Picking Prosecutors

I.     Introduction

The United States imprisons its citizens at such a remarkable rate ––unprecedented in American history and without international parallel ––that critics have characterized the country as a carceral state.1 Despite its size, the “carceral state has been a largely invisible feature” of American politics: Much of the rise of mass incarceration over the last five decades has happened without significant public debate.2 This is in part because there is no single, national criminal justice system, but rather a patchwork of local systems. That patchwork frustrates efforts not only at coordination but also observation.3

Within the past decade or so, though, a dedicated group of academics has offered multiple, sometimes competing, lenses to understand why we experienced surging incarceration rates.4 These include racial control,5 racial backlash,6 racial wedges,7 racial liberalism,8 intra-class black politics,9 “pendulum justice,”10 political economic crises,11 a punitive public,12 and a unique institutional landscape that fundamentally transformed modern social movements.13 This work is nothing less than crucial. But nearly all prominent work on the politics of mass incarceration has focused, at least primarily, on a national story.14 As a result, the local prosecutor has been largely absent from the project of explaining why we experienced surging incarceration rates and how to unwind it.15

Our Article contributes to a small, but growing literature that acknowledges the centrality of local prosecutors to criminal justice outcomes.16 The reality has always been that the vast majority of people under criminal supervision are arrested by the local police and charged by the thousands of local prosecutors who dot the United States.17 William Stuntz famously characterized these local prosecutors as “the criminal justice system’s real lawmakers.”18 More recently, John Pfaff has provided empirical support for Stuntz’s account, using existing but under-utilized data to try to persuade others that prosecutors are “the most powerful actors in the criminal justice system.”19

This recent appreciation of prosecutorial power has extended beyond the pages of law reviews and academic press books. In the last few years, non-academics have begun to appreciate how much power prosecutors yield and to use that insight as the basis for more effective criminal justice reform.20

Most notably, a motivated group of advocates and their supporters have started a movement to elect progressive prosecutors.21 The term “progressive prosecutor” does not have a single, accepted meaning. It has been used to describe both Democrats and Republicans.22 And some have questioned whether the label “progressive” may alienate some voters who support criminal justice reform, but who do not identify as politically progressive or liberal.23 In any event, we use the term to refer to prosecutors who have specifically championed or adopted prosecutorial practices that are intended to make the criminal justice system less punitive.24

The rise of mass incarceration has highlighted how “[o]ver the last forty years, prosecutors have amassed more power than our system was designed for.”25 But the progressive prosecutor movement does not seek to reform prosecutorial power, at least immediately, so much as to capitalize on it. Reformers argue that electing progressive prosecutors will help to reverse the punitive trends of the past several decades.26 Put differently, they hope to harness the prosecutorial power which helped to create our system of mass incarceration in order to dismantle it. This movement has included efforts to both recruit progressive challengers in local prosecutor elections and to channel campaign donations to those challengers.27 Advocacy organizations have also launched national campaigns to better educate voters.28

Whether these efforts to elect progressive prosecutors will succeed remains to be seen. But the movement has already scored some high-profile victories in Chicago,29 Philadelphia,30 San Francisco,31 and Boston,32 as well as some smaller cities.33 The movement’s success so far is the product of a welcome change in the “surface politics” of criminal law.34 As Stuntz noted in 2001, the “tough-on-crime phase of our national politics will someday end; indeed it seems to be ending already.”35

This popular movement to elect progressive prosecutors defies academic conventional wisdom. It is a widely shared view among law professors that prosecutor elections are rarely more than empty exercises: Prosecutor elections are rarely contested, and incumbents virtually always win.36 As one law professor recently put it, “mostly, incumbents just win until they quit.”37

But our understanding of prosecutor elections is surprisingly thin. There appears to be only one empirical study to date that surveys prosecutorial elections.38 This study by Professor Ronald Wright concluded that elections are not very effective at holding prosecutors accountable,39 but also suggested that nonpartisan elections and term limits might be promising reforms to make prosecutor elections more effective. Wright’s study, while incredibly important as foundational scholarship, has several limitations. For one, the data itself is limited, covering only 15 states and including elections only up until 2008 or 2012.40 And because Wright sought to answer questions about accountability for incumbent prosecutors, the study focused on the limited question of local prosecutors’ incumbency advantage, and it did not examine open-seat elections. It also does not consider the extent of competition, which offers an arguably more faithful assessment of critics’ concerns with prosecutorial elections in the first place. Most importantly, because Wright treated the primary and general elections as two independent contests, rather than one democratic exercise, it may lead readers to draw incomplete or misleading inferences about how often prosecutor elections are contested.

This Article makes two distinct contributions to the literature. First, it catalogues for the first time, in detail, how local prosecutors are selected in each state. It highlights the great similarity and important variation between the states. For example, a sizable minority of states have combined multiple counties into prosecutorial “districts,” rather than electing a separate prosecutor for each county. A few have taken the opposite tack, carving out particular cities within counties as a separate jurisdiction. It also uncovers variation within states regarding prosecutorial selection because a handful of states leave it to their counties to decide how to select their local prosecutors.

The second major contribution of this Article is to present the results of the first comprehensive study of prosecutor elections across the country. Our original data set supports a more complete accounting of prosecutor elections. Our study both confirms and complicates the conventional wisdom. Our study confirmed that when incumbents seek reelection, they win an astonishing 95 percent of the time. What is more, once incumbents serve more than five years in office—which for most prosecutors means more than a single term in office—their chances of facing a challenger decrease significantly, and their chances of winning a contested election increase significantly. But our study also revealed that incumbency is only part of the story. For example, we discovered that even when an incumbent was not running in an election, it was still likely that only a single candidate ran for the office. Put differently, it isn’t just that incumbents are unlikely to run in contested elections, it is that all prosecutor elections are unlikely to be contested. In addition to providing a comprehensive look at which elections were contested, we also looked at the competitiveness of those elections. By collecting the vote totals and vote share for each candidate, we were able to calculate how competitive a race was as well.

In addition to providing new data, our study also deployed a new framework for analyzing these local elections: Rather than considering general elections and primary elections in isolation, we analyzed whether the candidate who won the prosecutorial election faced an opponent in either the general or the primary election. This framework allowed us to better capture whether voters had a choice––at any point––in the selection of their local prosecutor.

Most importantly, we found that elections looked very different depending on the type of jurisdiction: the larger the jurisdiction, the more likely the election was contested. Contested elections were also more likely to be competitive as the jurisdictions grew larger.41

All told, urban and suburban prosecutors were far more likely to face challengers than were rural prosecutors. Urban and suburban prosecutors, when challenged, were also more likely to find themselves in a competitive election. Put simply, elections pose a greater threat to urban and suburban prosecutors than to rural prosecutors. This pattern is important because prosecutorial boundaries are not set in stone but reflect a structural choice. More than a dozen states have chosen to aggregate their counties into more populous districts for prosecutor elections. States sometimes draw new district boundaries,42 but those decisions appear to have escaped the notice of activists or academics. Our findings suggest that drawing new district lines could potentially increase the likelihood of contested and competitive prosecutor elections, though it is not clear how large the effect would be.

The stark differences that we found between urban and rural counties led us to hypothesize that many districts do not have a sufficient supply of candidates for prosecutor elections. It is not enough to have multiple people who are interested in becoming the local prosecutor in order to have a contested election, those people must also be eligible for the position—that is to say, those interested people must be attorneys who are active members of the state bar.

We confirmed that there is, indeed, a very small pool of candidates for prosecutor elections. Using data from a subset of ten states, we found a large number of districts with very few lawyers—almost 40 percent of the districts in those ten states had 15 lawyers or fewer living within their boundaries. The supply problem is most acute in smaller districts. In those small population districts, we also found a correlation between the number of lawyers living in a district and the likelihood of a contested election. In particular, we found that jurisdictions with fewer than five lawyers see less than ten percent of their elections contested. Indeed, as a jurisdiction’s population falls below 7,000, we begin to see elections where not even a single candidate runs for district attorney.43 Importantly, those jurisdictions that are most likely to have an insufficient supply of candidates—small and rural counties—are the places where incarceration rates are the highest.

This is a key finding for the criminal justice reform community. Those jurisdictions that have done the least to reform their high incarceration rates are also the least likely to be able to present their voters with any real choice about who will serve as their local prosecutor. Put simply, in order for prosecutor elections to serve as an antidote to mass incarceration, more candidates will have to run in rural districts.

This Article proceeds in three parts. Part II catalogues how local prosecutors are selected across the country. It notes that, although the vast majority of states elect their local prosecutors, there is variation in how those elections occur. In particular, there is variation in the length of terms that prosecutors serve and the timing of prosecutor election cycles. There is great variation in the geographic unit and population that each prosecutor serves. There is even variation in the legal rules surrounding the election of unopposed candidates. Part II also identifies the extent to which some states permit counties to make their own rules about prosecutorial selection.

Part III moves from the legal to the empirical. It presents, in detail, the data from our national study of prosecutor elections. The study collected data about the most recent election of every local prosecutor in the country, including under-studied details, such as incumbents’ tenure in office and candidates’ gender. Part III describes how our data were collected, and it provides an analysis of the data. Our study confirms that incumbents enjoy a significant advantage in prosecutor elections—an advantage that is especially pronounced for incumbents who have served for more than five years. But our study also indicates that prosecutor elections are more contested than previously understood, and that elections are more likely to be contested in jurisdictions with larger populations. Our study also reveals a significant amount of turnover during prosecutorial terms, as sitting prosecutors resign to seek higher office or to return to private practice.

Part III also uses data about local incarceration rates to show that it is precisely those rural areas where prosecutors are least likely to face challengers where incarceration rates continue to increase. While we do not make any causal claims about the relationship between prosecutor elections and incarceration rates, we believe that this correlation between uncontested elections and punitiveness deserves further study.

Part IV explores what these findings tell us about the promise of prosecutor elections to achieve criminal justice reform. Our data allows us to offer informed suggestions about how to increase the number of contested elections but, as Part IV explains, we do not know how more competitive elections will affect the criminal justice system.

The nascent movement to elect progressive prosecutors has taken advantage of the current low-contestation environment to unseat incumbents in some more populous districts.44 But if prosecutor elections become more contested and more competitive as a matter of course, and if the profile of those elections continues to increase, it is far from clear that progressive candidates will routinely win. Once voters see prosecutors as political actors who ought to be more accountable to the public, what stops other groups from funding “law and order” prosecutors to run for office?45 In this sense, the task of those who would elect progressive prosecutors is persuasive as much as anything else.46 Importantly, though, their audience is two-fold: They must convince progressives to be prosecutors, and they must convince the public to be progressive.

Given the higher rate of contested elections in urban areas, and given how many Americans live in urban and other highly populated areas, the reality is that many (if not most) Americans have the ability to dramatically change criminal justice policy through the ballot box. Indeed, if progressive prosecutors were to win elections in the 150 largest of the more than 2,300 prosecutor offices, then they could institute reforms that would affect more than half of all Americans. But the mere fact that voters can change criminal justice policy through prosecutor elections does not tell us whether they will. Nor does it tell us what those changes are likely to be; elections could result in more progressive criminal justice policies or in more punitive policies.

II.     Methods of Prosecutorial Selection

The United States is unique in electing its prosecutors.47 Most Americans, though, don’t realize that we do.48 Although the Attorney General and the United States Attorneys are appointed in the federal system,49 the vast majority of states elect their attorney general and hold separate elections for local prosecutors. This was not always the case. From the colonial era until the Jacksonian era, local prosecutors were often appointed by state authorities.50 But the centralization of power and patronage politics that ensued51 ultimately stoked such discontent that three-quarters of states between 1832 and 1860 fashioned the distinctive office of the local elected prosecutor.52

Although their mandates vary, local prosecutors generally prosecute state felonies within a specified geographic area.53 There are more than 2,300 separate state felony prosecutor’s offices in the United States.54 To be clear, there are at least two other types of prosecutors in the states: Most states have an Attorney General, who is the state’s chief prosecutor, but who is not the primary prosecutor of felonies.55 Some jurisdictions also have city attorneys who operate on the municipal level but focus on misdemeanors, not felonies. This Article focuses on local prosecutors who prosecute state felonies within a specific subset of the state.

One source of confusion is that this single office goes by many different names. Elected local prosecutors may be officially referred to as the “district attorney,” the “county attorney,” the “prosecuting attorney,” the “state’s attorney,” or the “commonwealth attorney,” among other titles.56 Although the jurisdiction of a “county attorney” extends only to a single county, the jurisdiction of a “district attorney” is not as clear. Mississippi, for example, divides its 82 counties into 22 judicial districts;57 while some districts contain only one county, others are comprised of up to seven counties.58 The local prosecutor for each of these districts is referred to as the “district attorney.” New York also calls the office “district attorney,” but it does not combine its counties into districts; instead it selects one district attorney per county.59

To avoid confusion, this Article will use the term “local prosecutor” to refer to the office. It also uses the general term “district” to describe a prosecutor’s area of jurisdiction regardless of whether the jurisdiction contains multiple counties, a single county, or a city.

Table 1 identifies how local prosecutors are picked in different states.60 As it illustrates, the vast majority of states—45 of 50—elect prosecutors at the local level.61 Most of those 45 states elect their prosecutors for four-year terms with no term limits in partisan elections.

Table 1. Prosecutor Selection in the United States


Only four states follow the federal government in appointing rather than electing their local prosecutors.62 Alaska, Connecticut, Delaware, and New Jersey appoint their local prosecutors. New Jersey is the only state whose appointment process resembles the federal model: County prosecutors are appointed by the governor with the advice and consent of the state senate.63 Alaska and Delaware have their Attorneys General appoint the local prosecutors.64 Connecticut creates a commission to select and appoint a State’s Attorney for each district.65

The state of Rhode Island does not select local prosecutors. All criminal cases are instead handled by the Attorney General, who is elected.66

Previous characterizations of how states select their prosecutors have focused on differences between states. But there is also sometimes variation within states. Three states—Hawaii, Montana, and North Dakota—leave the decision whether to appoint or to elect a local prosecutor to the counties themselves. Notably, the vast majority of counties in these states still rely on elections. Three of four counties in Hawaii elect their prosecutors.67 But in Maui County, the Mayor appoints the Prosecuting Attorney, who is confirmed by the County Council.68 Montana allows counties to either appoint or elect county attorneys, and all appointed prosecutors “serve[] at the pleasure of the [county] commissioners.”69 Two counties in Montana—Carter County and Petroleum County—appoint their prosecutors. Carter County shares its County Attorney with a nearby county, Fallon County. But while the Fallon County Attorney stands for election, the position of County Attorney does not appear on the Carter County ballot.70 North Dakota allows counties that wish to appoint their state’s attorney to “place the question of appoint[ment] . . . on the ballot”;71 a majority vote changes the selection method “from electi[on] to appoint[ment].”72 Two counties—Golden Valley County and Steele County—have chosen to appoint their prosecutors.73

Somewhat surprisingly, local prosecutor is overwhelmingly a partisan office. In only five states is the local prosecutor a nonpartisan office. It is surprising that so many states rely on partisan elections to select their local prosecutors because most local elections in America are nonpartisan.74 Hawaii and Montana allow counties to decide whether the office is partisan or nonpartisan.75

Only one state—Colorado—imposes term limits on its local prosecutors. The state constitution limits district attorneys to serving two consecutive four-year terms, but districts are permitted to “lengthen, shorten, or eliminate the limitations on terms of office.”76 Some districts have used this power to increase the number of terms that their district attorneys may serve.77

While most states elect their prosecutors to serve four-year terms, a handful of states have chosen different term lengths. Alabama,78 Kentucky,79 and Louisiana80 have six-year terms; Tennessee has eight-year terms.81 New Hampshire is the only state that has its local prosecutors serve less than a four-year term; the state elects its District Attorneys to two-year terms.82

The timing of the election cycle also varies. Most state prosecutor elections occur in even-numbered years, and thus coincide with the congressional election cycle. But some states hold their elections in odd-numbered years.83 And several states do not hold all of their prosecutorial elections in the same year.84

The biggest source of variation in how we pick prosecutors is how the state defines the polity—that is to say, what defines the boundaries of the prosecutor’s jurisdiction. For the most part, elections are county-centric. About two-thirds of states use single-county districts.85 Sixteen states use multi-county districts, aggregating counties together to form a district.86

Aggregation allows states to distribute population more evenly across districts. States that do not aggregate, but rather elect one local prosecutor per county, tend to have large variations in constituency size. That variation is a result of the unequal distribution of populations within a state. For example, the district attorney in Los Angeles County, California has nearly ten million constituents, while the district attorney in Alpine County, California has barely more than one thousand.87

Not every state relies on county boundaries to form their jurisdictional boundaries. Some Texas districts partially reach into multiple counties. And Virginia has a large number of cities that elect their own prosecutors.88

States also vary in how they deal with the common occurrence of prosecutors who run unopposed in elections. A number of states do not require candidates to appear on the general election ballot if they are unopposed; the fact that they face no opposition results in them being declared the winner.89 This is the procedure in Arkansas,90 Florida,91 Hawaii,92 Louisiana,93 Oklahoma,94 and South Dakota.95 Texas gives local governments the option to declare unopposed candidates elected.96 In California, if a candidate receives more than 50 percent of the vote at the June nonpartisan primary, she is declared elected.97 If no candidate receives more than 50 percent in the June primary, then a runoff election is held between the top two candidates during the November general election.98 Minnesota and Oregon, which also hold nonpartisan elections, only hold a primary if more than two candidates file for the office.99 Many more states do not put unopposed candidates on the ballot for primary elections.100 Montana allows those local governments that have chosen to hold nonpartisan elections to cancel the primary election if fewer than three candidates have filed, instead allowing the election to be determined at the general election.101 And several states cancel local elections if all of the candidates on the ballot are unopposed.102

These laws may contribute to the fact that many Americans do not know their local prosecutor is an elected office.103 For example, when Louisiana held elections in 2014, 27 of the 42 district attorneys were unopposed.104 As a consequence, more than half of the district attorneys in the state did not appear on the ballot. And when prosecutors go unchallenged for multiple election cycles, the information deficit for voters can increase dramatically.105 For example, the State’s Attorney for Florida’s 20th Judicial Circuit did not face a challenger after his first election in 2002 until he retired in 2018.106 As a result, any person in that district under the age of 34 when the prosecutor retired had never seen the office of State’s Attorney appear on the ballot.

III.  Prosecutorial Election in the States

In this Part, we describe and analyze the data collected by the Prosecutors and Politics Project, a research initiative at the University of North Carolina School of Law.107 The Project collected election information from all local prosecutor races in a recent election cycle. We use that data to assess the amount of contestation and competition in those elections. In addition to contestation and competition, we offer new data on the length of incumbents’ time in office, the appointment of prosecutors to complete terms, and the gender makeup of prosecutor candidates. We also report variation in contestation and competition across states.

In addition to analyzing the data collected in our nationwide study, we briefly analyze data from Professor Ronald Wright’s study of prosecutor elections in 15 states. We also use data from the National Corrections Reporting Program, based on an analysis by the New York Times, with assistance from John Pfaff, to detail prison admission rates for different prosecutorial districts.

A.     Data

Beginning in February 2018, the Prosecutors and Politics Project began collecting information about the most recent election cycle in each state that elects its local prosecutors. For most states, that meant we collected data from the 2014 or 2016 election cycle.108 In total, we collected election results for 2,315 districts across 45 states.109

The dataset includes the name, party affiliation, and incumbency status of each candidate who ran for local prosecutor. For those candidates running as incumbents, we also identified the year that they first assumed office. For both the primary and the general elections,110 we indicate which candidates ran and which candidates won as well as the total number of votes received and the percentage of votes received.111 Table 2 offers a stylized example of this data.

Table 2. Stylized Example of Candidate Data


The data allow us to determine whether each district had an open seat, whether it was contested, and whether it was competitive. We define an open seat as those where the incumbent did not seek reelection.112 A contested election is simply one in which there are two or more candidates. We define competitive elections in reference to the vote share of the winning candidate. Following the convention adopted by others who have studied state and local elections, we classified an election as competitive if the winning candidate received less than 60 percent of the vote.113 In the Appendix, we provide data about competitiveness using alternative thresholds of 55 percent and 65 percent.114

The decentralized nature of prosecutor elections made it extremely difficult and time-consuming to collect this data. As a result of those difficulties, not all the information depicted above was available for all candidates. We have the names of all candidates who ran in the latest election cycle. For 99 percent of these candidates, we know whether they ran in the general election; for 96 percent we know the same for the primary. We know who won and who lost general elections for 99 percent of candidates. We have incumbent status for more than 99 percent of candidates. And we have population figures for all districts except for a few districts in Texas.115 We also endeavored to collect the year in which an incumbent first assumed office, but that proved to be one of the most difficult variables to obtain. We have that information for only 70 percent of candidates whom we have identified as incumbents.116 In the example provided in Table 2 above, the incumbent in District 1 is uncontested. In District 2, there is no incumbent, but the open seat is also uncontested. District 3 features a contested general election, with the incumbent winning by such a margin as to not be competitive, under any definition. In this case, both party primaries were uncontested. District 4 offers an example of a contested primary, where the incumbent loses, 41 percent to 59 percent. This would not be considered an open seat because the incumbent sought reelection, even if unsuccessfully. It would be considered competitive under the broad definition we adopt. Importantly, the example shows how the primary winner is uncontested in the general election, revealing how local political conflict can be structured in different ways.117 Relatedly, District 5 shows a three-way contested primary for an open seat, with the winner again automatically capturing the general election. Under the definition we adopt, the primary is considered competitive.

Beyond candidates’ election results, the dataset also includes information about the jurisdictions in which candidates ran. Jurisdictional data includes population;118 whether the jurisdiction is urban, suburban, or rural;119 and the number of state prison admissions per 10,000 residents in various years according to the National Corrections Reporting Program.120 Some of this data is available only at the county level and so it was aggregated in multi-county districts. For the prison admissions dataset, we drop counties that are marked as having an invalid measure in either 2006 or 2013.121 Table 3 offers a stylized example of this data. In general, it highlights how, for the purposes of understanding prosecutorial behavior, counties and districts may describe differing jurisdictions.

Table 3. Stylized Example of Jurisdiction Data


In addition, we also collected the name of the current office holder,122 allowing us to determine whether the current office holder was appointed after the most recent election. Finally, our analysis incorporates data collected by Professor Ronald Wright and that he generously shared with us.123

B.     Results

1.     Electoral Results

The academic literature on prosecutor elections is thin. To the extent there is one, it focuses on incumbents. The only published empirical information comes from Professor Ronald Wright, and his findings have been very influential in shaping the academic conventional wisdom. In 2009, Wright used data from ten states to first report a startling statistic: Incumbents were winning 95 percent of general elections for local prosecutor.124 In 2014, he expanded his sample, including five more states and a few more electoral cycles, and he found the same thing.125 Table 4 uses our national data to continue this inquiry and confirms what Wright initially found: Incumbents all across the country won 95 percent of their latest elections too.

Table 4. Incumbent Candidates in Prosecutorial Elections


The first two rows of Table 4 show that, in the election cycles we studied, incumbents ran in the general election 76 percent of the time, in 1,763 districts, and that they faced challengers just 15 percent of the time, or in 264 districts. These numbers look very similar for primaries. Like Wright, we found that incumbents won 95 percent of all general elections in which they ran, and they won a similar percentage of primary elections. In those few elections in which incumbents faced a challenger, they won the general election approximately 65 percent of the time. The incumbent win rate for contested primary elections was lower, barely above 50 percent.

The second panel of Table 4 considers how these statistics vary by the population of the district. Incumbents win everywhere almost all the time. But while incumbents were challenged in just 13 percent of general elections in districts with less than 100,000 people, they faced challengers more often in more populous districts. Incumbents ran in contested general elections 21 percent of the time in districts with at least 250,000 people and 34 percent of the time in districts with at least one million people. In other words, incumbents were least likely to face challengers in districts with the smallest populations.126

But Table 4, like existing academic work, is incomplete for at least three reasons. While the initial focus on incumbents was fruitful, it leaves uninterrogated the quarter of districts with an open seat. Further, the general election and the primary election are not independent contests, but rather two pieces of a single democratic effort. The partisan composition of some districts may be so lopsided that all important political fights occur in the primary; in others, those fights may occur in the general election. In only a few districts are both the primary and the general elections significant. Disaggregating the general election and the primary election may fail to capture these important local political dynamics. Finally, contested elections may not be competitive elections. That is to say, elections with two candidates might be extremely lopsided, with one candidate the clear winner well before votes are cast. In such elections, it is hard to say that the second candidate provides a viable alternative for voters, which is closer to the concern academics and activists alike have raised.

With this in mind, Table 5 presents a more complete account of the prosecutorial elections we studied. The table is divided into three sets of columns. The first set, labeled “All Districts,” considers the latest election in every district. The second set, “Open Seats,” specifically examines those where the incumbent did not seek reelection. The third set presents information for elections in which an incumbent did run.127 For each of these three categories, the table reports how often the election was contested and how often it was competitive.128 This information is also broken down by population.

Table 5. Contestation and Competition in Prosecutorial Elections


The initial rows in Table 5 correspond to different electoral contests. We report the outcome of the party primary of the current prosecutor rather than both primaries129 because we are concerned with whether the public had an electoral opportunity to choose their current prosecutor.130

Because we are most interested in determining whether there was an electoral opportunity to choose the current prosecutor, we also introduce an aggregate metric that examines whether there was contestation or competition in either the general or the primary election. This “either election” metric captures the political realities of many communities—namely that voters in local elections will sometimes only have an electoral choice in the primary election because voters in that district so heavily favor one party. Take, for example, the recent defeat of incumbent prosecutor Angela Corey in Florida’s 4th judicial district. The 4th district is a Republican jurisdiction.131 Corey was unseated in the Republican primary by challenger Melissa Nelson, who has been heralded as an example of a progressive prosecutor.132 Nelson’s subsequent general election was uncontested.133

Table 5 reports that about 18 percent of general elections are contested. This is slightly higher than the 15 percent incumbent contestation rate because open seats are more likely to be contested. In fact, comparing open-seat elections with elections in which an incumbent runs reveals that open elections are twice as likely to be contested: Open seat general and primary elections are both contested approximately 30 percent of the time, as compared to only 13 percent of primary elections in which an incumbent runs.

A focus on either the general election or the primary election in isolation, whether an open seat or not, has similar distortive effects. In addition to the 18 percent of general elections which were contested, 16 percent of primary elections were also contested. The difficulty is that, without more information, it’s not clear if those contested elections are occurring in the same jurisdiction. The third row, labeled “Either Election,” shows that they are often not.134 Almost a third of current prosecutors, and slightly more than half of current prosecutors who won an open seat, faced an opponent in either the general election or their own primary election. Even elections involving incumbents look different when we conceptualize contestation as a matter in either election. A full quarter of elections involving incumbents were contested either in the primary or in the general election.

This is a less dour portrait than has been commonly understood. But while the current stock of local prosecutors is more likely to have been contested at some point in the election, contestation is just a simple measure of whether there were at least two candidates in a given race. Competition offers a more robust, and perhaps realistic, assessment of the prospect for electoral change. Table 5 reveals that while the current prosecutor faced contestation 30 percent of the time, she faced some form of competition just 18 percent of the time, though this varies dramatically by the threshold for competition.135

It’s important to note that this definition of district competition, even expanded to include either the general election or the primary, does not capture some important electoral dynamics. Some local prosecutors may not have faced contestation or competition during their latest election, but they may have faced a challenger in the previous election. If, for example, a district had a contested election every other cycle, then we might still conclude that voters in that district had meaningful input into the selection of their prosecutor. Put differently, to understand whether voters have meaningful choices about their prosecutors, we must look at more than simply a single election cycle.

To get a sense of the variation in contestation and competition over time, we re-analyzed the Wright data to see how often a given district lacks any contestation or competition. Depending on the definition of competition, about half of the districts he studied were competitive, in either the general election or the primary, in at least one election spanning 1996 to 2012.136 This analysis of Wright’s limited data indicates that there are a significant number of districts where voters are not given any meaningful choice about who will serve as their local prosecutor.

Of course, looking only at general contestation and competition rates obscures significant variation. For example, states had very different rates of contested and competitive elections. (A table showing contestation and competition rates in each state is included in the Appendix.137) This variation between the states does not obviously appear to track any of the selection differences that we discuss above in Part I. For example, Wright found that nonpartisan states had more contested elections.138 But our data does not show the same. Some nonpartisan states, such as California, had high rates of contested elections. But others, such as North Dakota, had incredibly low rates of contested elections. Nor did we find that states with multi-county districts had significantly different contestation rates than states with single-county districts. Both multi-county district states and single-county district states had an average contestation rate of approximately one-third.139

We did not find a difference in the rate of contested elections in those districts with term limits—that is to say, districts in Colorado.140 To be sure, Colorado’s contestation rate is higher than the average rate across all districts—36 percent as compared to 30 percent. But there are several other states without term limits that have contestation rates that are as high, if not higher.

While we did not see an obvious relationship between contestation rates and legal rules, we did find great variation based on population. Returning to Table 5, the bottom panel shows dramatically different rates of contestation by population.141 As population size increases, so too does the likelihood of a contested election and a competitive election. For open-seat elections in the largest jurisdictions, we found that 100 percent of elections were contested, as compared to 45 percent of elections in the smallest jurisdictions. Contested elections with incumbents are also far more likely in the most populous jurisdictions—more than half of elections in the largest districts are contested as compared to 22 percent in the smallest. Overall, the local prosecutors in the smallest districts faced challengers 26 percent of the time as compared to nearly two-thirds of the time in the largest districts.

We also see significant differences in the number of competitive elections. In the smallest districts, only 16 percent of elections are competitive, as compared to 49 percent of elections in the largest jurisdictions.

As Table 5 demonstrates, voters in larger districts tend to have a choice more often at the ballot box about who will serve as their prosecutors than do voters in smaller districts. But that is far from the only difference between large and small districts. As others have noted, prosecutors in smaller districts also appear to be the most punitive.

Table 6 compares new prison admissions per 10,000 residents in 2006 and 2013, according to the National Corrections Reporting Program, as assembled by the New York Times, with assistance from John Pfaff. It shows incarceration is increasing in the smaller districts, while decreasing, dramatically so, in the largest.142

Table 6. Prison Admissions Per 10,000 People By District Population


Put simply, smaller districts—the districts that are least likely to have contested elections—have done less to curb their incarceration rates—or at least, had done less as of 2013.

2.     Tenure and Resignation

Closely related to the fact that incumbents usually win their elections is the question of how long they serve in office. Stuntz observed that local prosecutors had become “something akin to civil servants.”143 This prompted us to look at how long incumbents serve in office today. In 2007, the Bureau of Justice Statistics (“BJS”) surveyed all prosecutors’ offices in the country. Among other things, the survey asked how many years local prosecutors had been in office. As part of our data collection, we updated and supplemented the BJS’s effort with data on incumbents’ most recent tenure and resignation.

The BJS census found a median tenure of seven years.144 The prosecutor at the 25th percentile in the distribution had served for three years, while the prosecutor at the 75th percentile had served for 14. More than 200 local prosecutors had served for more than 20 years. There did not appear to be any particular correlation between the length of a local prosecutor’s tenure and the size of the constituency served.145

Our findings, approximately a decade later, are consistent with these numbers, reflecting little recent change in the tenure of elected prosecutors. Although we did not collect information about how long current office holders have been in office, we collected information about how long incumbents who stood for reelection had already served in office at the time of reelection. According to our data, the median incumbent has been in office for eight years when she runs for reelection, with a 25th percentile of four years and a 75th percentile of 14 years. Although our data is consistent with the BJS survey, we only have this information for about 70 percent of incumbents.146

We used this incumbents’ tenure data to assess whether incumbents were more or less likely to face a challenger the longer that they had been in office. We found that incumbents with less than five years in office were the most likely to face a challenger. The chances of facing a challenger decreased significantly for the cohort of prosecutors who had been in office for five to ten years, and then decreased modestly for those who had been in office longer. This data is presented in Table 7.

As Table 7 shows, a similar relationship exists for winning elections. The incumbents who were most likely to lose a general election were those who had been in office less than five years. An incumbent’s chances of winning reelection increased significantly when she served five to ten years. And additional time in office beyond ten years had the same modest relationship to winning the general election as it did for uncontested elections.

Table 7. Incumbent Tenure


In short, incumbents enjoyed a significant advantage—both in terms of running unopposed and in terms of winning elections—once they served for more than five years.

Although BJS did not collect local prosecutors’ demographics, an advocacy group recently collected such data. It found that 17 percent of prosecutors were women and five percent were prosecutors of color.147 We collected data on the gender of every candidate for prosecutor. Our data is not only consistent with that finding, but it also helps explain it. We found that 20 percent of incumbent prosecutors were women.148 The incumbency advantage coupled with lengthy tenure poses an obstacle to diversifying the nation’s prosecutors. That is because few incumbent prosecutors are women and because incumbents serve such long terms. Women were essentially proportionally represented in the pool of candidates challenging incumbents (21 percent) and slightly more so in the pool seeking an open seat (26 percent). Unless and until there is both significant turnover in the ranks of incumbent prosecutors and more female candidates, this large gender gap will persist.

Finally, an underappreciated aspect in the discussion of elected prosecutors is that about ten percent of local prosecutors resigned their office since the latest election. This is more common for those districts that held an election in 2014 (about 13 percent), rather than 2016 (about five percent), presumably because there was more time to prompt the need for an appointment.

We unfortunately do not know the reason for each appointment, nor did we systematically collect this information. When we did record a reason for appointment, a common reason was that the prosecutor had been elected or appointed as a judge.149 This is consistent with a recent report by the Cato Institute, which found that more than 38 percent of current federal judges are former prosecutors.150 It is also consistent with Jed Shugerman’s work on the “prosecutor politician.”151 Shugerman hypothesizes that the prosecutor’s office emerged in the last century or so as a stepping stone for higher office.152 Our study reveals that higher office is not the only reason that prosecutors failed to complete their terms. Some vacancies were caused by illness or death.153 One prosecutor resigned while publicly proclaiming that the salary provided for the position was too low.154

It is also important to note that when a prosecutor steps down, those who are appointed to complete the term often seek reelection to the position. Those appointed prosecutors then enjoy the advantages of incumbency during the election. Our data does not fully capture how often appointed prosecutors go on to win a subsequent election. But we have reason to believe that this happens with some regularity. For example, our study recorded that at least nine of the 44 incumbents who ran for Prosecuting Attorney in West Virginia during the 2016 election cycle began their service as an appointee serving out the remainder of another prosecutor’s term.

IV.     Prosecutorial Elections, Population, and the Supply of Lawyers

Our study confirms the conventional wisdom that incumbents, when they run, virtually always win.155 This finding is critical. It tells us that the most powerful actors in the criminal justice system face almost no threat of losing their office in an election. But this finding alone does not tell us that elections do not allow voters to hold their sitting prosecutors accountable—a claim that is oft-repeated in the academic literature.156 If voters are not given a choice in an election—that is to say, if their incumbent prosecutor is constantly running unopposed—then it would be fair to say that elections do not allow voters to hold their prosecutors accountable.157 But the mere fact that prosecutors are almost always being reelected, standing alone, does not tell us that. A prosecutor might be reelected because more of her constituents voted for her than for a challenger. In that instance, we can’t say that voters were not able to hold the prosecutor accountable. But when voters do not have a choice in their elections—that is to say, when elections are uncontested—then it is appropriate to say that voters do not have an opportunity to hold their prosecutors accountable.

Discussions of reform might lead us to reconsider the current model —direct election of prosecutors for four-year terms in partisan elections.158 Indeed, a number of academics have argued that prosecutors ought to be appointed, rather than elected.159 Of course, an appointments-based model raises its own problems of patronage and lack of accountability.160 In any event, elections without contestation are the worst of both worlds. And so we agree with Wright (and others) that uncontested elections—especially uncontested elections over multiple election cycles—are a serious problem. Therefore, our discussion of reforms will focus on how to increase the number of contested prosecutor elections.

A.     Contestation Rates and Population

One clear finding from our study was that elections looked very different depending on the size of jurisdiction. In general, the larger the population in the jurisdiction, the more likely that the election was contested. Compare the 26 percent contestation rate in the districts with the smallest population with a 65 percent rate in the largest districts.161 Indeed, when it comes to open-seat elections, we found that every single jurisdiction with a population of more than one million held a contested election—a 100 percent contestation rate.162 Contested elections were also more likely to be competitive if the population of the jurisdiction was larger. Contested elections in the smallest districts were competitive 59 percent of the time, as compared to 75 percent in the largest districts.163

The difference between large and small population districts is so stark that we might better think of small-population and large-population prosecutors as being different offices. The significant differences between these offices extend beyond elections. Large population prosecutors preside over large offices containing multiple supervising attorneys, more than a hundred career attorneys, and an equivalent number of non-attorney support staff.164 In contrast, small-population offices often have only a few additional attorneys and support staff.165 The budgets and caseloads of these offices are also dramatically different.166

When thinking about large-population and small-population prosecutors, we should keep in mind that very few Americans live in small-population districts. While there are more than 2,300 elected prosecutors across the country, more than a quarter of the entire U.S. population lives in only 35 prosecutor districts. And more than half of the entire U.S. population lives in only 146 prosecutor districts.167 The largest of those districts is Los Angeles County, California, which has a population of nearly ten million. In contrast, the smallest prosecutor district is Arthur, Nebraska, which has a population of 460.168 Figure 1 gives a visual representation of this skew, where each district is plotted as a point and arrayed in order of smallest to largest. The two dotted boxes visually differentiate two sets of districts, each of which represent half of the U.S. population.

Figure 1. Population Skew in Districts


As the figure illustrates, there are far more small-county prosecutors than large-county prosecutors. There are more than 1,700 local prosecutor offices in jurisdictions with less than 100,000 people, as compared to only 43 prosecutor offices in jurisdictions with more than one million people. And so, in discussing whether prosecutor elections provide voters with a robust opportunity to alter criminal justice policy, it may make less sense to speak in terms of the percentage of contested elections, and more sense to speak in terms of percentage of voters who had choices in those elections.169 Imagine, for example, a state with five districts: four of those districts are rural and they almost never see a contested prosecutor election, but the fifth district is an urban area, and the election for that prosecutor’s office is routinely contested. If 80 percent of the people who live in that state live in the urban district, then electing a new prosecutor in that district would have a significant effect on the state. The lack of contestation in the remaining districts, while still worthy of study and concern, would be less consequential for those who view criminal justice reform as a national or state (rather than only a local) issue.

This population skew explains why criminal justice reform advocates have largely concentrated their election efforts on urban and suburban districts. To the extent that urban and suburban prosecutors are responsible for such a large portion of the population, it makes sense to focus on those elections. But if the reason for electing progressive prosecutors is to reverse mass incarceration, then focusing on urban and suburban races will ultimately be insufficient. That is because, as Table 6 showed, these smallest districts have the most worrisome incarceration trends. The large districts in which progressive reformers are targeting elections are the districts that have already begun to reduce their incarceration rates. And the smaller districts—the districts that reformers are largely neglecting and the districts that are least likely to have contested elections—continue to see incarceration rates rise.

Another clear finding of our study is that the problem of uncontested elections is not simply a function of incumbency advantage. Although elections are more likely to be contested if an incumbent decides not to run, half of all open seats—that is, elections without an incumbent—had only a single candidate in both the general election and the party primary.170 To be sure, this problem varies significantly based on population. All open-seat elections in districts with more than one million people were contested.171 But in districts with 250,000 to 1,000,000 inhabitants, open seats were uncontested nearly 20 percent of the time.172 To put that in context, a recent study of municipal elections in cities across the country found that incumbents ran unopposed only 15 to 20 percent of the time, and that uncontested open seats were exceptional.173 Unsurprisingly, contestation in open-seat elections was lowest in the smallest jurisdiction. More than 50 percent of open-seat elections in jurisdictions with under 100,000 inhabitants were uncontested.174

B.     The Supply Problem Hypothesis

These two findings—namely, that contestation increases as population increases and when incumbents do not run for reelection¾led us to hypothesize that prosecutor elections suffer from a supply problem. That is to say, prosecutor elections may be uncontested because it is difficult to find candidates to run.

We are not the first to suggest that there is a problem with the supply of prosecutor candidates. Ron Wright has concluded that uncontested prosecutor elections are attributable to an insufficient pool of interested candidates. He studied 54 contested elections and created professional profiles of the challengers. Approximately 20 percent of those challengers worked in the incumbent’s office at the time of the election, while the rest largely worked as defense attorneys. As Wright noted, these challengers would “pay a price well beyond election day” if they failed to defeat the incumbent: Those working in the incumbent’s office would either have to leave the office or face professional stagnation, while those working as defense attorneys run the risk of getting less favorable treatment from the incumbent in plea negotiations and other matters.175

Wright’s analysis is quite persuasive. But we believe that interest is only part of the story. It does not, for example, explain why so many open seats are uncontested. Nor does it explain those districts where not even a single candidate ran for local prosecutor. Our nationwide study revealed 16 jurisdictions that were unable to recruit even a single person to run for local prosecutor, instead requiring the appointment of a local prosecutor.176 In an additional four counties, no candidates registered to run and the local prosecutor election ended up being decided by write-in votes.177 None of these 19 counties without a candidate has a population larger than 7,000, and nearly half of them have fewer than 1,000 inhabitants.

We hypothesized that the supply of prosecutor candidates is not merely a lack of interested candidates, but also a lack of eligible candidates. Prosecutors—unlike city councilmembers, school board members, and other traditionally local offices—must be licensed to practice law in order to run for office. The supply of prosecutors is limited, in most jurisdictions, to the active attorneys who live in the district.178 While that might not seem like a significant barrier to entry in the largest districts, the paucity of lawyers in rural areas is one of several access to justice problems facing the country.179 We suspected that the supply problem was most acute in small population districts—we assumed the smaller the population a district, the smaller the pool of qualified candidates.

Conversations with officials in small-population districts initially alerted us to the supply problem. For example, the only person interested in serving as Prosecuting Attorney in Custer County, Idaho had graduated from law school and taken the bar examination. But because she failed the bar exam, she could not serve as Prosecuting Attorney. There were no other attorneys within the county itself.180 Similarly, Hamlin County, South Dakota has not held a State’s Attorney election for “many years.”181 Only one attorney lives in the county, but he works for a bank and is uninterested in the position.

We sought to confirm our supply problem hypothesis by determining the number of active attorneys in each district, as measured by their registered address with the state bar. Because this data was often not readily available, we collected it only for a subset of states. We made a special effort to obtain this data in the six states with districts that did not have a single candidate run for local prosecutor, and we were ultimately able to obtain it for five of those six states. We supplemented this data with data from an additional five states that made the information easy to obtain.182

The data from these ten states confirms that many districts have an insufficient supply of qualified candidates for local prosecutor—that is to say, they have very few inhabitants who are active members of the bar. Table 8 reports the number of lawyers per district by percentile.183 The first row shows that of the 563 districts for which we have data, a quarter of them have six lawyers or less (the 25th percentile) and half of them have 31 lawyers or less (the 50th percentile). The remaining rows break down the number of lawyers per district by the population of the district, making clear that the lack of lawyers is a feature of small-population districts. In these districts with fewer than 100,000 inhabitants, a quarter of districts have four lawyers or less (the 25th percentile) and half of these districts have 16 lawyers or less (the 50th percentile). However, because we made a special effort to obtain data from states with districts that were unable to field candidates,184 Table 8 likely overstates the number of districts with small lawyer populations nationwide.185

Table 8. Number of Lawyers per District


Nonetheless, consistent with our hypothesis, the number of lawyers in these small population districts is correlated with the percentage chance that either the primary or the general election is contested. Table 9 focuses on districts with less than 100,000 inhabitants and groups each district into four approximately equal bins: districts with fewer than five lawyers, between five and 15 lawyers, between 16 and 45 lawyers, and more than 46 lawyers. Of the 114 small districts with fewer than five lawyers, less than ten percent had a contested election. By contrast, small districts with slightly more lawyers, between five and 15, had almost a 20 percent chance of contestation, and small districts with around 50 or more lawyers had almost a 30 percent chance of contestation.

Table 9. Contestation in Small-Population Districts by Number of Lawyers


C.     Possible Solutions to the Supply Problem

One possible solution to the acute supply problem would be to modify or abolish residency restrictions, adding a state-wide element to the office of local prosecutor. Of the 563 districts we studied, 19 had no active members of the bar living within their boundaries. Nebraska, for example, has 12 counties without any lawyers living within their borders. But the state permits people who live in adjoining counties to stand for election as county attorney in those counties that do not have any cities within their borders.186 That residency exception allowed six Nebraska counties with no lawyer residents to nonetheless elect a local prosecutor. While these residency exceptions appear moderately successful in getting an out-of-district lawyer to run for prosecutor, they do not appear well-suited to increasing the number of contested elections.

Another possible solution to the supply problem would be to aggregate counties into larger districts. Fifteen states have adopted this approach, creating districts out of one or more counties.187 In addition, Texas uses a mixture of counties and districts as the geographic unit for local prosecutors, and it creates some of those districts by not just combining counties, but also splitting them.188 But, as noted above, not all states with multi-county districts have high contestation rates.

This aggregation helps to avoid districts with very low numbers of attorneys. For example, Kentucky consolidates its 120 counties into 57 districts.189 Kentucky has 20 counties with fewer than ten lawyers and an additional 36 counties with between 10 and 19 lawyers. But only one of its consolidated districts has fewer than 20 lawyers.

South Carolina’s aggregation is even more dramatic. While both Kentucky and South Carolina have approximately 4.5 million inhabitants, South Carolina has only 46 counties, as compared to Kentucky’s 120. South Carolina aggregates counties to create 16 “judicial districts.”190 The difference in variation for the population per elected prosecutor is stark. Kentucky prosecutors serve jurisdictions as small as 25,000 people,191 while the smallest district in South Carolina has more than 130,000 inhabitants. South Carolina has, on average, one prosecutor for every 289,000 people, while Kentucky has one for approximately every 79,000 people.

As Kentucky and South Carolina illustrate, consolidated districts can eliminate some of the smallest prosecution offices and increase the average population within districts. But comparing Kentucky and South Carolina shows that states must consider not only whether to aggregate counties to create districts, but also how to aggregate. South Carolina’s decision to create 16 districts, as compared to Kentucky’s 57, resulted in dramatically different numbers of attorneys within districts. Kentucky has one district with fewer than 20 lawyers and 16 districts with fewer than 50 lawyers. In contrast, South Carolina has at least 120 lawyers living in each of its districts.

Given the apparent relationship between the number of lawyers in a district and contested elections reflected in Table 9, it seemed plausible that states with large numbers of lawyers in each district would have much higher rates of contested elections. But the effect was not as strong as we expected. To return to the comparison between Kentucky and South Carolina, we do see a difference between the percentage of contested elections in those states: South Carolina had contested elections in 25 percent of its districts, while Kentucky had contested elections in only 19 percent.192 But that difference is not as large as we expected given the significantly larger supply of lawyers in South Carolina’s districts.

Our study of these ten states also showed us that increasing the population within a district will not always dramatically increase the number of lawyers in that district. Our subset of states showed that districts with the smallest populations tended to have very small numbers of lawyers living within their boundaries, and districts with the largest populations tended to have much larger numbers of lawyers. But the relationship between population and number of lawyers is not linear, presumably because lawyers tend to cluster geographically. Some counties have more attorneys living within their boundaries than counties with larger populations.193 What is more, some states simply have smaller numbers of lawyers relative to their overall population. For example, New York has one lawyer for every 153 inhabitants, as compared to Nebraska, which has only one lawyer for every 339 people.194 While the ultimate effect may be modest, aggregating counties—especially rural counties—provides the simplest solution to the supply problem. And the more aggressive states are in their aggregation choices, the less of a problem the supply of lawyers will pose.

But aggregation comes at the cost of local control. One benefit of smaller jurisdictions is that they allow for greater participation and control by inhabitants.195 Imagine, for example, a district attorney in a county with a population of less than 10,000 people. If that district attorney received phone calls from ten constituents arguing in favor of a particular enforcement priority, then he or she would likely take that concern seriously. Now imagine that a district attorney in a jurisdiction of two million inhabitants received the same ten phone calls. The phone calls are likely to have a much smaller impact in the large-population jurisdiction than the small-population jurisdiction.

Political scientists and those who study local government law have long discussed the tradeoffs associated with jurisdiction size.196 As a general matter, the tradeoff is between participation and control on the one hand, and resources on the other. People can participate more in small, local governments; but the smaller the government, the fewer resources are available.

The lack of resources in smaller jurisdictions is ordinarily discussed in terms of economic or physical resources, such as funding for a water treatment plant197 or for high quality libraries and museums.198 Our study highlights a lack of human resources—namely, qualified candidates for a particular office—in smaller jurisdictions. The insufficient supply of active attorneys in smaller jurisdictions is yet another resource tradeoff in jurisdiction size—one that matters for prosecutor elections.

We are not the first to discuss the question of jurisdiction size in prosecutor elections. A desire to increase control and participation prompted other commenters to criticize certain urban districts for being too large and thus diluting the voting power of African Americans and other racial minorities. Specifically, John Pfaff has argued for dividing certain prosecutorial districts to give greater local control to urban voters.199 Bill Stuntz pointed to the decline of local control—specifically to the rise of white suburban control of district attorneys elected in “metropolitan counties [which] typically include both cities and close-in suburbs”—as one of several causes of harsh and racially disparate criminal justice policy in the late twentieth century.200 These arguments are reminiscent of a more general argument in favor of local control—namely, that groups in larger jurisdictions would constitute only minorities can rule as majorities under local control.201

We do not have a definitive answer about the optimal-sized jurisdiction for the selection of prosecutors, but there are several reasons why we are skeptical that electing prosecutors on a county-by-county basis is the best choice. For one thing, our study makes clear that, on average, the smaller the district population, the less likely it is to have a contested election.202 While sparsely populated counties doubtlessly give constituents more “voice,”203 they are also less likely to give them a meaningful choice in a prosecutor election —most of those counties only have one candidate running on the ballot. And in some counties, they have no candidates on the ballot.204

For another, even when counties select their own prosecutors, there is still a non-trivial amount of centralized control over criminal justice issues. Even today, states continue to play a role in the setting of local criminal justice policy. Not only are criminal codes enacted by state legislatures, but “virtually every state has a law that empowers the supersession of local prosecutors by state officials.”205

In addition, there is nothing inherent about the structure of the county that makes it an optimal jurisdiction for selecting prosecutors. Local prosecutors began as officers appointed largely at the state level.206 And some of the nineteenth century conventions that chose single-county rather than multi-county prosecutors did so not for some grand reason about the nature of community but because they feared it would take too long to travel between counties, presumably by horse.207 Those same considerations seem far less persuasive today.208

More fundamentally, as local government scholars have noted, counties are artificial constructs. County boundaries were often drawn before settlement patterns emerged, and county governments exist largely to help administer certain services.209 As a consequence, county boundaries do not delineate community ties, and inhabitants rarely feel any sort of attachment to or derive any identity from their county.210 And so there may not be significant resistance if the county stopped being the relevant jurisdiction for prosecutors.

Even if states were to aggregate counties into larger districts, that would not guarantee we will have high rates of contested elections. The supply of prosecutor candidates is sometimes low even in large-population districts. To be sure, in large-population districts the supply problem looks different than in small-population districts: Large-population districts have a large pool of eligible candidates—that is, people who live in the district and who are admitted to practice law in the jurisdiction—but they do not have a reliable supply of candidates who are willing to run.

This is Wright’s account—that not enough people are interested in running for local prosecutor.211 This account likely explains at least some of the uncontested races in jurisdictions with large numbers of attorneys. We have seen accounts of assistant district attorneys who were terminated because they were going to run against their boss.212 But we have also seen challengers come from within an office—and remain working there during the campaign.213

But the theory that challengers can’t afford to upset incumbent prosecutors does not explain why so many open seats remain uncontested. Even in the districts with 250,000 to one million people—districts where we know that there are large numbers of people admitted to the bar and who possess relevant practice experience—these open seats are uncontested nearly 20 percent of the time.214 Even when they are contested, the lack of competition in these elections suggests a potential lack of high quality candidates.

Those in the progressive prosecutor movement are attempting to recruit challengers.215 And some of those recruits have gone on to win. Take, for example, the astounding election of criminal defense attorney and civil rights lawyer Larry Krasner as district attorney of Philadelphia. As someone who was known for suing the police, “Krasner was the unlikeliest of candidates for district attorney in Philadelphia.”216 Krasner won despite vocal opposition from law enforcement and despite receiving no endorsements from local newspapers.217

While progressive prosecution advocates often hold up Krasner’s victory as a shining example of progressive victory,218 the circumstances of Krasner’s election may also suggest reasons to be cautious about how successful such candidates are likely to be in the future. Philadelphia is not your typical jurisdiction. Philadelphia is a “consolidated city-count[y],” which means that the city’s boundaries and the county’s boundaries are the same.219 Philadelphia is also a heavily Democratic city.220 And so, although Krasner defeated a Republican candidate in the general election, his victory was all but guaranteed after he won the Democratic primary.

Krasner’s was also not your typical election. The incumbent district attorney, Seth Williams, resigned in the face of corruption charges; and the person who was appointed interim DA chose not to run.221 The open seat drew seven candidates in the Democratic primary, including Krasner.222 A seven-person primary is highly unusual. And although Krasner beat the field by 18 points, he still received less than 40 percent of the vote.223 Perhaps most importantly, Krasner benefitted tremendously by nearly $1.5 million in campaign spending from George Soros,224 who has begun to finance a number of progressive prosecutor candidates in recent years.225 That spending gave him a distinct advantage as compared to other candidates.

While the crowded primary distinguishes Krasner’s victory from some of the other progressive prosecutor elections, there are also clear similarities. The progressive candidates who have prevailed in prosecutor elections have, as a group, taken advantage of the typically low resource environments for these races. Take, for example, when Eric Gonzalez beat out a number of candidates to become the district attorney in Brooklyn, he did so by raising more than six times more money than any other candidate.226 Similarly, when Scott Colom unseated long-time incumbent Forrest Allgood in Mississippi’s 16th District, Colom raised more than twice as much as Allgood in campaign contributions.227

The progressive candidates also changed the nature of campaign rhetoric. Until recently,228 modern prosecutor elections have tended to focus on the candidates’ individual qualifications, rather than the performance of the office.229 And when campaign rhetoric did include office results, those results were framed almost entirely in terms of conviction rates or the quantity of cases processed rather than “the practices of the office related to plea bargaining or screening of cases that the local police recommend for prosecution.”230 But the progressive candidates ran on policies, not just personalities and conviction rates, advocating for reducing racial disparities in sentencing and diversion programs for drug offenders.231 In other words, progressive candidates literally changed the public discussion surrounding prosecutor campaigns.

But neither of these tactics—increased resources and substantive campaign rhetoric—are inherently progressive. Candidates who do not support progressive criminal justice reforms can also raise money, and they can also change the rhetoric of their campaigns.232 Indeed, this seems to have occurred in the 2018 election cycle, in which several Soros-backed progressive candidates lost primary elections in California. For example, law enforcement groups gave hundreds of thousands of dollars to support Anne Marie Schubert in the 2018 election cycle.233 Schubert, the incumbent Sacramento district attorney, used those funds to successfully fight off a progressive primary challenger who had received financial support from Soros and other progressive prosecutor groups.234 And Summer Stephan, the incumbent prosecutor in San Diego, soundly defeated a progressive challenger who received millions of dollars in support from Soros by painting the challenger as “anti-prosecutor” and claiming that her “reform agenda endangered public safety.”235

Put simply, the same tactics that have proven effective at electing progressive prosecutors could equally support “law and order” prosecutors.236

What is more, there is some reason to think that more contested and competitive elections might lead prosecutors to behave more punitively—that is, to dismiss fewer charges and send more people to prison. There is a small, but growing literature that identifies an “election effect” on prosecutors. That literature documents changes in prosecutorial behavior in the time period leading up to elections. The election effect is premised on the “political business cycle theory,”237 which tells us that elected officials alter their behavior in advance of an election in order to maximize their chances of reelection. The theory suggests that voters evaluate prosecutors based on the convictions they obtain, and a prosecutor will be deemed more successful if she obtains more convictions and more punishments.238 And at least one study documented stronger election effects in districts that have historically seen more electoral competition.239

Importantly, not all studies of the election effect have found that prosecutors change their behavior in advance of elections.240 Even if the election effect does cause prosecutors to act more punitively in the time period leading up to an election, our findings suggest that it is the prosecutors who are least likely to face challengers—those in small and rural jurisdictions—who act more punitively as a general matter. In other words, in our study, those prosecutors who faced the smallest chances of a contested election sent a higher percentage of people to prison. In any event, the elections effect literature is still too small to draw any firm conclusions. But our broader point remains—a contested election will give voters a choice about who will serve as their local prosecutor, but it does not tell us who voters will choose.

V.     Conclusion

The Jacksonian era ushered in a sea change in the nature of the local prosecutor. Partly in response to the scourge of political patronage, the states remade the office into a uniquely American institution—a locally elected office. If the antebellum effort in popular democracy can be understood as essentially a defensive one to protect and insulate local governance from statewide political power,241 the rise of mass incarceration has highlighted how “[o]ver the last forty years, prosecutors have amassed more power than our system was designed for.”242

Today, our scourge is mass incarceration. As Marie Gottschalk has chronicled, “[t]he U.S. penal system has grown so extensive that it has begun to metastasize . . . alter[ing] how key governing institutions . . . operate . . . .”243 But we have largely not revisited how we pick our prosecutors.244 This is despite the fact that our method of selection for other local criminal justice actors is quite different.245

The movement for progressive prosecution might change that. Although it may be radical in its ends––to dismantle the carceral state—it has so far been realistic, even ordinary, in its means—to garner more votes for different prosecutors. The movement has helped to focus greater attention not only on the enormous discretion prosecutors wield, but also on the fact that most prosecutors are selected through local elections. By taking advantage of the traditionally low resource environment of these campaigns, progressive prosecutors have been able to defeat several notable incumbents. The energized movement may also prompt a renewed debate about the deeper politics of criminal law, including whether to elect prosecutors at all and, if so, how to best do that.

More work needs to be done on this local dimension of mass incarceration. To that end, this Article offers a more complete and nuanced understanding of how we pick prosecutors today.



Table 10. Contestation and Competition by State



Table 11. Expanded Version of Contestation and Competition in Prosecutorial Elections


  1. [1]. See, e.g., Marie Gottschalk, The Prison and the Gallows: The Politics of Mass Incarceration in America 1 (2006) (introducing the term and offering a comparative analysis of the U.S. incarceration rate at the time of 714 per 100,000 people).

  2. [2]. Id. at 19.

  3. [3]. See John F. Pfaff, Locked In: The True Causes of Mass Incarceration—and How to Achieve Real Reform 13 (2017) (“A major barrier to reform ... is the fractured nature of our criminal justice system.”).

  4. [4]. For background on incarceration rates, particularly how growth in the federal and state prison populations, local jail populations, and probation and parole populations began in the 1970s but vastly accelerated in the 1980s and 1990s, see Nat’l Research Council, The Growth of Incarceration in the United States: Exploring Causes and Consequences 39 fig.2-3, 41 fig.2-4, 43 fig.2-5 (Jeremy Travis et al. eds., 2014).

  5. [5]. See, e.g., Michelle Alexander, The New Jim Crow: Mass Incarceration in the Age of Colorblindness (2010).

  6. [6]. See, e.g., Michael Tonry, Punishing Race: A Continuing American Dilemma (2011).

  7. [7]. See Vesla M. Weaver, Frontlash: Race and the Development of Punitive Crime Policy, 21 Stud. Am. Pol. Dev. 230 (2007).

  8. [8]. See, e.g., Naomi Murakawa, The First Civil Right: How Liberals Built Prison America (2014).

  9. [9]. See, e.g., James Forman Jr., Locking Up Our Own: Crime and Punishment in Black America (2017).

  10. [10]. See William J. Stuntz, The Collapse of American Criminal Justice (2011).

  11. [11]. See, e.g., Ruth Wilson Gilmore, Golden Gulag: Prisons, Surplus, Crisis, and Opposition in Globalizing California (2007).

  12. [12]. See, e.g., Peter K. Enns, Incarceration Nation: How the United States Became the Most Punitive Democracy in the World (2016).

  13. [13]. See, e.g., Gottschalk, supra note 1.

  14. [14]. The lack of attention to the local dimension of mass incarceration is likely due to a lack of available data. Most of the data presented on the criminal justice system is at the national or state level. This is one reason why experts convened by the National Academy of Science have suggested that a bold research agenda for the field requires “significant new data collection.” See Nat’l Research Council, supra note 4, at 431.

  15. [15]. For example, prosecutors play only a cursory role in the recent tome by experts convened by the National Academy of Science. See id. at 336, 343, 347, 356 (mentioning prosecutors only a handful of times in this over 400-page report); see also Pfaff, supra note 3, at 127–28, 127 n.3 (making this observation as well).

  16. [16]. See, e.g., Emily Bazelon, Charged: The New Movement to Transform American Prosecution and End Mass Incarceration (2019); Pfaff, supra note 3; Stuntz, supra note 10; Stephanos Bibas, Prosecutorial Regulation Versus Prosecutorial Accountability, 157 U. Pa. L. Rev. 959 (2009); Angela J. Davis, The Progressive Prosecutor: An Imperative for Criminal Justice Reform, 87 Fordham L. Rev. Online 8 (2018); Adam M. Gershowitz, Consolidating Local Criminal Justice: Should Prosecutors Control the Jails?, 51 Wake Forest L. Rev. 677 (2016); Erik Luna & Marianne Wade, Prosecutors as Judges, 67 Wash. & Lee L. Rev. 1413 (2010). But cf. Jeffrey Bellin, Reassessing Prosecutorial Power Through the Lens of Mass Incarceration, 116 Mich. L. Rev. 835, 845 (2018) (“Contrary to the consensus, prosecutors are not the most powerful actors in the criminal justice system.”).

  17. [17]. See, e.g., E. Ann Carson, U.S. Dep’t of Justice, Bureau of Justice Statistics, Prisoners in 2016, at 3 tbl.1 (2018), available at p16.pdf [] (comparing the number of prisoners under the jurisdiction of state and federal correctional authorities for the period 2006 to 2016); Patrick A. Langan et al., U.S. Dep’t of Justice, Bureau of Justice Statistics, Historical Statistics on Prisoners in State and Federal Institutions, Yearend 1925–86, at 15 tbl.3 (1988), available at [ AAM9-4TH8] (reporting the total number of sentenced prisoners in state and federal institutions for the period 1925 to 1986).

  18. [18]. William J. Stuntz, The Pathological Politics of Criminal Law, 100 Mich. L. Rev. 505, 506 (2001).

  19. [19]. See Pfaff, supra note 3, at 133; see also Bibas, supra note 16, at 960 (“No government official in America has as much unreviewable power and discretion as the prosecutor.”).

  20. [20]. E.g., David Brooks, Opinion, The Prison Problem, N.Y. Times (Sept. 29, 2015), [https://]; Eric Gonzalez & Miriam Krinsky, Opinion, How a New Generation of Prosecutors Is Driving Criminal Justice Reform Outside of Congress, Hill (Feb. 26, 2018, 3:40 PM), []; Justin Jouvenal, From Defendant to Top Prosecutor, This Tattooed Texas DA Represents a New Wave in Criminal Justice Reform, Wash. Post (Nov. 19, 2018, 5:25 PM), /e1dca7cc-d300-11e8-83d6-291fcead2ab1_story.html []; Juleyka Lantigua-Williams, Are Prosecutors the Key to Justice Reform?, Atlantic (May 18, 2016), https:// 3252 []; Brandon E. Patterson, Philadelphia’s District Attorney Just Showed America’s Prosecutors How to End Mass Incarceration, Mother Jones (Mar. 15, 2018), []; Taylor Pendergrass, How Bad Prosecutors Cause Bad Policing, Slate (Aug. 16, 2016, 2:09 PM), html []; Jeffrey Toobin, The Milwaukee Experiment: What Can One Prosecutor Do About the Mass Incarceration of African-Americans?, New Yorker (May 4, 2015), [https://].

  21. [21]. For example, the Real Justice Political Action Committee was created in February 2018 “to help elect ‘reform-minded prosecutors’ at the county and city levels” by providing significant financial support to “campaigns by progressives running for district attorney offices.” Daniel Marans, Black Activist Starts Group that Aims to Elect Progressive Prosecutors, HuffPost (Feb. 15, 2018, 1:35 PM), b64ee4b0058d55670e4f []. Another political action committee, Color of Change, and billionaire George Soros have also spent significant amounts of money to elect progressive prosecutors. See Matt Ferner, George Soros, Progressive Groups to Spend Millions to Elect Reformist Prosecutors, HuffPost (May 12, 2018, 7:00 AM), /george-soros-prosecutors-reform_n_5af2100ae4b0a0d601e76f06 []. “[T]he ACLU launched Vote Smart Justice, a nonpartisan voter education drive to give Americans information about where candidates for state and federal office stand on key criminal justice reform issues, like bail reform, the war on drugs, and police accountability.” Udi Ofer, Common Wants You to Vote Smart Justice in 2018, ACLU (Oct. 11, 2018, 10:30 AM), [https://]. Although the ACLU does not endorse candidates, the voter education initiative has as its explicit goal to “cut the incarceration rate in this country by 50 percent and reduce the racial disparities in our prisons and jails.” Id.

  22. [22]. See Liane Jackson, Change Agents: A New Wave of Reform Prosecutors Upends the Status Quo, ABA J. (June 1, 2019, 12:00 AM), agents-reform-prosecutors [] (identifying several newly-elected prosecutors as “progressive,” including Melissa Nelson, a Florida prosecutor who is a Republican).

  23. [23]. E.g., Lars Trautman, Opinion, The Criminal Justice Reforms Pushed by ‘Progressive Prosecutors’ Are Surprisingly Conservative, Wash. Examiner (Nov. 26, 2019, 6:00 AM), https://www. -are-surprisingly-conservative [] (“Vast swathes of the country have no interest in anything remotely associated with the phrase ‘progressive.’ Thus, prosecutorial reform will only reach at least half the country if it has a more conservative cast and bent.”).

  24. [24]. See Note, The Paradox of “Progressive Prosecution, 132 Harv. L. Rev. 748, 751–52 (2018) (“Fundamentally, progressive prosecutors seek to rebalance the use of prosecutorial discretion. Where traditional prosecutors have used their enforcement powers in a heavy-handed manner to punish marginalized individuals, progressive prosecutors institute practices that pull back on those punitive measures, or, at least, divert them. And where traditional prosecutors have refused to exercise their expansive powers to hold police accountable for misdeeds, progressive prosecutors (sometimes) actively prosecute police officers.”).

  25. [25]. See Bazelon, supra note 16, at xxv.

  26. [26]. See, e.g., Marie Gottschalk, Caught: The Prison State and the Lockdown of American Politics 267 (2015) (“[G]etting deeply involved in electoral contests for local district attorneys ... should be a top priority for civil rights and other groups committed to dismantling the carceral state. If the aim is to shift penal policy in a less punitive direction, these local electoral contests are arguably as important—or even more important—than mobilizing for the quadrennial presidential elections.”); see also Bibas, supra note 16, at 983 (“[M]ost chief prosecutors are elected at the county, judicial circuit, or district level. Assuming that head prosecutors want to win reelection and move on to higher office, they should tailor their behavior to serve voters’ interests and opinions.” (footnote omitted)).

  27. [27]. See Bazelon, supra note 16, at 81–88.

  28. [28]. E.g., About Color of Change, Color Change, []; Smart Justice, ACLU, [].

  29. [29]. See Steve Schmadeke, Newly Elected Kim Foxx Details Plans to Reshape State’s Attorney’s Office, Chi. Trib. (Dec. 5, 2016, 7:39 PM), [].

  30. [30]. See Chris Brennan & Aubrey Whelan, Larry Krasner Wins Race for Philly DA, Phila. Inquirer (Nov. 7, 2017), [].

  31. [31]. See Evan Sernoffsky, Chesa Boudin, Reformer Public Defender, Wins Election as San Francisco’s New DA, S.F. Chron. (Nov. 9, 2019, 5:54 PM), /article/Chesa-Boudin-reformer-public-defender-wins-14823166.php [].

  32. [32]. See Maria Cramer & Jackson Cote, Rachael Rollins Wins Suffolk DA Race, Bos. Globe (Nov. 7, 2018, 12:21 AM), for-decided-mass-counties/ZqYhxN69Yx6hOwHPGp7ZtJ/story.html [].

  33. [33]. See, e.g., Matt Ferner, Progressive Prosecutors Win Primaries in North Carolina, HuffPost (May 8, 2018, 11:03 PM), _n_5af22ea3e4b0aab8a78a067e?guccounter=1 [] (detailing the North Carolina elections of “Satana Deberry in Durham County and Faris Dixon in Pitt County”); Justin Jouvenal & Rachel Weiner, Progressive Challengers Oust Longtime Prosecutors in Northern Virginia, Wash. Post (June 11, 2019, 9:03 PM), local/public-safety/northern-virginia-democrats-to-deliver-referendum-on-criminal-justice-reform /2019/06/11/1fbbcd2a-8bac-11e9-8f69-a2795fca3343_story.html [] (recounting the “hotly contested Democratic primaries for prosecutor in Arlington and Fairfax counties”).

  34. [34]. Stuntz, supra note 18, at 510 (introducing the term “[s]urface politics” and defining it as “the sphere in which public opinion and partisan argument operate, ebb and flow”); see Daniel Gotoff & Celinda Lake, Voters Want Criminal Justice Reform. Are Politicians Listening?, Marshall Project (Nov. 13, 2018, 7:00 AM), 2018/11/13/voters-want-criminal-justice-reform-are-politicians-listening [ J8AD-CHXB] (summarizing broad support for state-level criminal justice reforms during the 2018 election). But see David Alan Sklansky, The Changing Political Landscape for Elected Prosecutors, 14 Ohio St. J. Crim. L. 647, 667–68 (2017) (“Specific, local circumstances played a part in many of these election results: the corruption allegations against Charles Hynes in Brooklyn, the inflammatory rhetoric of Forrest Allgood in Mississippi, the family connections of Marco Serna, the high-profile shootings by the police in Chicago and Cleveland, and the opposition of gun rights activists to Angela Corey. All of these factors make it hazardous to draw broad conclusions from this relatively small set of races.”); id. at 672 (“The elections I have discussed are, to a great extent, exceptions that prove the rule ... .”).

  35. [35]. Stuntz, supra note 18, at 510.

  36. [36]. See Pfaff, supra note 3, at 134 (“[B]y and large, district attorneys are reelected with unfailing regularity. It’s hard to view elections as a way to systematically regulate prosecutorial behavior.”); id. at 139 (“Almost all prosecutors are elected, but those elections often seem like foregone conclusions. Incumbents rarely face challengers, and when they do they usually win.”); Shima Baradaran Baughman, Subconstitutional Checks, 92 Notre Dame L. Rev. 1071, 1103 (2017) (“[I]ncumbent prosecutors rarely face real accountability... . [N]inety-five percent of the incumbents who want to return to prosecutorial office are reelected.” (footnotes omitted)); Stephanos Bibas, Restoring Democratic Moral Judgment Within Bureaucratic Criminal Justice, 111 Nw. U. L. Rev. 1677, 1690 (2017) (“Prosecutorial elections are notoriously uncompetitive ... .”); Stephanos Bibas & William W. Burke-White, International Idealism Meets Domestic-Criminal-Procedure Realism, 59 Duke L.J. 637, 660 (2010) (“[Prosecutor] races are distorted by huge incumbency advantages and driven by occasional scandals and unrepresentative, high-profile celebrity trials.”); R. Michael Cassidy, Character and Context: What Virtue Theory Can Teach Us About a Prosecutor’s Ethical Duty to “Seek Justice, 82 Notre Dame L. Rev. 635, 657 (2006) (“It is exceptionally rare in this country for an incumbent prosecutor to be voted out of office.”); Roger A. Fairfax, Jr., Prosecutorial Nullification, 52 B.C. L. Rev. 1243, 1269 (2011) (“Prosecutorial re-election rates are high and the vast majority of incumbents go unchallenged.”); Eric S. Fish, Against Adversary Prosecution, 103 Iowa L. Rev. 1419, 1478 (2018) (“Several studies suggest that prosecutorial elections are low-information affairs in which incumbency is an overwhelming advantage (indeed, many are uncontested), and in which conviction rates and other metrics of adversarial success bear little apparent relationship to electoral success.”); Janet C. Hoeffel & Stephen I. Singer, Elections, Power, and Local Control: Reining in Chief Prosecutors and Sheriffs, 15 U. Md. L.J. Race, Religion, Gender & Class 319, 323 (2015) (“Unlike sitting mayors, governors, and state and local legislators, the vast majority of incumbent prosecutors and sheriffs run unopposed.”); Bidish Sarma, Using Deterrence Theory to Promote Prosecutorial Accountability, 21 Lewis & Clark L. Rev. 573, 591–92 (2017) (“Prosecutorial elections are historically low-information, low-turnout affairs. Often times, there are no opponents to vote for; even when other candidates materialize, incumbents win so often that ‘retention rates ... would make a candidate for the Supreme Soviet blush.’” (footnotes omitted) (quoting Ronald F. Wright & Marc L. Miller, The Worldwide Accountability Deficit for Prosecutors, 67 Wash. & Lee L. Rev. 1587, 1606 (2010))); Note, Restoring Legitimacy: The Grand Jury as the Prosecutor’s Administrative Agency, 130 Harv. L. Rev. 1205, 1220 (2017) (“[I]ncumbent prosecutors win reelection the vast majority of the time ... .”).

  37. [37]. Bazelon, supra note 16, at 80 (quoting Yale Law Professor David Schleicher).

  38. [38]. See Ronald F. Wright, Beyond Prosecutor Elections, 67 SMU L. Rev. 593, 601–03 (2014) [hereinafter Wright, Beyond Prosecutor Elections] (updated study); Ronald F. Wright, How Prosecutor Elections Fail Us, 6 Ohio St. J. Crim. L. 581, 595–96 (2009) [hereinafter Wright, How Prosecutor Elections Fail Us] (original study); see also Pfaff, supra note 3, at 141, 141 n.46 (describing the Wright study as “[o]ne of the only studies to look at prosecutorial elections”).

  39. [39]. See Wright, Beyond Prosecutor Elections, supra note 38, at 600 n.33, 601 tbl.1, 602 tbl.2, 603 tbl.3 (finding that incumbent prosecutors rarely faced opposition and that prosecutors usually won reelection even when opposed).

  40. [40]. The Wright study covered California, Colorado, Florida, Georgia, Hawaii, Idaho, Indiana, Maine, Massachusetts, Minnesota, New Mexico, North Carolina, Oregon, Texas, and Wisconsin. See id. at 600 n.33. Wright also used vote totals as a proxy for population. But because voter turnout can vary from place to place and election to election, it may not accurately capture population.

  41. [41]. By competitive, we mean the top vote-getter received less than 60 percent of the vote, though we explore alternative definitions as well.

  42. [42]. See, e.g., Shea Denning, New Districts and District Attorneys in 2019, N.C. Crim. L.: UNC Sch. Gov’t Blog (Nov. 27, 2018, 4:10 PM), [] (describing the statutory changes to the prosecutorial districts in the state of North Carolina).

  43. [43]. These jurisdictions had very small populations, and so they were unable to find anyone who was eligible for the job, satisfied the residency restrictions, and who wanted to run. See discussion infra Part III.

  44. [44]. See, e.g., Sklansky, supra note 34, at 651–67 (collecting examples).

  45. [45]. Indeed, this appears to have occurred in the 2018 California primaries. See infra notes 240–42 and accompanying text; see also William L. Riordon, Plunkitt of Tammany Hall: A Series of Very Plain Talks on Very Practical Politics 22–23 (1948) (contrasting the short tenure of reform movements and the endurance of particular interests with the pithy observation that reform movements of the time “were mornin’ glories—looked lovely in the mornin’ and withered up in a short time, while the regular machines went on flourishin’ forever, like fine old oaks”).

  46. [46]. See Jonathan Simon, Beyond Tough on Crime: Towards a Better Politics of Prosecution, in Prosecutors and Democracy: A Cross-National Study 250, 254 (Máximo Langer & David Alan Sklansky eds., 2017) (“The politics of prosecution then consists not mostly of running for office, but in influencing how citizens and other actors in the carceral state imagine crime and the penal system and further in promoting a specific set of normative expectations about what political institutions can do about crime and how they should be judged.”).

  47. [47]. Darryl K. Brown, Free Market Criminal Justice: How Democracy and Laissez Faire Undermine the Rule of Law 28 (2016); Michael J. Ellis, Note, The Origins of the Elected Prosecutor, 121 Yale L.J. 1528, 1530 (2012).

  48. [48]. Bazelon, supra note 16, at 78 (“In a poll paid for by the ACLU, half of sixteen hundred likely voters said they didn’t know the D.A. was elected.”).

  49. [49]. See 28 U.S.C. § 541(a) (2012) (“The President shall appoint, by and with the advice and consent of the Senate, a United States attorney for each judicial district.”).

  50. [50]. See Ellis, supra note 47, at 1537 (“Who appointed the district attorney varied from state to state: in Kentucky and New York, for example, it was the judge of the county court; in Alabama, Georgia, North Carolina, and Tennessee, it was the state legislature; in Massachusetts and New Hampshire, it was the governor, assisted by his council of advisors; in Michigan, the governor with the advice and consent of the state senate.” (footnotes omitted)).

  51. [51]. See id. at 1536 (“[V]oters became dissatisfied with the appointment process. Governors gained new powers, giving one man unchecked appointment authority in many states, while in states where the legislature selected prosecutors, political parties commandeered the appointment process to reward their allies and punish their enemies.”); see also id. at 1531 (“Supporters of elected prosecutors argued that popular election would give citizens greater control over government, eliminate patronage appointments, and increase the responsiveness of prosecutors to the communities they served.”).

  52. [52]. See id. at 1530, 1569 app. (listing the “chronology of the elected district attorney” in various states).

  53. [53]. See, e.g., Steven W. Perry & Duren Banks, U.S. Dep’t of Justice, Bureau of Justice Statistics, Prosecutors in State Courts, 2007—Statistical Tables 1 (2011), available at [] (“State court prosecutors ... handle felony cases in state courts of general jurisdiction.”).

  54. [54]. See id.

  55. [55]. See Rachel E. Barkow, Federalism and Criminal Law: What the Feds Can Learn from the States, 109 Mich. L. Rev. 519, 545 (2011) (“Most states vest state-level prosecutors––typically the attorney general––with exclusive or concurrent jurisdiction over just a handful of areas that repeat themselves in state after state.”).

  56. [56]. See Perry & Banks, supra note 53, at 1–2.

  57. [57]. Miss. Const. art. VI, § 174 (“A district attorney for each circuit court district shall be selected ... .”); see Miss. Code Ann. § 9-3-1 (West 2016) (listing the 82 counties in Mississippi); see also id. §§ 9-7-3 to -57 (setting boundaries for the 22 circuit court districts).

  58. [58]. Compare Miss. Code Ann. § 9-7-23(1) (“The Seventh Circuit Court District shall be Hinds County.”), with id. § 9-7-5 (listing the seven counties in the First Circuit Court District).

  59. [59]. N.Y. County Law § 700 (McKinney 2017).

  60. [60]. For the relevant sources of state constitutional and statutory law, see Alaska Const. art. III, § 25; Ariz. Const. art. XII, § 3; Ark. Const. amend. LXXX, § 20; Colo. Const. art. VI, § 13; Conn. Const. art. XXIII, which amended Article IV to include the process for appointing prosecutors; Del. Const. art. III, § 21; Fla. Const. art. V, § 17; Ga. Const. art. VI, § 8, para. 1; Ill. Const. art. VI, § 19; Ind. Const. art. VII, § 16; Ky. Const. § 97; La. Const. art. V, § 26; Md. Const. art. V, § 7; Mich. Const. art. VII, § 4; Miss. Const. art. VI, § 174; N.M. Const. art. VI, §24; Pa. Const. art. IX, § 4; R.I. Const. art. IV, § 1; Vt. Const. ch. II, § 43; W. Va. Const. art. IX, § 1; Ala. Code § 12-17-180 (LexisNexis 2012); Cal. Gov’t Code §§ 24200–01 (West Supp. 2019); Conn. Gen. Stat. § 51-278 (2019); Del. Code Ann. tit. 29, § 2505 (2017); Idaho Code § 31-5312 (2019); Iowa Code Ann. § 331.751 (West 2013); Kan. Stat. Ann. § 19-701 (2019); Me. Rev. Stat. Ann. tit. 30-A, § 251 (2011); Mass. Gen. Laws Ann. ch. 54, § 154 (West 2007); Minn. Stat. Ann. § 382.01 (West 2015); Mo. Ann. Stat. § 56.010 (West 1998); Mont. Code Ann. § 7-4-2203, -2205 (West 2019); Neb. Rev. Stat. Ann. § 32-522 (LexisNexis 2014); N.H. Rev. Stat. Ann. § 7:33 (2013); N.J. Stat. Ann. § 2A:158-1 (West 2011); N.Y. County Law § 400; N.C. Gen. Stat. § 7A-60 (Supp. 2018); N.D. Cent. Code § 11-08-07 (2012); Ohio Rev. Code Ann. §309.01 (LexisNexis 2016); Okla. Stat. Ann. tit. 19, § 215.20 (West 2018); Or. Rev. Stat. Ann. § 8.610 (West 2019); S.C. Code Ann. § 1-7-310 (2005); S.D. Codified Laws § 7-7-1.1 (2004); Tenn. Code Ann. § 8-7-102 (2016); Tex. Gov’t Code Ann. §§ 43–45 (West Supp. 2019); Utah Code Ann. § 17-53-101 (LexisNexis 2017); Va. Code Ann. § 15.2-1626 (2018); Wash. Rev. Code Ann. §§ 36.16.020 to .030 (LexisNexis 2018); Wis. Stat. Ann. § 978.01 (West Supp. 2018); and Wyo. Stat. Ann. § 9-1-802 (2019).

  61. [61]. We define “local” as a sub-state or district-level office.

  62. [62]. Published work has produced inconsistent accountings of which states use which method. Compare Wright, How Prosecutor Elections Fail Us, supra note 38, at 589 (reporting that Alaska, Connecticut, and New Jersey do not “elect their prosecutors at the local level” but omitting Delaware), with Ellis, supra note 47, at 1530 n.3 (reporting that Connecticut, Delaware, New Jersey, and Rhode Island “do not elect district attorneys” but omitting Alaska).

  63. [63]. N.J. Stat. Ann. § 2A:158-1 (“There shall be appointed, for each county, by the governor with the advice and consent of the senate ... some fit person ... who shall be known as the county prosecutor ... .”).

  64. [64]. Del. Code Ann. tit. 29, § 2505(i) (“The Attorney General may appoint ... a lawyer resident in this State who shall be designated as the ... Chief Prosecutor of a particular county and who shall serve on a full-time basis under the direct control of the Attorney General. Such persons shall have such responsibilities, powers and duties as the Attorney General shall designate.”). Alaska assigns the power to prosecute state crimes to the Attorney General. See Alaska Stat. § 44.23.020(b)(4) (2018). The Attorney General appoints individuals to serve as District Attorneys in various regional offices (Alaska does not have counties), and those DA offices are part of the state government. See Regional District Attorneys’ Offices, St. Alaska Dep’t L., [].

  65. [65]. Conn. Const. art. XXIII (adding “There shall be a commission composed of the chief state’s attorney and six members appointed by the governor and confirmed by the General Assembly, two of whom shall be judges of the Superior Court. Said commission shall appoint a state’s attorney for each judicial district and such other attorneys as prescribed by law” to Article IV of the constitution).

  66. [66]. See Perry & Banks, supra note 53, at 2.

  67. [67]. Haw. Rev. Stat. Ann. § 46-1.5(1), (17) (LexisNexis 2019) (“Each county shall have the power to frame and adopt a charter for its own self-government that shall establish the county executive, administrative, and legislative structure and organization, including but not limited to the method of appointment or election of officials, their duties, responsibilities, and compensation, and the terms of their office; ... Each county shall have the power to provide by charter for the prosecution of all offenses and to prosecute for offenses against the laws of the State under the authority of the attorney general of the State . . . .”). Kauai County, Hawaii County, and Honolulu County all elect their local prosecutors. Kalawao County, which has a population of 88, does not have a local prosecutor. Carissa Byrne Hessick, The Prosecutors and Politics Project: National Study of Prosecutor Elections 53–54 (2020) [hereinafter The Prosecutors and Politics Project], available at /2020/01/National-Study-Prosecutor-Elections-2020.pdf [].

  68. [68]. Prosecuting Attorney, County Maui Haw., [].

  69. [69]. Mont. Code Ann. §§ 7-4-2203, -2205(2) (West 2019).

  70. [70]. Telephone call with Carter County Clerk’s Office (Oct. 2018).

  71. [71]. N.D. Cent. Code § 11-10-02.3 (2012) (stating the county “shall place the question of appointing the state’s attorney on the ballot at the next regular election” if the electorate satisfies certain petition requirements).

  72. [72]. Id. (“If a majority of the qualified electors of the county voting on the question approves the change from elective to appointive, the change is effective at the end of the term of office of the state’s attorney holding office at the time of the election.”).

  73. [73]. Both counties have populations under 2,000. QuickFacts: Steele County, North Dakota; Golden Valley County, North Dakota; North Dakota, U.S. Census Bureau, /quickfacts/fact/table/steelecountynorthdakota,goldenvalleycountynorthdakota,ND [https://].

  74. [74]. Nancy Northup, Note, Local Nonpartisan Elections, Political Parties and the First Amendment, 87 Colum. L. Rev. 1677, 1677 (1987). “Nationwide, about two-thirds of municipal elections are nonpartisan and 93% of western cities have nonpartisan ballots. Small communities, with populations under 5,000, have the highest percentage of nonpartisan elections (88%). Large cities, with populations over 500,000, have the lowest percentage of nonpartisan elections (58%).” Id. at n.1 (citation omitted); see also Brian E. Adams, Campaign Finance in Local Elections: Buying the Grassroots 20–35 (2010) (describing the election rules in several American cities, the majority of which hold nonpartisan elections).

  75. [75]. See Haw. Rev. Stat. Ann. § 46-1.5(1) (LexisNexis Supp. 2019); see also Mont. Code Ann. § 7-3-413 (West 2009) (“Local government elections shall be conducted on a: (1) partisan basis; or (2) nonpartisan basis.”).

  76. [76]. Colo. Const. art. XVIII, § 11(2).

  77. [77]. See, e.g., Boulder County Issue 1D: District Attorney Term Limit Extension, Times-Call (Oct. 15, 2016, 4:00 PM), [] (discussing a proposal to increase the number of terms that the Boulder County district attorney is able to serve from three to four; the number of terms had been increased from two to three in 2009).

  78. [78]. Ala. Code § 36-3-3 (LexisNexis 2013) (“District attorneys shall hold office for the term of six years from the first Monday after the second Tuesday in January next after their election and until their successors are elected and qualified.”).

  79. [79]. Ky. Const. § 97 (“In the year two thousand, and every six years thereafter, there shall be an election in each county for a ... Commonwealth’s Attorney, in each circuit court district, unless that office be abolished, who shall hold their respective offices for six years ... .”).

  80. [80]. La. Stat. Ann. § 16:1(A) (2013) (“In each judicial district and in the parish of Orleans, there shall be a district attorney. He ... shall serve a term of six years ... .”).

  81. [81]. See, e.g., Tenn. Code Ann. § 16-2-506(31)(B)(i) (Supp. 2018) (“At the regular ... election ... the qualified voters of the thirty-first judicial district shall elect a person to the position of district attorney general for a full eight-year term.”).

  82. [82]. N.H. Rev. Stat. Ann. § 7:33 (2013) (“There shall be a county attorney for each county, who shall be a member of the New Hampshire bar, elected biennially by the voters of the county.”).

  83. [83]. For some of these states, such as Mississippi, the election cycle for prosecutors lines up with the state governor’s race. See supra Table 1. But this is not always the case. New York, for example, holds at least some prosecutor elections in each cycle. See supra Table 1.

  84. [84]. See supra Table 1.

  85. [85]. See supra Table 1.

  86. [86]. See supra Table 1. We have classified Indiana as a multi-county jurisdiction even though only one of its districts contains more than a single county.

  87. [87]. See U.S. Census Bureau, 2010 Census Summary File 1—California SE:T1 (2011).

  88. [88]. We include those city prosecutors here because they have felony jurisdiction. City or county attorneys who handle only misdemeanors were excluded from our study.

  89. [89]. For more on how states deal with unopposed candidates, see Noah B. Lindell, One Person, No Votes: Unopposed Candidate Statutes and the State of Election Law, 2017 Wis. L. Rev. 885 (2017).

  90. [90]. Ark. Code Ann. § 7-5-315(a)(1)–(2) (West 2017)(“The votes received by an unopposed candidate for the office of mayor, governor, or circuit clerk shall be counted and tabulated by the election officials. All other unopposed candidates shall be declared and certified as elected in the same manner as if the candidate had been voted upon at the election.”).

  91. [91]. Fla. Stat. § 101.151(7) (2019) (“[T]he names of unopposed candidates shall not appear on the general election ballot,” and deems the unopposed candidate to have won based on the assumption that the “candidate shall be deemed to have voted for himself or herself”).

  92. [92]. Haw. Rev. Stat. Ann. § 12-41(a) (LexisNexis 2017) (“[A]ny candidate for any county office who is the sole candidate for that office at the primary or special primary election, or who would not be opposed in the general or special general election by any candidate running on any other ticket, nonpartisan or otherwise, and who is nominated at the primary or special primary election shall, after the primary or special primary election, be declared to be duly and legally elected to the office for which the person was a candidate regardless of the number of votes received by that candidate.”). Notably, counties appear to have the ability to set their own election rules. See infra note 100.

  93. [93]. La. Stat. Ann. § 18:511(B) (2012) (“If, after the close of the qualifying period for candidates in a primary election, the number of candidates for a public office does not exceed the number of persons to be elected to the office, the candidates for that office, or those remaining after the withdrawal of one or more candidates, are declared elected by the people, and their names shall not appear on the ballot in either the primary or the general election.”).

  94. [94]. Okla. Stat. Ann. tit. 26, § 6-102 (West 2019) (“Any candidate who is unopposed in any election shall be deemed to have been nominated or elected ... and his name will not appear on the ballot at any election in which he is so unopposed.”).

  95. [95]. S.D. Codified Laws § 12-16-1.1 (2018) (“Any candidate who has been duly nominated to an elective office except State Legislature, political or nonpolitical, having no opposing candidate at the general election shall automatically be elected ... and his name shall not be printed on the general election ballot.”).

  96. [96]. Tex. Elec. Code Ann. § 2.053(a) (West 2010) (“On receipt of the certification [that a candidate is running unopposed], the governing body of the political subdivision by order or ordinance may declare each unopposed candidate elected to the office. If no election is to be held on election day by the political subdivision, a copy of the order or ordinance shall be posted on election day at each polling place used or that would have been used in the election.”). Indiana’s statute is ambiguous, but appears similar because it permits, but does not require, county officials to hold an uncontested primary election. Ind. Code Ann. § 3-10-1-5(a) (LexisNexis 2018) (“Whenever there is no contest in a political party for the nomination of a candidate or candidates for an office, the party may hold a primary election for that nomination. The appropriate election board shall certify the names of the candidates for each nomination for which there is no contest as though a primary election had been held.”).

  97. [97]. See Cal. Elec. Code § 8140 (West 2019).

  98. [98]. See id. §§ 8141, 8141.5. Hawaii County, HI appears to take a similar approach. County of Haw. Charter § 13-27(c) (2018) (“The first special election shall be held with the regularly scheduled primary election. Any candidate for mayor, prosecuting attorney or council receiving a majority of all votes validly cast for candidates for that office, excluding blank and spoiled ballots, shall be deemed elected.”).

  99. [99]. See Minn. Stat. Ann. § 204D.07(3) (West 2009) (“If not more than twice the number of individuals to be elected to a nonpartisan office file for the nomination, their names and the name of the office shall be omitted from the state and county nonpartisan primary ballot and the candidates who filed shall be the nominees.”); Or. Rev. Stat. § 249.091(1) (2019) (“If a nominating petition or declaration of candidacy is filed by no more than two candidates for the office of sheriff, county treasurer or county clerk or by no more than two candidates to fill a vacancy in a nonpartisan office: (a) The candidate or candidates are nominated; and (b) The name or names of the candidate or candidates may not be printed on the ballot at the nominating election.”).

  100. [100]. See, e.g., Ala. Code § 17-13-5(c) (LexisNexis 2007) (“If a legally qualified candidate for nomination to an office is unopposed when the last date for filing declarations of candidacy has passed, his or her name shall not appear on the ballots to be used in the primary election, and he or she shall be the nominee of the party with which he or she has qualified for the office.”); Ark. Code Ann. § 7-7-313 (2018) (“If there is a primary election in which only one (1) candidate has filed for the position by a filing deadline and there are no other ballot issues to be submitted for consideration, the county board of election commissioners may declare and certify the candidate as elected in the same manner as if the candidate had been voted upon at the election.”); Fla. Stat. § 101.252(1) (2019) (“Any candidate for nomination who has qualified as prescribed by law is entitled to have his or her name printed on the official primary election ballot. However, when there is only one candidate of any political party qualified for an office, the name of the candidate shall not be printed on the primary election ballot, and such candidate shall be declared nominated for the office.”); Ky. Rev. Stat. Ann. § 118.185 (LexisNexis 2015) (“If it appears, after the expiration of the time for filing nomination papers, that there is only one (1) candidate who has filed the necessary papers for a place on the ballot of any party on whose ballot he is entitled to have his name printed, the officer with whom the papers are filed shall immediately issue and file in his office a certificate of nomination, and send a copy to the candidate.”); Mich. Comp. Laws Ann. § 168.540 (West 2008) (“If, upon the expiration of the time for filing petitions for any nonpartisan primary election, it shall appear that as to any office on any nonpartisan ticket there are not to exceed twice the number of candidates as there are persons to be elected, then the officer with whom such petitions are filed shall certify to the proper board of election commissioners the names of such candidates whose petitions have been properly filed and such candidates shall be the nominees for such offices and shall be so certified. As to such offices, there shall be no primary election and such offices shall be omitted from the primary ballot.”); Miss. Code. Ann. § 23-15-299(7) (2018) (“Where there is but one (1) candidate for each office contested at the primary election, the proper executive committee or the Secretary of State ... shall declare such candidates the nominees.”); Neb. Rev. Stat. Ann. §32-401 (LexisNexis 2014) (“The statewide primary election shall be held for the purposes of (1) nominating all candidates to be voted for at the statewide general election except (a) candidates who were unopposed at the primary election and not required to be on the ballot ....”); Nev. Rev. Stat. Ann. § 293.260(1) (LexisNexis 2017) (“If there is no contest of election for nomination to a particular office, neither the title of the office nor the name of the candidate may appear on the ballot at the primary election.”); id. § 293.1715(1) (“The names of the candidates for partisan office of a minor political party must not appear on the ballot for a primary election.”); N.Y. Elec. Law § 6-160(2) (McKinney 2018) (“All persons designated for uncontested offices or positions at a primary election shall be deemed nominated or elected thereto ... without balloting.”); N.C. Gen. Stat. § 163A-983 (2017) (“If a nominee for a single office is to be selected and only one candidate of a political party files for that office, or if nominees for two or more offices (constituting a group) are to be selected, and only the number of candidates equal to the number of the positions to be filled file for a political party for said offices, then the appropriate board of elections shall, upon the expiration of the filing period for said office, declare such persons as the nominees or nominee of that party, and the names shall not be printed on the primary ballot, but shall be printed on the general election ballot as candidate for that political party for that office.”); Okla. Stat. Ann. tit. 26, § 6-102 (West 2019) (“Any candidate who is unopposed in any election shall be deemed to have been nominated or elected ... and his name will not appear on the ballot at any election in which he is so unopposed.”); S.C. Code Ann. § 7-11-90 (2018) (“After the closing of entries if any candidates shall be unopposed, the state committee in the case of state offices and the county committees in the case of county offices shall declare such unopposed candidates as party nominees, and the names of unopposed candidates shall not be placed upon the primary election ballots but shall be certified for the general election ballots.”); S.D. Codified Laws § 12-6-9 (2018) (“A candidate for nomination to an office, ... having no opposing candidate within his party, shall automatically become the nominee of his party ... for said office, and his name shall not be printed on the primary election ballot.”); Utah Code Ann. § 20A-9-403(5)(c)(ii) (LexisNexis 2019) (“A candidate who is unopposed for an elective office in the regular primary election of a registered political party is nominated by the party for that office without appearing on the primary election ballot.”); Va. Code Ann. § 24.2-526 (2016) (“Whenever ... there is only one declaration of candidacy in a political party for the nomination for any office, the person filing the declaration shall be declared the nominee of the party for the office for which he has announced his candidacy and his name shall not be printed on the ballot for the primary.”).

  101. [101]. Mont. Code Ann. § 13-14-115(2)(b) (West 2019) (“The election administrator may determine that a primary election for a nonpartisan county office need not be held if fewer than three candidates have filed for that office.”).

  102. [102]. Ga. Code Ann. § 21-2-285(j)–(k) (2019); Miss. Code. Ann. § 23-15-361(6). Colorado, Illinois, Michigan, Montana, and Ohio have a similar procedure when no offices are contested in primary elections. Colo. Rev. Stat. § 1-4-104.5 (2019); 10 Ill. Comp. Stat. 5/7-5(b) (2018); Mich. Comp. Laws Ann. § 168.541; Mont. Code Ann. § 13-10-209(2); Ohio Rev. Code Ann. § 3513.02 (West 2018). See generally Lindell, supra note 89, at 900–02 (discussing these cancellation statutes).

  103. [103]. See supra note 48.

  104. [104]. The Prosecutors and Politics Project, supra note 67, at 125–31.

  105. [105]. To be clear, these statutes are not limited to prosecutors; they apply to other state and local offices as well. And their effect on elections for those offices can also be dramatic. See, e.g., Lindell, supra note 89, at 887 (“[I]n the 2016 general election, over one-third of the Florida state legislature was ‘elected’ without receiving a single vote[.]”).

  106. [106]. State Attorney Russell Will Not Seek Re-election, News-Press (June 22, 2017, 12:06 PM), 9321001 [] (“Except for opposition in his first election for state attorney in 2002, Russell has run unopposed in three other elections in the partisan, five-county race.”).

  107. [107]. Founded in 2018, the Prosecutors and Politics Project studies the role of prosecutors in the criminal justice system, focusing on both the political aspects of their selection and their political power. The Project endeavors to “[b]ring scholarly attention to the democratic accountability of elected prosecutors,” to “[i]ncrease our understanding of the relationship between prosecutors and politics through empirical study,” and to “[p]ublicly share research ... [in order] to increase voters’ knowledge about their elected prosecutors and broader criminal justice issues.” Prosecutors and Politics Project, UNC Sch. L., academics/centers-and-programs/prosecutors-and-politics-project [ 9V]. Author Carissa Hessick is the Director of the Prosecutors and Politics Project. Id.

  108. [108]. Our data has 1,038 elections from 2014 and 961 from 2016. Some jurisdictions held elections in 2015 and 2017: We have 187 elections in 2015 (Virginia, Pennsylvania, New York, Mississippi, and one district in California), and 62 in 2017 (Virginia, New York, Pennsylvania, and one district in Texas). We also have 57 elections in 2012 from Kentucky. Recall that Kentucky elects prosecutors to six-year terms. See supra Table 1. We are missing the precise date for fewer than five elections.

  109. [109]. All of the data was obtained from public sources. Where available, the data was obtained from official election records. When official election records were not available or when they did not include the relevant information, data was obtained from websites or media accounts. Occasionally data had to be obtained via email or telephonic inquiries.

  110. [110]. As noted below, because of variation in state law, it was not always clear whether to classify a particular election as a primary or a general election. See infra note 134.

  111. [111]. Because write-in candidates were not included in the data, a given candidate’s percentage of total votes received might differ from the candidate’s total votes divided by the overall number of votes cast in the district. Information about primary elections and party affiliation proved particularly difficult to collect, and so our data is less complete on those issues.

  112. [112]. To be clear, we define an open seat as a race in which the incumbent did not seek reelection at all, in either the primary or in the general election. If an incumbent ran in the primary, but lost, and thus did not run in the general election, it is not counted as an open seat.

  113. [113]. See, e.g., David M. Konisky & Michiko Ueda, The Effects of Uncontested Elections on Legislator Performance, 36 Legis. Stud. Q. 199, 201 (2011); Richard G. Niemi et al., Competition in State Legislative Elections, 1992–2002, in The Marketplace of Democracy: Electoral Competition and American Politics 53, 65 (Michael P. McDonald & John Samples eds., 2006).

  114. [114]. See infra Table 11.

  115. [115]. We used data from the 2010 Census to determine district population. In this Article, we exclude population figures for two districts in Texas because the boundaries for those districts do not follow county boundaries, and so we were unable to use U.S. Census data to determine the population within those districts.

  116. [116]. We will continue to update the dataset as more information becomes available. The data set is available at Carissa Hessick & Michael Morse, Local Prosecutor Elections, 2012–2017, Dataverse, []. But the missing data and difficulties in coding because of state law variation may have created inconsistencies in our results.

  117. [117]. See infra notes 124–26, 129–31 and accompanying text.

  118. [118]. The population data is from the 2010 Census. We divide the data into four categories, following the Bureau of Justice Statistics’ convention. See, e.g., Perry & Banks, supra note 53, at 1 tbl.1, 4 tbl.2 (reporting various quantities by officers serving 99,999 or less people, 100,000 to 249,999 people, 250,000 to 999,999 people, and 1,000,000 or more people).

  119. [119]. The data on urbanicity comes from the National Center for Health Statistics Urban-Rural Classification Scheme for Counties, as used and implemented by the Vera Institute of Justice. See Vera Inst. of Justice, Incarceration Trends Dataset 7, 12 (2020), available at book.pdf?raw=true []. When aggregating county-level urbanicity to the district-level, we limit our data to districts with the same county urbanicity.

  120. [120]. We used summary data for calendar years 2006 and 2013, as collected and posted publicly by the New York Times. For the data, see Prison-Admissions, Github: TheUpshot (Sept. 13, 2016), []; for the reporting, see Josh Keller & Adam Pearce, A Small Indiana County Sends More People to Prison than San Francisco and Durham, N.C., Combined. Why?, N.Y. Times (Sept. 2, 2016), [ S5H7-YEJA].

  121. [121]. As a result, we have complete data for 74 percent of districts, with the smallest districts (also the most numerous districts) having the most missing variables.

  122. [122]. The name of the current holder was determined from visiting the office website. These visits occurred at different times for different offices. Some occurred as early as February 2018; others as late as February 2019.

  123. [123]. For the purpose of our analysis, some inconsistent observations in the Wright data are dropped, for reasons including: no district was listed (64 districts, all in Texas in 2002); the same candidate was listed twice with different vote totals in the same election (13 candidates in general election, all in California, all but one in 2002); two candidates for the primary election (in the same district in Oregon in 2000); and different candidates were listed with the same vote totals in the same election (one district in New Mexico). In addition, some variables that appeared to be transposed were corrected.

  124. [124]. See Wright, How Prosecutor Elections Fail Us, supra note 38, at 592, 595 tbl.1.

  125. [125]. See Wright, Beyond Prosecutor Elections, supra note 38, at 601 tbl.1 (finding that the incumbent was winning 94 percent of general elections).

  126. [126]. This positive correlation between population and contestation can also be seen in Ron Wright’s original work, though he used how many total votes were cast in the district as a rough proxy for jurisdiction size and did not report results for the smallest districts, which, in his case, were those with less than 10,000 total votes cast. See Wright, Beyond Prosecutor Elections, supra note 38, at 602 tbl.2; Wright, How Prosecutor Elections Fail Us, supra note 38, at 596 tbl.2.

  127. [127]. Some of the data presented in this third set of columns is duplicative of the data presented in Table 4.

  128. [128]. Table 5 defines a competitive election as one where the winner received less than 60 percent of the vote. Table 11 in the Appendix reports results with alternative definitions. See infra Table 11.

  129. [129]. In theory, there could be separate Democratic and Republican primaries for local prosecutor. In practice, this happens less than 15 percent of the time. See Hessick & Morse, supra note 116.

  130. [130]. This choice may lead us to undercount contested primaries. Imagine, for example, two candidates who run in a general election. The candidate who wins the general election did not face a primary challenger, but the general election candidate who loses did have a primary challenger. Because we record only the primary for the winning candidate in the general election, our data would indicate that the primary election was uncontested. However, this undercounting does not ultimately matter for our analysis because, as explained below, we define a contested election as an election in which either the general or the primary election is contested. We do not base any of our analysis on the sole question of whether there was a contested primary election. Also, in light of the different legal rules that states have for their primary elections, we had to make various judgment calls about how to code the data we did obtain. For example, Louisiana does not hold a traditional primary election and general election. Instead it holds an open primary—in which any candidate may participate, and any voter can cast a vote, regardless of party affiliation—that can determine the entire race. If a candidate in the open primary obtains a majority of all votes cast, then he or she is deemed elected. If no candidate receives a majority, then the top two candidates advance to a run-off election, which the state calls a general election. How Are Statewide and Local Candidates Elected?, La. Secretary St., ElectionsAndVoting/GetElectionInformation/HowAreCandidatesElected/Pages/default.aspx []. We treated Louisiana’s “open primary” as the general election if it was the determinative election. However, if it led to a run-off, then we treated the “open primary” as the primary and the run-off as the general election. This choice had further consequences for our primary data analysis.

  131. [131]. Florida’s 4th judicial district contains three counties: Clay County, Duval County, and Nassau County. To get a sense of the partisan make up of those counties, consider how the votes in those counties were distributed in the 2016 presidential election: Clay County—70.4% Trump, 26.1% Clinton; Duval County—49.0% Trump, 47.5% Clinton; Nassau County—73.5% Trump, 23.3% Clinton. 2016 Florida Presidential Election Results, Politico, 2016-election/results/map/president/florida [] (last updated Dec. 13, 2016, 1:57 PM).

  132. [132]. See, e.g., Bazelon, supra note 16, at 148–49, 155, 169–72; Sklansky, supra note 34, at 666–67.

  133. [133]. Kenny Leigh was slated to run against Nelson in the general election, but he decided to withdraw, making Nelson the elected state attorney. See Andrew Pantazi, Opponent’s Dropout Makes Melissa Nelson Florida’s New AG, Governing (Aug. 31, 2016, 5:00 PM), https://www. [https://perma. cc/E7GC-UK5C].

  134. [134]. Because there are districts where we have incomplete information about either the general or the primary, it is possible for more contests to be classified as either contested or competitive under “Either Election” than the sum total of the general and primary election.

  135. [135]. See supra Table 5; see infra Table 11.

  136. [136]. This data is more suggestive than anything else. For one, it only includes 15 states. For another, more electoral cycles were collected for some states than others, and thus there is more opportunity for at least one competitive race in some states than others. Finally, some data had to be dropped in order to standardize it with our election results, and we did not attempt to verify or supplement the historical data. Still, the data and re-analysis highlight the importance of longitudinal data in the continued study of prosecutorial elections.

  137. [137]. See infra Table 10.

  138. [138]. Wright, How Prosecutor Elections Fail Us, supra note 38, at 593–94.

  139. [139]. See supra Table 1; see infra Table 10.

  140. [140]. This is consistent with Wright’s finding that incumbents faced challengers in similar rates in Colorado districts. Wright, Beyond Prosecutor Elections, supra note 38, at 602.

  141. [141]. See id. at 602–03.

  142. [142]. This table was inspired by the New York Times’ analysis of the same data. See Keller & Pearce, supra note 120. The only modification was that counties were aggregated to the district level. We caution though that complete data is missing for approximately a quarter of districts, more so in small districts than in larger ones. As a result, we need to assume that the incidence of missing data is not related to the incarceration trends. Nonetheless, the result accords with the New York Times’ county-level analysis.

  143. [143]. Stuntz, supra note 10, at 250.

  144. [144]. See Perry & Banks, supra note 53, at 6 tbl.5; see also id. at 6 tbl.5, n.b (“Data on the tenure of the chief prosecutor were missing for 7.4% of offices surveyed.”).

  145. [145]. See id. at 6 tbl.5.

  146. [146]. This information was particularly challenging to obtain because, unlike election results and population data, there is no official source of information for when an incumbent prosecutor took office. Some local prosecutors include this information in the biographies that they publish on their office websites. But for other prosecutors, we had to search media reports. Sometimes we were unable to locate information about when the prosecutor first took office.

  147. [147]. See Nicholas Fandos, A Study Documents the Paucity of Black Elected Prosecutors: Zero in Most States, N.Y. Times (July 7, 2015), [].

  148. [148]. We coded gender based on candidates’ first names; names not easily identified as male or female were not coded.

  149. [149]. See, e.g., Katelyn Hibbard, Assistant DA to Take Reins, Complete Term, Williamsport Sun-Gazette (Dec. 8, 2017), [] (reporting that the Lycoming County District Attorney is “prepar[ing] to take the oath of office as judge, [and that] a longtime assistant district attorney ... has been appointed to Linhardt’s unexpired term”); Pat Robinson Named Wayne County District Attorney, Tri-County Indep. (Jan. 6, 2018, 12:01 AM), https:// [] (reporting that Wayne County Prosecutor stepped down from the position to become a judge on the Wayne County Court of Common Pleas).

  150. [150]. Clark Neily, Cato Inst., Are a Disproportionate Number of Federal Judges Former Government Advocates? tbl.1 (2019), available at publications/studies/are-disproportionate-number-federal-judges-former-government-advocates [].

  151. [151]. Jed Shugerman, “The Rise of the Prosecutor Politicians”: Database of Prosecutorial Experience for Justices, Circuit Judges, Governors, AGs, and Senators, 1880-2017, Shugerblog (July 7, 2017), -experience-for-justices-circuit-judges-governors-ags-and-senators-1880-2017 [ 25KU-VEFF].

  152. [152]. Id. Shugerman is currently working on a book project about the hypothesis, and he has made a database of politicians who began their careers as prosecutors publicly available. Id. (linking to the database at the bottom of the page at 1E6Z-jZWbrKmit_4lG36oyQ658Ta6Mh25HCOBaz7YVrA/edit#gid=224452527).

  153. [153]. See, e.g., Boris Krawczeniuk, Judge Swears in New Susquehanna DA, Wyo. County Press Examiner (Feb. 7, 2018), [] (reporting on an appointment for district attorney to fill a vacancy caused by the incumbent’s death).

  154. [154]. See Shane Sanderson, Howard County Prosecutor Cites Poor Pay in Decision to Resign, Missourian (Feb. 3, 2017), 9093374fc1.html [] (“Stephen Murrell, the Howard County prosecutor who resigned Monday, said he simply could not financially afford to continue as the Howard County Prosecuting Attorney.”).

  155. [155]. See supra Table 4.

  156. [156]. See, e.g., Baughman, supra note 36, at 1103 (“[A] national sample of outcomes in prosecutor elections reveals that incumbent prosecutors rarely face real accountability.”); Fairfax, supra note 36, at 1269 (“[T]he democratic accountability of prosecutors—even those who are directly elected—is fairly weak ... .”); Sarma, supra note 36, at 591–92 (stating that prosecutor “elections do not serve accountability purposes well”).

  157. [157]. Of course, one could argue that the lack of contestation is the result of a shadow effect —that is to say, no one runs because sitting prosecutors already account for all of their constituents’ preferences. But that seems highly unlikely given that so many voters do not even know that their local prosecutor is an elected official. See Bazelon, supra note 16, at 78.

  158. [158]. See supra notes 61–62 and accompanying text.

  159. [159]. See, e.g., Michael Tonry, Prosecutors and Politics in Comparative Perspective, in 41 Prosecutors and Politics: A Comparative Perspective 1 (Michael Tonry ed., 2012); see also David Alan Sklansky, Unpacking the Relationship between Prosecutors and Democracy in the United States, in Prosecutors and Democracy: A Cross-National Study, supra note 46, at 276, 277–78 (laying out the argument against politics as a method to control prosecutors).

  160. [160]. Concerns about patronage originally prompted states to move from appointments to elections, see Ellis, supra note 47, at 1550–57, and in those states that retained appointments, some have documented that political connections appear to be more important than experience or qualifications in selection. See generally Malcolm M. Feeley, The Process Is the Punishment: Handling Cases in a Lower Criminal Court 70–76 (1992) (describing the selection and performance of prosecutors in New Haven, Connecticut).

  161. [161]. See supra Table 5.

  162. [162]. See supra Table 5.

  163. [163]. Of the 453 contested elections in the smallest jurisdictions, 268 were competitive. Of the 28 contested elections in the largest jurisdictions, 21 were competitive. See supra Table 5.

  164. [164]. See Perry & Banks, supra note 53, at 4 tbl.2 (reporting a median of 22 supervising attorneys, 133 assistant prosecutors, and 139 support staff in prosecutor offices serving populations of more than one million inhabitants).

  165. [165]. See id. (reporting a median of zero supervising attorneys, two assistant prosecutors, and three support staff in prosecutor offices serving populations of fewer than 100,000 inhabitants).

  166. [166]. See id. (reporting a median budget of more than $35 million for prosecutor offices serving populations of more than one million inhabitants and less than $350,000 for offices serving populations of fewer than 100,000 inhabitants); id. at 5 tbl.4 (comparing caseloads).

  167. [167]. The Prosecutors and Politics Project, supra note 67, at 357–60.

  168. [168]. Our population-based analysis uses 2010 census data; current population levels may differ.

  169. [169]. We are, however, mindful that doing so might contribute to the already significant bias in the legal literature of speaking about prosecutors’ offices almost solely based on what happens in large, urban offices. Ronald F. Wright & Kay L. Levine, Place Matters in Prosecution Research, 14 Ohio St. J. Crim. L. 675, 678–81 (2017) (documenting that large urban prosecutor offices are significantly overrepresented in the empirical studies of prosecutors and that those studies have shaped the academic conversation about prosecutors).

  170. [170]. See supra Table 5.

  171. [171]. See supra Table 5.

  172. [172]. See supra Table 5.

  173. [173]. See Adams, supra note 74, at 42–45, 55.

  174. [174]. See supra Table 5.

  175. [175]. Ronald F. Wright, Public Defender Elections and Popular Control over Criminal Justice, 75 Mo. L. Rev. 803, 808 (2010).

  176. [176]. Those jurisdictions are: Camas, Idaho (population 1,117); Clark, Idaho (population 982); Custer, Idaho (population 4,368); Sullivan, Missouri (population 6,714); Worth, Missouri (population 2,171); Arthur, Nebraska (population 460); Grant, Nebraska (population 614); Hooker, Nebraska (population 736); Keya Paha, Nebraska (population 824); Logan, Nebraska (population 763); McPherson, Nebraska (population 539); Billings, North Dakota (population 783); Hamlin, South Dakota (population 5,903); Daggett, Utah (population 1,059); Piute, Utah (population 1,556); and Rich, Utah population (2,264). See Hessick & Morse, supra note 116. This does not include counties in Hawaii, Montana, and North Dakota that have chosen to adopt an appointment scheme, rather than to select their local prosecutor via an election. See supra notes 67–71 and accompanying text. Nor does it include Santa Cruz County, California, Harrison County, Missouri, or District 451 in Texas. During the 2014 cycle, the incumbent district attorney in Santa Cruz County was running unopposed for reelection. But the unopposed incumbent became ill and died in October, so a replacement had to be appointed and no one was elected. See Samantha Clark & Stephen Baxter, Jeff Rosell Appointed Santa Cruz County District Attorney, Santa Cruz Sentinel (Nov. 18, 2014, 12:00 AM), 11/18/jeff-rosell-appointed-santa-cruz-county-district-attorney []. A candidate ran for and won the office of Prosecuting Attorney in Harrison County, but did not take office after pleading guilty to felony charges in federal court. See Nixon Appoints Keedy as PA in Sullivan, NEWS/TALK 1150 AM KRMS 97.5 FM 103.3 FM (Dec. 31, 2014), []. District 451 was a newly created district in Texas; the Governor appointed the initial district attorney who served until the election in 2018. See Judge and D.A. Appointed for Kendall County’s New State District Court, mySanAntonio article/Judge-and-d-a-appointed-for-Kendall-County-s-10786607.php [] (last updated Dec. 9, 2016, 4:47 PM).

  177. [177]. Those jurisdictions are Hayes, Nebraska (population 967); Bowman, North Dakota (population 3,151); Divide, North Dakota (population 2,071); and Grant, North Dakota (population 2,394). See Hessick & Morse, supra note 116.

  178. [178]. See, e.g., Tex. Gov’t Code Ann. § 44.108(a)(3) (West 2004) (“The criminal district attorney of Austin County must ... have been a resident of Austin County for at least two years before his election or appointment.”). Some states allow people from outside of the district to run for local prosecutor. See infra note 194 and accompanying text.

  179. [179]. See Lisa R. Pruitt et al., Legal Deserts: A Multi-State Perspective on Rural Access to Justice, 13 Harv. L. & Pol’y Rev. 15, 120–26 (2018) (discussing rural attorney shortages).

  180. [180]. E-mails from Lura Baker, Custer Cty. Clerk, to Carissa Byrne Hessick, Professor of Law & Assoc. Dean, Univ. of N.C. Sch. of Law (Oct. 25–26, 2018) (on file with authors).

  181. [181]. Telephone call with Hamlin County Auditor’s Office (Feb. 6, 2019).

  182. [182]. The ten states from which we were able to obtain information are: Arizona, California, Kentucky, Missouri, Nebraska, New York, North Dakota, South Carolina, South Dakota, and Utah.

  183. [183]. Our data about the number of lawyers in Kentucky identified two counties as having “1000+” lawyers, rather than a specific number. We treated these districts as having 1001 lawyers, even though they almost certainly have more than 1,001 lawyers. Because our analysis largely focuses on districts with small numbers of lawyers, we do not think this missing information would change our recommendations.

  184. [184]. Six states had at least one election in which no candidate ran for prosecutor: Idaho, Missouri, Nebraska, North Dakota, South Dakota, and Utah. Our subset included all of those states except Idaho. See supra notes 182–83 and accompanying text.

  185. [185]. For example, Table 8 contains data from only two states with multi-county districts, but a third of states that elect local prosecutors use multi-county districts. See supra Table 1. The two multi-county states in our subset—Kentucky and South Carolina—had significantly larger numbers of lawyers in each district. See infra text accompanying notes 198–201.

  186. [186]. Neb. Rev. Stat. Ann. § 23-1201.01 (LexisNexis 2018).

  187. [187]. See supra Table 1.

  188. [188]. See supra Table 1.

  189. [189]. See Ky. Const. § 97 (amended 1992) (“[T]here shall be an election ... for a Commonwealth’s Attorney, in each circuit court district, unless that office be abolished.”); Ky. Const. § 112 (amended 2002) (renaming circuit court districts “Judicial Circuits” and setting certain rules about the composition of such circuits); Ky. Rev. Stat. Ann. § 23A.020 (West 2015) (listing the 57 judicial circuits and identifying which counties comprise each circuit).

  190. [190]. See S.C. Code Ann. § 1-7-310 (2005) (“There is one solicitor for each judicial circuit, to be elected by the qualified electors of the circuit ... .”); S.C. Code Ann. § 14-5-610(A) (2016) (listing the 16 judicial circuits and identifying which counties comprise each circuit).

  191. [191]. District 47 in Kentucky, which consists of only a single county (Letcher County), has a population of 22,773.

  192. [192]. By a contested election, we mean either a contested primary election or a contested general election. See infra Table 10.

  193. [193]. This is true in both counties with relatively small populations and counties with relatively large populations. For example, there are twice as many lawyers living in Mono County, California (33 lawyers) than in Mariposa County, California (15 lawyers) even though Mariposa’s population is larger than Mono (18,251 as compared to 14,202). And there are nearly 20 times as many lawyers living in Marin County (2,111 lawyers) than in Merced County (165 lawyers) despite similar populations. Marin has a population of 252,409 as compared to Merced’s population of 255,409.

  194. [194]. New York also has far fewer counties than Nebraska—62 as compared to 93—despite having a population that is more than ten times larger.

  195. [195]. See David L. Shapiro, Federalism: A Dialogue 92–94 (1995); Barry Friedman, Valuing Federalism, 82 Minn. L. Rev. 317, 389–90 (1997).

  196. [196]. E.g., Gerald E. Frug, City Making: Building Communities Without Building Walls 19–22 (1999) (discussing and critiquing the idea that devolving power to local governments is undesirable because of a perceived need in modern society for specialization and expertise); Robert A. Dahl, The City in the Future of Democracy, 61 Am. Pol. Sci. Rev. 953, 953–56 (1967) (discussing the tradeoffs in selecting “the appropriate unit for a democratic political system”).

  197. [197]. See Richard Briffault, Our Localism: Part II—Localism and Legal Theory, 90 Colum. L. Rev. 346, 392–93 (1990) (noting that two fundamental values in the discourse about local autonomy are “participation in public life ... and efficiency in the provision of public sector goods and services”).

  198. [198]. See Dahl, supra note 196, at 966.

  199. [199]. Pfaff, supra note 3, at 214–16.

  200. [200]. William J. Stuntz, Unequal Justice, 121 Harv. L. Rev. 1969, 2002 (2008).

  201. [201]. E.g., Michael W. McConnell, Federalism: Evaluating the Founders’ Design, 54 U. Chi. L. Rev. 1484, 1509 (1987).

  202. [202]. See supra Table 8.

  203. [203]. See Heather K. Gerken, Foreword: Federalism All the Way Down, 124 Harv. L. Rev. 4, 7–8 (2010) (reconceptualizing the value of federalism and localism as “voice” rather than “exit” or “autonomy” and defining “voice” as “the power not just to complain about national policy, but to help set it”).

  204. [204]. See E-mails from Lura Baker, supra note 180.

  205. [205]. Tyler Q. Yeargain, Comment, Discretion Versus Supersession: Calibrating the Power Balance Between Local Prosecutors and State Officials, 68 Emory L.J. 95, 98 (2018).

  206. [206]. See Ellis, supra note 47, at 1537 (“Who appointed the district attorney varied from state to state: in Kentucky and New York, for example, it was the judge of the county court; in Alabama, Georgia, North Carolina, and Tennessee, it was the state legislature; in Massachusetts and New Hampshire, it was the governor, assisted by his council of advisors; in Michigan, the governor with the advice and consent of the state senate.” (footnotes omitted)).

  207. [207]. See id. at 1561 (“The Illinois delegates feared that a prosecutor responsible for more than one county might be unable to travel to court proceedings, and, even if present, might be unprepared.”).

  208. [208]. Of course, if defendants are forced to stand trial too far away from their communities, that can have some ill-effects. See Kevin K. Washburn, American Indians, Crime, and the Law, 104 Mich. L. Rev. 709, 764–74 (2006) (discussing how holding trials involving crimes that occurred in Indian country in federal courthouses imposes costs on the defendant and the community).

  209. [209]. See Clyde F. Snider, Local Government in Rural America 20 (1957); Rick Su, Democracy in Rural America, 98 N.C. L. Rev. (forthcoming 2020) (manuscript at 19–24) (on file with authors).

  210. [210]. See 1 James Bryce, The American Commonwealth 586 (1888).

  211. [211]. See Pruitt et al., supra note 179 and accompanying text.

  212. [212]. See Tammy Joyner, Clayton Prosecutor Says He Was Fired Over Rumored Run for DA, Atlanta J.-Const. (Mar. 28, 2014),—politics/clayton-prosecutor-says-was-fired-over-rumored-run-for/HCQAdeuLElOURh3fVLZgGP [].

  213. [213]. See Annie Knox, Salt Lake District Attorney’s Race Pits County’s Top Prosecutor Against One of His Deputies, Deseret News (Oct. 24, 2018, 5:11 PM), article/900038624/salt-lake-district-attorneys-race-pits-countys-top-prosecutor-against-one-of-his -deputies.html [].

  214. [214]. See supra Table 5.

  215. [215]. See Scott Bland, George Soros’ Quiet Overhaul of the U.S. Justice System, Politico (Aug. 30, 2016, 5:25 AM), [] (“[A] broader national push into local prosecutor campaigns is expected to intensify in the next few years, thanks to longer-term planning and candidate recruitment.”).

  216. [216]. See Chris Brennan, Krasner Declared Winner of Democratic Primary for DA in Philly, Phila. Inquirer (May 16, 2017, 9:06 PM), [].

  217. [217]. Jennifer Gonnerman, Larry Krasner’s Campaign to End Mass Incarceration, New Yorker (Oct. 22, 2018), [].

  218. [218]. Id.; see also Note, supra note 24, at 750–51.

  219. [219]. Lauren M. Ouziel, Democracy, Bureaucracy, and Criminal Justice Reform, 61 B.C. L. Rev. 523, 537 n.47 (2020).

  220. [220]. Brennan, supra note 216 (“Philadelphia’s registered Democratic voters outnumber Republicans 7-1.”).

  221. [221]. See id.; Phillip Jackson, Hodge Reflects on Short Tenure as Trailblazing D.A., Phila. Trib. (Nov. 3, 2017), [].

  222. [222]. See Gonnerman, supra note 217.

  223. [223]. See Brennan, supra note 216.

  224. [224]. Tommy Rowan & Emily Babay, 2017 Pennsylvania Primary Election Roundup: Who Won and Lost, Phila. Inquirer (May 17, 2017, 7:12 AM), politics/2017-primary-election-roundup-Philadelphia-Pennsylvania.html [ 7P-R2FT].

  225. [225]. See Bland, supra note 215.

  226. [226]. Carissa Byrne Hessick, The Prosecutors and Politics Project: Study of Campaign Contributions in Prosecutorial Elections 104 (unpublished report), available at https:// [] (last updated June 9, 2019). Gonzalez was also appointed Acting District Attorney when the previous district attorney, Ken Thompson, died suddenly from cancer. As a result, Gonzalez likely had many of the advantages of an incumbent running for reelection. See Alan Feuer, Ken Thompson’s Successor: A ‘Pure District Attorney’ Working Under the Radar, N.Y. Times (Nov. 28, 2016), [].

  227. [227]. Hessick, supra note 226, at 76.

  228. [228]. David Sklansky identifies Ken Thompson’s successful 2013 challenge to Brooklyn District Attorney Charles Hynes as the turning point. Sklansky notes that “[n]o incumbent district attorney had lost an election in Brooklyn for over a century,” and that “Thompson’s campaign focused on allegations of prosecutorial misconduct under Hynes’s watch and Hynes’s failure to rein in aggressive police tactics that disproportionately impacted minority youth.” Sklansky, supra note 34, at 651–52.

  229. [229]. Wright, How Prosecutor Elections Fail Us, supra note 38, at 600–02.

  230. [230]. Wright, Beyond Prosecutor Elections, supra note 38, at 604–05.

  231. [231]. See Bland, supra note 215.

  232. [232]. See Note, supra note 24, at 766–67 (arguing that “voters are notoriously fickle” and explaining that “the election of a progressive prosecutor in a particular jurisdiction should not be seen as a one-way ratchet toward leniency: nothing prevents localities that favor progressive prosecution today from electing a traditional tough-on-crime prosecutor tomorrow”).

  233. [233]. See Shaun King, Law Enforcement Groups Gave $420,000 to DA Deciding Whether to Bring Charges Against Cops Who Killed Stephon Clark, Intercept (Apr. 25, 2018, 2:24 PM), https:// [ 2MCR-44DH].

  234. [234]. See Marcos Bretón, Opinion, Sacramento’s DA Race Is Done: Why Progressives Never Should Have Backed Noah Phillips, Sacramento Bee (June 15, 2018, 3:55 AM), https://

  235. [235]. See Greg Moran, DA Race: Stephan Easily Defeats Challenger Jones-Wright, Earns Full Term, San Diego Union-Trib. (June 6, 2018, 5:50 AM), news/public-safety/sd-me-elex-da-20180531-story.html [].

  236. [236]. Judicial elections might serve as a cautionary tale for those who wish to achieve progressive criminal justice goals through elections. As Michael Kang and Joanna Shepard have documented, the number of contested judicial elections increased dramatically in the late twentieth century. Michael S. Kang & Joanna M. Shepherd, The Partisan Price of Justice: An Empirical Analysis of Campaign Contributions and Judicial Decisions, 86 N.Y.U. L. Rev. 69, 81 (2011). As the number of contested elections rose, incumbents became less likely to win reelection. Id. at 82. This might seem like precisely what those in the progressive prosecutor movement hope to achieve. But Kang and Shepherd also document that, as judicial elections became more competitive, campaign spending in those elections increased dramatically, which allowed “wealthy interest groups ... [to] influence the outcomes of judicial elections by contributing substantial campaign funds in support of favored candidates.” Id. at 84. Using a large dataset, Kang and Shepherd found that increased spending in partisan judicial elections led to more judicial decisions in favor of business interests. Id. at 99–100. To the extent that law enforcement groups and others who are opposed to progressive criminal justice reforms are willing to spend more in prosecutor elections, then more contested and competitive elections may actually hurt criminal justice reform.

  237. [237]. Andrew Dyke, Electoral Cycles in the Administration of Criminal Justice, 133 Pub. Choice 417, 419 (2007).

  238. [238]. E.g., Sanford C. Gordon & Gregory A. Huber, Citizen Oversight and the Electoral Incentives of Criminal Prosecutors, 46 Am. J. Pol. Sci. 334, 335 (2002) (explaining why “optimal voter strategy ... is always to reelect prosecutors who obtain convictions”); Eric Rasmusen et al., Convictions Versus Conviction Rates: The Prosecutor’s Choice, 11 Am. L. & Econ. Rev. 47, 61–62 (2009) (explaining why elected prosecutors might prefer to increase their conviction rates at the expense of sentence lengths).

  239. [239]. Dyke, supra note 237, at 431–34.

  240. [240]. For example, one study documented a decrease in case dismissals when prosecutors run for reelection. Id. Dismissing fewer cases is consistent with the election effect theory because it reduces conviction. But a separate study showed that prosecutor elections were correlated with an increase in case dismissals. Bryan C. McCannon, Debundling Accountability: Prosecutor and Public Defender Elections in Florida 15–17 (Feb. 3, 2018) (unpublished manuscript), available at []. And a third study found only a very small reduction in case dismissals. Siddhartha Bandyopadhyay & Bryan C. McCannon, Queuing Up for Justice: Prosecutor Elections and Case Backlogs 25 (Sept. 25, 2017) (unpublished manuscript), available at sol3/papers.cfm?abstract_id=3097736 [] (finding that “[e]lections coincide with a reduction in case dismissals” but noting that “this effect is weak”).

  241. [241]. See Daniel C. Richman, Accounting for Prosecutors, in Prosecutors and Democracy: A Cross-National Study, supra note 46, at 40, 63 (“[I]t bears remembering that the idea behind the move to popular elections—widespread in the second quarter of the nineteenth century —was more defensive than offensive: to deflect efforts of governors and legislators to use appointments for political patronage.”).

  242. [242]. See Bazelon, supra note 16, at xxv.

  243. [243]. See Gottschalk, supra note 26, at 2.

  244. [244]. The National Commission on Law Observance and Enforcement, also known as the Wickersham Commission, did propose to reverse the Jacksonian-era shift and revert back to appointing a local prosecutor at the state level. The proposal went nowhere. See Ronald F. Wright, The Wickersham Commission and Local Control of Criminal Prosecution, 96 Marq. L. Rev. 1199, 1199, 1217–19 (2013) (explaining the Commission’s proposal).

  245. [245]. Sheriffs followed the trajectory of prosecutors, from appointed at the state level to elected at the county level during the nineteenth century. See James Tomberlin, Note, “Don’t Elect Me”: Sheriffs and the Need for Reform in County Law Enforcement, 104 Va. L. Rev. 113, 121–23 (2018) (explaining the history). In contrast, (the more modern invention of) police chiefs are generally appointed at the municipal level. See Lawrence M. Friedman, Crime and Punishment in American History 247–89 (1993); Wright, supra note 175, at 811–12. Selection methods for local trial judges continue to be an area of legislative ferment. States have taken a wide variety of approaches, not easily summarized as including appointment as well as partisan, nonpartisan, and retention elections. See generally History of Reform Efforts, Nat’l Ctr. for St. Cts., [] (cataloging “[f]ormal [c]hanges” in each state’s law “[s]ince [i]nception”).


Anne Shea Ransdell and William Garland “Buck” Ransdell, Jr., Distinguished Professor of Law and Director of the Prosecutors and Politics Project, University of North Carolina School of Law.


Ashford Fellow and Ph.D. Candidate, Department of Government, Harvard University. J.D., Yale Law School.

This manuscript greatly benefitted from the comments of Shima Baradaran Baughman, Jack Chin, Beth Colgan, Don Dripps, Cynthia Godsoe, Russell Gold, Bruce Green, Eve Hanan, Irene Joe, Peter Joy, Sam Kamin, Anders Kaye, Richard Leo, Ben Levin, Kate Levine, Cortney Lollar, Justin Marceau, Marc Meredith, Janet Moore, Justin Murray, Lauren Ouizel, Ellen Podgor, Anna Roberts, Addie Rolnick, Rebecca Roiphe, Maybell Romero, Jessica Roth, Jocelyn Simonson, David Schleicher, Anne Traum, and Ron Wright. This research was made possible by a generous gift from the Vital Projects Fund, Inc. The data to reproduce this analysis is available at: Carissa Hessick & Michael Morse, Local Prosecutor Elections, 2012–2017, Dataverse, https:// [ ME-8SC4].