Violent crime and gun violence are top concerns for the majority of Americans. Over 80% of Americans view gun violence reduction as an important priority, placing it above several other issues, including the economy and immigration. The social costs of gun violence are staggering, amounting to over $250 billion annually (Cook & Ludwig, 2000). However, despite significant public concern and the substantial economic burden, there is a critical need for research to examine the effect of the rapidly changing landscape of public policies across the United States on violence. Compounding this issue is the lack of a robust and comprehensive data infrastructure capable of accurately measuring violence and gun violence outcomes. This dissertation aims to address gaps in causal empirical analysis by evaluating the effects of broad-reaching public policies on violence and gun violence outcomes. In addition, I apply machine learning methods to improve measurement of gun violence outcomes. In the first chapter of this dissertation, I examine permitless concealed carry, a law that since 2010 has been adopted by over 20 U.S. states. Permitless carry removes the need for a firearm owner to undergo a background check and obtain a license to carry a concealed handgun in public. I find that adoption of permitless carry increases serious assaults with firearms without evidence of change in non-violent or non-firearm violent crime relative to states that did not enact a permitless carry law. In part, substantial decreases to arrests for weapons law violations may indicate that decreases in police enforcement reduce the cost of carrying a concealed weapon. In the second chapter, we use medical claims data spanning over 100 million patients in 48 states to measure changes in domestic violence injuries treated in emergency departments during the Covid-19 pandemic. We find that visits for domestic violence injuries declined considerably in the early phase of the pandemic from March to June of 2020 and remained below prior year levels in later months of the year. Relative to accidental injury visits, the declining domestic violence injury visits are smaller in percentage terms and comprised almost entirely of less severe injuries. Visits for severe or life-threatening injuries remain almost entirely consistent through the duration of the pandemic. In the third chapter, I examine whether natural language processing and machine learning methods can be used to predict the locations (e.g. home, store, car) of nonfatal shootings and improve the accuracy of existing national estimates of where nonfatal gunshot injuries occur. This cross-sectional study of 22 years of data from the National Electronic Injury Surveillance System Firearm Injury Surveillance Study used unstructured medical text combined with language representation and machine learning models. Contrary to existing national estimates indicating that nonfatal firearm injuries occur most often at home, I find that these injuries occur most often in public places outside the home, including the street or highway. Natural language processing and machine learning may be used to predict gunshot injury locations. Taken together, this dissertation presents essays that provide significant empirical evidence regarding the effects of public policies on violence and firearm-related incidents. In addition, this research also advances the application of natural language processing techniques that enhance the accuracy and reliability of what is known about firearm injury.