Women farmers’ participation in the agricultural research process: implication for agricultural sustainability in Ethiopia Annet A. Mulema (ILRI), Wellington Jogo (CIP), Elias Damtew (ILRI), Kindu Mekonnen (ILRI) and Peter Thorne (ILRI) Seeds of Change Conference, University of Canberra, Australia, 2-4 April 2019 There MUST be a CGIAR logo or a CRP logo. You can copy and paste the logo you need from the final slide of this presentation. Then you can delete that final slide   To replace a photo above, copy and paste this link in your browser: http://www.flickr.com/photos/ilri/sets/72157632057087650/detail/ Find a photo you like and the right size, copy and paste it in the block above.

Presentation outline Context Research issue Research objectives Methods Results Conclusion

Ethiopian context Ethiopia is a highly agrarian and densely populated country with fragile natural resource Technological change is struggling to keep pace with the rapidly growing population The country is prone to droughts and about 10% of the population are chronically food insecure Women contribute about 43% to the agricultural labor force, managing calorie generating plots. Women also manage calorie generating plots which are key to the food security of the household. Women’s contribution to agriculture limited by Social norms Limited scope of agricultural policy Limited metrics and indicators to guide policy towards women empowerment in agriculture

Research issue Women are under represented in agricultural research, extension and governance systems. Inclusion of women in agricultural research may enhance food security in a country that is prone to droughts and market inefficiencies. Enhancing women’s role as innovators. agricultural producers and care takers is critical to sustain agriculture. Women are under represented in agricultural research and governance systems and often given inadequate attention by external agencies Limited women representation contributing to food insecurity

Research issue Empowerment of women is central to their participation in agricultural research and achieving sustainable development. Limited understanding of the relationship between women’s empowerment and participation in agricultural research processes. Social aspects of agricultural sustainability have received limited attention. Although new agricultural development policies are focusing on improving women engagement in agriculture, there is limited literature on women empowerment in relation to agricultural research. Agricultural sustainability is multi-faceted with important drivers operating in the human, social and environmental domains Social aspects have received limited attention and limited research undertaken to understand their implications

Objectives of the study Understand how women farmers are involved in the agricultural research process Determine the socio-economic factors that influence women’s participation in different stages of the agricultural research process. This approach would enable entry points to be identified and the design of effective strategies for women’s empowerment

Research sites – Africa RISING project sites Description of Africa RISING

Methods Mixed methods 230 individual interviews with women farmers 16 FGDs with men and women farmers Africa RISING project participants and non- participants Africa RISING is a USAID funded project under the feed the future initiative, aimed at creating opportunities for smallholder farm households to move out of hunger and poverty through sustainably intensified farming systems. Empowerment is central to women’s participation in agricultural research and to boost their role in agriculture and contribution to food security. To do so it is important to understand the current level of their participation and the factors that influence their participation in the agricultural research process. This enables identification of entry points and design of effective strategies for women empowerment.

Methods… The Africa RISING project research process Identification of problems and opportunities Presentation of potential solutions and selection of farmers to test/validate technologies Groups farmers into farmer research groups Implementation on action research Monitoring, Evaluation and Learning Capacity development

Data analysis Qualitative data were analyzed using line-by-line coding. Quantitative data were analyzed using binary and multivariate probit models. Generated a composite women’s empowerment index based on the WEAI domains Included this index together with other socio- economic variables into econometric models Included this index together with other socio-economic variables in a multiple regression analysis to assess how women’s empowerment affects their participation in agricultural research while controlling for other factors Understand how the specific individual empowerment indicators are associated with women’s participation in each stage of the agricultural research process

The stages of the agricultural research process Stage of the agricultural research process Description Identification and prioritization of research and development problems (design stage) Identifying and prioritizing agricultural problems or opportunities in the kebele Identifying possible solutions to be tested Identification and testing of potential technology options (testing) Identifying and selecting farmers to participate in testing different technologies Organizing farmers in to farmer research groups based on preferences Conducting demonstrations or experiments to test technology options   Dissemination of tested and validated technologies (diffusion) Creating awareness of recommended solutions among future users e.g. hosting farmer field days Monitoring and evaluation Data collection, reflection and information sharing to decide on actions to be taken Technology assessments e.g. Participatory Variety Selection (PVS)

Women participation in agricultural research stages Africa RISING project participants (n=118) Africa RISING project non-participants (n=112) Total (N=230) Pearson Chi-square (p-value) Did not participate in any stage 36 21 0.000 Identification and prioritization of research and development problems (design) 55 30 42 0.001 Identification and testing of potential technology options (testing) 58 17 34 Dissemination of tested and validated technologies (diffusion) 38 27 0.002 Monitoring and evaluation 32 12 22 The results reveal a positive and significant (p<0.01) relationship between participation in the Africa RISING project and participation in all the stages of the research process. Women participation is highest in identifying and prioritizing research problems and identifying and testing of technology options. The level of women participating in dissemination of tested technologies and monitoring and evaluation is low.

Results from binary probit model for women participation in each stage (Decomposed model)   Explanatory variables STAGES Did not participate in any stage Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated technologies Monitoring and evaluation AGE 0.401 (0.03) -0.014 (0.02) 0.000 (0.02) 0.005 (0.02) 0.001 (0.21) EDUCATION 0.405 (0.29) 0.260 (0.22) 0.212 (0.17) 0.150 (0.17) 0.121(0.17) MARITAL STATUS -0.007 (0.68) -0.450 (0.44) -0.033 (0.37) -0.123 (0.14) -0.171 (0.37) LAND SIZE 0.003 (0.07) -0.039 (0.04) -0.016 (0.03) 0.099 (0.05)** -0.042 (0.33) AR PARTICIPATION -0.915 (0.59) 0.069(0.39) 1.711 (0.39) *** 0.690 (0.37)* 0.186 (0.35) INFOR SOURCE -0.196 (0.16) 0.246 (0.14)* 0.093 (0.11) -0.111 (0.12) 0.208 (0.13)* EXTENSION ACCESS -1.444 (0.56)*** 2.141 (0.64) *** -0.179 (0.53) -0.162 (0.54) 1.165 (0.67)* FAMER GROUP 0.189 (0.48) 1.399 (0.48) *** -0.004 (0.36) 0.094 (0.36) 0.717 (0.36)** PUBLIC SPEAK -0.875 (0.56) 0.487 (0.41) 0.699 (0.39) 1.077 (0.39)*** 0.832 (0.39)** CREDIT DECISION 0.542 (2.82) -1.510 (1.22) -1.086 (0.76) 0.833 (0.78) -1.042 (0.64) INCOME DECISION -0.970 (1.39) 1.918 (1.524) 0.385 (0.95) 0.123 (0.24) 0.607 (1.07) PROD DECISION 0.903 (2.70) 2.027 (1.44) * 0.119 (0.92) 0.001 (0.45) -0.498 (0.90) PLOT INDIVIDUAL -1.845 (0.63)*** 1.604 (0.47) *** 0.260 (0.40) 1.008 (0.47)*** 0.402 (0.41) LABOR SUFFICIENT 1.10 (0.97) -1.444 (0.64) -0.526 (0.59) -0.270 (0.42) -0.068 (0.44) Pseudo R2 0.52 0.50 0.35 0.29 0.27 P-value 0.000 0.002

Results from binary probit model for women participation in each stage (Decomposed model)   Explanatory variables STAGES Did not participate in any stage Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated technologies Monitoring and evaluation AGE 0.401 (0.03) -0.014 (0.02) 0.000 (0.02) 0.005 (0.02) 0.001 (0.21) EDUCATION 0.405 (0.29) 0.260 (0.22) 0.212 (0.17) 0.150 (0.17) 0.121(0.17) MARITAL STATUS -0.007 (0.68) -0.450 (0.44) -0.033 (0.37) -0.123 (0.14) -0.171 (0.37) LAND SIZE 0.003 (0.07) -0.039 (0.04) -0.016 (0.03) 0.099 (0.05)** -0.042 (0.33) AR PARTICIPATION -0.915 (0.59) 0.069(0.39) 1.711 (0.39) *** 0.690 (0.37)* 0.186 (0.35) INFOR SOURCE -0.196 (0.16) 0.246 (0.14)* 0.093 (0.11) -0.111 (0.12) 0.208 (0.13)* EXTENSION ACCESS -1.444 (0.56)*** 2.141 (0.64) *** -0.179 (0.53) -0.162 (0.54) 1.165 (0.67)* FAMER GROUP 0.189 (0.48) 1.399 (0.48) *** -0.004 (0.36) 0.094 (0.36) 0.717 (0.36)** PUBLIC SPEAK -0.875 (0.56) 0.487 (0.41) 0.699 (0.39) 1.077 (0.39)*** 0.832 (0.39)** CREDIT DECISION 0.542 (2.82) -1.510 (1.22) -1.086 (0.76) 0.833 (0.78) -1.042 (0.64) INCOME DECISION -0.970 (1.39) 1.918 (1.524) 0.385 (0.95) 0.123 (0.24) 0.607 (1.07) PROD DECISION 0.903 (2.70) 2.027 (1.44) * 0.119 (0.92) 0.001 (0.45) -0.498 (0.90) PLOT INDIVIDUAL -1.845 (0.63)*** 1.604 (0.47) *** 0.260 (0.40) 1.008 (0.47)*** 0.402 (0.41) LABOR SUFFICIENT 1.10 (0.97) -1.444 (0.64) -0.526 (0.59) -0.270 (0.42) -0.068 (0.44) Pseudo R2 0.52 0.50 0.35 0.29 0.27 P-value 0.000 0.002

Results from binary probit model for women participation in each stage (Decomposed model)   Explanatory variables STAGES Did not participate in any stage Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated technologies Monitoring and evaluation AGE 0.401 (0.03) -0.014 (0.02) 0.000 (0.02) 0.005 (0.02) 0.001 (0.21) EDUCATION 0.405 (0.29) 0.260 (0.22) 0.212 (0.17) 0.150 (0.17) 0.121(0.17) MARITAL STATUS -0.007 (0.68) -0.450 (0.44) -0.033 (0.37) -0.123 (0.14) -0.171 (0.37) LAND SIZE 0.003 (0.07) -0.039 (0.04) -0.016 (0.03) 0.099 (0.05)** -0.042 (0.33) AR PARTICIPATION -0.915 (0.59) 0.069(0.39) 1.711 (0.39) *** 0.690 (0.37)* 0.186 (0.35) INFOR SOURCE -0.196 (0.16) 0.246 (0.14)* 0.093 (0.11) -0.111 (0.12) 0.208 (0.13)* EXTENSION ACCESS -1.444 (0.56)*** 2.141 (0.64) *** -0.179 (0.53) -0.162 (0.54) 1.165 (0.67)* FAMER GROUP 0.189 (0.48) 1.399 (0.48) *** -0.004 (0.36) 0.094 (0.36) 0.717 (0.36)** PUBLIC SPEAK -0.875 (0.56) 0.487 (0.41) 0.699 (0.39) 1.077 (0.39)*** 0.832 (0.39)** CREDIT DECISION 0.542 (2.82) -1.510 (1.22) -1.086 (0.76) 0.833 (0.78) -1.042 (0.64) INCOME DECISION -0.970 (1.39) 1.918 (1.524) 0.385 (0.95) 0.123 (0.24) 0.607 (1.07) PROD DECISION 0.903 (2.70) 2.027 (1.44) * 0.119 (0.92) 0.001 (0.45) -0.498 (0.90) PLOT INDIVIDUAL -1.845 (0.63)*** 1.604 (0.47) *** 0.260 (0.40) 1.008 (0.47)*** 0.402 (0.41) LABOR SUFFICIENT 1.10 (0.97) -1.444 (0.64) -0.526 (0.59) -0.270 (0.42) -0.068 (0.44) Pseudo R2 0.52 0.50 0.35 0.29 0.27 P-value 0.000 0.002

Results from binary probit model for women participation in each stage (Decomposed model)   Explanatory variables STAGES Did not participate in any stage Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated technologies Monitoring and evaluation AGE 0.401 (0.03) -0.014 (0.02) 0.000 (0.02) 0.005 (0.02) 0.001 (0.21) EDUCATION 0.405 (0.29) 0.260 (0.22) 0.212 (0.17) 0.150 (0.17) 0.121(0.17) MARITAL STATUS -0.007 (0.68) -0.450 (0.44) -0.033 (0.37) -0.123 (0.14) -0.171 (0.37) LAND SIZE 0.003 (0.07) -0.039 (0.04) -0.016 (0.03) 0.099 (0.05)** -0.042 (0.33) AR PARTICIPATION -0.915 (0.59) 0.069(0.39) 1.711 (0.39) *** 0.690 (0.37)* 0.186 (0.35) INFOR SOURCE -0.196 (0.16) 0.246 (0.14)* 0.093 (0.11) -0.111 (0.12) 0.208 (0.13)* EXTENSION ACCESS -1.444 (0.56)*** 2.141 (0.64) *** -0.179 (0.53) -0.162 (0.54) 1.165 (0.67)* FAMER GROUP 0.189 (0.48) 1.399 (0.48) *** -0.004 (0.36) 0.094 (0.36) 0.717 (0.36)** PUBLIC SPEAK -0.875 (0.56) 0.487 (0.41) 0.699 (0.39) 1.077 (0.39)*** 0.832 (0.39)** CREDIT DECISION 0.542 (2.82) -1.510 (1.22) -1.086 (0.76) 0.833 (0.78) -1.042 (0.64) INCOME DECISION -0.970 (1.39) 1.918 (1.524) 0.385 (0.95) 0.123 (0.24) 0.607 (1.07) PROD DECISION 0.903 (2.70) 2.027 (1.44) * 0.119 (0.92) 0.001 (0.45) -0.498 (0.90) PLOT INDIVIDUAL -1.845 (0.63)*** 1.604 (0.47) *** 0.260 (0.40) 1.008 (0.47)*** 0.402 (0.41) LABOR SUFFICIENT 1.10 (0.97) -1.444 (0.64) -0.526 (0.59) -0.270 (0.42) -0.068 (0.44) Pseudo R2 0.52 0.50 0.35 0.29 0.27 P-value 0.000 0.002

Results from binary probit model for women participation in each stage (Decomposed model)   Explanatory variables STAGES Did not participate in any stage Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated technologies Monitoring and evaluation AGE 0.401 (0.03) -0.014 (0.02) 0.000 (0.02) 0.005 (0.02) 0.001 (0.21) EDUCATION 0.405 (0.29) 0.260 (0.22) 0.212 (0.17) 0.150 (0.17) 0.121(0.17) MARITAL STATUS -0.007 (0.68) -0.450 (0.44) -0.033 (0.37) -0.123 (0.14) -0.171 (0.37) LAND SIZE 0.003 (0.07) -0.039 (0.04) -0.016 (0.03) 0.099 (0.05)** -0.042 (0.33) AR PARTICIPATION -0.915 (0.59) 0.069(0.39) 1.711 (0.39) *** 0.690 (0.37)* 0.186 (0.35) INFOR SOURCE -0.196 (0.16) 0.246 (0.14)* 0.093 (0.11) -0.111 (0.12) 0.208 (0.13)* EXTENSION ACCESS -1.444 (0.56)*** 2.141 (0.64) *** -0.179 (0.53) -0.162 (0.54) 1.165 (0.67)* FAMER GROUP 0.189 (0.48) 1.399 (0.48) *** -0.004 (0.36) 0.094 (0.36) 0.717 (0.36)** PUBLIC SPEAK -0.875 (0.56) 0.487 (0.41) 0.699 (0.39) 1.077 (0.39)*** 0.832 (0.39)** CREDIT DECISION 0.542 (2.82) -1.510 (1.22) -1.086 (0.76) 0.833 (0.78) -1.042 (0.64) INCOME DECISION -0.970 (1.39) 1.918 (1.524) 0.385 (0.95) 0.123 (0.24) 0.607 (1.07) PROD DECISION 0.903 (2.70) 2.027 (1.44) * 0.119 (0.92) 0.001 (0.45) -0.498 (0.90) PLOT INDIVIDUAL -1.845 (0.63)*** 1.604 (0.47) *** 0.260 (0.40) 1.008 (0.47)*** 0.402 (0.41) LABOR SUFFICIENT 1.10 (0.97) -1.444 (0.64) -0.526 (0.59) -0.270 (0.42) -0.068 (0.44) Pseudo R2 0.52 0.50 0.35 0.29 0.27 P-value 0.000 0.002

Results from a multivariate probit model of women participation in each stage (Decomposed model)   Explanatory variables STAGES Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated technologies Monitoring and evaluation AGE -0.152 (0.02) 0.00 (0.02) 0.007 (0.19) 0.005 (0.02) EDUCATION 0.231 (0.22) 0.256 (0.17) 0.143 (0.17) 0.098 (0.17) MARITAL STAT -0.431 (0.43) -0.045 (0.37) -0.278 (0.37) LAND SIZE -0.037 (0.04) -0.013 (0.03) 0.102 (0.05)** -0.031 (0.03) AR PARTICIP 0.045 (0.39) 1.678 (0.381)*** 0.662 (0.36)* 0.083 (0.34) INFOR SOURCE 0.257 (0.14)* 0.059 (0.11) -0.147 (0.11) 0.144 (0.11) EXTEN ACCESS 2.109 (0.63)*** -0.176 (0.54) -0.286 (0.37) 1.016 (0.61)* FAMER GROUP 1.349 (0.48)*** 0.042 (0.36) 0.062 (0.35) 0.713 (0.34)** PUBLC SPEAK 0.471 (0.41) 0.656 (0.39) 1.22 (0.38)*** 0.763 (0.37)** CREDITDEC -1.431 (1.27) 0.742 (0.67) 0.525 (0.73) -1.113 (0.62) INCOME DEC 1.820 (1.51) 0.269 (0.93) 4.389 (0.181) 0.750 (1.28) PROD DEC 1.988 (1.17)* 0.129 (0.96) 3.981 (1.79) -0.177 (0.91) PLOT INDIV 1.577 (0.47)*** 0.267 (0.41) 1.045 (0.46)** 0.460 (0.39) LABORSUFF 1.325 (0.62) -0.659 (0.45) -0.236 (0.41) -0.022 (0.39) A comparison of the probit model results presented in Table 7 with those of the multivariate probit model that takes into account correlations between different stages (Table 8) shows that the significant variables are by and large the same for the two estimation procedures. The general similarity of the results across the two estimation procedures attest to the robustness of our results, which is further validated using the qualitative data.

Results from a multivariate probit model of women participation in each stage (Decomposed model)   Explanatory variables STAGES Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated technologies Monitoring and evaluation AGE -0.152 (0.02) 0.00 (0.02) 0.007 (0.19) 0.005 (0.02) EDUCATION 0.231 (0.22) 0.256 (0.17) 0.143 (0.17) 0.098 (0.17) MARITAL STAT -0.431 (0.43) -0.045 (0.37) -0.278 (0.37) LAND SIZE -0.037 (0.04) -0.013 (0.03) 0.102 (0.05)** -0.031 (0.03) AR PARTICIP 0.045 (0.39) 1.678 (0.381)*** 0.662 (0.36)* 0.083 (0.34) INFOR SOURCE 0.257 (0.14)* 0.059 (0.11) -0.147 (0.11) 0.144 (0.11) EXTEN ACCESS 2.109 (0.63)*** -0.176 (0.54) -0.286 (0.37) 1.016 (0.61)* FAMER GROUP 1.349 (0.48)*** 0.042 (0.36) 0.062 (0.35) 0.713 (0.34)** PUBLC SPEAK 0.471 (0.41) 0.656 (0.39) 1.22 (0.38)*** 0.763 (0.37)** CREDITDEC -1.431 (1.27) 0.742 (0.67) 0.525 (0.73) -1.113 (0.62) INCOME DEC 1.820 (1.51) 0.269 (0.93) 4.389 (0.181) 0.750 (1.28) PROD DEC 1.988 (1.17)* 0.129 (0.96) 3.981 (1.79) -0.177 (0.91) PLOT INDIV 1.577 (0.47)*** 0.267 (0.41) 1.045 (0.46)** 0.460 (0.39) LABORSUFF 1.325 (0.62) -0.659 (0.45) -0.236 (0.41) -0.022 (0.39) A comparison of the probit model results presented in Table 7 with those of the multivariate probit model that takes into account correlations between different stages (Table 8) shows that the significant variables are by and large the same for the two estimation procedures. The general similarity of the results across the two estimation procedures attest to the robustness of our results, which is further validated using the qualitative data.

Results from a multivariate probit model of women participation in each stage (Decomposed model)   Explanatory variables STAGES Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated technologies Monitoring and evaluation AGE -0.152 (0.02) 0.00 (0.02) 0.007 (0.19) 0.005 (0.02) EDUCATION 0.231 (0.22) 0.256 (0.17) 0.143 (0.17) 0.098 (0.17) MARITAL STAT -0.431 (0.43) -0.045 (0.37) -0.278 (0.37) LAND SIZE -0.037 (0.04) -0.013 (0.03) 0.102 (0.05)** -0.031 (0.03) AR PARTICIP 0.045 (0.39) 1.678 (0.381)*** 0.662 (0.36)* 0.083 (0.34) INFOR SOURCE 0.257 (0.14)* 0.059 (0.11) -0.147 (0.11) 0.144 (0.11) EXTEN ACCESS 2.109 (0.63)*** -0.176 (0.54) -0.286 (0.37) 1.016 (0.61)* FAMER GROUP 1.349 (0.48)*** 0.042 (0.36) 0.062 (0.35) 0.713 (0.34)** PUBLC SPEAK 0.471 (0.41) 0.656 (0.39) 1.22 (0.38)*** 0.763 (0.37)** CREDITDEC -1.431 (1.27) 0.742 (0.67) 0.525 (0.73) -1.113 (0.62) INCOME DEC 1.820 (1.51) 0.269 (0.93) 4.389 (0.181) 0.750 (1.28) PROD DEC 1.988 (1.17)* 0.129 (0.96) 3.981 (1.79) -0.177 (0.91) PLOT INDIV 1.577 (0.47)*** 0.267 (0.41) 1.045 (0.46)** 0.460 (0.39) LABORSUFF 1.325 (0.62) -0.659 (0.45) -0.236 (0.41) -0.022 (0.39) A comparison of the probit model results presented in Table 7 with those of the multivariate probit model that takes into account correlations between different stages (Table 8) shows that the significant variables are by and large the same for the two estimation procedures. The general similarity of the results across the two estimation procedures attest to the robustness of our results, which is further validated using the qualitative data.

Results from a multivariate probit model of women participation in each stage (Decomposed model)   Explanatory variables STAGES Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated technologies Monitoring and evaluation AGE -0.152 (0.02) 0.00 (0.02) 0.007 (0.19) 0.005 (0.02) EDUCATION 0.231 (0.22) 0.256 (0.17) 0.143 (0.17) 0.098 (0.17) MARITAL STAT -0.431 (0.43) -0.045 (0.37) -0.278 (0.37) LAND SIZE -0.037 (0.04) -0.013 (0.03) 0.102 (0.05)** -0.031 (0.03) AR PARTICIP 0.045 (0.39) 1.678 (0.381)*** 0.662 (0.36)* 0.083 (0.34) INFOR SOURCE 0.257 (0.14)* 0.059 (0.11) -0.147 (0.11) 0.144 (0.11) EXTEN ACCESS 2.109 (0.63)*** -0.176 (0.54) -0.286 (0.37) 1.016 (0.61)* FAMER GROUP 1.349 (0.48)*** 0.042 (0.36) 0.062 (0.35) 0.713 (0.34)** PUBLC SPEAK 0.471 (0.41) 0.656 (0.39) 1.22 (0.38)*** 0.763 (0.37)** CREDITDEC -1.431 (1.27) 0.742 (0.67) 0.525 (0.73) -1.113 (0.62) INCOME DEC 1.820 (1.51) 0.269 (0.93) 4.389 (0.181) 0.750 (1.28) PROD DEC 1.988 (1.17)* 0.129 (0.96) 3.981 (1.79) -0.177 (0.91) PLOT INDIV 1.577 (0.47)*** 0.267 (0.41) 1.045 (0.46)** 0.460 (0.39) LABORSUFF 1.325 (0.62) -0.659 (0.45) -0.236 (0.41) -0.022 (0.39) A comparison of the probit model results presented in Table 7 with those of the multivariate probit model that takes into account correlations between different stages (Table 8) shows that the significant variables are by and large the same for the two estimation procedures. The general similarity of the results across the two estimation procedures attest to the robustness of our results, which is further validated using the qualitative data.

Results from binary probit models with a composite empowerment index (Composite model)   Explanatory variables STAGES Did not participate in any stage Identification and prioritization of research and development problems Identification and testing of potential technology options Dissemination of tested and validated Monitoring and evaluation AGE 0.016 (0.01) 0.015 (0.01) -0.004 (0.01) -0.005 (0.01) -0.015 (0.01) EDUCATION -0.093 (0.12) 0.108 (0.10) 0.199 (0.10) * 0.057 (0.10) 0.147 (0.11) MARITAL STATUS 0.525 (0.26) -0.499 (0.20) -0.327 (0.21) -0.293 (0.21) 0.162 (0.23) LAND SIZE 0.014 (0.01) -0.01 (0.01) 0.020 (0.01) * 0.012 (0.13) -0.035 (0.02) AR PARTICIPATION -0.617 (0.23) *** 0.405 (0.19) ** 0.786 (0.19) *** 0.394 (0.20) * 0.403 (0.22) * INFOR SOURCE -0.122 (0.08) -0.04 (0.07) 0.112 (0.67) * -0.059 (0.07) 0.119 (0.08) *** EXTENSION ACCESS -1.033 (0.25) *** 0.859 (0.27) *** 0.614 (0.28) ** 0.496 (0.28) * 0.426 (0.29) EMPOWERINDEX -0.327 (0.09) *** 0.350 (0.07) *** 0.204 (0.07) *** 0.330 (0.07)*** 0.312 (0.07) *** Pseudo R2 0.28 0.23 0.20 0.16 0.17 P-value 0.000 Women empowerment is positively and significantly (p<0.01) associated with women participation in all four stages. Consistent with the decomposed model results (Table 7 & 8) which showed that access to extension and several empowerment indicator variables significantly increase women participation in problem identification and prioritization of research problems, the composite model results also confirms that women empowerment and access to extension enhance women participation in this stage. While the decomposed model results show that participation in the Africa RISING project is the only significant factor influencing participation in the technology identification and testing stage, the composite model results show that several other factors, including education, land size, number of information sources, access to extension and the level of empowerment affect women participation in this stage. Like the other two earlier stages, access to extension, level of empowerment and participation in Africa RISING project are key determinants of women participation in the dissemination of tested and validated technologies. For the monitoring and evaluation stage, participation in Africa RISING project, number of information sources and empowerment enhance women participation in this stage.

Qualitative vs quantitative data Production decision making Lack of decision making power over productive resources Land size and location of land Female household heads with higher chances Household head regardless of gender enjoys most of the decision making power Women’s decision making ability is very key for sustainable intensification Farmers with larger farm sizes are more flexible in implementing environmentally friendly technologies e.g. crop rotation, crop diversification and agro-forestry

Qualitative vs quantitative data Leadership Membership to farmer based groups (design and M&E) Ability to speak up ( diffusion and M&E) > 50% of surveyed women not members of farmer-based groups Selection of household head, active and innovative women, women who are active in community politics Africa RISING approach which was less discriminatory – organize research participats irrespective of their membership to grous and organized them into farmer research groups based on their technological preferences Membership of farmer research groups lowers transaction costs, provides farmers with an array of services, build trust among members

Qualitative vs quantitative data Time allocation Labor sufficiency not sig in quant Women’s workload frequently cited in FGDs “Women farmers are limited to participate in the different extension events and trainings because they are too busy with reproductive work. They do not have enough time to participate in different trainings and farmer field days as compared to men (women’s FGD, non-participants, Ilu-Sambitu, Sinana)”.

Other socio-economic factors Information & knowledge Mostly cited by women Ranked amongst the top 5 influencing factors Access to extension agent Key in early and later stages Cultural norms Cultural norms Slightly mentioned more by men Men considered farmers and women as helpers Men are more knowledgeable and capable Define technologies that are appropriate for women

Conclusion Empowerment increases women’s participation in agricultural research processes Different domains and indicators influence their participation in each of the stages (early or later) Addressing empowerment factors that influence women’s participation in earlier stages is Participation in earlier stages requires more empowerment aspects which prepare them for later stages Any technology added to the system should fit into the social domain – women’s empowerment as a key social indicator

We would like to acknowledge all CGIAR Research Programs and Centers for supporting the participation of their gender scientists to the Seeds of Change conference. Photo: Neil Palmer/IWMI