| Description |
The thesis work developed and validated a system for scoring the quality of household grocery purchases, the Grocery Purchase Quality Index-2016 (GPQI-2016). A grocery sales data set (2012-13) without individual household shopper attributes was provided by a national grocery chain, and a sample of 4,000 households in each of four geographic locations was drawn. The 1,887 categories, known as "subcommodities," in the database were classified into the 29 food categories of the US Department of Agriculture's Food Plan market baskets. A standardized expenditure share for each category was calculated. Quality of food purchases was evaluated by comparing the observed to the standardized USDA expenditure share for each category and then grouping the categories into 11 components, based on the Healthy Eating Index-2010 (HEI-2010). Processed meat expenditures were also assessed. Households that never purchased tobacco (n=12,460) had higher (6%) median total quality scores (30.7 of a possible 75 points) than those who did purchase tobacco (n=3,540, median score 26.3, p<0.01), as well as higher scores for each of the 11 components (p<0.01). Tobacco users typically have poorer diets than nonusers; therefore, construct validity of the GPQI-2016 in assessing grocery food purchases was established. The scoring design was further applied to evaluate food purchases from another data set, the USDA's 2012 Food Acquisition and Purchase Survey (FoodAPS), and to conduct a correlation analysis with results using the HEI-2010 reference standard. A discussion of consumer health informatics application use cases concludes the thesis. |
