Farm production diversity and its influence on diet quality in South Eastern Kenya

Abstract

Agriculture nutrition linkages have long been recognized as a potentially viable pathway of attaining food and nutrition security. However, these insights have started to influence mainstream thinking on agricultural development only recently and the empirical evidence in sub-Saharan Africa is non-conclusive. We evaluate the influence of farm production diversity on diet diversity in two semi-arid Counties of South Eastern Kenya by employing a sample of 830 smallholder farmers selected using a three-stage sampling procedure and the data were analyzed using a Poisson regression model. On the average, households consumed food from seven food groups out of the 12 possible food groups. We show that farm production diversity has a positive and significant influence on dietary diversity, a proxy for diet quality. Other interventions that can improve dietary diversity include commercialization and wealth creation. Thus, pro-farm diversification interventions are recommended as potential strategies for improving the dietary diversity of rural farming households. Moreover, improving market infrastructure to enhance commercialization and supporting wealth creation through savings and asset accumulation can contribute towards more diversified diets.

Keywords:

• Production diversity
• diet quality
• market access
• Poisson regression
• Nutrition security

REVIEWING EDITOR:

• M. Luisa Escudero-Gilete
• Universidad de Sevilla
• Spain

SUBJECTS:

• Development Studies
• Rural Development
• Economics and Development

Ethical approval

Since this study used a survey to collect cross-sectional data of farmers’ socio-economic characteristics, such as their farm production diversity and diet diversity, major ethical concerns do not arise as would be the case with randomized control trials, where for research purposes, the control group may be denied treatment that they actually need. However, we ensured to eliminate any further ethical concerns in three ways. First, before the start of the interview, enumerators briefed the farmers including mentioning that the data collected was for academic purposes and would not be used for commercial gains. Farmers were also briefed on some of the key variables of interest and that the data analysis would be anonymized. After the brief, the farmers were asked whether they were willing to proceed with the interview. Two options were possible, if a farmer was willing, she or he was interviewed, otherwise, the enumerator terminated the interview and proceeded to the next farmer on the list. All the farmers in the sample agreed to be interviewed. Secondly, the preliminary findings of the study were shared in a dissemination workshop that included a group of randomly selected farmers from among those interviewed. The workshop was organized by the funder of the study, the African Economic Research Consortium (AERC).

Author contributions

Jonathan Makau Nzuma conceptualized the study, designed the survey, developed the questionnaire, supervised data collection, analyzed and interpreted the data, and wrote the paper. Dasel Wambua Mulwa Kaindi digitized the questionnaire using kobotoolbox, supervised data collection, and computed the model variables. Henry Muli Mwololo designed the survey tool, analyzed the data, and wrote the paper.

Disclosure statement

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

Data availability statement

Data is available from the authors on request.

Notes

1 https://ieg.worldbankgroup.org/sites/default/files/Data/reports/ppar_kenyaagriculture.pdf. Accessed on 8th November 2022.

2 https://projects.worldbank.org/en/projects-operations/project-detail/P109683. Accessed on 8th November 2022.

3 Inclusion of the two counties (Machakos and Makueni) was not for comparison purposes per se but rather to enhance the external validity from a geographic scope point of view since the study uses cross section data that suffer from weak external validity.

4 The average VIF was 1.19 and ranged from 1.05 to 1.34 for all the variables whereas the partial correlation coefficients were in the range of 0.01–0.14 thus the model did not exhibit multicollinearity.

5 All the models returned low McFadden’s Pseudo R² values ranging between 1.1 and 5.9%. Although a higher value of the McFadden’s Pseudo R² is desirable, it is not uncommon for nonlinear models to return such ratios. Chaudhry et al. (2020) provide a plausible explanation that the McFadden’s Pseudo R² is not an equivalent of the standard R² found in linear models, i.e. the proportion of variance in the dependent variable that can be explained by the independent variables included in the model. Therefore, the McFadden’s Pseudo R² should not be interpreted directly to imply the model’s goodness of fit.

Additional information

Funding

This study was supported by the African Economics Research Consortium (AERC) under Grant No. AE/FAC/19-002 (ERP Award 1080).

Notes on contributors

Jonathan Makau Nzuma

Jonathan Makau Nzuma is an Associate Professor at the Department of Agricultural Economics, University of Nairobi, Kenya.

Dasel Wambua Mulwa Kaindi

Dasel Wambua Mulwa Kaindi is a Senior Lecturer at the Department of Food Science, Nutrition and Technology, University of Nairobi, Kenya.

Henry Muli Mwololo

Henry Muli Mwololo is a Research Associate at the Department of Agricultural Economics, University of Nairobi, Kenya.