This dissertation seeks to understand the determinants and motivations of citizens' individual and collective participation in the process of public service delivery through the lens of citizen co-production. While citizen participation has been highlighted particularly in the administrative decision-making process, citizen co-production literature emphasizes the role of citizens in the public service delivery process. This body of literature argues that citizens may contribute to public service outcomes by providing time, efforts, knowledge, and by cooperating with professional public service providers. Utilizing both a relatively uncommon machine learning technique in public administration, random forest regression, and traditional statistical approaches, I examine various factors shaping citizens' individual and collective participation in the process of emergency service delivery before, during, and after tornadoes. Three empirical chapters suggest that public trust in issue-specific agencies and social capital play significant roles in structuring citizens' individual and collective co-production of emergency service. The analyses of two methods utilized in this dissertation also suggest further investigation in quantitative methodology for a better understanding of citizen participation in public service delivery.