Explaining food insecurity among indigenous households of the Sierra Tarahumara in the Mexican state of Chihuahua

• Main Authors: Otilia-Vanessa Cordero-Ahiman, Eduardo Santellano-Estrada, Alberto Garrido
• Format: Article
• Language: English
• Published: Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria 2017-04-01
• Series: Spanish Journal of Agricultural Research
• Subjects: ELCSA; measurement; poverty; human health; logit model
• Online Access: http://revistas.inia.es/index.php/sjar/article/view/10151

Numerous studies have analyzed the factors that determine food security and explored the problem from regional or national points of view. However, there has been less research targeting an understanding of the food security problem at the household level in specific rural locations like indigenous communities. Indigenous groups are recognized as priority groups in Mexico, because they live in a situation of poverty. For this reason, the objective of this research was to investigate the determinants of food insecurity among the indigenous communities of the Sierra Tarahumara in Mexico. We used the Latin American and Caribbean Household Food Security Measurement Scale (ELCSA). This scale is useful for measuring food insecurity levels in households. A questionnaire was administered to 123 households. We employed the method based on Cronbach's alpha to measure internal consistency, which was 0.96. In addition, we estimated the main determinants of household food insecurity using both ordered logit model and binomial logit model. We found that approximately 59.35% of households were living in a situation of severe food insecurity. The two predictive models applied suggest that: i) income is the most important determinant of access to food; ii) increased maize production improves food security; iii) farmers consume their seed stocks in times of food scarcity, and iv) households are food insecure when the householders are in casual employment. Akaike's information criterion and the Bayesian information criterion suggest that the goodness of fit to the data was better for the ordered logit model.