Empirical Analyses Of Queues With Applications To Elections And Healthcare

In Chapter 1, we conduct a dynamic panel data study of voting resource allocation within Florida counties. We find that a 1\% increase in the percentage of voters registered as Democrat in a county results in a 2.8\% increase in the number of registered voters per poll worker. Furthermore, using a queue simulation, we estimate that a 5\% increase in voters registered as Democrat in a county could increase the average wait time to vote from 40 minutes (the estimated average wait time to vote in Florida in 2012) to approximately 100 minutes. Our study recommends that states regulate the number of voters per poll worker or voting machine in polling locations so that wait times are equated across all voters. In Chapter 2, we perform a differences-in-differences analysis on cross-sectional voter wait time data across the 2006, 2008, 2012, and 2016 Georgia elections. We estimate that polling place closures increased Georgia’s average wait time to vote in the 2016 election by 7 minutes or approximately 78\% (based on Georgia’s average wait time of 16.5 minutes in the 2016 election). This increase in the average wait time to vote suggests that in the 2016 election, Georgia may have idled its spare capacity (e.g., voting machines) following polling place closures. As a result, we suggest that states implement policies that require the redistribution of all functioning voting machines from closed polling places or at least increase transparency in how voting resources are used in elections. In Chapter 3, we use an instrumental variable estimation and find that patients prefer waiting for endoscopies in pre-op rather than reception. Additional experiments suggest that pre-op is less favorable due to its intensity (e.g., clinical, emotional). We also find that in transparent, shared waiting areas where patients do not observe the doctor assignment of others, patients may still monitor queue discipline. Finally, patients may be negatively impacted by waits that conclude after a scheduled appointment time but care less about waits that conclude early. This study emphasizes the importance of businesses managing queues where customers wait in multiple locations with different attributes