Year of Publication


Degree Name

Doctor of Philosophy (PhD)

Document Type

Doctoral Dissertation


Arts and Sciences



First Advisor

Dr. Carrie Oser


Mounting evidence of the deleterious health effects of poor diet, obesity, and correlated conditions underscore the need to understand how social factors influence food choices. A variety of factors contribute to the diets that Americans consume including limited time, limited income, lack of cooking skills, food deserts, and cheap, convenient foods in abundant portions in grocery stores and restaurants. These contextual factors serve as constraints that impact an individual’s ability to prioritize health when shopping for food. Using the three paper dissertation format, this project will utilize a Constrained Choice Theory (CCT) framework to investigate sociodemographic trends in priorities in shopping for food, the prevalence of shared priorities within social networks, and the outcomes of those priorities as measured by the nutritional quality of grocery purchases.

Constrained Choice Theory provides a useful framework for examining health behavior. It conceptualizes individual health decisions as a product of social forces acting upon individual priorities. These forces exist on three levels: social policy, community actions, and work/family. This paper will focus in particular in on the community action level by examining how network and group membership impacts priorities, and the work/family level by investigating how factors like requiring convenient foods, being married, or having children impact food choices. This research represents a new application of CCT as grocery priorities have not been previously examined under this framework.

Data are derived from a spring 2011 survey of 410 Lexington, KY families with children. The 20 minute, IRB approved paper survey was distributed via area churches, daycares, and community organizations in socioeconomically diverse neighborhoods. Respondents were directed to return the survey via mail or to a locked box at the location from which they obtained it. Though responses initially included a large number of high socioeconomic status families, results were balanced by recruiting additional participants from low income areas. Participants were offered their choice of a check for $15 or a $15 donation to the location where they obtained the survey.

Paper one identifies and interprets sociodemographic variation in food priorities such as budget, taste, and health among families with children. The unique application of the CCT framework allows for an analysis of sociodemographic trends in food shopping priorities, which will provide direction for public health planners seeking to reduce incidence of diseases with a dietary component. Data are analyzed using STATA 13 for logistic regression. Overall, this study concludes that income is an important consideration in how constrained a family feels by budget and how able they are to choose foods based on pleasurable taste. Income, therefore, is an important factor in the food choices a family makes, and a critical point of intervention. Being married/cohabitating and having kids are also contexts in which constraints on food shopping become apparent with partners wanting to prioritize nutrition, but finding that more difficult with children in the household. These results also indicate a critical juncture at which food constraints should be addressed.

Paper two addresses the following question: do respondents perceive those in their social network as having priorities that reflect the same contextual constraints, personal preferences, or health priorities? Network agreement is examined first among intimate partners, then among other adult family members, and finally, among friends. Correlation is investigated using ordinal logistic regression in STATA 13. The results of this study confirm that there is a correlation between respondent food priorities when shopping for food and partner, family, and friend food priorities. The relationships are correlated across budget, taste, convenience, nutrition and health. These results have important implications for dietary intervention programs that focus on the individual instead of larger social networks.

Paper three examines the impact of prioritizing budget, convenience, or health on the nutritional quality of grocery purchases. As part of the initial survey, grocery store receipts were collected and coded for nutritional value based on the NuVal® food scoring system, which considers over thirty dimensions of nutritional quality. Ordinary least squares regression analyses and mediation analyses are performed using STATA 13 to assess correlation between shopping priority and NuVal® score. Prioritizing budget reduces the overall nutritional quality of a parent’s purchases, an effect partially mediated by income. Prioritizing convenience has no effect, while prioritizing health results in an increased NuVal® score.

Taken as a whole, the three papers reveal the presence of constrained choices operating in several ways. Financial factors, relationship factors, and sociodemographic trends are of particular importance. First, financial considerations reduce the likelihood of prioritizing nutrition when shopping for food. Further, those with higher incomes make more healthful purchases overall than those with lower incomes. Public policy that reduces the cost of healthy food or increases individual income would be beneficial in helping families make more nutritious grocery purchases. Secondly, relationship factors like social network membership and familial status play an important role in food choice. Social norms established among partners, friends, and other family members may influence choices made at the grocery store, while shopping for a spouse or children is shown to make healthy choices a greater priority, but make those choices more difficult to achieve with children in the family. Programs that address community level dietary patterns or particularly target the needs of families could prove useful. Finally, unique constraints among sociodemographic groups also merit policy attention. In addition to variation by income group and familial status, variation is also present by sex, race, and weight category. Future research should identify constraints unique among those groups for more targeted dietary intervention.

Digital Object Identifier (DOI)