Validity of the Food Insecurity Experience Scale (FIES) for use in Sub‐Saharan Africa (SSA)

Abstract

BackgroundThe Food Insecurity Experience Scale (FIES), an 8‐item questionnaire is the latest experience‐based metric of severity of food insecurity recently developed by the United Nations Food and Agricultural Organization Voices of the Hungry project (FAO‐VOH), thus its latent dimensional structure and psychometric properties haven't been fully explored for SSA. The objectives of this study were to assess the psychometric properties of FIES for measuring food security in SSA using the Rasch measurement model and to examine the prevalence of food insecurity (FI) among adults ages 19 and older.MethodsTo assess the intended validity of the FIES, the Rasch modeling procedure was applied to the SSA survey data collected in 2014 and 2015 by the Gallup World Poll (GWP) and FAO‐VOH. Data were provided by 58, 324 respondents aged 19 years and above. All respondents with missing food security information on any of the FIES questions were omitted from this analysis. SAS (version 9.3) and R (version 3.1.3) were used to perform the analysis.ResultsUsing the derived FI thresholds for SSA, the prevalence of FI (including moderate and severe FI) was 57%. There was higher prevalence of FI among older adults ≥ 50 years than younger adults 19–49 years (59.1% and 56.5% respectively, p<0.0001), and among women than men (58.4% and 55.6% respectively, p<0.0001). The severity of the items at country level estimated under the Rasch‐model assumptions ranged from 1.7 logistic units in Mauritania to 4.63 logistic units in Namibia. There was good reliability of the scale for all countries (≥ 0.7). Most of the item‐infit statistics were <1.4, except for the item “worrying about not having enough food to eat at any time in the past 12 months due to lack of money or other resources” in Namibia and Sudan. Additionally, several countries in the Western region and Islands showed high item‐outfit (≥ 1.8) for the item “going without eating for a whole day”. This was also true in the responses by older adults. Four countries showed significant correlations between residuals for the questions “being hungry but not eating” and “running out of food”. Finally, there were differences in the order of severity for most of the items of SSA countries compared to the order seen in the combined SSA metric. However, there were no differences in the order of item severity between older and younger adults.ConclusionsFIES demonstrated acceptable levels of internal validity, although the question “not eating for whole day” may need to be cognitively tested in a few countries. The similarity in response patterns by younger and older adults suggests that food security is experienced similarly in the two age groups, thus the overall prevalence of FI in these two age groups can be directly compared. Because the response patterns were different by country level, it is recommended that a SSA standard excluding the correlated items be used and a strict significance level be employed in cross‐country FI prevalence comparisons.Support or Funding InformationNone