Dissertation Defense Announcements

Exclusive breastfeeding in the first six months of life and continued complementary breastfeeding up to 24 months is encouraged to ensure optimal infant and young child nutrition and health. The WHO and UNICEF emphasize these optimal Infant and Young Child Feeding (IYCF) practices, especially for regions of the world where extensive child nutrition and healthcare support is lacking or inaccessible.
This dissertation explores the epidemiology of IYCF practices and child health outcomes in sub-Saharan Africa. It also examines the status of IYCF policies and programs in this relatively less studied region of the world. I use publicly available data from the Nigerian Demographic and Health Survey (NDHS) and the WHO/UNICEF Breastfeeding Collective scorecard to answer important questions explored across three studies.
Findings from the first study suggest that longer durations of breastfeeding are associated with fewer reported acute illnesses post-infancy at 24 to 59 months; demonstrating the long-term protective effect of breast milk from illnesses that contribute to the high under-five mortality rates recorded for decades in sub-Saharan Africa. Another important finding from the second study is that the relationship between exclusive breastfeeding, household living environmental conditions, and acute health outcomes in infancy is complex. The results suggest that the efficacy of exclusive breastfeeding in reducing the incidence of diarrhea and acute respiratory illness is strongest for infants living in households with poor sanitation facilities and inadequate building materials respectively. Lastly, findings from the third study indicate that sub-Saharan Africa as a region is yet to meet global and World Health Assembly targets for the implementation of recommended IYCF policies and programs. These findings have implications for child nutrition and health outcomes in especially for a region already disproportionately impacted by high under-five mortality rates.

Recent advances in the field of Natural Language Processing, specifically in Natural Language Generation (NLG) towards Dialogue Systems have focused mainly on two-party conversations. However, group conversations or multi-party conversations (MPC) are just as prevalent in our everyday lives. While the area of multi-party conversation modeling has received some attention in recent times, MPC lacks resources for 1) corpora in differing settings (formal/informal, synchronous/asynchronous), 2) dialogue models which can participate in informal open-domain settings while maintaining speaker information, and 3) evaluation metrics which provide better insights into the performance of MPC models when it comes to operating in groups and interacting with multiple participants. We thus take a three-pronged approach towards contributing to research in the MPC modeling research area. For corpora collection, we contribute a mock social media tool that can be utilized for collecting asynchronous MPC conversations called Community Connect and utilize it for 3 experiments to collect everyday talk. For MPC modeling, we propose a response generation model, using large language models (LLMs) and graph structured networks, which is capable of taking participant relations into account towards maintaining multiple persona profiles and generating responses keeping the speaker characteristics in mind and responding accordingly. Lastly, for MPC evaluation, we present an expansion to the taxonomy of errors which specifically contributes MPC-specific metrics to the overall NLG errors. In addition to the taxonomy, we contribute to better evaluation standards across which progress in the tasks within MPC can be tracked more saliently. Through these contributions, we aim to fill the necessary gaps towards advancing MPC understanding and modeling, while also providing the tools to gauge progress until now.

The increased reliance on Big Data Analytics (BDA) in society, politics, policy, and industry has catalyzed conversations related to the need for promoting ethical reasoning and decision-making in the mathematical sciences. While the majority of professional data scientists today come from privileged positions in society, those processed by the decisions made using data science are more often members of one or more marginalized social groups, translating into disproportionately negative outcomes for these individuals in society. Thus, it is argued that future citizens must develop an ethical mathematics consciousness (EMC) that human beings do mathematics; thus, there are potential ethical dilemmas and implications of mathematical work which may affect entities at the individual, group, societal, and/or environmental level. Drawing from this conjecture, the purpose of this Design-based research study was to develop a local instruction theory and materials that promote students’ ethical mathematics consciousness in a high school Ethical Data Science (EDS) course grounded in a feminist, relational ethic of caring and social response-ability. Outputs include the identification of design heuristics, including the task structures, participation structures, and discursive moves that supported students' development of EMC and equitable participation in classroom activities, an initial curriculum for the EDS course, and a student-use protocol and corresponding analytic framework for making critically conscious ethical decisions in data science.

Community violence occurs primarily in public settings, frequently involves high-risk behaviors such as firearm use, and is often geographically concentrated as a result of racial and economic segregation enforced through policy and practice. Community violence has risen in Mecklenburg County, North Carolina over the past five years, with a plurality of incidents concentrated in neighborhoods which also have high rates of social, economic, and health-related risk factors. This dissertation builds on my work with the City of Charlotte and Mecklenburg County as part of a multi-sector collaboration intended to leverage resources and align programs and policies to disrupt, reduce, and prevent community violence. In this dissertation, guided by the Ecological Systems Theory and Social Determinants of Health Framework for Action, I used qualitative, quantitative, photographic, and geospatial data to (1) explore residents’ perceptions of safety and experiences of community violence; (2) describe an integrated, place-based methodology that can be used in community violence research; and (3) explore how positionality informs cross-sector, collaborative data sharing efforts to address community violence.

In study one, participants identified neighborhood features across ecological levels that contributed to them feeling safe or unsafe. Notably, participants perceived historical and on-going disinvestment, enacted through structural racism, as contributing to unsafe conditions. In study two, which grows out of study one, we found that walking interviews generated more findings specific to place and situated within the micro-, meso-, and exosystem levels, while more traditional, semi-structured sedentary interviews yielded results that were largely centered within the individual and microsystem levels. In addition, using an integrated methodology highlighted gaps in the publicly available quantitative data and demonstrated the utility of employing multiple methods to capture data related to place, most notably by generating data that informed actionable insights across ecological levels. In study three, we found that individuals’ and organizations’ social identities (e.g., individuals’ level of data knowledge and data sharing experiences, and organizations’ use of formal data sharing processes) as well as power (specifically, individuals’ sense of empowerment, and organizations’ use of resources and data sharing capacity) interacted to influence barriers and facilitators to data sharing.

Findings point to areas for future research and suggest local implications including (a) the need for increased attention in research and practice related to how structural racism contributes to unsafe neighborhood conditions; (b) the potential benefits of considering how the described integrated, place-based methodology can be scaled to capture residents’ perceptions of safety and experience of violence across neighborhoods; and (c) the salience of attending explicitly to how the positionality of the individual and organization contributes to barriers and facilitators to cross-sector data sharing. Results from my dissertation can be used locally to inform cross-sector, collaborative solutions to community violence that incorporate residents’ perspectives and address risk factors across ecological levels. While conducted in Mecklenburg County, results also have implications for community violence prevention and intervention efforts in communities across the country.