Direct and Comprehensive Measure of Child Food Security: Reliability, Accuracy, and Concordance With Parental Report

Year: 2011

Research Center: Southern Rural Development Center, Mississippi State University

Investigator: Fram, Maryah, and Edward Frongillo

Institution: University of South Carolina

Project Contact:
Maryah Fram
College of Social Work
University of South Carolina
Columbia, SC 29208
Phone: 803-777-5945


Household food insecurity is associated with deficits in children’s physical, psycho-social, and educational development. These associations are evident not only in the small proportion of households in which parents report that children do not always have enough to eat, but among food-insecure households more generally. This suggests that the mechanisms through which children are harmed by household food insecurity extend beyond child hunger—or at least beyond child hunger of which parents are aware. This idea is supported by recent qualitative research revealing that children have worries, stresses, fears, discomforts, and food management strategies of their own. Child experiences fall into two general components and six distinct domains:

  1. awareness of household food insecurity (involving cognitive, emotional and physical awareness)
  2. taking responsibility for managing household food resources (involving participation in parental food management strategies, initiation of child’s own strategies, and generation of resources to contribute to household food resources).

These child experiences are distinct from parental experiences, reflecting children’s different household roles and status, their developmental level, and the unique opportunities and constraints that children have in observing, understanding, and responding to family processes and problems.

Despite the uniqueness of children’s food-related experiences, most research about child food insecurity has relied on parent reporting on the household food environment to establish whether or not the child experiences low or very low food security. This is problematic for two main reasons. First, because parents and children experience food insecurity in different ways, a parent perspective on what happens in the household is not necessarily inclusive of the most salient aspects of what the child feels, sees, thinks, and does. Second, substantial research on other topics indicates that children, in general, are in the best position to report accurately and reliably on their own experiences. As much as a parent may want and try to understand what a child is experiencing, a parent cannot have complete information about the child’s internal processes. For these reasons, reliance on parental report alone may be a barrier to accurate assessment of childhood food insecurity and hunger.

To investigate this barrier, the present study developed, refined, and validated a new survey tool designed to accurately assess child food insecurity using child self-report. An initial set of survey items was developed based on prior research; these items addressed all six domains of childhood food insecurity. These items were refined through cognitive interviews with a sample of 24 diverse children in South Carolina. The revised items were then field-tested with a sample of 100 South Carolina children between the ages of 6 and 17 who were at high risk of food insecurity. Subsequent to completion of the survey, sample children were interviewed in depth about their experiences of the household food environment. Parents of sampled children also participated in the study, completing the full 18-item USDA Household Food Security Survey Module (HFSSM) and responding to several demographic questions and questions about children’s health and development. Analyses of the in-depth interviews were conducted to definitively classify children as experiencing or not experiencing each of the six domains of childhood food insecurity. Both child and parent survey responses were then compared to the definitive classification to assess the sensitivity and specificity of child and parent reports. Finally, child and parent responses on substantively similar items were compared and level of agreement was calculated.

The coding of in-depth interview data demonstrated that the six dimensions of childhood food insecurity were prevalent in the field test sample, ranging from 0.29 (resource generation) to 0.66 (cognitive awareness). The comparison of child items to the definite classification yielded child-report sensitivities ranging from 0.70 (resource generation) to 0.91 (emotional awareness). Specificities ranged from 0.53 (participation) to 0.83 (physical awareness). The area under the receiver operating characteristic (ROC) curve that quantifies jointly sensitivity and specificity was good for four of the six dimensions (cognitive, emotional, physical, and initiation). The comparison of parent items to the definitive classification yield parent-report sensitivities of 0.59 (cognitive awareness) and 0.35 (physical awareness) and specificities of 0.58 and 0.94, respectively, resulting in poor ROC curve areas; because the USDA HFSSM does not include items on other domains of children’s experiences, only cognitive and physical awareness items were substantively similar enough to support comparison. Of the items for which direct parent-item to child-item comparisons could be made, even the best pairing had only 63.5 percent agreement, with a Cohen’s kappa coefficient of 0.25, which is poor.

Taken together, these findings demonstrate that children are accurate reporters of their own food-security experiences related to cognitive, emotional, and physical awareness and to the initiation of strategies to help manage household food resources. Parents on average are not good reporters, and tend to under-report children’s hardships. This fits well with existing research suggesting that the “gold standard” for assessing child experiences is child self-report.

Overall, this study suggests that child self-report is the appropriate assessment approach when the purpose is to understand what children experience rather than what is happening in the household in general, or what adults in the household experience. Additional research is needed to apply the new child self-report measure with a larger and more representative sample. Such research would lead to more accurate and complete understanding of the nature, extent, and distribution of children’s individual experiences of food insecurity.