S. Agarwal, V. Sethi, Palak Gupta
Aug 4, 2009
One-third of India’s urban population resides in extreme poverty, in slums and squatters. Food insecurity remains a visible reality among this segment. Yet, it is scarcely documented. This paper describes levels and determinants of experiential household food insecurity (HFI) in an underserved urban slum of Delhi (India) and reports the internal validity and reliability of the measure used to assess experiential HFI. A four-item scale was adapted from the U.S. six-item short-form food security scale and was administered in Hindi through household interviews with 410 female adults. Association of HFI with household economic and socio-demographic characteristics were examined using multiple logistic regression. Cronbach’s alpha and Rasch-model-based item fit statistics were used to assess reliability and internal validity. Fifty-one percent of households were food insecure. Significant HFI predictors were unemployed to employed family members’ ratio of > 3:1 (Odds Ratio 2.1, Confidence Interval 1.2 – 3.4) and low household standard of living (OR 4.9, C.I. 2.7 – 8.9). Cronbach’s alpha was 0.8. Item severities as estimated under Rasch model assumptions spanned 9.7 logits. Item infit statistics (0.77 – 1.07) indicated that the Rasch model fit the data well. Item outfit statistics suggested that one item was inconsistently understood by a small proportion of respondents. For improving HFI among the urban poor, in addition to improving behaviors/entitlement access, programs should consider linkage of urban poor to existing employment schemes, upgrading of their skills and linkage to potential employers. The adapted scale was reliable and easy to administer. However, being a subjective assessment, its sensitivity to social expectation and its association with nutrition security require examination.