INTERNAL VALIDITY AND RELIABILITY OF EXPERIENCE-BASED HOUSEHOLD FOOD SECURITY SCALES: EVIDENCE FROM INDIA

At the household-level, food insecurity refers to the inability of a household to acquire or consume adequate quantity and quality of food as a result of inadequate food supplies or lack of resources. As the severity of food insecurity increases, steps taken to cope with it also become intense (e.g., progressing from consuming a less variety of foods to decreasing the portion size of the meals to reducing the number of meals consumed in a day)

Household food insecurity (HFI) is one of the key determinants of chronic undernutrition in Indian children, particularly among children living in resource-poor settings. However, we do not understand the extent of and severity of food insecurity experiences in Indian households. To date, none of India’s health and nutrition surveys include globally-recommended indicators for measuring food insecurity. The National Sample Survey Organization (NSSO) survey is the only national survey that includes a measure of households’ access to foods. This single question is inadequate to capture the frequency and intensity of food insecurity and is not a globally recommended measure. The National Family Health Survey (NFHS) on the other hand, only includes indicators on diet diversity and neglects food security.

There are four globally validated experience-based HFI questionnaires, all at different scales: 1) USDA 18-item combined adult-child scale; 2) USDA six-item adult scale; 3) FANTA nine-item Household Food Insecurity Access Scale (HFIAS); and 4) eight-item Food Insecurity Experience Scale (FIES). It is, however, unclear which one of these might be the most relevant scale for the Indian context. With this in mind, UNICEF and FAO-India led a review of these scales in collaboration with seven researchers from different organizations, including IFPRI, with experience on use of experience-based HFI scales in Indian settings. Between June and August, the authors reviewed 19 published and unpublished studies using experience-based HFI (access) scales, which were conducted in India between 2000 and 2015. As part of this exercise the authors gathered field experiences of 31 researchers who used the HFI scales and conducted psychometric tests conducted where raw data were available, to examine the internal validity and reliability of scales used in these studies.

Preliminary highlights of the paper:

1. HFI experiential scales used in India are internally reliable: Internal reliability across scales ranged between 0.75-0.94; however certain items (‘balanced meal’, ‘preferred food’, ‘worried food would run out’) had poor in-fit and out-fit statistics, indicating the need for robust pretesting and contextualizing.
2. HFI experiential scales used in the Indian context have weak internal validity. Across the HFI scales, some items compared well while others did not. Items under the domain of being worried or anxious about getting enough food, perceptions of the quality or quantiy of the food eaten, and reduced food intake were comparable across the scales. However, comparison for items on reduced food intake for children in the household was not possible.
3. Different experience-based HFI scales are being used in India, limiting comparability: Of the 19 studies reviewed, three studies used the 18-item USDA scale, 3 studies used the eight-child items only of the USDA scale, five studies used the nine-item HFIAS scale, six studies used the six-item short-form scale and two studies used the eight- item FIES. HFI prevalence varied depending on the type of scale used.
4. To ensure comparability of findings across multiple studies, it is important to identify and use a standard scale that is contextually relevant for India. An Indian adaptation of the FIES, should involve testing for validity across Indian states in small-scale studies before inclusion in national demographic surveys, such as NFHS. The inclusion of FIES scale in nationally representative comprehensive micronutrient survey of Ministry of Health and Family Welfare, with support of UNICEF is to start in the beginning of 2016. This is a welcome move to test its validity and reliability across Indian states.

It is timely for India to open an evidence-based policy dialogue to revisit the current measures of household food insecurity experiences in India’s national surveys, make efforts for inclusion of appropriate, reliable and valid measures of household food insecurity in Indian surveys and to invest in harmonizing the measures across multiple surveys, to aid comparability over time, to enhance understanding of the impact of food insecurity on families and children, and to support decision making.

_________________________________________________________________________________________________________

Authors: Chandana Maitra¹, Vani Sethi², Sayeed Unisa³, Rasmi Avula⁴, Bhaskar Goswami⁵, Kevin Gallagher⁵ and Saba Mebrahtu²

¹School of Economics University of Queensland, Brisbane, Australia, ²United Nations Children Fund, New Delhi, India, ³Department of Mathematical Demography and Statistics, Indian Institute of Population Sciences, Mumbai, India, ⁴International Food Policy Research Institute, New Delhi, India and ⁵Food and Agricultural Organization of the United Nations, New Delhi, India

For any queries on this paper or for the full paper, please contact: Dr. Vani Sethi, Nutrition Specialist, Child Development and Nutrition Section, UNICEF India Country Office, 73 Lodhi Estate, New Delhi. vsethi@unicef.org
_________________________________________________________________________________________________________