The spaces and places of food security: learning from spatial, hierarchical, and econometric models in urban data-poor areas.

In data poor areas, the use of statistical models is often determined by the quantity and quality of the data.. Here, we explore the pros and cons of three model outcomes, which allow us to evaluate the range of predictions and how they would significantly influence our research conclusions. Using food security survey data for Accra, Ghana collected in 2003, we examine the information derived from spatial, hierarchical, and econometric models respectively. While the data source is the same, the outcomes are different, highlighting the caution researchers must use when determining an appropriate statistical approach. The spatial model delivered vital information on the geographic distribution of food security across the urban landscape, highlight areas of particular concern “hotspots” with statistically significant values. Our use of the hierarchical, or multi-level, model separated the effects of household versus neighborhood variables, allowing us to distinguish the level at which variables were most influential. Lastly, our econometric model emphasized the economic trends among household based on estimated values of household wealth. Together, these three models allow us to draw a more complete picture of food security patterns in Accra, and to draw important and more comprehensive conclusions for policy recommendations.