Debunking the one-size-fits-all approach: Synergistic and trade-off effects of collective action on household food security among the smallholder farmers in central Kenya
Description
The dataset was used to evaluate the effect of collective action on household food security (HFS). It is cross-sectional data randomly and proportionately drawn from 532 smallholder collection action participants and non-participants in Murang’a County, Kenya. A well-structured questionnaire was employed to draw the data from the respondents from the month of August to October 2022. We sampled 289 members of livelihood CAIs, 64 members of efficiency CAIs and 179 non-members from the neighbouring households as a control. The data covered the production year of 2021. The variables collected included the household demographics, farm and farming factors, institutional factors and the characteristics of the collective action initiatives (CAIs). In particular, the study evaluated the effects of efficiency and livelihood CAIs on household food security across the production and income pathways. Production indicator included the quantity of maize and beans (major staple food in the study area) produced in a year. The income indicators included food expenditure and household income. The HFS was measured using Food Consumption Score (FCS) which was calculated by multiplying the frequency of foods consumed in the last seven days (before the survey) with the weights of each food group for staples, pulses, vegetables, fruits, meat/egg/fish, sugar and oil. Consumption scores of 0 to 21 were stratified as chronic food insecurity, 21.5 to 35 were classified as transitory food insecurity and more than 35 were categorized as food secure. The study was subjected to a probit model to provide a real account of the effect of interactions between the two forms of CAIs and the household socio-economic attributes on HFS. In addition, the multinomial endogenous treatment effects model was also employed to allow us to disaggregate how efficiency and livelihood CAIs influence HFS outcomes, wherein non-members served as a reference category. The approach controls for the unobservable factors that may influence HFS and the endogeneity problems that may arise due to reverse causality between collective action participation and HFS outcomes. Before the analysis, the dataset was subjected to pre-tests including the multicollinearity test and the Hausman’s test to check for the independence of Irrelevant Alternative Assumption (IIA). The data was analysed in Stata 17 (see the accompanying do file herein). The study found that the effects of HFS were context-specific depending on the socioeconomic characteristics of the households and the organizational characteristics of the CAIs.
Files
Steps to reproduce
The data constituted 532 smallholder households from Murang’a County in central Kenya. The county was selected due to its prominence in agricultural collective action initiatives (CAIs) and high prevalence rates of food insecurity particularly staples i.e., maize and beans. The sample size was computed following the normal binomial distribution approximation approach. A multistage stratified and random proportionate sampling techniques were employed to draw the sample. In the first stage, two sub counties namely Kandara and Gatanga were selected following their dominance in the CAIs. In the second stage, five wards including two wards from Gatanga sub-county and five wards from Kandara sub-county were selected based on a proportionate random sampling procedure. In the third stage, with the aid of the collective action lists generated by the agricultural officers at the ward level, we sampled 289 members of livelihood CAIs, 64 members of efficiency CAIs and 179 non-members from the neighbouring households as a control. A well-structured and pre-tested questionnaire was employed by trained enumerators to draw cross-sectional data from the respondents from the month of August to October 2022. Detailed information was collected from the respondents covering the year 2021-2022. The data included household demographics, farm and farming factors, infrastructural and institutional factors and the organizational characteristics of the CAIs. To test for the accuracy of the tool, pre-testing was done by administering about 20 questionnaires to the randomly sampled households. The split-half method was employed to test for the validity of the questionnaire. The correlation coefficient of the responses was estimated by the Pearson Product linear correlation method. The correlation coefficient approaching one implies stronger reliability and otherwise. In ensuring the validity of the tool, the items found to be inadequate and ambiguous were correctly rephrased to avoid misinterpretation by the respondents. Before data analysis, data was cleaned for consistency. Data analysis involved descriptive and inferential statistics to explore the patterns of the data and to test for the appropriateness of the data for econometric analysis. Before the analysis, the dataset was subjected to pre-tests including the multicollinearity test using the variance Inflating Factor (VIF) and the Hausman test to check for the independence of Irrelevant Alternative Assumption (IIA). The data was analysed in Stata 17 using the ordered probit model and multinomial endogenous treatment effects model.