes it difficult for farmers to predict optimal planting and harvesting times. Additionally, temperature rise and pests are exacerbating the challenges. Beans, while relatively resilient to moderate climate stresses, are highly susceptible to waterlogging and drought. This has led to a decline in yields and the increasing risk of crop failures (Mubiru et al., 2018). Furthermore, access to Weather and Climate Information Services (WCIS) and Impact-Based Early Warnings (IBEWs) in Uganda is inconsistent, particularly in rural areas where farmers often rely on informal sources of information (Muyiramye, 2020). These barriers hinder farmers’ ability to make timely, informed decisions, further compounding climate-related vulnerabilities. Gender disparities are also notable in Uganda, with women often having less access to resources, training, and information, despite their key roles in bean cultivation (J et al., 2021; Nakazi et al., 2017; Vincent, 2022). 3. Kenya One of the largest and most diverse agricultural producers in East Africa is Kenya where Beans are a key crop farmed across the country, particularly in the central, eastern, and western areas. They are facing a number of climatic threats, such as recurring droughts, flooding, and rising temperatures that have an impact on rain-fed agriculture (Tasokwa, 2011). Beans are very vulnerable to soil moisture stress, and problems with water availability are made worse by irregular rainfall (Omwenga, 2015). Furthermore, yields are greatly impacted by the lengthier dry spells that occur during crucial growing seasons. Even though Kenya has made progress in strengthening its agricultural resilience, the nation continues to struggle with timely, localized climate information availability. High temperatures and dry periods are getting worse in places like the country's east and northeast, which limits beans' capacity to thrive. One of the key factors influencing climate resilience in Kenya is the gender gap in decision-making within farming households, especially in regions where men dominate control over land allocation and crop variety decisions (Ochieng et al., 2017). Despite this, women, who are the primary labor force in agriculture, often face barriers to accessing relevant climate information and advisory services (Ngigi et al., 2017). 13 4. Tanzania Tanzania's agricultural economy is primarily reliant on rain-fed systems, and beans are a key crop in many areas. Drought, flooding, temperature extremes, and irregular rainfall patterns are among the major climatic concerns that the country faces (Brown et al., 2023). These dangers are especially acute in locations such as the north and center, where beans are an important crop for food security and revenue. Drought is a significant threat to bean production in Tanzania, particularly during the lengthy dry seasons, whilst floods during the rainy season can bring waterlogging and damage to bean crops (Jha et al., 2023). Pests and diseases, especially during prolonged wet spells, heighten the hazards. Nonetheless, Tanzania has some of the highest levels of participation in cooperatives and advisory groups, which could play a key role in enhancing the delivery of weather and climate services. However, while Tanzania has access to WCIS, there are still substantial gaps in terms of timeliness, accuracy, and agro-climatic relevance. Dissemination issues, such as poor radio coverage in remote areas and low literacy rates, limit the impact of climate information in some sections of the country (Mtega & Ronald, 2013). Climate risks and agriculture (Beans) The geographical and climatic diversity of Rwanda, Uganda, Kenya, and Tanzania creates a complex environment for agricultural production. Beans, being a major staple crop in all four countries, are sensitive to rainfall patterns, temperature fluctuations, and soil health. The key climatic risks, such as drought, flooding, and erratic rainfall, are exacerbated by the diverse topography, which affects water retention and soil erosion The agricultural sector in East Africa, particularly in the growing of beans, faces multiple climate risks that affect production, productivity, and food security. Key risks include: ● Drought: All the four countries experience prolonged dry spells that impact the ability of beans to grow effectively. Beans require sufficient moisture, and drought leads to reduced germination rates and stunted growth. ● Flooding: Intense rainfall events and flooding disrupt the growing conditions for beans, especially in poorly drained areas. ● Pests and diseases: Climate-induced changes in temperature and rainfall patterns foster the spread of pests and diseases that affect bean crops, further reducing yields. 14 ● Unpredictable rainfall: Beans, which rely on predictable seasonal rainfall, are particularly vulnerable to irregular weather patterns. This unpredictability makes it difficult for farmers to plan planting and harvesting schedules effectively. Given the reliance on agriculture for livelihoods in these countries, strengthening climate resilience through better access to weather and climate information, as well as gender-inclusive and context- specific advisory services, is crucial for improving the adaptive capacity of farming communities in the study areas. 2.3. Research design The design used a quantitative approach, ensuring that the data collected using the Interactive Voice Response (IVR) and that collected from face-to-face workshops provide a comprehensive understanding of the issues related to weather and climate information services (WCIS) and Impact Based Early Warnings (IBEWs) and their impact on climate resilience of the communities in the ECREA intervention areas. Figure 2 explains a brief logical flow of research steps in the study. The research aimed to evaluate the accessibility, usability, and impacts of WCIS and IBEWs on agricultural resilience among smallholder farmers in the four target countries. Key areas of investigation included the accessibility and utilization of CIS and IBEWs, the effectiveness of communication channels, changes in farm management practices, behavioral changes, barriers and challenges in accessing and using the information, and the overall level of agricultural resilience. Data were collected through structured surveys and questionnaires using the 5Q approach. Procedures at enrollment involved random farmer selection, obtaining consent, and informing farmers about the study. Measurement of exposures and confounders includes farmer demographics, socio-economic factors, and farming practices that may influence decision- (ECREA Project baseline study) Research questions Literature review from similar studies conducted Research methodology Develop research questions `Develop the Question tree- data collection questionnaire conduct household survey (5Q)/ Interactive Voice Response Paper based Survey during the PICSA Workshop Data Analysis Findings and recommendations report writing 15 making. The study's outcomes were measured through key indicators, including access and use of CIS and IBEWs, adaptive capacity, adoption of climate-smart practices, productivity, income, and avoided losses 2.3.1. Sampling Techniques and Sampling Frame In the context of the ECREA project, the chosen sampling technique and sampling frame are designed to align with the project’s logical framework (log-frame), ensuring the study’s outcomes are relevant, reliable, and representative of the targeted population. (i) Sampling Technique: Stratified Random Sampling and Purposive Stratified random and purposive sampling techniques were employed to ensure a comprehensive representation of different subgroups (strata) within the population. The purposive sampling technique specifically targeted key demographics, such as gender (men and women), age groups (youth and elderly), and marginalized populations (people with disabilities and ethnic minorities), to ensure their inclusion in the data. Stratified random sampling facilitated the capture of variations across bean crop value chain groups and geographical regions, ensuring that the findings were relevant and applicable to all stakeholders involved. By applying stratified random and purposive sampling, we ensure that the data is not biased towards one group, and the specific needs of vulnerable populations. (ii) Sampling Frame The sampling frame consists of a list of all the key groups identified within the project’s geographic scope (e.g., regions of Rwanda, Tanzania, Uganda, and Kenya): ● Farmers, particularly smallholder bean farmers. ● Key bean value chain actors in agriculture, such as cooperative members and local traders. ● Local communities, focusing on gender and marginalized groups. ● The sample strata contain men, women, and youth in proportions determined by the project. The determination of an appropriate sample size is crucial in research endeavors to ensure reliability and validity of findings. In this study, Yamane's formula was employed to calculate 16 the sample size. According to Yamane's formula, the sample size (n) is calculated using the equation: n = N/(1+N(e)2) (Ramadhani & Matoka, 2024) Where: n represents the sample size, N denotes the population of the study and e signifies the margin error in the calculation In this study, the target population (N) is 3,000,000 people, which is the indirect reach for the ECREA project. With a margin of error of ±5%, the sample size was calculated as follows: n = 3,000,000 / (1+(3,000,000*0.05*0.05)) = 399.94 Rounding up, the initial sample size was determined to be 400 smallholder farmers across the four countries. Since larger sample sizes generally provide greater precision (Lakens, 2024), the decision was made to increase the sample size. The adjustment was intended to better address the connectivity issues and ensure a higher level of accuracy in the collected data. This expansion was achieved by implementing the 5Q survey to the 2000 selected farmers in each country. (iii) Key Considerations for Sampling Across All Countries 1. Proportional Sampling: Ensured that regions experiencing high levels of climate vulnerability (e.g., areas prone to droughts, floods, or soil degradation) were adequately represented. 2. Inclusivity: Efforts were made to include youth, women, and people with disabilities in the sample, particularly focusing on groups that are typically marginalized in climate adaptation discussions. 3. Geographical Diversity: Sample selection ensured that data were collected from highland, and lowland regions where beans are cultivated, reflecting the diverse climatic zones across the ECREA project’s countries. 4. Sectoral Representation: In addition to farming communities, the study targeted key stakeholders in the bean value chain (producers, traders, and cooperatives), ensuring comprehensive coverage of all actors. 17 2.4. Collection of data Data on key indicators were systematically gathered from a sample of farmers across all the ECREA intervention areas. The data collection process employed a designed structured questionnaire (see Appendix 1). The study aimed to compile foundational information on critical ECREA project indicators, including the fundamental socio-economic characteristics of farmers. The areas covered included climate-related risks, access to and use of climate information services and Impact-Based Early Warnings, the influence of various information types and sources on agricultural practices, decisions related to farm management, behavioral changes, and the perceived impact resulting from the use of climate services to boost productivity, profitability, and avoid losses. The focus of the data collection process was on Seasonal Forecasts and Impact-Based Early Warning (IBEWs) products or services. This method was developed by the International Centre for Tropical Agriculture (CIAT) to promote a two-way communication loop to stakeholders segregated by their geographical location and influence quick decision-making. The 5Q approach efficiently gathers massive data within a short timeframe and at minimal cost, and systematically analyzes the data to generate results, often in a graphical form. Through the implementation of this approach, a robust evidence base is established, bolstering decision-making, adoption, and impact assessment processes. The systematic collection and analysis of data allow for the visualization of results via graphics, thereby fostering discussions. Moreover, this method enables the systematic tracking of near- real-time changes, thus supporting adaptive management in development project (Eitzinger, 2021). During the survey, the selected farmers received training on how to respond to five questions using Interactive Voice Response (IVR) technology. After this training, the survey questions were translated and recorded in Swahili for Tanzania and Kenya, Luganda and Runyakitara for Uganda, and Kinyarwanda for Rwanda. These recorded questions were then delivered to the sampled farmers across all the project regions via a digital platform using IVR. Additionally, further data was gathered through PICSA workshops conducted in each country, using the same paper-based questionnaire. 18 2.5. Data analysis techniques The survey used descriptive and inferential statistics to analyze collected data on IBEWs and WCI related trends in the surveyed populations, revealing areas of interventions and promising strategies to improve the resilience of the farming communities. Excel functions were used to perform descriptive statistics analysis of a 2304 x 85 (rows x columns) clean data set, capturing surface-level trends on access to and use of WCI and IBEWs among the farmers, WCI-and IBEWs-related preferences, barriers, and challenges of the farming communities, and WCI-and IBEWs-informed decisions and their outcomes among the farming communities. To enhance the granularity of the observed descriptive statistics trends, binary and multinomial logistic regression models based on a 554-participant sample (Kate et al., 2023) were used to explore the socio-economic-and demographic-group disaggregated analysis of the observed descriptive statistics trends. The responses of the 554 participants (approximately 50% women) were converted to the appropriate format that can be handled by MATLAB (see the Excel sheet). (The percentage of female respondents in initial data sets was very low; a systematic elimination of male participants was used to “amplify the voice of women.”) The MATLAB functions—fitmnr and fitglm—were used to conduct one-dependent-variable-on-one-independent-variable analyses to select the best socioeconomic and demographic explanatory variables of the responses (Wang et al.,2020). Education, gender, farmer group membership status, and household land size were found to be the best independent variables for the binary and multinomial logistic regression models: 1. Pr(yi = j/xi) =exp (XBj)/ (1 +∑ij XBj) where i and j =1, 2,...(A multinomial logistic regression model: “Pr” stands for the probability that a respondent belongs to the “j” response category given an “X” socio-economic or demographic characteristic.) 2. Logit(P(j='2')) ~ 1 + XBj (A binary logistic regression model: “XBj” represents the explanatory variables. Here “2” represents “No.” “P” stands for the probability that a respondent with an “X” socio-economic or demographic characteristic belongs to the “No” response category.) https://journals.sagepub.com/doi/full/10.1177/09622802231151220 https://journals.sagepub.com/doi/full/10.1177/09622802231151220 https://docs.google.com/spreadsheets/d/1fBa_Mv-yP7cnGaJPQ-6fBQBCFsBR3eXjiCtkGjzj-BQ/edit?gid=0#gid=0 https://academic.oup.com/jrsssb/article/82/5/1273/7056114 19 The fitmnr and fitglm functions were also used to produce table-format results of the models, showing the strength (regression coefficients) of positive or negative associations between the levels of the explanatory variables and those of the response variables and the probabilities that the observed relationships could have been due to random effects (pValue). 2.6. Research limitations Although the research employed various strategies to enhance participation, such as training the participating study farmers on the 5Q approach, notifying them about the survey one day in advance, sending reminder SMS messages 30 minutes before the survey through the 5Q approach, and making multiple follow-ups calls to those who had not completed the survey, several limitations were encountered during the data collection process. Some farmers did not pick up their phones, while others had their phones turned off during both the training and the survey. Additionally, some farmers chose not to participate in the survey. There were also cases where farmers began the survey but did not complete it. These challenges may be attributed to cultural attitudes toward technology or digital literacy gaps, which can significantly influence farmers' ability and willingness to participate in the 5Q survey process. Issues related to phone connectivity may have also limited the farmers’ participation in the survey. The low response rates on some interview questions may have skewed the socio-economic and demographic-group disaggregated results. Due to low responses, some standard error terms are large, making some results almost unreliable. Nevertheless, most of the models’ results would reasonably probabilistically characterize the study population. 20 Chapter 3: Results This section presents the findings of the study, providing insights into the accessibility, usability, and impacts of WCIS and IBEWS, as well as the barriers limiting the access to and use of the information. It also examines the WCIS-and-IBEWs-related preferences of the farmers who do not receive or do not use the information. The results are analyzed to identify key trends, patterns, and observations, while the discussion explores the implications of these findings for agricultural resilience, highlighting their significance for smallholder farmers and other stakeholders in the target regions. 3.1. The socio-economic characteristics of surveyed farmers The baseline household survey results from the ECREA project, summarized in Table 1, highlight variations in gender, age, education levels, livelihood activities, and roles in the bean value chain among respondents from Rwanda, Uganda, Kenya, and Tanzania. The results offer a lens through which the socio-economic-and demographic-related climate-change-and- variability vulnerabilities and adaptive capacities of the East African farming communities can be understood to guide the climate-change-and variability adaptation and resilience interventions of the project. Gender distribution across the countries reveals that men are generally more represented, with the highest percentage observed in Rwanda (75%) and the lowest in Kenya (51%). Women's representation, while lower in some countries, shows a notable presence, particularly in Uganda and Kenya where nearly half of the respondents are female. Age distribution highlights a young adult to middle-aged population in most countries, with a significant proportion of farmers falling within the 18-35 and 35-55 age ranges. The educational background of the farmers varies significantly across countries, with Rwanda and Tanzania showing a higher proportion of respondents with primary education (59%). In contrast, Uganda and Kenya have more respondents with secondary and tertiary education, with Uganda particularly showing a notable proportion of individuals with university education (24%). 21 Table 1: Demographic information for the farmers who participated in the baseline household survey Countries Rwanda Uganda Kenya Tanzania Variables Results Gender 537 229 140 260 Male 75% 66% 51% 72% Female 25% 34% 49% 28% Age 483 164 188 257 Between 18-35 30% 41.50% 33.00% 43% Between 35-55 53% 32.90% 48.40% 42% Between 55-65 10% 20.10% 14.90% 9% Above 65 2% 3.00% 2.70% 4% I don't know/I don't want to say 5% 2.40% 1.10% 2% Education 462 221 190 264 No formal education 3% 2.30% 1.60% 4% Primary 59% 30.80% 18.90% 59% Lower secondary 17% 24.90% 19.50% 24% Upper Secondary 13% 5.90% 21.60% 3% College 4% 12.20% 22.10% 8% University and above 4% 24.00% 16.30% 2% Livelihood activities 469 215 186 274 Livestock 10% 5.10% 22.60% 11% Crop farming 82% 80.00% 56.50% 74% Transporter 2% 0.00% 0.00% 1% Off taker 3% 3.70% 2.20% 3% Small business 2% 6.00% 14.00% 11% Other 1% 5.10% 4.80% 1% Main role in bean value chain 456 173 169 302 Farmer producing beans for market 24% 54.30% 45.60% 52% Farmer producing beans for home Consumption 56% 31.20% 33.70% 36% Input supplier 15% 8.10% 17.20% 6% Distributor/Aggregator 5% 6.40% 3.60% 5% Other 2% 22 Livelihood activities reveal that crop farming is the dominant activity across the region, with over 70% of respondents in all countries engaged in it. However, there are noticeable differences in the diversification of livelihoods. For example, livestock farming is a significant activity in Kenya (22.6%) but less so in the other countries. Small businesses also play a notable role in Kenya (14%), Tanzania (11%), and Uganda (6%), indicating that some farmers may rely on non-agricultural income sources. The role of farmers in the bean value chain offers insight into their market orientation. Most farmers in Rwanda, Uganda, and Tanzania are involved in growing beans primarily for home consumption, with market-oriented bean production being more prominent in Uganda (54.3%) and Tanzania (52%). The results from Table 2 highlight critical insights into land ownership, land use for bean production, and decision-making processes across Rwanda, Uganda, Kenya, and Tanzania. These findings are significant for the Enhancing Climate Change Resilience in East Africa (ECREA) project, which seeks to bolster resilience and adaptive capacities through improved access to Weather and Climate Information Services (WCIS) and Impact based Early Warnings (IBEWS). The analysis of land ownership reveals substantial variations in land sizes across the four countries. In Rwanda, 56% of respondents own less than 0.5 hectares of land, a proportion significantly higher than in Uganda (29.5%), Kenya (14.9%), and Tanzania (9%). On the other hand, ownership of larger parcels of land (above 5 hectares) was most prevalent in Tanzania at 16%, compared to 10.5% in Uganda, 6.9% in Kenya, and 0% in Rwanda. The analysis of land allocation for bean production reveals significant variations in the proportion of land devoted to this crop across the study countries. In Tanzania, 24% of respondents allocate more than 50% of their land to bean cultivation, a figure notably higher than in Rwanda (13%), Uganda (14.2%), and Kenya (14%). Decision-making around land allocation and crop selection further illustrates gendered and household dynamics. Decisions on land allocation to farming systems were most frequently made jointly across all the countries, with joint decision-making highest in Rwanda (55%) and lowest in Kenya (48%). However, the role of men in these decisions was most pronounced in Tanzania (39%) and least in Rwanda (29%). Similarly, decisions about crop or variety selection were largely made jointly in Rwanda (58%) but showed more prominent female involvement in Tanzania (50%), contrasting with Kenya (29%) and Uganda (28%). 23 Factors influencing decision-making reveal the role of social dynamics and knowledge in agricultural choices. Across the countries, social roles were a significant determinant, particularly in Kenya (54%) and Tanzania (49%). Knowledge and experience were crucial in Uganda (47%), Tanzania (41%), and Kenya (30%), whereas Rwanda exhibited a balance between these two factors. Table 2: Land ownership and decision-making factors in bean production Countries Rwanda Uganda Kenya Tanzania Variables Results Land size they own 427 200 175 254 Below 0.5ha 56% 30% 14.9% 9% Between 0.5ha-1ha 31% 22% 27.4% 24% Between 1ha-2ha 12% 20% 31.4% 31% Between 2ha-5ha 1% 19% 19.4% 19% Above 5ha 0% 11% 6.9% 16% Land size used for bean production 411 197 173 243 Less than 5% 43% 31.50% 30% 13% Between 5%-20% 22% 26.40% 30% 24% Between 20%-35% 14% 17.80% 12% 17% Between 35%-50% 9% 10.20% 16% 22% Above 50% 13% 14.20% 14% 24% Household-level land allocation to farming systems decision-making scheme 453 203 181 254 Man 29% 33.50% 36% 39% Women 11% 17.70% 16% 7% Jointly 55% 44.30% 48% 51% Other HH member (Man) 2% 3.00% 1% 2% Other HH member (Woman) 2% 1.50% 1% 1% The determinants of the decision-maker 431 185 175 246 Main breadwinner (resources) 16% 23.20% 15% 10% Social role 48% 29.70% 54% 49% Knowledge/Experience 35% 47.00% 30% 41% Other 0.60% Household-level crop/variety selection decision-making scheme 419 193 171 252 Man 19% 25.9% 21% 19% women 21% 28.0% 29% 50% Jointly 58% 44.6% 49% 30% Other HH member (Man) 1% 0.5% 1% 0% Other HH member (Woman) 2% 1.0% 1% 0% The determinants of the decision-maker 425 192 167 136 Main breadwinner (resources) 13% 19.3% 9% 14% Social role 49% 25.5% 32% 35% Knowledge/Experience 37% 52.1% 59% 51% 24 The results presented in Table 3 highlight the participation of study farmers in farmer associations and cooperatives across Rwanda, Uganda, Kenya, and Tanzania. A significant proportion of households across the study countries reported being members of farmer associations or cooperatives. Membership was particularly high in Tanzania (84%) and Rwanda (79%), followed closely by Uganda (79%), however, only 59% of respondents indicated membership, reflecting a comparatively lower level of organized farmer participation. The type of associations household heads and spouses belong to also varies across the countries, reflecting diverse priorities and agricultural needs. Farmer cooperatives had the highest membership in Tanzania (42%) and Kenya (36%). In Rwanda and Uganda, agricultural advisory committees or extension services were the most common groups, with 42% and 32.6% participation, respectively. Credit and microfinance groups, which facilitate financial access, were particularly prominent in Uganda (21%) compared to lower levels in Tanzania (10%), Rwanda (11%), and Kenya (15%). The participation of spouses in farmer associations mirrors the overall trends but with some differences. Spousal membership was highest in Rwanda (78%) and Tanzania (75%). Uganda (72.9%) followed closely, while Kenya had the lowest spousal participation at 52.5%. Among the groups spouses joined, farmer cooperatives featured prominently, particularly in Tanzania (36%) and Kenya (30%). 25 Table 3: Participation of the surveyed farmers in farmer associations/cooperatives Countries Rwanda Uganda Kenya Tanzania Variables Results Farmer associations membership status of household heads 478 229 163 288 Yes 79% 79% 59% 84% No 21% 21% 41% 16% The type of association or community groups the household head belongs to 378 181 142 243 Agricultural advisory committee or extension services 42% 32.60% 28% 26% Marketing group 8% 10.50% 9% 9% Credit/microfinance group 11% 21.00% 15% 10% Civic/charitable group 13% 7.70% 8% 13% Farmer cooperative 26% 26.50% 36% 42% Other 1.70% 4% The farmer associations membership status of spouses 418 225 158 284 Yes 78% 72.90% 52.50% 75% No 22% 27.10% 47.50% 25% The type of association or community groups spouses belong to 327 164 122 213 Agricultural advisory committee or extension services 28% 27.40% 30% 23% Marketing group 15% 10.40% 12% 13% Credit/microfinance group 10% 26.80% 14% 8% Civic/charitable group 21% 5.50% 8% 20% Farmer cooperative 20% 26.20% 30% 36% Other 5% 3.70% 6% 0 26 3.2. Accessibility of Weather and Climate Information (WCI) and Impact- Based Early Warnings (IBEWs) among the study farmers 3.2.1. Delivery of WCI and IBEWs through available communication channels In the baseline household survey, farmers who agreed to participate were asked about their access to Weather and Climate Information (WCI) and Impact-Based Early Warnings (IBEWs). The survey explored the frequency, type, sources, and level of sharing of the information, providing insights into the existing WCI-and IBEWs-related gaps and opportunities to strengthen climate resilience through tailored communication strategies. The results in Table 4 reveal varied levels of access to WCI and IBEWs among farmers in Rwanda, Uganda, Kenya, and Tanzania. Regular access to WCI was most consistent in Kenya (44.9%) and Tanzania (44.4%), where nearly half of respondents reported receiving information most of the time. Rwanda and Uganda had lower proportions at 29.5% and 31.8%, respectively. Notably, a significant proportion of farmers in Rwanda (34.8%) and Uganda (39.5%) reported not receiving WCI at all. Regarding IBEWs, Tanzania (42%) and Uganda (40.7%) led consistent access, while Kenya had the highest percentage of farmers (36.1%) who did not receive the information. Table 5 presents the type of information farmers received. Across all the countries, short-term weather forecasts were the most received WCI, especially in Rwanda (50%) and Tanzania (39%). Seasonal forecasts were also significant, with Uganda (40.8%) and Kenya (38.6%) showing strong reliance on this type of information. Tailored WCI for bean production was particularly high in Rwanda and Tanzania, where 84% of respondents confirmed receiving such information, compared to 76.4% in Kenya and 61% in Uganda. For IBEWs, the focus was predominantly on extreme rainfall and its possible impacts, as reported by nearly half of the respondents in all countries. Table 6 shows the primary sources of WCI and IBEWs, indicating a strong dependence on radio as the dominant medium. Radio was cited by over 64% of farmers in Rwanda and slightly lower percentages in Uganda, Kenya, and Tanzania for both WCI and IBEWs. Other notable sources included government extension services and meteorological departments, though their reach was comparatively lower. In Tanzania, television emerged as a more prominent source, utilized by 19% of farmers for WCI and 18% for IBEWs. 27 Table 7 presents the level of sharing of the information, showing high levels of knowledge exchange among farmers. For WCI, over 84% of respondents in all the countries shared the information, primarily with fellow farmers, reflecting a strong culture of peer collaboration. Similarly, IBEWs were widely shared, with Tanzania leading at 92% of respondents engaging in information dissemination. Family members were the second most common group to receive information. The logistic regression models identified several socio-demographic and farm-related factors that significantly influenced farmers’ regular access to Weather and Climate Information (WCI) and Impact Based Early Warnings (IBEWs). These key determinants included education level, membership in farmer associations, landholding size, and the proportion of land allocated to bean cultivation. The analysis revealed that farmers with lower levels of education were significantly less likely to report regularly receiving WCI or IBEWS compared to those with higher levels of education (Table 8). In particular, low education level farmers were also less likely to report access to historical climate information (Table 9), which is crucial for making informed decisions regarding agricultural planning and risk management. Landholding size emerged as another critical factor. Smallholder farmers, defined by limited land size and a smaller share allocated to bean production, were generally less likely to receive both short-term and seasonal weather and climate information (Table 10.1). They were also significantly less likely to report regular access to IBEWS compared to farmers with larger landholdings (Table 10.2). These findings underscore the vulnerability of smallholder farmers to climate variability due to inadequate access to timely and relevant climate information. Membership in farmer associations also played a pivotal role. The models indicated that households whose heads were not members of any farmer association were consistently less likely to report receiving WCI and IBEWS regularly (Tables 11.1 and 11.2). This suggests that farmer associations may serve as important channel for the dissemination of climate-related information and early warning messages. Gender disparities were also evident in access to climate services. According to the regression analysis, female farmers were less likely than their male counterparts to report receiving IBEWS (Table 12.2). This finding highlights the need for targeted interventions to ensure equitable access to climate information services across gender groups. Overall, these results point to systemic inequalities in the access of WCI and IBEWS, with low-education farmers, smallholders, women, and non-members of farmer organizations being at a particular disadvantage. These insights should inform the design of future interventions 28 aimed at improving the reach, inclusivity, and effectiveness of climate information services in bean-growing communities. Table 4: Access to WCIS and IBEWs among the surveyed farmers Countries Rwanda Uganda Kenya Tanzania Variables Results Regularity of access to Weather and Climate information Services 414 337 176 232 Yes, most of the times 29.5% 31.8% 44.9% 44.4% Yes, sometimes 35.7% 28.8% 33.5% 34.5% Not at all 34.8% 39.5% 21.6% 21.1% Impact-Based Early Warnings information received 337 204 191 192 Yes, most of the times 28% 40.7% 25.1% 42% Yes, sometimes 48% 37.7% 38.7% 41% Not at all 23% 21.6% 36.1% 17% Table 5: Types of WCI and IBEWs received by surveyed farmers Countries Rwanda Uganda Kenya Tanzania Variables Results Type of WCI Received 270 152 140 176 Historical climate information 4.1% 3.9% 2.1% 20% Seasonal forecast 30.7% 40.8% 38.6% 29% Short term forecast 50% 32.9% 28.6% 39% Seasonal and short-term forecast 7.8% 19.7% 17.9% 1% All of them 7.4% 2.6% 12.9% 11% WCI received tailored to bean production 241 136 165 207 Yes 84% 61.0% 76.40% 84% No 16% 39.0% 23.60% 16% Type of IBEWs received 258 120 128 170 Extreme rainfall and their possible impacts 48% 39.2% 46.9% 44% Information on wind and their possible impacts 10% 9.2% 7.8% 8% Information on thunderstorms and their possible impacts 6% 7.5% 4.7% 4% Information on both extreme rainfall and wind and their possible impacts 12% 17.5% 17.2% 11% Information on both extreme rainfall, thunderstorms and their possible impacts 9% 11.7% 9.4% 12% All of them 16% 15.0% 14.1% 22% 29 Table 6: Sources of information for WCI and IBEWs among the surveyed farmers Countries Rwanda Uganda Kenya Tanzania Variables Results Did you share the Weather and Climate Information? 222 137 135 165 Yes 89% 87% 84.40% 87% Not at all 11% 13% 15.60% 13% Whom did you share W&C information with? 198 119 114 142 Fellow farmers 68% 59.70% 67.50% 73% Family members 21% 21% 17.50% 18% Others 11% 19.30% 14.90% 9% Did you share the IBEWs information? 224 100 117 181 Yes 78% 90% 88% 92% No 22% 10% 12% 8% Whom did you share IBEWs information with? 175 90 103 167 Fellow farmers 72% 66.70% 63.10% 66% Family members 18% 22.20% 15.50% 28% Others 10% 11.10% 21.40% 7% Table 7: WC& IBEWs information sharing among the surveyed farmers Countries Rwanda Uganda Kenya Tanzania Variables Results Source of Weather and Climate Information 270 204 183 235 Radio 68% 37.30% 30.60% 42% Government extension services 15% 13.70% 13.10% 11% NGOs 3% 5.90% 3.80% 4% Meteorological department 9% 7.40% 13.70% 15% SMS and toll-free number 3% 6.90% 12.60% 2% Internet/websites 1% 9.80% 7.70% 6% Television 2% 17.20% 13.10% 19% Peer farmers 0.00% 5.50% 0% Source of Impact Based Early Warnings information 258 160 168 198 Radio 64.70% 39.40% 30.40% 43% Government extension services 15.50% 18.10% 12.50% 14% NGOs 2.30% 5.00% 6.00% 6% Meteorological department 8.50% 7.50% 13.10% 13% SMS and toll-free number 4.30% 7.50% 10.10% 4% Peer farmer 0.80% 3.80% 1.80% 2% Television 3.90% 13.10% 16.10% 18% Internet/Websites 0% 5.60% 10.10% 0% 30 Table 8: Effect of education on regularity of access to weather and climate information Variable Regression co-efficient Standard error T-static pValue (Intercept_1) 18.773 18.471 1.0164 0.30945 Lower-secondary education vs. most-of-the- time WCI access level -1.7899 13.441 -0.13317 0.89406 Upper-secondary education vs. most-of-the- time WCI access level 8.2753 23.337 0.35459 0.72289 College education vs. most-of-the-time WCI access level 11.083 19.93 0.55608 0.57816 University-and-above education vs. most-of- the-time WCI access level 1.7893 20.502 0.087274 0.93045 No-formal-education vs. most-of-the-time WCI access level 13.789 60.003 0.2298 0.81825 (Intercept_2) 18.876 18.472 1.0219 0.30683 Lower-secondary education vs. some-times CI access level -1.6723 13.442 -0.12441 0.90099 Upper-secondary education vs. some-times WCI access level 9.057 23.339 0.38807 0.69797 College education vs. some-times WCI access level 11.391 19.931 0.57152 0.56765 University-and-above vs. some-times WCI access level 2.5375 20.503 0.12376 0.9015 No-formal-education vs. some-times WCI access level 13.698 60.009 0.22827 0.81944 (Intercept_3) 16.862 18.485 0.9122 0.36166 Lower-secondary education vs. not-at-all WCI access level -1.525 13.448 -0.1134 0.90971 Upper-secondary education vs. not-at-all WCI access level 8.983 23.345 0.38479 0.7004 College education vs. not-at-all WCI access level -73.93 27.812 -2.6581 0.007857 University-and-above vs. not-at-all WCI access level -152.62 27.528 -5.5444 2.95E-08 No-formal-education vs. not-at-all WCI access level -12.007 97.403 -0.12327 0.90189 *Reference category: primary education level. 31 Table 9: Effect of education on types of WCI received by surveyed farmers Variable Regression coefficient Standard error T-static pValue (Intercept_3) 11.118 15.489 0.71779 0.47289 Lower secondary education vs. historical climate information -22.134 17.391 -1.2727 0.20312 Upper secondary education vs. historical climate information -22.563 25.441 -0.8869 0.37515 College education vs. historical climate information 0.006247 4 15.448 0.0004 0.99968 University-and-above education vs. historical climate information -54.2 15.784 -3.4339 0.00059 No-formal-education vs. historical climate information -16.02 58.097 -0.2758 0.78274 Table 10: Effect of household land size on the regularity of access to IBEWs and the types of weather and climate information received by surveyed farmers Table 10.1: Land size and WCI received Variable Regression coefficient Standard error T-static pValue (Intercept_8) 3.424 20.806 0.16457 0.86929 0.5 ha - 1 ha vs. seasonal, short- term, and both seasonal and short- term forecasts -53.709 13.846 -3.8789 0.0001 1 ha - 2ha vs. seasonal, short- term, and both seasonal and short- term forecasts 5.8991 13.104 0.45019 0.65258 2ha - 5 ha vs. seasonal, short- term, and both seasonal and short- term forecasts 4.3281 13.341 0.32441 0.74563 Greater than 5 ha vs. seasonal, short-term, and both seasonal and short-term forecasts -22.866 26.846 -0.8518 0.39435 32 Table 10.2: effect of land size and regularity of access to IBEWs Variable Regression coefficient Standard error T-static pValue (Intercept_1) 1.1783 0.47678 2.4713 0.01346 0.5 ha - 1 ha vs. most of the time 1.419 0.64847 2.1882 0.02866 1 ha - 2 ha vs. most of the time 1.7053 0.82015 2.0793 0.03759 2 ha - 5 ha vs. most of the time 2.7861 1.0897 2.5569 0.01056 Greater than 5 ha vs. most of the time 25.256 82839 0.0003 0.99976 (Intercept_2) 1.7563 0.4521 3.8847 0.0001 0.5 ha - 1 ha vs. sometimes 1.1655 0.61684 1.8895 0.05882 1 ha - 2 ha vs. sometimes 1.3319 0.79921 1.6666 0.0956 2 ha - 5 ha vs. sometimes 1.7515 1.0789 1.6235 0.10448 Greater than 5 ha sometimes 24.466 82839 0.0003 0.99976 Table 11: Effect of farmer associations membership status on the regularity of access to weather and climate information and the types of information received by surveyed farmers Table 11.1: Types of weather and climate information received Variable Regression coefficient Standard error T-static pValue (Intercept_8) -48.044 12.8 -3.7533 0.00017 No farmer association for household head vs. seasonal, short-term, and both seasonal and short-term forecasts -111.91 8.9839 -12.457 1.28E-35 Table 11.2: Regularity of access to weather and climate information Variable Regression coefficient Standard error T-static pValue (Intercept_1) 1.6121 0.48964 3.2923 0.00099 No farmer association for household heads vs. most of the time -1.8977 0.56411 -3.3641 0.00077 (Intercept_2) 1.9194 0.47342 4.0543 5.03E-05 No farmer association for household heads vs. some times -1.2929 0.53173 -2.4315 0.01504 33 Table 12: Gender and farmer groups membership status and types of Impact-Based Early Warnings received Table 12.1: Farmer groups membership status and types of IBEWs received Variable Regression coefficient Standard error T-static pValue (Intercept_4) 10.573 65.346 0.1618 0.87147 No farmer association for spouses vs. extreme rainfall and wind and their possible impacts -2.7148 1.6344 -1.661 0.09671 (Intercept_5) 11.372 65.341 0.17404 0.86183 No farmer association for spouses vs. extreme rainfall and thunderstorms and their possible impacts -2.8169 1.6137 -1.7456 0.08089 (Intercept_2) 3.0117 0.7377 4.0826 4.45E- 05 No farmer association for household heads vs. wind and possible impacts -1.8396 0.70336 -2.6154 0.00891 Table 12.2: Gender and types of IBEWs received Variable Regression coefficient Standard error T-static pValue (Intercept_2) 8.5362 11.48 0.74359 0.45712 Female vs. information on wind and possible impacts 4.5339 2.4535 1.8479 0.06461 (Intercept_4) -0.919 0.71036 -1.2937 0.19577 Female vs. both extreme rainfall and wind and their possible impacts 1.3208 0.66625 1.9824 0.04744 3.2.2. Preferences and barriers/challenges in accessing WCI and IBEWs To identify the barriers and preferences of respondents who do not have access to the information, farmers were asked to specify why they were not receiving WCI and IBEWs, as well as their preferences for receiving such information. Table 13 explores similar barriers for farmers who did not receive IBEWs, highlighting both infrastructural and knowledge-based challenges. In Rwanda, 46% of farmers did not know the information was available, while in Tanzania, this figure was even higher at 57%. Furthermore, 35% of farmers in Rwanda and 43% in Kenya cited a lack of access to channels as a reason for not receiving IBEWs. Similar to WCI, the majority of respondents expressed interest in receiving IBEWs, with 94% in Rwanda, 97% in Kenya, and 85% in Tanzania. The preferred channels for IBEWS were similar to those for WCI, with a notable preference for SMS and toll-free numbers, radio, and meteorological departments. 34 Table 14 illustrates the reasons why farmers in different countries did not receive WCI, and it reveals both knowledge gaps and infrastructural challenges. In Rwanda, the majority (53%) of farmers did not know that the information was available, while in Tanzania, this was true for 64% of respondents. In addition, a notable proportion of farmers across the countries reported that the information was disseminated at inappropriate times (e.g., 19% in Rwanda and Kenya). The multinomial logistic regression analysis revealed that education, gender, and farmer association membership status factors determined barriers faced by farmers. The models showed that low-education-level farmers were more likely than their high-education-level counterparts to report lacking awareness of availability of IBEWs and access to the channels of the information (Table 15). Additionally, the analysis highlighted that women (Table 16.1) and farmers who were not members of farmer associations (Table 16.2) were more likely than their counterparts to report the problem. Despite these barriers, there was a strong desire for access to WCI. The majority of respondents expressed interest in receiving this information, with 91% in Rwanda, 95% in Uganda, 97% in Kenya, and 98% in Tanzania. When asked about their preferred channels for receiving WCI, farmers showed a strong preference for radio, especially in Rwanda (30%) and Kenya (37%). Other preferred channels included SMS and toll-free numbers, government extension services, and meteorological departments, with varying preferences across the countries. Finally, Table 17 addresses the concerns of farmers who were not interested in receiving WCI or IBEWs. In Rwanda and Uganda, some farmers found the information complicated to understand, with 45% in Rwanda and 43% in Uganda citing this as a barrier. However, the primary concern in Kenya was trust in the information, with 100% of non-recipients indicating that they did not trust it. 35 Table 13: Preferences and barriers for farmers who do not receive WCI Countries Rwanda Uganda Kenya Tanzania Variables Results The reasons for not receiving WCI 144 50 26 44 The inaccessibility of channels used 28% 52% 30.80% 16% The lack of awareness of the availability of the information 53% 36% 50% 64% The information was disseminated at inappropriate time 19% 12% 19.20% 20% Would you like to receive WCI 144 133 39 47 Yes 91% 95% 97% 98% No 9% 5% 3% 2% WCI channels preferred 131 126 38 44 Radio 30% 25% 16% 20% Government extension services 10% 12% 11% 11% NGOs 2% 7% 0% 5% Meteorological department 20% 10% 24% 7% SMS and toll-free number 37% 22% 37% 36% Internet/websites 1% 10% 11% 5% Television 1% 10% 3% 5% Peer farmers 0 5% 0 11% Types of information that would be helpful for the farmers 128 75 37 42 Historical climate information 2% 4% 8% 2% Seasonal forecast 48% 37% 16% 26% Short term forecast 20% 27% 30% 31% Seasonal and short-term forecast 5% 21% 22% 2% All of them 25% 11% 24% 38% 36 Table 14: Preferences and barriers for farmers who do not receive IBEWs Countries Rwanda Uganda Kenya Tanzania Variables Results The most likely reasons why the farmers do not receive IBEWs information 74 44 69 28 Inaccessibility of channels used 35% 47.7% 43% 32% Lack of awareness of the availability of the information 46% 43.2% 21% 57% Disseminated at inappropriate time 19% 9.1% 36% 11% Would you like to receive information on IBEWs? 72 0 69 25 Yes 94% 97% 85% No 6% 3% 15% IBEWs channels preferred by the farmers 67 0 67 22 Local NGOs 3% 6% 0% Radio 32.8% 24% 40.90% Government extension services 3% 7% 4.50% Internet(websites) 1.5% 10% 4.50% Meteorological department 22.4% 13% 4.50% SMS & Toll-free number, WhatsApp 34.3% 34% 36.40% Peer farmers 0% 4% 4.50% Television 3% 0% 4.50% Others 0% 0% 0% Types of IBEWs that would be helpful for the farmers 66 0 54 22 Extreme rainfall and their possible impacts 48.50% 38.90% 50.00% Information on wind and their possible impacts 7.60% 7.40% 0%% Information on thunderstorms and their possible impacts 4.50% 3.70% 18.20% Both extreme rainfall and wind and their possible impacts 9.10% 14.80% 9.10% Extreme rainfall and thunderstorms and possible impacts 7.60% 22.20% 18.2 All of them 22.70% 13.00% 22.70% 37 Table 15: Effect of education levels on farmers’ barriers to IBEWs access Variable Regression coefficient Standard error T-static pValue (Intercept_1) 1.8089 0.60948 2.9679 0.002998 Lower secondary education vs. lack of access to channels used -0.36313 0.66741 -0.54409 0.58638 Upper secondary education vs. lack of access to channels used -2.6096 0.95156 -2.7425 0.006098 College education vs. lack of access to channels used 0.033397 1.2772 0.026148 0.97914 University-and above education vs. lack of access to channels used 0.53712 1.2506 0.42948 0.66757 No formal education vs. lack of access to channels used 10.013 109.57 0.091383 0.92719 (Intercept_2) 0.055598 0.75187 0.073946 0.94105 Lower secondary education vs. lack of awareness about the availability of information 0.042209 0.79859 0.052854 0.95785 Upper secondary education vs. lack of awareness about the availability of information -3.5219 1.3931 -2.5281 0.011468 College education vs. lack of awareness about the availability of information 0.89823 1.4388 0.62428 0.53245 University-and above education vs. lack of awareness about the availability of information 1.416 1.3413 1.0557 0.29112 No formal education vs. lack of awareness about the availability of information -0.0094079 152.04 -6.19E-05 0.99995 38 Table 16: Gender and farmer association membership status and barriers to accessing IBEWs Table 16.1: Gender and barriers to accessing IBEWs Variable Regression coefficient Standard error T-static pValue (Intercept_2) -0.24163 0.61855 -0.39064 0.69606 Female vs. lack of awareness of the availability of the information 1.8287 0.64807 2.8217 0.00478 Table 16.2: Farmer association membership status and barriers to accessing IBEWs Variable Regression coefficient Standard error T-static pValue (Intercept_2) -0.24163 0.61855 -0.39064 0.69606 No farmer association group for household heads vs. lack of awareness about the availability of information 1.221 0.72467 1.6849 0.092 (Intercept_3 No farmer association group for household heads vs. untimely dissemination of information 2.575 1.4348 1.7947 0.0727 Table 17: Concerns and barriers for farmers who would not like to receive the WCIS and IBEWs Countries Rwanda Uganda Kenya Tanzania Variables Results The most likely reasons why the farmers do not like to receive WCI 11 7 1 1 The perceived untrustworthiness of the information 18% 0% 100% 0 The technical/linguistic inaccessibility of the information 45% 43% 0 100% They do not understand how to use it 36% 57% 0 0 The most likely reasons why the farmers don’t like to receive IBEWs 3 0 2 3 It's complicated to understand 33.30% 50% 33.30% They do not understand how to use it 66.70% 50% 66.70% 39 3.3. Use of Weather and Climate Information (WCI) and Impact-Based Early Warnings (IBEWs) among the study farmers 3.3.1. Utilization, Usefulness, and Accuracy of WCI and IBEWs among the study Farmers Farmers who had access to information were surveyed regarding the utilization of the information provided to anticipate, prepare, and make informed decisions in coping with hazards. They were also asked about the usefulness and accuracy of the information received in predicting weather events and hazards. Table 18 reveals that most farmers across the four study countries reported using WCI to some extent. In Tanzania, 54.88% of farmers reported that they used WCI most of the time to plan and make timely decisions, followed by Kenya (48.6%) and Rwanda (40.91%). While a significant proportion of farmers used WCI occasionally (43.29% in Tanzania, 52.27% in Rwanda, 48% in Kenya, and 43.2% in Uganda), a smaller group of farmers did not use the information at all, with the lowest percentage found in Tanzania (1.83%). When asked about the usefulness of WCI in anticipating and preparing for climate events and disasters, a large majority of farmers found the information helpful. In Rwanda, 58% of farmers reported that WCI was useful most of the time, while 37% found it useful sometimes. Similar patterns were observed in Uganda, Kenya, and Tanzania, with Tanzania again showing the highest percentage (47%) of farmers who found the information useful most of the time. In all the countries, most farmers (70% for Rwanda, 63.1% for Uganda, 55.4% for Kenya, and 42% for Tanzania) reported that WCI was accurate sometimes. Similarly, the utilization of IBEWs was assessed as shown in Table 19. A significant proportion of farmers across all the countries used IBEWS information, with the highest usage reported in Kenya (58.3%) and Tanzania (56%). In contrast, Rwanda and Uganda had slightly lower usage rates, but still, most farmers used the information at least sometimes. The usefulness of IBEWs in anticipating and preparing for climate events was also high, with Kenya and Tanzania again showing the most positive responses (61% and 61.5%, respectively). Rwanda (56%) and Uganda (45.6%) followed, indicating that IBEWs were seen as a valuable tool for farmers in these regions. 40 The logistic regression models further showed that households that did not belong to farmer associations were less likely than their counterparts to report that WCI accurately predicted the actual event (Table 20). Additionally, the models revealed that these households were less likely to report that WCI helped them anticipate and make timely response plans (Table 21). Furthermore, according to the models, female farmers were less likely than male counterparts to report that IBEWs were accurate. In terms of accuracy, the results were like those of WCI, with a high percentage of farmers reporting that IBEWs were accurate at least some of the time. In Kenya and Tanzania, 44.5% and 47% of farmers, respectively, reported that IBEWs were accurate most of the time. Logistic regression analysis revealed that farmers who allocated large pieces of their land to bean production were more likely to report that IBEWs helped them make preparedness plans (Table 20). These farmers were also likely to report that IBEWs accurately predicted the actual event (Table 21.1). Additionally, the models highlighted that female farmers were less likely to report IBEWs were accurate. Nevertheless, women were more likely than men to report that IBWs were suitable for beans (21.2). Furthermore, the models showed that farmers who did not belong to farmer associations were less likely than their counterparts to report that IBEWs were accurate (Table 22). Table 18: Perception of the surveyed farmers on use, usefulness and accuracy of WCI Countries Rwanda Uganda Kenya Tanzania Variables Results The usefulness of WCI for helping the farmers anticipate and devise a preparedness plan for making timely informed decisions 220 148 148 164 Yes, most of the times 40.91% 32.4% 48.6% 54.88% Yes, sometimes 52.27% 43.2% 48% 43.29% Not at all 6.82% 4.1% 3.4% 1.83% The usefulness of WCI for helping the farmers anticipate and prepare for a climate event and disaster 239 130 126 174 Yes, most of the time 58% 36.9% 50.8% 47% Yes, sometimes 37% 55.4% 45.2% 51% Not at all 5% 7.7% 4% 3% The accuracy of WCI to help the farmers predict the actual weather events and hazards 230 130 130 173 Yes, most of the time 21% 28.5% 41.5% 54% Yes, sometimes 76% 63.1% 55.4% 42% Not at all 3% 8.5% 3.1% 4% 41 Table 19: Perception of the surveyed farmers on use, usefulness, and accuracy of IBEWs Countries Rwanda Uganda Kenya Tanzania Variables Results The regular use rates of IBEWs 247 132 108 147 Yes, most of the times 49% 51.50% 58.30% 56% Yes, sometimes 45% 35.60% 38.90% 39% Not at all 7% 12.90% 2.80% 4% The usefulness of IBEWs for helping the farmers anticipate and prepare for a climate event or disaster 230 103 109 155 Yes, most of the times 56% 45.60% 61.50% 61% Yes, sometimes 40% 42.70% 36.70% 38% Not at all 4% 11.70% 1.80% 1% The accuracy of IBEWs to help the farmers predict the actual weather events and hazards 228 98 110 159 Yes, most of the times 29% 28.60% 44.50% 47% Yes, sometimes 71% 60.20% 50.90% 50% Not at all 1% 11.20% 4.50% 3% Table 20: Effect of land allocated to beans on perceived usefulness of IBEWs for making preparedness plans Variable Regression coefficient Standard error T-static pValue (Intercept_1) 1.7684 0.51925 3.4057 0.00066 5% - 20 % vs. most of the time 1.1432 0.70701 1.6169 0.10589 20% - 30 % vs. most of the time 1.7418 1.0877 1.6014 0.10928 30% - 50 % vs. most of the time 25.429 76165 0.00033 0.99973 Greater than 50% vs. most of the time 1.9367 1.0879 1.7802 0.07504 (Intercept_2) 1.4359 0.52481 2.7361 0.00622 5% - 20 % vs. sometimes 1.254 0.70446 1.7801 0.07507 20% - 30 % vs. sometimes 1.6499 1.0895 1.5144 0.12994 30% - 50 % vs. sometimes 25.525 76165 0.00034 0.99973 Greater than 50% sometimes 1.9213 1.0912 1.7607 0.07829 *Reference categories: less than 5% and not at all. 42 Table 21: Effects of gender and land-size allocated to beans on perceived predictive accuracy of IBEWs and gender on perceived suitability of WCI for bean production Table 21.1: Land allocated to beans and perceived accuracy of IBEWs Variable Regression coefficient Standard error t-static pValue (Intercept_1) 2.4151 1.1479 2.104 0.03538 5% - 20 % vs. most of the time -0.66722 1.0282 -0.6489 0.5164 20% - 30 % vs. most of the time -1.4885 1.1622 -1.2808 0.20027 30% - 50 % vs. most of the time -1.9486 1.1222 -1.7364 0.0825 Greater than 50% vs. most of the time -1.2486 1.1716 -1.0657 0.28656 (Intercept_2) 3.034 1.1197 2.7097 0.00674 5% - 20 % vs. some times -1.05 1.0099 -1.0398 0.29845 20% - 30 % vs. some times -1.8104 1.1371 -1.5922 0.11135 30% - 50 % vs. some times -1.9298 1.0958 -1.7611 0.07821 Greater than 50% vs. some times -1.7585 1.155 -1.5226 0.12786 Table 21.2: gender and perceived accuracy of IBEWs Variable Regression coefficient Standard error T-static pValue (Intercept_2) 3.5813 0.71034 5.0417 4.61E- 07 Female vs. some times -1.0668 0.60506 -1.7631 0.07788 Table 21.3: effect of gender on farmers' perceived suitability of WCI for bean production Variable Regression coefficient Standard error T-static pValue (Intercept) -0.48714 0.55622 -0.8758 0.38114 Female vs. No -0.79442 0.34575 -2.2977 0.02158 43 Table 22: Effect of farmer group membership status of spouses and household heads on perceived anticipatory accuracy of IBEWs and types of IBEWs received by surveyed farmers Table 22.1: effect of farmer group membership status of spouses and household heads on perceived anticipatory accuracy of WCI Variable Regression coefficient Standard error T-static pValue (Intercept_1) 2.1788 0.58777 3.7069 0.00021 No farmer association for household heads vs. most of the time -1.155 0.67257 -1.7173 0.08593 No farmer association for spouses vs. most of the time 2.0492 1.1005 1.8621 0.06259 (Intercept_2) 1.6252 0.59933 2.7117 0.00669 No farmer association for spouses vs. some times 2.5451 1.1002 2.3132 0.02071 Table 22.2: effect of farmer group membership status of spouses and household heads on perceived predictive accuracy of WCI Variable Regression coefficient Standard error T-static pValue (Intercept_1) 3.4693 1.182 2.935 0.00334 No farmer association for spouses vs. most of the time -1.3022 0.70533 -1.8462 0.06486 *Reference category: a member of the farmer association. 3.3.2. Preferences and Barriers/Challenges in Using Information on W&C and IBEWs Farmers who did not use the received WCI and IBEWs were asked to elaborate on the reasons for their non-use, as shown in Table 12. The results reveal several factors that hinder the effective use of these critical climate services, offering insights into challenges that must be addressed to improve climate resilience in East Africa. In terms of WCI, the primary reason for non-utilization was a lack of understanding of the information. In Rwanda, 47% of respondents cited this as the reason for not using the WCI, a significantly higher proportion than in Uganda (6.7%), Kenya (28.6%), and Tanzania (33%). The second most common reason was the information arriving too late, which was reported by 27% of respondents in Rwanda, 33 % in Kenya and 40% in Uganda, indicating delays in receiving critical climate information during crucial farming periods. 44 A smaller proportion of farmers in Rwanda (13%) and Uganda (26.7%) also mentioned insufficient resources as a barrier, suggesting that even when information is available, farmers may lack the means to act upon it. Regarding IBEWS, the most common reason for non-utilization across the countries was a lack of trust in the information, particularly in Uganda, where 80% of farmers did not trust the IBEWS information received. In Rwanda, 76.47% of respondents stated that they did not use IBEWS due to a lack of understanding, indicating that similar issues of communication and clarity affect both WCI and IBEWS. Additionally, insufficient resources were cited by 33.33% of Tanzanian farmers, and late information was reported by 5.88% in Rwanda. A notable finding from the IBEWs results is the absence of decision-making involvement for some farmers, with 50% of Kenyan respondents indicating that they were not the ones making decisions about the use of IBEWs. Table 23: Reasons for non-utilization of received WCI & IBEWs Countries Rwanda Uganda Kenya Tanzania Variables Results The most likely reasons why the farmers do not use the WCI information received 15 15 7 3 Not trusting the information 13% 20% 28.60% 0% Not understanding the information 47% 6.70% 28.60% 33% Information received too late 27% 40% 0% 33% Lack of household-level decision-making authority 0% 6.70% 0% 0% Insufficient resources 13% 26.70% 42.90% 33% The reason they didn't use IBEWs received 17 40 2 6 Did not trust the information 11.76% 80% 50% 16.67% Did not understand the information 76.47% 10% 0% 33.33% The information came late 5.88% 2.50% 0% 16.67% Insufficient resources 5.88% 5% 0% 33.30% Lack of household-level decision-making authority 0% 2.50% 50% 0% 45 3.4. The Impact of WCI and IBEWs on Climate Change Resilience among the Study Farmers 3.4.1. Farmers' Adaptive Capacity to Climate Risks using WCI and IBEWs To assess how weather and climate information (WCI) and Impact-Based Early Warning Systems (IBEWS) are helping farmers in Eastern Africa enhance their adaptive capacity to climate-related risks, survey data was collected to understand the decisions influenced by such information. Farmers were asked about the types of decisions made, the costs associated with implementing these decisions, and the climate shocks or stresses they experienced between 2022 and 2023. Tables 24, 25, and 26 provide valuable insights into how farmers are responding to the impacts of extreme weather and the role of WCI and IBEWs in guiding these actions. The Results in Table 24 show that the most common decisions influenced by WCI and IBEWs across the four countries were related to pl