CO-PRODUCTION OF COURT SERVICES: BRINGING THE PUBLIC AND THE COURTS 
TOGETHER 

By 

Matthew P. Galasso 

A DISSERTATION 

Submitted to 
Michigan State University 
in partial fulfillment of the requirements   
for the degree of 

Criminal Justice – Doctor of Philosophy 

2024

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ABSTRACT 

Courts at every jurisdictional level are tasked with managing conflict and settling 

disputes and disagreements between individuals, and further, between individuals and the 

government. In these environments, cooperation may seem to be a difficult, if not 

impossible, task. However, previous research in the field of public administration is quite 

clear in that directly engaging the public in the public services process improves the overall 

experience for the service user, the service provider, and the public at large through the 

concept of co-production. While co-production research has been applied to policing for 

over fifty years, scholars have only recently begun applying the concept to courts. Using 

survey data gathered as part of the Public Engagement Pilot Projects conducted by the 

University of Nebraska Public Policy Center and the National Center for State Courts, this 

dissertation investigates which factors may encourage people to engage in co-production 

with the courts in their communities. Findings indicate that levels of trust in courts, as well 

as education and status as leaders in the community influence how people engage in co-

production with courts, comporting with the overall literature on co-production and the 

justice system and other forms of public services. Practical implications of this research, 

among others, may include more efficient use of public resources and overall 

improvements in process and outcomes for those involved in the court system. 

 
 
This dissertation is dedicated to the memory of our angel, Winnie Grace. Even though 
you’re not here to share this with us, you’ve inspired every word.

iii 

 
 
 
 
ACKNOWLEDGEMENTS 

First, I would like to thank Christina DeJong, my chair and my mentor. You have 

shown ultimate patience with me as I have worked through this process, sometimes 

running and sometimes crawling. I know I would not have gotten it done without your 

guidance and assistance. You were one of the first faculty members I met at Michigan State 

when I sat in on your methods class before enrolling in the program, and that positive 

experience was a sign of good things to come. 

I would also like to extend my heartfelt thanks to the members of my dissertation 

committee: Drs. Joe Hamm, Chris Smith, and Cory Smidt. Each of you provided a significant 

amount of time and expertise at a time when I needed it the most. Your comments and 

critiques have been invaluable in helping me craft what I believe is a valuable piece of 

scholarship of which we can be proud. To Dr. Hamm in particular, I want to thank you for 

being so willing to act as a point of contact with the PEPP team and assist me with getting 

access to the data. Finding data sources can be difficult for people in my situation, and you 

made it much less stressful. 

Thanks also go out to the researchers, staff members, and officials that make up the 

PEPP team at the University of Nebraska Public Policy Center and the National Center for 

State Courts. I appreciate the assistance with getting access to the data, as well as your help 

in getting early context for my study. I would also like to thank all of the respondents 

behind the data. Even though I will never get to meet any of you, I hope that your time with 

the courts or court administrators involved proved valuable and I thank you profusely for 

your time and participation.  

iv 

 
 
 
 
 
To the faculty and staff of the School of Criminal Justice at Michigan State, I cannot 

express in any way how much you have helped me get through the past eight years and 

finally reach this point. It has definitely been a long and winding road—cue McCartney—

but you should know that I credit many of you with my successful completion of this 

project. The skills I have learned both in the classroom and in the field have made this an 

obtainable feat. To Jennifer Cobbina-Dungy, being able to work with you during the first 

few years of my time at Michigan State set me on the path both toward success on this 

dissertation as well as my chosen career path as a teaching-focused faculty member. You 

were also one of the faculty members who supported my wife and I as we suffered our loss 

in 2019, which likely kept me in the program, so once again, thank you. 

There were days, even weeks, when I struggled with completing this dissertation. 

Motivation would sometimes escape me. Professionally and academically, the one thing 

that was always constant and the one thing that kept me in the pocket was that I got to 

teach a fantastic group of students each semester, and completing this dissertation was the 

best way of assuring that I will continue to be able to teach fantastic groups of students in 

future semesters. To all of my students, being able to share my knowledge with you and 

being able to learn from you every time I step into the classroom has been a saving grace 

and has been one of the biggest motivators for me to finish this dissertation. 

Family matters. It takes a village. Pick your phrase. I have a terribly supportive 

community around me that has lifted me up on innumerable occasions throughout the past 

eight years, and many more beyond that. I still remember telling my in-laws that we were 

going to be coming up to Michigan State and the joy that induced in them that has yet to go 

away. I remember calling my parents down in Indiana in a rush to have them come up and 

v 

 
 
 
 
 
mind the house so that I could be with my wife at the hospital for the joy and sorrow of the 

birth and near-immediate loss of our daughter. Being able to visit with and love on the 

menagerie of nephews (and a niece) that we have is always joyful and a worthwhile 

diversion from my work. I also get inspiration from my sister who is doing what she loves 

in Bloomington and the redemption and success embodied by my cousin and her husband 

and their terribly strong boys. 

And for the strongest person I know, my wife Jennie. If there is anything I know, far 

more than I know anything in the pages of this dissertation, I know that there is no way on 

the Lord’s great Earth that I would be sitting in this basement writing these words without 

your presence in my life. You have celebrated me in my successes, corrected me in my 

wrongs, supported me in my sadness, and encouraged me in my doubts. Thank you is not 

enough, but it will do for now. 

vi 

 
 
 
 
 
TABLE OF CONTENTS 

Chapter 1: Introduction…………………………………………………………………………………………………...1 

Chapter 2: Co-Production………………………………………………………………………………………………...5 

Chapter 3: Co-Production in Criminology and the Law…………………………………………………….13 

Chapter 4: Data and Methods…………………………………………………………………………………………29 

Chapter 5: Results and Analysis……………………………………………………………………………………...40 

Chapter 6: Discussion and Conclusion…………………………………………………………………………….56 

REFERENCES…………….……………………………………………………………..…………………………………...67 

APPENDIX A: IMPUTATIONS………………………………………………………………………………………….74 

APPENDIX B: ALTERNATIVE MEASURES OF KEY CONCEPTS…………………………………………..76

vii 

 
 
Chapter 1: Introduction 

If one was to pick up any newspaper or load any news website, it would be quite 

easy to find stories on how the public and government interact with each other. The 

dominant image displayed by many of these stories is one of conflict or controversy (Diaz, 

2023, “Jan. 6 rioter…sentenced to prison”; Robertson, 2024, “…[N]onprofit led by lieutenant 

governor’s wife ‘seriously deficient’). One component of government that deals regularly 

with conflict and controversy is the judiciary. Inherent within the judiciary is a tension 

between the government, as a representative of or proxy for broader societal interests, and 

the public at large. The courts, at every jurisdictional level, are tasked with managing 

conflict and settling disputes and disagreements between individuals (i.e., civil cases) and 

further between individuals and the government (i.e., criminal cases). In an environment 

that deals with conflict regularly and is often characterized by power differentials among 

actors, cooperation may seem to be at best not possible and at worst, counter-productive. 

However, even in this environment, it is possible and even beneficial to have all the actors 

working together to plan the services, deliver the services, and determine whether the 

services have been effective. In other areas of public service, the literature is quite clear in 

that directly engaging the public in the public services process, or co-production, improves 

the experience for the service user, the service provider, and the public at large (Ugwudike, 

2017). 

Through the latter part of the 19th and first half of the 20th century, the American 

government was relied upon to guide the country through large-scale events like the 

Industrial Revolution, the Great Depression and two world wars (Bryson et al., 2014). 

Throughout that period, the government was viewed by the public as capable of handling 

1 

 
 
public problems and providing public services without citizen involvement (Bryson et al., 

2014). If anyone outside of the government was involved, their input was secondary to 

politics; the public’s role was narrowly confined to that of client and voter. Through the 

middle of the 20th century, the public administration field saw a shift from this 

bureaucratic mindset toward market-based or market-like solutions, including seeing the 

public as consumers and “participating ‘beyond the ballot box’” (Frederickson, 1996, p. 

265).  

Scholars have studied citizen participation in government dating back at least to the 

1970’s through the work of Elinor Ostrom and her colleagues at Indiana University. Within 

this research, service production and service provision were initially considered as 

separate, though the focus remained largely on supply over demand. The co-production 

scholarship eventually shifted to a greater focus on involving the public in co-production 

and the public service functions they could fulfill, which can include design, management, 

delivery, and evaluation (Osborne et al., 2016). 

There are several potential benefits to co-production, including therapeutic benefits 

such as determining needs and building trust and communication (Needham, 2008). 

Incorporating other members of the community, and incorporating service providers into 

the community, through co-production promotes community-centered interests such as 

hearing all voices (McCulloch & PPPF, 2016), efficiently using scarce resources (Boyle & 

Harris, 2009), and achieving just and fair outcomes (Loeffler & Timm-Arnold, 2020). The 

concept of procedural fairness also plays a role in that the public is more likely to cooperate 

with the service provider (in this case, accept and abide by the courts’ decisions) if they 

perceive the service provider to be fair and ethical (De Cremer & Tyler, 2007; Tyler, 1988). 

2 

 
 
As the public services field evolves, new understanding and new values will lead to 

changing practices. The importance of co-production lies in its inclusion of everyone 

involved in the public services process in determining what that process should look like 

and how it should operate. These benefits have been extensively studied in many public 

services contexts such as health care, education, disaster relief, and public utilities 

(McLennan, 2020; Van Eijk & Steen, 2016). Much of the early co-production literature 

actually focused on policing and public safety, and several have applied co-production to 

correctional environments, but few have studied this issue in relation to courts. The 

research that does exist shows some valuable exchange of feedback, but an accompanying 

reluctance to include lay people in the co-production process as a response to the potential 

loss of control of the format and results of the co-productive processes (Van Gils et al., 

2021). 

If part of the goal of co-production is to involve the public more fully in the public 

service process, and to do so in a sustainable way, then a relationship between the public 

and the service provider (i.e., the government in most cases) needs to be established. Trust 

can be also part of the transformation of behavior from the acrimonious nature of 

bureaucracy—and the actual or perceived adversarial nature of the courts—to that of co-

production, though this relationship does not always have to be perfectly equitable to be 

sustainable. For example, if the public sees that the endeavors in which they are 

participating have a tangible effect on their lives, they may be more willing to accept some 

responsibility and some risk for the public service process (Loeffler, 2021). 

This dissertation will investigate the role that co-production might play in the 

design, management, delivery, and evaluation of court services by examining data collected 

3 

 
 
 
as part of the Public Engagement Pilot Projects in several courts in different states. The 

dissertation will proceed as follows: Chapter 2 contains a review of the foundational 

literature on co-production, including a chronological look at the development of the co-

production literature as well as the primary conceptualizations of co-production used 

throughout the proposal.  

Chapter 3 contains an overview of the more limited literature applying co-

production to criminology and the law. Included in Chapter 3 is the progression from the 

early focus on policing and public safety and limited public involvement to increased public 

involvement and importance of public perceptions of what co-productive activities they 

can accomplish. Also emphasized is the fact that the research on co-production and the 

courts, as well as trust’s involvement in co-production and the courts, is limited. Chapter 3 

closes with a recitation of the research opportunities presented in the extant research and 

a presentation of the research questions to be examined in this dissertation.  

In Chapter 4, I begin by introducing the Public Engagement Pilot Project initiative, the 

engagement efforts conducted with the courts, and the data that were collected that are the 

focus of the dissertation. I propose three hypotheses generated from my research 

questions and prior research centered around the concepts of co-production, procedural 

fairness, risk, and trust. I then walk through the process of generating the measures I used 

to answer the research questions and hypotheses, as well as the analysis plan for the 

results that are presented in Chapter 5. Ordinary least squares (OLS) was used as the 

statistical technique, and, after accounting for missing values through multiple imputation, 

findings were mixed. Finally, Chapter 6 closes with discussion of those results, implications 

. 

of my research, and concluding remarks

4 

 
 
Chapter 2: Co-Production 

History and Definitions

One of the first notable mentions of citizen participation in government comes from 

Ostrom (1972). Through the dual lenses of reform and economics, Ostrom discusses the 

relationship between the size, number, and professionalism of governmental units in each 

jurisdiction and how those inputs affect outputs such as efficiency, responsibility of public 

officials, and most applicable to the current study, citizen participation.  

According to Ostrom, the reform tradition proposes that increasing the size of 

governmental units, professionalism, and the reliance upon hierarchy will all have a 

positive relationship with the dependent variables of interest. Interestingly for the 

purposes of co-production, increasing the number of locally elected public officials is 

proposed to have a negative relationship with citizen participation and responsibility of 

public officials. Compared with this perspective, political economy views urban problems 

in a significantly different manner. Instead of pairing a sizeable and professional 

government with participatory citizens, political economists see citizen participation 

occurring when the smallest possible governmental unit is used that can efficiently 

perform the task at hand (Ostrom, 1972). 

However, Ostrom does not provide a clear definition of what citizen participation is 

in this article beyond mentioning “the participation of citizens in political life” (p. 480, 

quoting Stigler, 1962) and the fact that it is among the concepts that may lack consistent, 

careful, and valid definitions (p. 487-488). Ostrom et al. (1978, p. 383) continue this 

examination of the relationship between the government and the public by looking at 

police services, in which the authors explicitly use the term “coproducers” to describe how 

5 

 
 
 
the public’s activities related to safety (i.e. locking doors, reporting crime, providing 

evidence) can contribute to police outputs such as arrests and the return of stolen 

property. Parks et al. (1981) similarly focus on the activities completed by public service 

agents and the public more so than the people engaged in the co-production. According to 

the authors, a combination of technological, economic, and institutional considerations 

determines whether mixing organized production of public services with consumer 

production is appropriate; co-production is “a mixing of the productive efforts of regular 

and consumer producers” (p. 1002). 

Another line of scholarship focuses more on the individuals or groups engaged in 

the co-productive process. Osborne et al. (2016, p. 640) define co-production as “the 

public service users 

voluntary or involuntary involvement of 

[emphasis added] in any of the 

design, management, delivery and/or evaluation of public services.” Alford (2014) notes 

that co-production can be a two-person event, but it is just as likely to be a set of 

interactions between individuals as well as groups. Van Eijk and Steen (2016) are heavily 

focused on the co-producers and why they engage in co-production, though interestingly 

their definition of co-production is based on a list of broad and under-defined activities in 

which both professionals and citizens can engage related to public services (i.e., co-

planning, co-prioritization, co-assessment). 

This breadth of conceptualization has been seen as both a limitation and a strength 

in the co-production literature. According to some scholars, co-production has been used in 

enough disciplines and applied to enough different activities, individuals, and groups that 

the term is too broad to be of practical use (Bovaird & Loef�ler, 2012). Durose et al. (2017) 

recognize this breadth of conceptualization, but also see a signi�icant practical bene�it. Co-

6 

 
 
 
 
production requires versatility in how public resources are used, the expertise of 

professionals and citizens is engaged, all to share both control and responsibility. According 

to the authors, the tangible bene�it to that versatility is that co-production then becomes 

appealing across the political spectrum. Brix et al. (2020, pp. 171-172) take a different 

approach to the breadth of conceptualizations of co-production, arguing that “[their] study 

does not consider it essential that co-production does not have a common theoretical 

de�inition….” Instead, they argue that the focus of co-production should be on the local 

context and what activities and forms that context dictates.  
Co-Production in the Public Administration Field

Over the past forty years, the prevailing governance model has changed, along with 

the ability for members of the public to participate therein. Prior to the Civil War, 

governance was hyper-local, where people would individually settle new territory and 

have to create new social, political, and business structures for themselves (Seymour, 

1878). Government was increasingly relied upon through the latter part of the nineteenth 

century and first half of the twentieth century to guide the country through several 

character changing events, such as the Industrial Revolution, the Great Depression, and two 

world wars (Bryson et al., 2014). In the eyes of politicians, though, the public’s role in 

governance within this “traditional public administration” (Nederhand et al., 2019) was 

confined to its status as clients, voters (Kim, 2021), and constituents, or what Meinke 

(2008, p. 447) calls a “mass electoral audience.” 

th

However, by the middle of the 20

 century, a paradigm shift was already underway 

in the public administration field. The primacy of bureaucracy began to be questioned 

(Bryson et al., 2014) and criticized as too insular, too focused on efficiency, and too 

7 

 
 
 
 
 
dismissive of discretion (Denhardt & Denhardt, 2000; Lynn, 2001). During this period of 

New Public Management (NPM) (Bryson et al., 2014; Hood, 1991; Osborne et al., 2016), 

members of the public were increasingly considered as consumers, befitting the market-

based tendencies of the field. Members of the public had the right to express their 

preferences and complaints on how the government provided public services. According to 

Kim (2020), this customer-based orientation through expressed preferences and 

complaints did not allow for direct participation in public services.  

It is at this point that the direction of the public administration field becomes a little 

less clear. Frederickson (1996) presents as an alternative to NPM the “reinventing 

government movement.” Importantly for the study of co-production, reinventing 

government takes the “citizen as consumer” concept even further than NPM by “breaking 

the bureaucratic service monopoly” (p. 265). Not only are citizens consumers of public 

services, but they also make individual choices that work best for them even to the 

detriment of others. Any consideration of consensus, social equity, or collective democracy 

is secondary to individual interests (Kim, 2020). 

Moore (1995) looks at creating public value both from the citizen’s as well as the 

public manager’s perspective. Citizens want to see three things from governmental 

organizations: high-performing bureaucracy, efficient and effective achievement in 

attaining desired social outcomes, and just and fair operations leading to just and fair 

outcomes for society at large. From the manager’s perspective, success comes from a 

“strategic triangle”: achieving something substantively valuable; sustainable both through 

legitimacy and politics; and operational and administrative feasibility (pp. 22-23).  

8 

 
 
 
Any overlap between these perspectives, whether it is in feasibility, performance or 

outcomes, can only contribute to the co-productive relationship. If the relationship 

between service providers and service users is sufficiently co-productive, the public will be 

able to hold public agencies and administrators accountable more directly as the public 

develops a closer relationship with them (Loeffler & Timm-Arnold, 2020). Co-production 

brings the citizenry and nonprofit sector to public problems for which government may not 

have ideal answers and sometimes involves a level of “desperation” on the part of the 

government due to an inability to secure sufficient resources to produce the desired public 

services (Cheng, 2019, p. 206; see also Haeffele & Storr, 2019). 

Most of the co-production literature acknowledges that the exact co-productive 

relationship between service user and service provider is going to be different in individual 

contexts. Weaver (2011) draws a connection between co-production and the concept of 

“personalization,” which by nature is focused on the service user. The levels of 

personalization can differ, as more shallow levels of personalization may be limited to the 

service provider giving individuals some choices. However, she indicates that co-

production is a form of “deep” personalization focused on regular and long-term 

relationships. Bovaird (2007) provides an in-depth consideration of the professional-user 

relationship demonstrated by case studies on a continuum from sole professional delivery 

to full co-production and then to traditional self-organized community delivery.  

9 

 
 
 
 
Table 1. Case studies from Bovaird (2007) and the design, planning and delivery of services. 

Porto Alegre, Brazil 
budgeting process 
Gateshead, England 
family support 
initiative 
Caterham, England 
community trust 
Falmouth, England 
community partnership 
France, Villa Family 
elderly housing 
Tackley, England village 
shop 

Design 
User/community 
consultation 
Professional 

Planning 
User/community 
consultation 
Professional? 

Provision/delivery 
Traditional professional 

User co-delivery 

Full co-production 

Full co-production 

Full co-production 

No formal process 

No formal process 

Professional? 

Professional 

User/community co-
delivery 
User/community delivery 

Co-designed 

Co-planned 

User/community delivery 

Two examples from Table 1 illustrate how the responsibilities for delivering public 

services can shift between the public and professionals, all based on the needs of the 

program and the public. The Gateshead family support initiative was designed by 

professionals to promote the health of children under the age of four using counselor visits 

and publicity campaigns. Parents were trained as peer counselors and course organizers as 

a form of co-delivery. In contrast, the Tackley village shop was planned from the outset by 

the community based on a need they saw for a central point from which postal services, 

leisure facilities, a café, and a community meeting room could spin out (Bovaird, 2007).
Benefits of Co-Production

Each individual is going to have a different willingness to participate in co-

productive relationships. Some researchers see co-production as being more individualistic 

or reciprocity-dependent. For example, Bovaird and Loeffler (2012) emphasize reciprocity 

in the use of resources between the public sector and citizens without a focus on equality in 

any long-term relationship. In contrast, Needham (2008) notes that co-production can 

produce therapeutic as well as traditional diagnostic benefits that can apply to individual 

10 

 
 
 
 
 
 
 
citizen-provider relationships as well as system-level relationships between the 

community and the public sector. Therapeutic benefits include making sure everyone’s 

voice is heard, generating “credible commitments” (p. 223), and building trust; regular 

advisory board meetings or focus groups could be examples of co-productive efforts along 

these lines. Diagnostic benefits involve making sure that the community has a higher 

quality of life, making sure that scarce public resources are used to help as many people as 

possible, and being able to identify and resolve issues as they occur. A successful co-

productive effort can also serve as an example to the community of how being “civically 

minded” can be extended to other areas in life (Needham, 2008, p. 223, see also Ostrom, 

1996). Successful efforts at co-production with housing authorities may translate to a 

facility at co-production with health care workers and local politicians, increasing the 

quality of services the obtain beyond the initial co-productive encounter (Ostrom, 1996). 
Why Co-Production Matters

Public services—whether provided by public agencies, private agencies, or private 

individuals—require significant financial and human capital. Co-production on balance is 

an effective way to harness many different sources of capital but sustaining co-productive 

efforts may be difficult. Addressing these difficulties means that “we need to understand 

how institutional, cultural, and biophysical contexts affect the types of individuals who are 

recruited into and leave particular types of collective action situations” (Ostrom, 2000, p. 

154, see also Van Eijk & Steen, 2016). Public agency (or government more broadly) reform 

has been a recent hot button topic for numerous reasons, including the Supreme Court’s 

rejection of long-standing judicial deference to federal agency interpretations of relevant 

Loper Bright Enterprises v. Raimondo

law (

, 2024); however, this “frustration with the 

11 

 
 
 
 
functioning of public agencies” is almost as much a constant as the existence of those 

agencies (Ostrom et al., 1978, p. 381; see also Adams, 1984). In an environment of reform, 

co-production may play a role in establishing more long-term public service regimes. Part 

of this reform is making sure that the public services stakeholders participate in designing 

and delivering those services, making or modifying rules, and determining what constitutes 

public value (Alford, 2008; Boyle & Harris, 2009; Ostrom, 2000). And according to Osborne 

et al. (2016, p. 640), “co-production is currently one of [the] cornerstones of public policy 

reform across the globe.” 

This cornerstone role can be seen in a number of different applications: public 

utilities, schools, neighborhood councils, care for the elderly, among others. One of the 

areas of public life in which co-production has not been as regularly applied is the justice 

system, whether civil or criminal. This issue will be discussed in more detail in Chapter 3, 

where I will discuss the extant research on applying co-production to justice-related 

organizations and where that extant research may be lacking. 

12 

 
 
 
 
 
 
Chapter 3: Co-Production in Criminology and the Law 

As discussed in Chapter 2, the co-production literature is quite robust and rich. 

However, the applications of co-production that dominate the discussion do not often 

involve the loss of property or loss of freedom through criminal or civil sanction. It seems 

counterintuitive that a person who might lose some of their property in a civil suit, or 

someone who is facing a criminal sanction, would be offered or willing to accept a co-

productive role in the system that is set to deprive them of their liberty or property. 

However, many of the benefits that result from co-production in other public service fields 

(i.e., health care, public utilities, disaster management, and schools; see McLennan, 2020; 

Van Eijk & Steen, 2016) may well lead to similar positive outcomes for many stakeholders 

in the justice process, including those who have been subjected to coercive action in areas 

of public services such as criminal justice and mental health (Osborne et al., 2016). 

Beginning in earnest with the work of Elinor Ostrom and colleagues in 1978, the 

potential role of co-production has been seen in police services, community justice 

(Ugwudike, 2017), community policing (Scott, 2002), encouraging desistance from crime 

and contribution to restorative justice (Loeffler & Bovaird, 2020), and family law (Van Gils 

et al., 2021). However, the extant literature on co-production and how it can be applied to 

the justice system is still underdeveloped as related to courts. The first section of this 

chapter will review the early literature on co-production and the justice system, with an 

emphasis on establishing which justice services might be appropriate for co-productive 

activity. The second section of this chapter will review the more recent literature on who 

engages in co-production with justice-related agencies and why they do so, with the 

13 

 
 
 
chapter closing with a discussion of the questions left in the literature that this dissertation 

is designed to answer. 
Early Developments 

Throughout the course of American history, citizens have made contributions to 

public safety. Early versions of what we would call a “neighborhood watch” came as early 

as the immediate post-colonization period. Members of a community would rotate as the 

watch person whose duty was simply to notify the community of the risk of danger. Even as 

governmental entities involved in legal and public safety concerns came into play, citizens 

contributed to their own public safety through self-protective actions such as securing their 

property (Ostrom et al., 1978) and contributing to informal controls such as faith 

communities and family units. 

It is this type of citizen action that generates the first significant research on the 

criminal justice system and co-production. Ostrom et al. (1978, p. 389) claim that 

“[v]iewing citizens as co-producers of police (and other social) services is a rather novel 

and important aspect of our approach.” The aforementioned self-protective actions were a 

pre-existing way that citizens could contribute to their own public safety, but Ostrom et 

al.’s work makes that connection more direct and explicit in the literature. 

Their subjective outcomes (i.e., perceptions and evaluations) provide another point 

of connection between co-production and the justice system. Not only is it appropriate to 

look at the actions taken by the public toward their own safety, it is also appropriate to 

consider how the public feels about their interactions with criminal justice or legal 

agencies, programs, and processes. The justice system overall is not always front of mind 

for a significant portion of the American public, so the limited interactions that members of 

14 

 
 
 
 
the public have with the system can have a profound impact on those perceptions 

(Longazel et al., 2011). Courts are not often mentioned in glowing terms (Wegman, 2024, 

“The Supreme Court is gaslighting us all”). For the majority of the public that has limited 

contact with the courts, that contact is often through a jury summons or in the context of a 

dispute, whether it is a divorce proceeding, a property dispute, or an allegation of criminal 

activity. The courts may have a role to play in improving the public’s perceptions of what 

they are and what they do (Sun & Wu, 2006), thus contributing to the efficient function of 

the courts much as the public may also contribute to a better experience using court 

services through co-productive action (i.e., getting through proceedings in a timely manner, 

abiding by the court’s decisions). 

Considering Ostrom et al. (1978) and Parks et al. (1981), the question then becomes 

which public services within the realm of criminal justice or the courts might be 

appropriate for co-productive activity. Whitaker (1980) illustrates the initially pessimistic 

outlook that law enforcement had on their ability to control crime without any public 

assistance through an excerpt from a police administration textbook: “if there are no 

effective forces of community social control at work, there is little if anything the police can 

do to deal with crime and lawlessness” (p. 243). However, in the next thought, Whitaker 

mentions how members of the public will easily turn to the courts—often perceived as a 

force for imposed or mandatory behavior change through punishment or civil damages—to 

seek assistance with unjust laws or regulations. Whitaker, even though he sees less co-

productive potential when forced choice or behavior change is the focus, still acknowledges 

non-cooperation as a legitimate method of influencing policy: “Citizens increasingly 

influence public policy by their non-cooperation whether it is recognized formally through 

15 

 
 
court suits or, more commonly, through the acquiescence of public officials when citizens 

fail to comply” (p. 244). 
Modern interpretations 

Policing and public safety 

The connections between co-production and the law have continued in more recent 

years, building on the work established by Ostrom et al. (1978), Whitaker (1980), Parks et 

al. (1981), and others. One of the disciplines in which co-production has been more 

regularly applied is policing and public safety, with mixed results. Within their broader 

conceptual model of outcomes in policing and criminal justice, Loeffler and Bovaird 

(2020) have divided the purpose of policing and the criminal justice system into two 

parts: helping the community feel safe and achieving justice in the community. To make 

their communities feel safer, the public can engage in co-productive activities to reduce 

opportunities for crime, deter crime, encourage desistance, and remove criminals from the 

community. Interestingly, reducing the social causes of crime is something that Loeffler 

and Bovaird see as being outside the influence of policing and the criminal justice system 

entirely, even though they recognize it as a “critically important driver of crime reduction” 

(p. 209)

1
. Of these, Loeffler and Bovaird see reducing opportunities for crime as ripe for 

citizen contribution, much along the lines of Ostrom et al. (1978) (i.e., police encouraging 

the public to lock their homes and vehicles). There are also opportunities for co-

production in encouraging desistance, often including some form of peer support, as well 

as reporting crime and giving evidence as witnesses (Loeffler & Bovaird, 2020). 

1

 While other co-productive activities could target these social causes of crime (i.e., housing, education, health 
care), they are beyond the scope of this dissertation. 

16 

 
 
 
 
However, each of these co-productive activities shares a common characteristic: 

each primarily, if not exclusively, involves the public in the co-delivery of services. Any 

involvement of the public in the co-commissioning, co-design, or co-assessment of public 

safety, policing, and criminal justice services has been minimal (Bovaird, 2007; Loeffler & 

Bovaird, 2020; Nabatchi et al., 2017). Taking a different approach, Uzochukwu and 

Thomas (2018) look at the co-production of public services in the city of Atlanta among a 

sample of participants in the city’s neighborhood planning units: a set of 25 citizen 

advisory councils that make recommendations to the city on neighborhood-level issues 

(Atlanta City Council, 2024). Among their sample, many participated in the typical co-

delivery of services; in their study, this took the form of community clean-ups or 

neighborhood patrols. Other reported co-productive activities have more indirect impacts 

on criminal justice, including sharing feelings about a policy or project concerning the 

community, attending a training session, reporting service malfunctions (i.e. potholes and 

streetlight outages), and reporting neighbors for code violations (Uzochukwu & Thomas, 

2018). 
Courts 

Van Gils et al. (2021) conducted a study of how Dutch courts engaged in co-

production with the public by gathering feedback during “mirrormeetings,” where 

professionals in a field meet with their clients and ask them to talk about their experiences 

in the presence of the professionals

2
. These mirrormeetings shared many of the same 

characteristics as other feedback mechanisms such as focus groups or surveys would, 

2

 In the article title, these are described as “mirrormeeting-focusgroups.” They share many of the same 
characteristics, except that in mirrormeetings, the people being discussed are in the room but most often are 
nonparticipants in the discussion. As the authors do, I will simplify by using only mirrormeetings. 

17 

 
 
 
 
including the same subjects (experiences of court users), the same content (court 

limited 

function), and the same targeted audiences (court professionals and in 

examples, 

lay people).  

The value of Van Gils et al.’s (2021, p. 164) work to the present study comes in 

what they find missing from the mirrormeetings and what they mention as future 

directions for research. From the beginning of their work, they note that “[mirrormeetings 

and focus groups] are almost absent in literature on court administration.” In a significant 

departure from other studies on co-production, the mirrormeetings examined in the Van 

Gils et al. (2021) study for the most part excluded lay people, with lawyers, bailiffs, and 

other public authorities as the primary participants. While co-production can occur 

between professionals and public officials without any involvement of the public (Alford, 

2014), most research on co-production includes the public or citizens as an integral 

component (e.g., McCulloch & PPPF, 2016; Uzochukwu & Thomas, 2018; Van Eijk & Steen, 

2016). The authors note as a significant limitation of the data the fact that the judges select 

both the topics to be discussed in the mirrormeetings and the participants, leading to the 

regular occurrence of no change needing to happen; in the alternative, the judges may be 

too cautious to seek the input of lay people and prefer the relative safety of legal 

professionals: 

That evokes the question why mirrormeetings with non-professional court 
users have not been organised more often, considering the positive 
experiences in Amsterdam and Rotterdam. There seems to be some 
reluctance in many courts to communicate directly with lay court users 
about their experiences. This shows a tension between the need for control 
on the content and selection of participants by the organisers of 
mirrormeetings in the courts and the possible benefits of a more open 
approach with the possibility of unforeseen feedback (Van Gils et al., 2021, p. 
176-177). 

18 

 
 
 
 
As the authors note, limiting feedback to areas the courts find important may exclude areas 

that would be of significant value to the public, with whom the courts could co-produce the 

necessary services. This limitation has a practical rationale; courts in the Van Gils et al. 

study limit feedback to issues for which implementation is easy. Also, they may limit 

feedback on issues for which they simply do not want suggestions (p. 177). 
Trust and Co-Production 

To this point, co-production has been described in more transactional terms: goods 

and services, feedback, evaluation, delivery, and more, or what Needham (2008) would call 

diagnostic co-production. However, for co-production to be a viable long-term strategy to 

improve the delivery of public services, whether in criminology and the courts or in other 

venues, there has to be a continuing relationship between the parties that regularly 

encourages both the public entity and the citizens to continue to co-produce (Bovaird, 

2007; Needham, 2008). Many consumer-producer relationships are categorized in 

adversarial ways: each party is trying to get the most benefit out of the relationship, 

thereby determining who at least feels like they got the “win” from the deal (Needham, 

2008). This can even be seen in the provision of public safety; the horse trading that goes 

on between the public, law enforcement, and elected officials for services, influence, and 

limited financial resources often results in acrimony. One of the most prevalent examples of 

this over the past five years has been the “defund the police” movement. It became unclear 

what “defunding” actually meant, what disbanding police would entail, what police reform 

might look like, and what the public’s role in the process would be. All of these 

uncertainties in many cases lead to abandoned efforts, continued tensions, and public 

safety issues that “remain raw and unresolved” (Londoño, 2023). 

19 

 
 
 
However, the “transformation of citizen behavior” (Whitaker, 1980, p. 243) that 

comes with service delivery changes can be a starting point toward Needham’s therapeutic 

co-production. Therapeutic co-production helps to build communication between 

participants, and it helps to build trust between participants as well. Trust is an often-

discussed and often-misused concept in overall society, but it has a valuable place within 

the study of co-production. First, several of the most common definitions of trust will be 

presented. This will be followed by a discussion of how trust has been included in the 

broader co-production literature and how trust has been included in the literature on co-

production in criminology and the law. 
Defining trust 

Given the scope of the trust literature—and what would seem to be common 

sense—it might be surprising to think that early trust research focused so strongly on the 

trusting party over the party to be trusted (Mayer et al., 1995). According to the authors, 

trust is “the willingness of a party to be vulnerable to the actions of another party based 

on the expectation that the other will perform a particular action important to the trustor, 

irrespective of the ability to monitor or control that other party” (p. 712). For the authors, 

the vulnerability component is key. Without the ability to control the other party or the 

ability to monitor whether they are taking the agreed upon action, the trustor may be 

placing themselves at some level of risk. Similarly, Rousseau et al. (1998, p. 395) define 

trust as “a psychological state comprising the intention to accept vulnerability based upon 

positive expectations of the intentions or behavior of another.” Both definitions share the 

psychological component—vulnerability—but Rousseau et al. do not focus on the actions 

of another, just on positive expectations of their intentions or behavior. There is an 

20 

 
 
 
 
interesting comparison to be drawn here between the trust literature and the public 

administration literature when it comes to risk. As seen here, trust and risk—or the 

willingness to accept risk—are closely linked. However, in the public administration 

literature, the connection is not present or not as strong (e.g., Brown & Osborne, 2013; 

Loeffler, 2021). Loeffler presents low trust or a lack of trust as a barrier to co-production; 

risk management is tied closely to co-production without any connection to or mention of 

trust.  Similarly, Brown and Osborne (2013) draw a close positive connection between 

innovation and risk without envisioning any role for trust. 

Both definitions concentrate on the perspective of the trustor. Additionally, each of 

these foundational works examines what is needed of the trustee for trust to be 

established. Mayer et al. (1995) gives three essential characteristics of a trustworthy 

trustee: ability, benevolence, and integrity. Ability is a set of skills that enable the trustee 

to have influence within their domain. Benevolence is the extent to which they are 

believed to want to do good to the trustor, aside from any profit motive. And finally, 

integrity is the perception that the trustee adheres to an acceptable and consistent set of 

principles. For Rousseau et al. (1998), the trustee is either at risk of sanction for breach of 

the relationship or there is a pattern of interactions and a long-term trust relationship in 

play for trust to be established.  
Trust and co-production 

Throughout the co-production literature, there is a concern that the individuals or 

groups engaging in co-production might only treat the co-productive relationship in a 

transactional manner (Needham, 2008), similar to how Rousseau et al. see some forms of 

the trust relationship. Other scholars see this level of engagement as an entry point to a 

21 

 
 
 
co-productive relationship; Loeffler and Timm-Arnold (2020) progress through three 

different forms of governance (hierarchies, markets, and networks), showing how the 

potential for co-production increases as the engagement between citizen and traditional 

service provider (i.e., the government) increases. 

Similarly, Van Eijk and Steen (2016) view ad hoc engagement as the first rung of a 

metaphorical co-production ladder: the citizen must understand how the service will 

affect their family and their community. Those forms of salience—personal and social—

then lead into consideration of just how much effort it would take to engage in recurring 

co-production and what the results might be: the second rung of their ladder. First among 

these is ease, which is common sense: if the task requires too much effort, there is less 

likelihood of involvement. Another consideration they posit is trust, which they circularly 

operationalize as “trust in the ‘system’ perceived when deciding about whether to engage” 

(p. 40). However, looking at the other component at this stage, one can see much more of a 

progression toward co-production and trust. According to Van Eijk and Steen (2016), 

external efficacy in this context involves asking the following question: “Does the 

government provide room for my interaction, and will it matter?” (p. 31). The focus group 

participants they interviewed felt it was very important that they have input and that 

public officials were willing to listen to their concerns, bringing to mind notions of 

representation (Tyler, 1988) and procedural fairness (De Cremer & Tyler, 2007). This was 

combined with a significant presence of internal efficacy in successful co-productive 

relationships, where the governmental organization went beyond providing room for the 

interaction and progressed to allowing the respondents to effect change. 

22 

 
 
Fledderus et al.’s (2014) seminal work on trust and co-production provides a useful 

way of making this connection between efficacy, trust, and co-production. Their 

assumption throughout the work is that changing the relationship between service users 

and service deliverers (increasing co-production) will result in changing attitudes, 

specifically toward more trust (or at least less distrust). Without a feeling of self (or 

internal) efficacy expressed as the thought that the user’s actions have the potential to 

change policies or services, they may become disillusioned, disengaged from co-

production, and distrustful of the process and the institution. Conversely, Fledderus et al. 

note that if a service user does feel as if their actions influence their daily lives, they “will 

feel less risk in trusting others” because they are not as dependent on others (p. 433).  

This willingness to be vulnerable to the risk of alienation from the decision-making 

process (Mayer et al., 1995) can also be seen in more positive terms. If the expectations 

and goals are clearly understood from the beginning of the co-productive relationship, the 

change in attitude (Fledderus et al., 2014) should lead to more positive or benevolent 

expectations and interdependency (Rousseau et al., 1998). Weaver (2019) looks at 

prisoner-led councils engaged in co-productive partnership with prison administration in 

the United Kingdom. There are risks for both groups: the users risk significant effort and 

personal capital being expended with the potential for administration to backtrack or not 

provide accurate support, while the administration risks giving up some of their ability to 

control the users by granting limited authority to these councils. This co-productive 

relationship does not always have to be perfectly equitable to be sustainable, however. As 

noted by a prison officer, “[Prisoners] are not always gonna get what they want but they 

know things will be listened to and a decision will be made” (Weaver, 2019, p. 257).  

23 

 
 
 
Procedural fairness 

In the prison environment, maintaining order is the primary concern of the 

administration, along with the safety of the staff and the incarcerated individuals. 

However, maintaining order in prisons cannot be achieved by recourse to a system of 

rewards and punishments alone (Rottman, 2007). There needs to be another technique by 

which authorities gain compliance from the incarcerated. And it is not just prisons for 

which authorities need cooperation, compliance, and satisfaction from other parties. In 

1975, Thibaut and Walker hypothesized that people who were involved in dispute 

resolution decisions would be independently influenced by their judgments about how 

fair those processes were. They found strong support for the effect of procedural fairness 

on the assessment of dispute resolution decisions.  

This same effect has been proven across many different justice and non-justice 

environments. Rottman (2007) tracks the potential effects of procedural fairness 

throughout the criminal justice system from the family (if the parents treat their children 

in a fair manner, it should lead to less deviant and antisocial behavior), through to 

probation and parole (if the supervision officer or treatment provider treats the client in a 

fair manner, non-compliance should be reduced significantly) to the aforementioned 

corrections environment. Rottman notes that for courts, there are two publics or two ways 

to look at procedural fairness: one for those with court experience and one for those 

without experience. For those with experience, the focus is on indicators of 

trustworthiness of the judge; the impact of procedural fairness assessments is longer-

lasting regardless of whether the court proceeding was over a $500 car repair bill, child 

custody, adoption, or criminal prosecution. For those without experience, the focus tends 

24 

 
 
 
to be on issues of neutrality, political ideology, and media images of the justice system 

(Rottman, 2007, p. 838).  

 Early research on procedural fairness weighed which criteria were most important 

when assessing fairness, and research also asked whether everyone shares the same set of 

criteria or whether individual characteristics of the people involved, and the encounter 

would influence the choice of criteria (Tyler, 1988). According to Tyler, there are seven 

aspects of process that people consider when assessing fairness: authorities’ motivation, 

honesty, ethicality, quality of decisions, bias, any opportunities for representation, and any 

opportunities for error correction. Two positive findings were key. First is the most 

commonly stated finding among all procedural fairness research: judgments on how hard 

authorities try to be fair are the key overall factor in assessing procedural fairness. 

Additionally, one criterion from Tyler’s list that emerged as new at the time was ethicality, 

which he operationalized as “being treated politely and seeing one’s rights respected” (p. 

129). De Cremer and Tyler (2007) apply the concept of procedural fairness beyond the 

criminal justice system to social interactions and the workplace, noting that interactions 

can be more enjoyable and efficient when there is cooperation and a pursuit of collective 

goals. Procedural fairness can promote this cooperation, especially when the authority in 

question is trusted. 

There is one major caveat to the demonstrated effect of procedural fairness on how 

people assess their interactions with authority figures and people in social situations. In 

order for the authorities’ effort to be fair (Tyler, 1988) to have a positive effect on peoples’ 

evaluations and levels of cooperation (De Cremer & Tyler, 2007), those efforts have to be 

meaningful and sincere (Rottman, 2007; see also De Cremer & Tyler, 2007). Without 

25 

 
 
 
 
sincerity and meaning, people will not value the opportunities for interaction and 

cooperation and any reciprocity will be non-existent (Rottman, 2007). 
Research Opportunities

Since the 1970’s, a robust literature on co-production has developed on broad 

questions of how to examine citizen participation in government and metropolitan reform 

(Ostrom, 1972) to other questions of public governance and administration (e.g., 

Frederickson, 1996; Lynn, 2001, Nederhand et al., 2019). The literature has moved the 

citizen’s role in co-production from the limited, transactional, or passive, to the fully 

involved position of primary responsibility in some cases (see Table 1). However, within 

this robust literature there are opportunities for the co-production literature to develop 

even further, questions that have yet to be answered. 

Research on co-production began in the late 1970’s with how people could co-

produce public safety services, though those self-protective activities had been taking place 

in the United States for long before then (Ostrom et al., 1978; Whitaker, 1980; Parks et al., 

1981). This application of co-production to policing and public safety has continued in the 

ensuing decades (e.g., Loeffler & Bovaird, 2020; Nabatchi et al., 2017). In recent years, co-

production has also been applied in correctional contexts, such as community-based 

supervision (Ugwudike, 2017), prison councils (Weaver, 2019), and with people who are 

ex-offenders (McCulloch & PPPF, 2016). While there has been research on user 

participation in courts through the concept of compliance (e.g., McIvor, 2009; Tyler, 2006), 

applying co-production to courts remains an understudied part of the field.

26 

 
 
 
 
 
 
 
 
Court experience 

Courts are traditionally hierarchical regardless of jurisdiction (i.e., civil, criminal, 

community) (Loeffler & Timm-Arnold, 2020). Especially in criminal courts, but in any court 

environment, the citizen’s role is most passive. They are there to give testimony or they are 

a party to a case. Even then, they are typically represented by a professional attorney or 

there is a specific reason why an attorney is not participating. In some circumstances, their 

role is passive almost—if not fully—to the point of coercion. This comes out, of course, 

most commonly in criminal courts, but coercion can still come into play in other types of 

courts. As an example, participants in civil cases related to property (i.e., foreclosures and 

evictions) are likely to feel coerced into participation when they do not have the resources 

to properly defend their interests. Even when someone is serving on a civil or criminal jury, 

they are not there of their own free will; they were summoned to participation as part of 

their civic obligation. 

How willing might an individual who has had coerced encounters with the courts be 

to participate in any form of co-production with the courts they may feel have oppressed 

them? Conversely, could positive experiences with the courts lead to better co-productive 

efforts? Loeffler (2021) suggests that research should focus on perceived barriers to 

effective citizen co-production rather than actual barriers. It also remains to be what other 

factors may contribute to the influence of that experience, leading to Research Question 1: 
Research Question 1:  How does experience with courts affect the public’s co-

production of court services?

27 

 
 
 
 
 
 
Trust 

Trust, along with ease and efficacy, are among the motivations on Van Eijk and 

Steen’s (2016) ladder model of co-production. While they focus more closely on and find 

that more of their respondents mention efficacy as motivation for engaging in co-

production in their study, trust still has a role to play, especially once the decision to co-

produce has already been made. They invite further development of their model in other 

policy domains and other countries (i.e., other than the Netherlands) to look at 

motivational patterns across different types of co-production and different motivations and 

their impacts. This dovetails into the continued need in the literature to determine how the 

service experience integrates with a service user’s overall life experience, how that 

experience will affect their engagement with the service, and what they will bring to the co-

production table (Osborne et al., 2016). The centrality of psychological motivations for co-

production (Uzochukwu & Thomas, 2018; Van Eijk & Steen, 2016) ties into both trust and 

co-production but is not immediately considered when thinking about the courts. These 

questions and gaps in the literature lead to Research Question 2: 

Research Question 2: What role does trust play in the co-production of court services?

In the next chapter, I introduce the data I will use to answer these research 

questions, as well as the research project from which the data were gathered. After 

generating hypotheses from my research questions and the research opportunities in this 

chapter, I walk through the process of measurement and planning the analysis to follow. 

28 

 
 
 
 
 
 
 
 
 
Chapter 4: Data and Methods 

Data Set 

The Public Engagement Pilot Projects (PEPP) initiative chose six courts from across 

the United States and Puerto Rico to receive support from the National Center for State 

Courts (NCSC) and the University of Nebraska Public Policy Center in their efforts to 

initiate public engagement with the courts in their jurisdictions, all with the underlying 

goal of “improving public trust and confidence across minority and economically-

disadvantaged communities” (NCSC, 2024). Depending on the location, examples of 

engagement efforts included: small, facilitated discussions with the public, meetings with 

community leaders, and/or focus groups. In many of these projects, the PEPP team 

members actively worked with community leaders and local professionals to craft and 

deliver the engagement efforts (NCSC, 2024), a crucial component of co-production as 

viewed by Needham (2008). 

Data collected as part of PEPP generally included quantitative surveys of court 

actors and members of the public taken directly before and after each engagement effort, as 

well as qualitative, structured interviews with members of the PEPP teams who were 

responsible for crafting and delivering the engagement efforts with each of the six court 

entities. For this dissertation, the data set includes survey responses collected in person at 

five different sites, described throughout as: Midwest state, Southern state, East Coast state, 

Great Plains state, and Great Lakes state

3
. 

3

 To account for differences in engagement efforts across location, I am including data collection site as a 
control variable, see Table 6. 

29 

 
 
 
 
 
 
 
Research Questions and Hypotheses

Research question 1: How does experience with courts affect the public’s co-production of 
court services? 

Based on previous research, there are several ways that experience with courts 

could affect the public’s co-production of court services. Some members of the public have 

little contact with the courts, but some have significant contact. Experimental research 

indicates that when people have contact with legal authority, procedural fairness is 

positively associated with concepts of cooperation (De Cremer & Tyler, 2007), leading to: 

Increased perception of procedural fairness will increase individual 

Hypothesis 1 (H

co-production of court services.

1
): 

Engaging in co-production requires an outlay of resources on the part of the public agency 

typically tasked with providing the services as well as the members of the public newly 

participating in the co-productive efforts. With this changed relationship, there is the 

potential that people might see too much risk involved in the process and choose against 

engaging in co-production, leading to: 

Increased perception of risk to the community will decrease 

Hypothesis 2 (H

individual co-production of court services. 

2
): 

Research question 2: What role does trust play in the co-production of court services? 

This study seeks to extend previous literature on trust and co-production 

from more traditional public services such as housing and health care—and even 

other justice-related services such as policing—to court services, leading to the 

following hypothesis: 

Increased trust in the courts will lead to increased individual co-

Hypothesis 3 (H

production of court services. 

3
): 

30 

 
 
 
 
 
 
Measures

Co-production 

As this dissertation seeks to develop knowledge related to the concept of co-

production, it follows that co-production is the dependent variable of interest. I follow the 

definition of co-production used by Osborne et al. (2016, p. 640), which says that co-

production is “the voluntary or involuntary involvement of public service users

in any of 

the design, management, delivery and/or evaluation of public services.” As discussed in 

Chapter 2, this provides a reasonable balance between parsimony and breadth. I took 

component variables from the survey given to PEPP participants after each engagement 

event. Each component variable was measured using a Likert-type scale; however, since 

the scales were not consistently worded or of the same number of levels, instead of 

creating my own scale from the components, I created a factor score for co-production that 

was then used as the dependent variable in subsequent analyses. Table 2 displays the 

component variables in the factor analysis as well as other statistics. 
Table 2. Dependent variable factor analysis: Coproduction.

Component variable name 

Variable label 

increase2 

important2 

helpful2 

discusshelp2 

viewpoints2 

satis�ied2 

During today’s engagement, to what 
degree, if any, did your knowledge of 
the [courts in your area] increase? 
How important to you were the 
topics addressed during the 
engagement activities? 
How helpful were the engagement 
activities in making progress toward 
solving one or more problems? 
How much did the discussion help 
you see new viewpoints? 
How many different viewpoints were 
expressed in front of the whole 
group? 
How satis�ied or unsatis�ied were 
you with the engagement activities? 

Factor 
component 
loading 
.712 

.525 

.735 

.766 

.644 

.795 

KMO 

Overall

.815 

Eigenvalue 

2.958 

% Variance 

49.292 

31 

 
 
 
 
 
 
 
 
 
 
 
 
 
As constructed, the co-production variable �its well according to prior research as 

well as the measurable statistics. For example, being able to design a public service requires 

knowledge of that area, or what Van Eijk and Steen (2016) would call salience; the 

“increase” question directly asks whether the activity increased the respondent’s 

knowledge of the courts. Additionally, if the respondent indicates a high level of satisfaction 

with the engagement activity or that the activity was helpful in making progress toward 

solving a problem, this indicates a level of involvement on the part of the respondent with 

the activity, at minimum on an informational or co-evaluation level and probably on a co-

management or co-delivery level (Osborne et al., 2016). 

More so than the independent variables, co-production required more exploration 

before I was able to �ind a suitable measurement. There were several components that I had 

initially included that I was able to excise essentially out-of-hand. Those included the 

following: • 
• 

timediscuss2: “Was there time for discussion?” 
missing2: “Were any groups of people or viewpoints missing from today’s 
engagement?” 

In the early factor analyses, these components had loadings of .15 or less that persisted 

through the inclusion of other components. Without available follow-up information, they 

did not provide much in the way of theoretical value either, hence their elimination. Two 

additional components that required more exploration were the following: 

• 

• 

followup2: May the evaluation team contact you again later about your 
opinions? 
socnetsurvey2: Would you be willing to invite people you know to do a very 
short survey? 

32 

 
 
 
 
These two questions ask whether the respondent would be willing to continue PEPP-like 

efforts in the future, which does comport with co-production. I did consider leaving them 

in, which led to a reasonable two-factor solution. The “importance” component would have 

had to be eliminated in that instance, as it did not load well on either factor. Forcing a one-

factor solution was not a reasonable solution, as it led to a split between components with 

higher loadings and those two with loadings under .32. After returning to the research 

questions and the literature again, I felt that the focus is best placed on present co-

production and that eliminating these two components made sense on that basis. 
Procedural fairness 

Directly related to H

1
 is the independent variable of procedural fairness. Similar to 

co-production, the chosen component variables were both measured using Likert-type 

scales of different levels. As such, even though there are only two component variables, 

factor analysis was the technique of choice, as displayed in Table 3.   
Table 3. Independent variable factor analysis: Procedural fairness.

Component variable name 

Variable label 

fair1 

care1 

How fair or unfair do [courts in your 
area] treat people of different races, 
genders, ages, wealth, or other 
characteristics? 
How much do you feel the [courts in 
your area] care about the problems 
faced by people like you? 

Component 
factor loading 
.908 

KMO 

Overall

.500 

Eigenvalue 

1.650 

.908 

% Variance 

82.492 

This variable draws validity from Van Eijk and Steen’s (2016) concept of external political 

ef�icacy (“Does government…provide room for my interaction, and if so, will my interaction 

matter in their decision-making and service provision processes?”). Additionally, including 

both of these components draws validity from the work of De Cremer and Tyler (2007) as 

an example of the fair procedure of providing voice to the court users: “‘the authority 

33 

 
 
 
 
 
 
 
 
 
listens to me and uses this information to make the best possible decision’” (p. 640). Tyler’s 

(1988) ethicality criterion for assessing procedural justice includes a desire to see one’s 

citizen rights be respected as well as a concern with the “interpersonal aspects of 

encounters with authorities” (p. 129). In the PEPP context, both fair treatment and caring 

about the respondent’s problems address the same overall concepts. The components 

loaded well on one factor and the total variance explained (TVE) was high, though some of 

that is likely due to the fact that there are only two components. 
Risk 

Table 4. Independent variable factor analysis: Risk.

Component variable name 

Variable label 

neg_likely1 

neg_extent1 

In your opinion, how likely is it that 
[courts in your area] will have 
negative effects on your community? 
If negative effects happened, how 
negative would they be? 

Component 
factor loading 
.907 

KMO 

Overall

.500 

Eigenvalue 

1.644 

.907 

% Variance 

82.180 

Table 4 displays the factor analysis I ran that generated the factor score used to 

measure risk in the subsequent multivariate analyses in Chapter 5. Both of the component 

variables were measured using Likert-type scales of the same number of levels in the 

survey. Loef�ler (2021) similarly conceptualizes risk as having two elements: the magnitude 

of the negative effect of an event (“how negative”) and the likelihood that an event with 

adversarial effects is going to take place in the future (“how likely”), while Wilson et al. 

(2019) in general measure risk perception using probability and consequences. Where 

prior research diverges is in how to measure the independent variable. Wilson et al. (2019) 

see probability and consequences (in the data set: likely and extent) having a multiplicative 

effect; further, they view probability through exposure and vulnerability, and they view 

34 

 
 
 
 
 
 
 
 
 
consequences through severity and affect (emotion). I chose to measure risk using factor 

analysis and factor score as shown in Table 4 primarily due to the fact that prior research 

best supported that choice (Brown & Osborne, 2013; Loef�ler, 2021). Speci�ically, Loef�ler 

de�ines risk as having two elements: the magnitude of the negative effect of an event and 

the likelihood that an event with negative effects will take place in the future (p. 261). 

These are direct parallels to the components in Table 4. Additionally, using this technique 

allowed for consistency with the other variables of interest measured in that manner. As 

was the case with procedural fairness, both components loaded well on one factor and the 

total variance explained was over 80 percent.  
Trust 

I conducted a factor analysis using trust-related variables from the post-engagement 

surveys, incorporating the components seen in Table 5
Table 5. Independent variable factor analysis: Trust.

below: 

Component variable name 

Variable label 

trust1 

honest1 

ctcomm1 

comfort1 

pos_likely1 

pos_extent1 

How much do you trust or distrust 
the [PEPP] courts? 
How much do the [courts in your 
area] act with honesty and integrity? 
To what extent do you see the [PEPP] 
Courts as being part of your 
community? 
How comfortable would you feel 
letting the [courts in your area] 
decide a case that was important to 
you? 
In your opinion, how likely is it that 
[courts in your area] will have 
positive effects on your community? 
If positive effects happened, how 
positive would they be? 

Component 
factor loading 
.783 

KMO 

Overall

.869 

Eigenvalue 

4.724 

% Variance 

59.053 

.795 

.679 

.826 

.806 

.699 

35 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Table 5 (cont’d) 

resp_judges1 

resp_staff1 

judges – In the [courts in your area], 
how much are court personnel 
respectful and courteous to all 
members of the public? 
other court staff – In the [courts in 
your area], how much are court 
personnel respectful and courteous 
to all members of the public? 

.786 

.761 

None of the component factor loadings were any lower than 0.67, and with an eigenvalue of 

over 4.0 and a TVE of 59%, the factor score generated here was what I chose to use in my 

analyses. Further, these components �ind support in some of the foundational trust 

research: for example, the “comfort” and “community” questions both tie into 

identi�ication-based trust (Fledderus et al., 2014) and relational trust (Rousseau et al., 

1998). Additionally, the components that ask about the positive effects that the courts can 

have on the community speak directly to benevolence (Mayer et al., 1995) and the positive 

expectations that the respondent may have of the courts (Rousseau et al., 1998). 
Control variables 

The controls used throughout the analyses consist of a series of demographic 

variables, as well as binary variables related to prior experience with the courts and the 

community. Table 6 provides a list of the included control variables. 
Table 6. Independent variable factor analysis: Procedural fairness.

Variable name 

Please indicate if you have had each of the 
Variable label 
following experiences with the courts before 
today: 

Variable name 
education 

Served on a jury 

Defendant (in a civil or criminal case) 

Witness 

Plaintiff (who brought a case to court)

jury 

defend 

witness 

plaintiff 

What is the highest 
Variable label 
degree you have 
attained? 

No high school 
diploma 
High school 
diploma/GED 
Some college or post 
high school, but no 
degree 
Technical/associate, 
junior college (2 yr, 
LPN) 

36 

 
 
 
 
 
 
 
  
 
 
 
 
 
 
 
Table 6 (cont’d) 

juvijust 

probat 

public 

other 
workcts 

Participant in a juvenile justice or child welfare 
case 
Probationer (on probation) 

Engaged as a member of the public 

Other 
Do you currently work with or for the courts in 
any of�icial role or position? 

ideol_lib 

lead 

age 
gender_recode 

Do you play any leadership roles in your 
community? 
What is your age?
What is your gender? 
What race or races do you consider yourself to 
be?

American Indian or Alaska Native 
Asian
Black or African American 
Spanish, Hispanic, Latina/o/x, Puerto Rican
White (Caucasian) 
Other 

sitebinary_1 
sitebinary_2 
sitebinary_3 
sitebinary_4 
sitebinary_5 

race_AI 
race_asian 
race_black 
race_hisp 
race_white 
race_other 

Analysis plan

Bivariate analysis 

Bachelor’s degree (4 
yr, BA, BS, RN) 
Some graduate 
school 
Graduate degree 
(Masters, PhD, law, 
medicine) 

In general, how 
would you describe 
your ideological 
views? 

Very liberal 
Liberal 
Middle-of-the-road 
Conservative 
Very conservative 

Data collection site 
Midwest state 
Southern state 
East Coast state 
Great Plains state 
Great Lakes state 

The analysis plan begins with a series of independent-samples t-tests to evaluate 

whether there are any differences among the study sample when it comes to the binary 

control variables and co-production. For the categorical and ordinal control variables, I will 

conduct a series of analyses of variance (ANOVA) to make the same determination. If the 

ANOVA shows that there is at least one statistically signi�icant mean difference, post hoc 

analysis will be run to determine where that difference is. The bivariate analysis will be 

informative, but it will not be dispositive for my analysis and conclusions; any statistically 

37 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
signi�icant differences in means may help answer the hypotheses and, if necessary, modify 

the multivariate analyses I conduct. 
Multivariate analysis 

Once the factor scores for the dependent and independent variables have been 

constructed and the control variables have been chosen, I plan to run a series of ordinary 

least squares (OLS) regressions to test my hypotheses, using co-production as the 

dependent variable in each. Given their importance, the independent variables will be 

present in every model. What will be adjusted are the categories of control variables that 

are present in each OLS model. This should show how meaningful the relationship between 

the independent variables and co-production is, accounting for the effects of the other 

characteristics that the data set makes available. 
Missing data analysis 

The �inal step in the analysis will be to examine the data set for missing values and 

potentially re-run the OLS model(s) accounting for any missing data. First, I will use the 

missing value analysis function of SPSS to take an account of what data are missing on the 

dependent and independent variables, as well as the control variables. The method I will 

use to replace any missing data will be multiple imputation. Multiple imputation provides a 

way to generate statistical inferences on the entire study sample, based on the known 

relationships between measured variables and cases (Manly & Wells, 2015). Multiple 

imputation has been used as a primary technique for dealing with missing data in the field 

of higher education (Manly & Wells, 2015), and it is being used with more regularity in 

criminal justice research as well (e.g., Doherty & Bersani, 2020; Mitchell et al., 2022; Motley 

et al., 2020; Wadsworth & Roberts, 2008). Once any missing data are imputed, I would then 

38 

 
 
 
 
be able to re-run the OLS model(s) and compare the imputation models to the previous 

models to see if there are any differences. 

39 

 
 
 
Chapter 5: Results and Analysis 

Descriptive statistics 

0
Table 7 presents descriptive statistics on the overall study sample (N
=419). Gender 

is one of the few demographic variables that is somewhat skewed. Across the study sample, 

60% of respondents report as female. There is significant racial diversity across the study 

sample, with 30% of respondents reporting they are American Indian or Alaskan Native, 

26% Black, and 32% White. The sample is educationally diverse with at least 10 percent of 

respondents self-reporting in each educational category except for “some graduate school.” 

However, the sample does contain a significant number of respondents with graduate 

degrees (21%) or a 2- or 4-year degree (27.2%). In terms of political ideology, the sample 

is quite “middle-of-the-road,” as shown by their mean score of 3.21 on a 1-5 scale. 
Table 7. Sample descriptive statistics.

Age
Gender 

Female 
Male 

Race/ethnicity

White (Caucasian) 
Spanish, Hispanic, Latina/o/x, Puerto Rican 
Black or African American 
Asian 
American Indian or Alaska Native 

Education 

No HS diploma 
HS diploma/GED 
Some college 
Tech/Assoc/Jr college (2 yr) 
Bachelors (4 yr) 
Some graduate school 
Graduate degree 

N 

0

% N
(N=419) 

Mean 

SD  Missing 

43.94  17.1 
.49 

.62 

19 
12 

.33 
.16 
.27 
.01 
.31 
4.05 

.47 
.37 
.45 
.11 
.46 
2.05 

14 
14 
14 
14 
14 
15 

253 
154 

135 
64 
110 
5 
125 

50 
58 
77 
51 
63 
14 
91 

60.4 
36.8 

32.2 
15.3 
26.3 
1.2 
29.8 

11.9 
13.8 
18.4 
12.2 
15.0 
3.3 
21.7 

0

% N
(N=419) 

4.5 
2.9 

3.3 
3.3 
3.3 
3.3 
3.3 
3.6 

40 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Table 7 (cont’d) 
Ideology 

Very conservative 
Conservative 
Middle-of-the-road 
Liberal 
Very liberal 

Experience 

Court actor 
Community leader 
Served on a jury 
Defendant (civil or criminal case) 
Witness 
Plaintiff (brought case to court) 
Participant in juvenile or child welfare case 
Probationer (on probation) 
Engaged as member of public 
Other 

Data collection site 

Midwest state 
Southern state 
East Coast state 
Great Plains state 
Great Lakes state 

Bivariate analysis 

10 
56 
205 
90 
35 

68 
147 
116 
89 
96 
68 
109 
47 
114 
37 

115 
66 
91 
139 
8 

2.4 
13.4 
48.9 
21.5 
8.4 

16.2 
35.1 
27.7 
21.2 
22.9 
16.8 
26.0 
11.2 
27.2 
8.8 

27.4 
15.8 
21.7 
33.2 
1.9 

3.21 

.883 

23 

5.5 

19 
23 
14 
14 
14 
14 
14 
14 
14 
14 

4.5 
5.5 
3.3 
3.3 
3.3 
3.3 
3.3 
3.3 
3.3 
3.3 

.17 
.37 
.29 
.22 
.24 
.17 
.27 
.12 
.28 
.09 

.28 
.16 
.22 
.33 
.02 

.38 
.48 
.45 
.42 
.43 
.37 
.44 
.32 
.45 
.29 

.45 
.36 
.41 
.47 
.14 

Independent-samples t-tests were conducted on the dichotomous control variables 

as presented in Table 8. This bivariate analysis was conducted to determine whether there 

are any significant differences among the overall study sample on the mean value of the 

dependent variable of interest: coproduction. 
Table 8. Independent-samples t-tests.

Engaged as member of public 
Controls: Previous experiences with the courts before data collection 
Probationer (on probation) 
Participant in juvenile justice or child welfare case 
Plaintiff (who brought a case to court) 
Served on a jury 
Witness  

DV 

Mean
(X=1, 
yes) 
.137 
-.112 
.017 
.019 
-.027 
-.019 

DV 

Mean
(X=0, 
no) 
-.071 
.003 
-.025 
-.018 
-.007 
-.011 

t 
-1.557 
.664 
-.321 
-.224 
.150 
.058 

p 

.121 
.507 
.748 
.823 
.881 
.953 

41 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
 
 
 
Table 8 (cont’d) 
Defendant in a civil or criminal case 
Other 

Any leadership role in the community 
Currently work with or for the courts in any official role or position 
Gender 

(0=female, 1=male)

-.007 
-.077 

.109 
-.132 
.017 

-.014 
-.007 

-.075 
.013 
-.018 

-.053 
.325 

.958 
.746 

-1.411 
.884 
-.273 

.159 
.377 
.785 

Three of the variables merit some discussion. First is the subset of the study sample 

who had previous experience with the courts through engagement as a member of the 

public, either through other surveys or input or learning about the courts. Additionally, 

there were slight mean differences in the sample for those who worked with or for the 

courts at the time of data collection, as well as those who played a leadership role in their 

community. Each of the mean differences was considerable in the sample group (.208, -

.145, .184 respectively); however, none of these mean differences was statistically 

significant. Therefore, we cannot infer that these differences exist in the overall study 

sample. 
Table 9. Analyses of variance (ANOVA). 

1.00  Very conservative 
Political ideology 
2.00  Conservative 
3.00  Middle-of-the-road 
4.00  Liberal 
5.00  Very liberal 

1.00  No HS diploma  
Highest degree attained 
2.00  HS diploma/GED  
3.00  Some college  
4.00  Tech/Assoc/Jr college (2 yr)  
5.00  Bachelor (4 yr) 
6.00  Some graduate school 
7.00  Graduate degree  

DV 

N  Mean
7 
39 
137 
62 
24 

-.198 
.168 
.113 
-.243 
-.164 

DV

N  Mean
43 
40 
57 
36 
43 
12 
44 

-.347 
.395 
.162 
-.009 
-.039 
.220 
-.279 

Bonferroni 

p 

Overall 

Group 
no sig. diff. 
differences 

F 
Sig. 

1.890 
.113 

Group 
1.00 & 2.00 
differences
2.00 & 7.00 

p 
.013* 
.036* 

F 
Sig. 

2.971 
.008** 

S.D. 
.947 
.841 
.981 
.899 
1.444 

S.D. 
.985 
.906 
.966 
.908 
.869 
.802 
1.208 

42 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Table 9 (cont’d) 

1.00  Midwest state 
Data collection site 
2.00  Southern state 
3.00  East Coast state 
4.00  Great Plains state 
5.00  Great Lakes state 
p 
p 

**

< .01; *

< .05 

DV

N  Mean
85 
57 
30 
105 
5 

-.251 
.255 
.467 
-.095 
.551 

Group 
1.00 & 2.00 
differences
1.00 & 3.00 
3.00 & 4.00 

p 
.027* 
.006** 
.054 

F 
Sig. 

4.755 
.001** 

S.D. 
1.069 
.847 
1.062 
.941 
.711 

Analyses of variance were conducted on the categorical control variables, as 

presented in Table 9, to examine whether there are any significant differences among the 

study sample on these variables as related to coproduction. Prior research indicates that 

political participation and political efficacy—both internal and external—may have a direct 

relationship with the choice to engage in coproduction (Uzochukwu & Thomas, 2018; Van 

Eijk & Steen, 2016), making political ideology a valid choice as a control variable. The mean 

value of coproduction was considerable for each category, though not in a consistent 

direction. However, there were no statistically significant differences shown among the 

political ideology categories on the mean value of coproduction. This lack of significant 

group differences was somewhat surprising given the historical and recent prior research. 

Level of education is also a commonly examined variable in other studies on 

coproduction (e.g., Kang & Van Ryzin, 2019; Loeffler, 2021). The overall result of the 

F 

p 

analysis of variance was statistically significant (

(6, 268) = 2.971, 

= .008). Bonferroni 

post hoc analysis shows that there are significant mean differences between those with no 

high school diploma as their highest level of education and those with a high school 

diploma or a GED, as well as those with a high school diploma or GED and those with a 

graduate degree. At the bivariate level, the mean values of coproduction relative to highest 

43 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
degree attained do not follow a consistent pattern; multivariate analysis may generate 

further information and relationships. 

The PEPP data were collected at several different sites across the United States; this 

F 

study used data from five of these sites as shown in Table 9. The analysis of variance (

p 

(4, 

277) = 4.755, 

= .001) again showed that there was at least one statistically significant 

mean difference in the study sample. Post hoc analysis showed two statistically significant 

p 

mean differences: Midwest and Southern states (-.251 & .255, 

p 

= .027) and Midwest and 

East Coast states (-.251 & .467, 
Multivariate analysis 

= .006).  

Analysis of the relationship between co-production and the main independent 

variables then moved to multivariate, using ordinary least squares (OLS) regression. Based 

on prior research and the bivariate analysis, five models were developed as shown in table 

10.  Table 10. OLS regressions on co-production. 

Procedural fairness 
Model 1 (n=131) 
Trust 
Risk 

Procedural fairness  
Model 2 (n=131) 
Trust 
Risk 
Data collection site: 

Procedural fairness 
Model 3 (n=131) 
Trust 
Risk 

B 
-.134 
.369 
.075 

-.124 
.406 
.060 

-.185 
-.115 
-.444 
.399 

-.088 
.365 
.123 

t 
-1.030 
2.754 
.829 

-.903 
2.683 
.644 

-.689 
-.499 
-1.362 
.678 

-.620 
2.575 
1.231 

p 

.305 
.007** 
.409 

.368 
.008** 
.521 

.492 
.619 
.176 
.499 

.537 
.011* 
.221 

Midwest state 
Southern state 
East Coast state 
Great Lakes state 

F 
Sig. 
R2 

Overall 

3.281 
.023* 
.072 

F 
Sig. 
R2 

1.781 
.097 
.092 

F 
Sig. 
R2 

1.239 
.261 
.125 

44 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Table 10 (cont’d) 
Experience: 

Work with the courts 
Community leadership 
Served on a jury 
Defendant in civil or criminal case 
Witness 
Plaintiff (brought case to court) 
Juvenile justice or child welfare case 
Probationer (on probation) 
Engaged as member of the public 
Other 

Procedural fairness 
Model 4 (n=126) 
Trust 
Risk 
Demographics: 

Age 
American Indian or Alaska Native 
Asian 
Black or African American 
Spanish, Hispanic, Latina/o/x, Puerto Rican 
Race, other 
Ideology (liberal is high)
Education level 
Gender 

Procedural fairness 
Model 5 (n=122) 
Trust 
Risk 
Demographics: 

Age 
American Indian or Alaska Native 
Asian 
Black or African American 
Spanish, Hispanic, Latina/o/x, Puerto Rican 
Race, other 
Ideology (liberal is high) 
Education level 
Gender 

Work with the courts 
Community leadership 
Served on a jury 
Defendant in civil or criminal case 

Experience: 

-.121 
.333 
-.180 
-.214 
.023 
-.130 
-.070 
.210 
.002 
B 
-.289 

-.096 
.347 
.072 

.005 
.149 
.084 
.272 
.266 
-1.063 
-.236 
-.122 
.097 

B 
-.036 
.389 
.092 

.008 
.202 
.282 
.071 
.531 
-1.570 
-.180 
-.113 
.210 

-.097 
.328 
-.071 
-.018 

-.530 
1.611 
-.910 
-.919 
.109 
-.545 
-.341 
.795 
.009 
t 
-.996 

-.721 
2.323 
.794 

.790 
.683 
.146 
1.148 
1.150 
-2.134 
-2.136 
-2.397 
.497 

t 
-.229 
2.365 
.862 

1.225 
.633 
.454 
.207 
1.790 
-2.544 
-1.438 
-1.931 
.939 

-.408 
1.540 
-.335 
-.079 

.597 
.110 
.365 
.360 
.914 
.587 
.734 
.428 
.993 
p 
.321 

.473 
.022* 
.429 

.431 
.496 
.884 
.253 
.253 
.035* 
.035* 
.018* 
p 
.620 

.819 
.020* 
.391 

.223 
.528 
.651 
.837 
.077 
.013* 
.154 
.056 
.350 

.684 
.127 
.738 
.938 

F 
Sig. 
R2 

2.836 
.002** 
.231 

F 
Sig. 
R2 

1.734 
.029* 
.322 

45 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Table 10 (cont’d) 

Data collection site: 

p 

p 

**
Model 1 

< .01; *

< .05 

Witness 
Plaintiff (brought case to court) 
Juvenile justice or child welfare case 
Probationer (on probation) 
Engaged as a member of the public 
Other 

Midwest state 
Southern state 
East Coast state 
Great Lakes state 

.200 
-.140 
-.006 
-.043 
.095 
-.375 

.002 
-.120 
-.014 
1.560 

.904 
-.575 
-.030 
-.160 
.398 
-1.268 

.005 
-.300 
-.032 
1.612 

.368 
.567 
.976 
.873 
.691 
.208 

.996 
.765 
.975 
.110 

I chose to present Model 1 in the most parsimonious way possible, using only the  

independent variables as predictors and co-production as the dependent variable, without 

controlling for any outside factors. Several of the limited extant studies on co-production 

and courts or other justice system-related issues also emphasize similar concepts, lending 

validity to their use (Loeffler & Bovaird, 2020; Van Eijk & Steen, 2016). 

Model 1, while parsimonious and direct, is not functionally practical. These data 

were not collected in an experimental context. Any conclusions made about the 

relationships between co-production, procedural fairness, trust, and risk would be suspect 

without considering the impact of the additional factors present in the additional models. 

While this model may not be functionally practical, it presents a reasonable baseline for the 

2
other regressions. The overall predictability of this model is low (R
=.072, F=3.281, 
p

=.023), and neither procedural fairness nor risk are statistically significant. However, 

trust as a predictor of co-production is robust, positive, and significant at the .01 level. This 

is consistent with the broader co-production literature; specific to Model 1, Ostrom (2000) 

views assuring mutual trust as an element of the basic structure of most contractual 

46 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
relationships and collective action. Additionally, Alford (2014) notes that building trust is a 

contextual matter, and co-production is the context. 
Model 2

Peeking back briefly to Chapter 4 to some of the components of the independent 

variables, we can see that many of them involve either the participant or the courts being 

part of the community or the effect that the courts can have on the community. Model 2 

looks at the geographic aspect of community by controlling for the data collection site in 

conjunction with the independent variables of interest. Overall predictability of model 2 is 

p

modest but not statistically significant (R

2
=.092, F=1.781, 

=.097). People in the East Coast 

state were considerably unwilling to engage in co-production, and people in the Great 

Lakes state were almost as equally willing to engage in co-production,

4
 however, none of 

the coefficients were statistically significant.  

Interestingly, the impact of trust on co-production maintains statistical significance, 

and there is a slight increase in the slope of the co-efficient. Procedural fairness and risk 

each lose some of the limited explanatory impact they had as compared to Model 1. At this 

point in the analysis, it is unclear

5
 whether this is due to the inclusion of the data collection 

site controls or the importance of trust. 
Model 3 

Prior research has also indicated that experience with public services can affect how 

a member of the public may engage in co-production (Longazel et al., 2011). In what might 

be the greatest difference with those using other public services (i.e., housing, health care, 

4

5

 Of the n=131 in Model 2, only 8 made up the Great Lakes subset. 
 Examination of the VIF and tolerance values indicated no issues with collinearity. 

47 

 
 
 
 
 
 
education) who may experience some level of coercion, those engaging in co-production 

with courts may be doing so under the threat of punishment (Weaver, 2011). Model 3 looks 

at the effect of experience by controlling for previous experience with the courts and the 

independent variables only to isolate this effect. Somewhat surprisingly, there was no 

statistically significant effect shown overall. The overall predictability of this model 

p

2
=.125), though it loses statistical significance (
improves modestly (R

=.261) due to the 

fact that enough variables without statistical significance were added to the model to make 

the entire model lose predictability. 

A pattern is developing regarding the relationships among the independent 

variables, and between the independent variables and co-production. Trust persists in its 

strong predictability of co-production (though there is a very modest deduction in the 

slope of the coefficient). One might have anticipated more of a change in this model, as 

several of the added controls could be related to trust. As an example, one might think that 

those who have previously been defendants or probationers, or those who have been part 

of a juvenile case, may have their trust in the courts affected. However, adding these 

controls to the model showed no effect. 
Model 4

Previous models have looked at the impact of where the respondents are located 

(Model 2) and their previous experience (Model 3) on levels of co-production. However, it 

is essential to also consider the importance of each individual’s characteristics on their 

levels of co-production. Ostrom and her colleagues (1978) included individual 

characteristics at multiple stages of their foundational study on policing and co-production, 

48 

 
 
 
 
 
as has much of the research on co-production in the decades since (e.g., Brudney & 

England, 1983; Loeffler, 2021; Van Eijk & Steen, 2016; Weaver, 2011).  

Personal characteristics, socio-demographic factors (Loeffler, 2021), or 

demographics as they are described in Table 7, cover a wide spectrum of influences and 

may be expected to have a wide range of statistical effects as shown in Model 4. Both age 

and gender have a very modest positive effect on co-production; however, those effects are 

not statistically significant. Relative to the reference category (white), none of the racial 

categories have a statistically significant effect on the respondents’ levels of co-production, 

p

save for those people that self-identify in the “other” category (B=-1.063, 

=.035). This is, 

including the trust variables, the greatest slope seen so far in any of the previous models, 

indicating that these people are highly disinclined to co-produce with courts.

6

Two additional personal characteristics in the data set are level of education and 

political ideology, both of which in Model 4 have a negative and statistically significant 

p

p

effect on respondents’ level of co-production (B=-.236, 

=.035; B=-.122, 

=.018 

respectively). As Loeffler (2021) states, those with higher levels of education may be less 

inclined toward co-production because they are busier with work commitments or that 

they live in neighborhoods where public services are of better quality. However, for these 

respondents there may be an additional explanation worth noting. While there is little 

measurable difference in the mean value for education in the overall study sample 

0
=4.05) versus Model 4 (μ

4
=4.24), the education variable is not normally distributed, 

(μ

with a significant number of people self-identifying as holding a graduate degree (7 on a 1-

7 Likert-type scale). Logically this makes sense, as many of the professionals involved in 

6

 Of the N=419 in the overall study sample, only 7 self-identified in this category. 

49 

 
 
  
 
the work of the courts might find it either helpful or necessary to have that level of 

education. 

The addition of ideology to the model theoretically could have had an impact on 

trust relative to co-production, as those on each end of the ideological spectrum seem to 

show a distrust of governmental entities of many types, including courts. However, even 

though ideology had a negative and statistically significant effect on co-production in Model 

p

4 (B=-.236, 
p

=.035), trust maintains its positive relationship with co-production (B=.347, 

=.022). 
Model 5

Model 1 was presented in a highly parsimonious, yet functionally impractical 

fashion, with only the independent variables of interest included. Each of the subsequent 

models has helped to develop the picture of how members of the public engage in co-

production with the courts. With a better understanding of these factors, it then becomes 

necessary to look at the entire picture at once, to consider each potential co-producer in 

full. Model 5 does this by including the primary independent variables, as well as the 

controls for data collection site, experience, and demographic factors. Concluding the 

pattern established in Models 1-4, including all of the control variables did not diminish the 

impact or statistical significance of trust on co-production in Model 5. Regardless of where 

they were, how they self-identified, or what they did, how they trusted the courts in their 

communities had the greatest impact on whether they would engage in co-production with 

those courts. 

At this point in the analysis, it is important to recognize that Model 5 is based on a 

total of 122 cases, while the number of cases in the overall study sample is 419. In order to 

50 

 
 
 
 
 
determine whether this discrepancy may have any effect on how trust affects how people 

engage in co-production with courts, I conducted a missing value analysis, including 

imputing values for any missing data. In addition, the number of independent variables 

with a small sample size can result in unreliable estimates. Multiple imputation can help 

avoid such issues. 
Table 11. Pooled imputation model with missing value analysis.

Variables with >5% missing values 
Trust 
Coproduction 
Risk 
Procedural fairness 
Community leadership 
Ideology (liberal is high) 
Procedural fairness 
Original OLS Model 5 (n=122) 
Trust 
Risk 
Demographics: 

Age 
American Indian or Alaska Native 
Asian 
Black or African American 
Spanish, Hispanic, Latina/o/x, Puerto Rican 
Race, other 
Ideology (liberal is high) 
Education level 
Gender 

Experience: 

Work with the courts 
Community leadership 
Served on a jury 
Defendant in civil or criminal case 
Witness 
Plaintiff (brought case to court) 
Juvenile justice or child welfare case 
Probationer (on probation) 
Engaged as a member of the public 
Other 

N 
211 
282 
334 
386 
396 
397 

B 
-.036 
.389 
.092 

.008 
.202 
.282 
.071 
.531 
-1.570 
-.180 
-.113 
.210 

-.097 
.328 
-.071 
-.018 
.200 
-.140 
-.006 
-.043 
.095 
-.375 

Missing N 
208 
137 
85 
33 
23 
22 

t 
-.229 
2.365 
.862 

1.225 
.633 
.454 
.207 
1.790 
-2.544 
-1.438 
-1.931 
.939 

-.408 
1.540 
-.335 
-.079 
.904 
-.575 
-.030 
-.160 
.398 
-1.268 

0

%N
49.6 
32.7 
20.3 
7.9 
5.5 
p 
5.3 

.819 
.020* 
.391 

.223 
.528 
.651 
.837 
.077 
.013* 
.154 
.056 
.350 

.684 
.127 
.738 
.938 
.368 
.567 
.976 
.873 
.691 
.208 

F 
Sig. 
R2 

Overall 

1.734 
.029* 
.322 

51 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Table 11 (cont’d) 
Data collection site: 

Midwest state 
Southern state 
East Coast state 
Great Lakes state 

Procedural fairness 
Pooled model (n=419) 
Trust 
Risk 
Demographics: 

Age 
American Indian or Alaska Native 
Asian 
Black or African American 
Spanish, Hispanic, Latina/o/x, Puerto Rican 
Race, other 
Ideology (liberal is high) 
Education level 
Gender 

Experience: 

Work with the courts 
Community leadership 
Served on a jury 
Defendant in civil or criminal case 
Witness 
Plaintiff (brought case to court) 
Juvenile justice or child welfare case 
Probationer (on probation) 
Engaged as a member of the public 
Other 

Data collection site: 

p 

p 

**

< .01; *

< .05 

Midwest state 
Southern state 
East Coast state 
Great Lakes state 

.002 
-.120 
-.014 
1.560 

B 
-.054 
.390 
.097 

.010 
.067 
.481 
.156 
.358 
-1.366 
-.119 
-.097 
.155 

-.259 
.290 
-.012 
-.016 
-.069 
.177 
.041 
-.192 
.229 
-.344 

-.459 
-.043 
.378 
.402 

.005 
-.300 
-.032 
1.612 

t 
-.337 
2.085 
1.041 

2.712 
.346 
.886 
.833 
1.630 
-2.771 
-1.806 
-2.990 
1.111 

-1.355 
2.348 
-.082 
-.087 
-.470 
1.171 
.292 
-.911 
1.548 
-1.643 

-1.897 
-.142 
1.192 
.748 

.996 
.765 
.975 
p 
.110 

.711 
.056 
.310 

.008** 
.730 
.379 
.408 
.114 
.008** 
.076 
.004** 
.275 

.186 
.021* 
.935 
.931 
.640 
.243 
.771 
.367 
.130 
.106 

.061 
.887 
.244 
.458 

After setting a conservative threshold of 5 percent for display purposes, Table 11 

shows the variables from Model 5 for which there are substantial missing values. 

Important for the analysis are that trust, co-production, and risk each have over 20 percent 

of cases with missing values. There seem to be no discernible explanations for these 

52 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
missing values either from the survey instruments themselves (i.e., skip patterns) or from 

the overall missing value patterns, leading to the thought that statistical correction in the 

form of either deletion of cases or some form of data imputation is necessary. Given that 

this study is already having to handle a lower number of cases than what would be ideal, 

using a technique such as listwise deletion to entirely eliminate the cases in the study 

sample without complete data would unreasonably reduce its explanatory power (Manly & 

Wells, 2015).  

impute missing data values

To set up a multiple imputation model, I used the 

command in SPSS version 29 to create 10 new datasets in addition to the original dataset 

(IBM, 2024). Multiple imputation can produce useful estimates with as few imputed data 

sets as two (Rubin, 2018); imputing 10 datasets should be sufficient to produce the 

estimates needed without the 20 or more datasets suggested by other researchers (Graham 

et al., 2007; White et al, 2011). Included in the imputation command were all of the 

variables from Model 5. After splitting the file by imputation, I then ran a new OLS 

regression which generated 12 different results: the original result from Model 5, 10 

imputed models

7
, and a pooled imputation model. Table 11 presents the results from 

Model 5 and the pooled imputation model for comparison; in short, from Model 5 through 

2
all of the imputed models and the pooled model, the R
 values do not significantly change, 

indicating that the explained variance does not change significantly. 

Before conducting the missing value analysis and the multiple imputation process, 

the importance of trust on how people engage in co-production with courts was quite clear 

across models. The trust factor score maintained significance regardless of which controls 

7

 See Appendix A for the results of the other 10 imputations relative to Model 5. 

53 

 
 
 
 
were added to or subtracted from the regression. This seemed to indicate that people’s 

trust in the courts in their communities was the dominant factor in how they engaged in co-

production with those courts. Comparing Model 5 to the multiple imputation, however, 

may change the analysis. Looking at the missing value patterns across all variables in the 

model, only half of the respondents have a valid value for the trust variable and only 122 

cases have complete data for all variables, leading to the n=122 for Model 5, as shown in 

Figure 1.  
                    Figure 1. Missing values summary.

After running the multiple imputation, comparing Model 5 to the pooled model leads to 

some interesting observations. First, the coefficient for education does not measurably 

change but it does become significant at the .01 level. As previously mentioned, this does 

seem to agree with previous research indicating that people with higher levels of education 

are less inclined to engage in co-production with the courts. A new entrant to significance is 

community leadership, which is positively associated with co-production and statistically 

p

significant (B=.290, 

=.021). Once considered across the entire study sample, this inference 

makes sense. Co-production is often considered a community-level concept (Osborne et al., 

2016; Van Eijk & Steen, 2016); these respondents are proxies for the rest of the community, 

54 

 
 
 
 
and they can act as examples of how co-production could work for the rest of the 

community. 

However, of primary interest for this study is what happens to trust when missing 

data are imputed. As a reminder, through the five OLS models in Table 10, trust maintained 

p

a 

-value of no more than .022 and averaged a coefficient of .375. When pooling all ten 

imputation models, however, trust loses statistical significance. This infers that, if the data 

were complete across all 419 respondents, trust might not hold the same importance as it 

did in Model 5 when there were only 122 complete cases. In the final chapter, I discuss 

some of the implications of these findings and how future research can expand upon them. 

55 

 
 
 
 
Chapter 6: Discussion and Conclusion 

The relationship between the public and the public sector has always existed in a 

state of change and, in some cases, a state of tension. Depending on the prevailing 

governance model, the public has been viewed by the government in a limited fashion from 

mere audience members (Meinke, 2008) to constituents and voters (Kim, 2021). It has 

taken several decades for the government to be willing to see the public’s ability to 

participate “beyond the ballot box” (Frederickson, 1996, p. 265) in providing public 

services. This is where the concept of co-production enters the discussion. As the 

“voluntary or involuntary involvement of public service users in any of the design, 

management, delivery and/or evaluation of public services” (Osborne et al., 2016, p. 640), 

co-production is a way that the public can participate in typically government-provided 

services that benefit them. 

In the public administration and economics literature from which much of the 

scholarship on co-production has been developed, we typically see it applied to services 

such as health care, education, and elder care, among others. The question then becomes 

which public services within the realm of criminal justice, justice-related services, or for 

the purpose of this dissertation the courts, might be appropriate for co-productive activity. 

The extant literature on co-production and the justice system largely has involved policing 

and how people can co-produce public safety with the police through neighborhood 

watches and citizen review boards among other opportunities. When the concept of co-

production has been applied to the courts, it has been limited and for example, in different 

jurisdictions that would not be generalizable to the American context (Van Gils et al., 2021). 

This dissertation has contributed to this discussion. The extant literature on co-production 

56 

 
 
 
 
has also invited application of its models and concepts in other policy domains that may 

not be immediately considered when thinking about co-production and public services 

(Loeffler, 2021; Van Eijk & Steen, 2016). This dissertation has contributed to those efforts. 
Discussion

Procedural fairness 

The literature on the effect of procedural fairness on cooperation, or the effects of 

procedural justice and legitimacy, is quite clear and quite robust (e.g., De Cremer & Tyler, 

2007; Tyler, 1988; Tyler, 2006). However, making the step forward from cooperation and 

legitimacy to Osborne et al.’s (2016) four components of co-production (design, 

management, delivery, evaluation) as measured by my dependent variable was shown by 

my models not to be affected by perceptions of procedural fairness. As these perceptions 

fall under the broader research question of experience with courts, it is interesting to note 

here the fact that the fact that demographic controls not having any significant effect on co-

production parallels Tyler’s (1988) result that a person’s characteristics do not influence 
not 

their criteria on whether a procedural decision is fair. Overall, though, Hypothesis 1 is 
supported

Risk 

. 

The public administration literature sees risk largely as a negative influence or a 

barrier against co-production, both from the service provider’s and the service user’s 

perspective. The service provider risks losing the control they have had for an extended 

period of time and the loss of reputation from getting the blame for poor results, even if the 

public co-producer may be to blame (Loeffler, 2021); it may take significant human and 

financial capital to convince them to participate in co-production (Kim, 2021). The service 

57 

 
 
 
 
 
user also will likely have to spend time and other resources to participate in co-production 

when the potential benefits may not be immediately apparent (Loeffler, 2021).  

In this study, even though never statistically significant, the results of the 

regressions suggested a direct relationship between risk and co-production. Why in a risk-

laden environment, when many of the stakeholders may be risk-averse and resistant to 

change, would risk be directly related to levels of co-production? Brown and Osborne 

(2013) provide some guidance: it may be cliché, but innovation requires stepping outside 

of one’s comfort zone through those risks to reputation, potential expenditure, or loss of 

resources and personal capital. Without a shared understanding among all of the involved 

stakeholders of the risks, fear and misapprehension However, even though the potential 

rewards of innovation through co-production may not be immediate, they are potentially 

significant. For example, if one community leader is encouraged enough by these pilot 

projects to continue to work with the courts in their community, and potentially they end 

up encouraging their own constituents (neighborhood members, church members) to do 

the same on an issue like getting members of the community into treatment courts or 
not supported

conflict resolution, that can only be of benefit. However, Hypothesis 2 is 
Trust 

. 

Given the significant literature supporting a connection between trust and the 

provision of other public services, I expected and hypothesized a connection between trust 

and court services. Whether it is a dispute over a $500 car repair bill, mediation concerning 

a child custody dispute, or criminal prosecution, to the individuals involved there is always 

an important matter at issue. The courts have to keep that in mind as they approach these 

issues, and the individual’s need to feel self-determination should motivate them to engage 

58 

 
 
 
in co-production, where they would have greater influence over “their own life politics” 

and make them a “key arbiter of service quality and performance” (Fledderus et al., 2014, p. 

434). Throughout the five OLS models, trust maintained a pattern of statistical significance 

to an extent that it seemed as if it was going to share the same importance as in the rest of 

the co-production literature; however, it lost statistical significance in the pooled 

imputation model. This suggests that trust may not be as important as originally suggested, 

but for some people it still influences their levels of co-production with the courts. 

partially supported

Therefore, Hypothesis 3 is 
Alternative measures 

. 

During the process of data analysis, there was some thought that trust and 

procedural fairness could be measured using single-item measures, instead of the two-item 

factor scores that I have used in my analysis. Appendix B contains the results of OLS 

regressions and imputations using single-item measures for those independent variables. 

While there were some increases in the number of cases in the OLS regressions and some 

variables that gained and lost statistical significance, none of the conclusions related to my 

hypotheses changed using those measures. 
Implications for Court Administration

“No government can be efficient and equitable without considerable input from 

citizens” (Ostrom, 1996, p. 1083). In the eyes of much of the public, courts’ reputations 

have suffered from the public’s perception of a lack of efficiency, equitability, and trust over 

the past several years (Shaw et al., 2024, ‘The justices dropped this bomb…’). Part of this 

negative perception comes from the fact that Supreme Court justices are appointed by 

political actors (Buchanan & Meller, 2019, “Brett Kavanaugh: A representation of the 

59 

 
 
 
 
damaged U.S. judiciary”); at other jurisdictional levels, many judicial actors are elected. And 

as part of the electoral process, there is a demonstrated connection between public opinion 

and judicial behavior (Nelson, 2014). Given this, there is a significant need for research and 

techniques to improve these relationships so that the function of the courts for all involved 

can improve. Hearings can take an appropriate amount of time and judgments can be 

handed down in an efficient manner (Van Gils et al., 2021). People can walk away from an 

experience with the courts feeling as if they were able to tell their story and plead their 

case, even if the result of the experience went against their interest (Longazel et al., 2011). 

This study contributes to these efforts by showing that increasing trust in courts increases 

co-production at the individual level. It is important to note that much of the literature 

emphasizes the fact that co-production is best accomplished in a personalized fashion 

(Weaver, 2011), and that “[w]hat works well in what set of circumstances may be 

unsuitable for another” (Alford, 2014, p. 312). This research could be used to build trust 

networks (Fledderus et al., 2014) as a way to sustain co-productive efforts in courts over 

time. Increasing trust works, at least in certain contexts. 

Everyone covered by the control variables in this dissertation (i.e., plaintiffs, 

defendants, court employees, members of the public) has a role to play in the function of 

the courts and could be part of the related trust network. Building this network requires 

making the relationships between co-producers explicitly known, establishing shared 

rights and obligations, communicating shared connections, and developing boundaries 

from outsiders. Brown and Osborne’s (2013) research on risk governance is instructive on 

this front. Public agencies can contribute by providing the enabling frameworks and 

convening the groups in which collaboration and innovation can take place. Court 

60 

 
 
 
administrators can provide grant writing and other resources to which members of the 

public likely do not have access. Members of the public in collaboration with those 

professionals then can use those resources to operate the courts, run advisory councils, 

connect probationers and other court users with community resources, and engage in 

other co-productive efforts (Fledderus et al., 2014). 

Further, increased trust in the courts allows court personnel to delegate court 

functions within the court system and also to actors not traditionally associated with the 

court system as a form of co-production. As the public increases trust in the courts, they are 

more likely to act in a manner comporting with any positive expectations that the courts 

have of them (Rousseau et al., 1998). For example, while the government may have full 

control over formal criminal sentencing and punishment, if the public trusts the 

government enough to engage in co-productive activity, court personnel may then delegate 

some court functions back to the public in the form of techniques like restorative justice, 

which according to Howard Zehr (2015) can be outlined as “…address[ing] the harms and 

needs of those harmed, hold those causing harm accountable to ‘put right’ those harms, and 

involve both of these parties as well as relevant communities in this process” (p. 35). This is 

also where the findings on risk come into play. Both court administrators and members of 

the public have to take on some level of risk in order to have co-production become 

successful, and also for any level of innovation to take place. However, potential benefits of 

more just outcomes in court proceedings, better use of financial resources, and better 

goodwill with the community may follow. 

61 

 
 
 
 
 
Limitations

The primary limitation to my study is one that cannot be avoided, and that is the low 

number of overall cases in the data set (N

0
=419). Beyond that, after accounting for all of the 

independent variables and the controls, Model 5 was based on 122 cases, which is 

significantly low. I ended up having to bring in a high number of cases through multiple 

imputation. I wanted to bring in the larger number of cases to avoid as much bias as 

possible from the low number of cases, which likely inflated some of the significance in my 

models. Even though I lost some contact with the actual data, my conclusions were 

improved through my use of multiple imputation, an accepted missing data replacement 

technique. 

When working through the measurement part of the dissertation process, my first 

effort was to go through the data set and find variables that seemed to intuitively fit the 

concepts I was trying to measure. This led to less focus and refinement than what was 

eventually used in the dissertation. I based each of the variables here on prior research, 

working for example from Osborne et al.’s (2016) definition of co-production and the dual 

definitions of trust from Mayer et al. (1995) and Rousseau et al. (1998), lending validity to 

each measure. Even though this dissertation is more exploratory in nature, grounding the 

measures in prior research is essential. Finding components for the trust variable was not 

difficult, which makes sense as the PEPP projects were targeted at “improving public trust 

and confidence” in the courts and not specifically at co-production. Again, though, prior 

research was instructive in finding enough component variables to create reasonable 

measurements of each variable of interest. Any future research based on this study should 

include a longitudinal component, as this study concentrated on present co-production. 

62 

 
 
 
 
Future research should design survey instruments specifically with these concepts in mind, 

so that these measurements can improve even more.   

It was important to include as much information about each case as the data set 

provided, given the low number of cases, hence the higher number of control variables. And 

for the most part in the models, they did not have a significant impact on the respondents’ 

level of co-production. Though not the subject of any of the hypotheses or the direct focus 

of the dissertation, several of the controls have been the focus of prior research on co-

production. Take, for example, education, which has been shown to have an effect on co-

productive activity (Kang & Van Ryzin, 2019; Loeffler, 2021). In my study, education had a 

considerable but not statistically significant effect on co-production before and after data 

imputation. Future research on co-production and the courts should be designed 

specifically to test the effect of education, as it was not a focus of mine. 

An additional limitation to my study is the lack of generalizability, including the fact 

that this is not an experimental study. Additionally, these data were collected at five 

different sites across the country. While they show geographic, cultural, political, and other 

forms of variety, they are not representative of the entire country. Future research could 

expand on my measures and models by using a more representative dataset for improved 

generalizability.  
Implications for Practice and More on Future Research 

This study also shows how more people could participate with the courts and with 

justice-related issues overall. As an example, one of the control variables in my models was 

whether the respondents had ever served on a jury. This is probably the most easily 

accessible example for the largest number of people participating with the courts. 

63 

 
 
 
 
 
Hannaford-Agor and Moffett (2024) say that in 2021, an estimated 33 million summonses 

were sent out for jury duty in state courts; far fewer than that were eventually qualified to 

serve on a jury, let alone actually served. Among their study sample, 27.7% replied they 

had served on a jury. Co-production through one of the four main methods could be 

another way to bring the community closer to the courts in addition to jury duty.  

The findings support the fact that those who identify as community leaders engage 

in more co-production. According to Tørfing et al. (2019, p. 817), part of advancing 

systemic change is transforming the public sector from professional knowledge being “the 

holy truth” to where there is a premium on “dialogue, curiosity, and openness,” which can 

be seen in design, management, delivery, and evaluation. If formal leaders are able to 

delegate the co-creative or co-productive process to “experienced and trusted participants” 

such as community leaders, they would better allocate their time to tasks for which co-

production is not an option. Durose et al. (2017) found that co-production is most likely to 

grow through the spread of ideas at the local level using locally appropriate practices. 

Community leaders are ideally positioned to know what those practices are, and if they are 

already engaged in co-production with the courts through participation on an advisory 

board or a peer support group, that makes entry for other members of the community even 

easier, starting with smaller-scale activities and growing to full engagement (Durose et al., 

2017). 
Conclusion 

My choice to study co-production and how it could apply to the provision of court 

services comes from its potential to bring more people into the function of a public service 

that is sometimes tension-filled and controversial. It is important to see co-production as 

64 

 
 
 
multi-faceted and individualized. Part of the value of the concept of co-production is that it 

provides an avenue for people to individualize their experience with public services 

(Alford, 2014). Going through the court process can be intimidating for anyone, especially 

for people who feel like, as Feeley (1979) describes it, the process itself is the punishment. 

Lower-level processes, whether they are misdemeanor criminal court cases (Natapoff, 

2018) or local eviction proceedings (Sudeall & Pasciuti, 2021), can treat the participants as 

impersonal widgets on an ever-running assembly line. Participants are not offered any 

choices or input, and when they are, these are limited. Any notions of fair procedures or 

treatment (Tyler, 1988) are minimized. 

While my hypotheses generally were not supported, it is still important that I found 

that for some people, their trust in the courts has a positive effect on their co-productive 

activity. Co-production offers an opportunity to “deep[ly]” personalize (Weaver, 2011) the 

courts for those who are involved with them. This deep personalization, according to 

Weaver, requires sustained negotiation, interaction and support (pp. 1042-1043) that can 

take the form of trust networks (Fledderus et al., 2014; see also Van Eijk & Steen, 2016).  

Co-production involves the actions of several or many individuals in a personalized 

fashion, yet it is an inherently social enterprise, impacting social order (Diver, 2017). Even 

though there may be some significant individual and administrative risks to finances and 

capital (Weaver, 2019), there also may be significant benefits. All co-productive processes, 

even the largest-scale, most sustained ones, must start from somewhere. This dissertation, 

and the opportunities for others to build upon its findings, represents a valuable 

contribution to the research in this field. Further, court administrators, community leaders, 

65 

 
 
and interested members of the public can use its findings to be encouraged to engage in co-

productive activities with their local courts to improve the courts for everyone. 

66 

 
 
 
 
 
REFERENCES 

Public Administration Review, 

44

Adams, B. (1984). The frustrations of government service. 

(1), 5-13. 

Alford, J. (2008). The limits to traditional public administration, or rescuing public value 

The Australian Journal of Public Administration, 67

from misrepresentation. 
366. 

Public Management Review, 16

(3), 357-

Alford, J. (2014). The multiple facets of co-production: Building on the work of Elinor 

Ostrom. 

(3), 299-316. 

Atlanta City Council (2024). Neighborhood Planning Unit (NPU). 

https://citycouncil.atlantaga.gov/other/npu-by-neighborhood/neighborhood-
planning-unit 

Bovaird, T. (2007). Beyond engagement and participation: User and community 

Public Administration Review, 67

coproduction of public services. 

(5), 846-860. 

Voluntas, 23, 

Bovaird, T. & Loef�ler, E. (2012). From engagement to co-production: The contribution of 

users and communities to outcomes and public value. 
between professionals and the public are crucial to improving public services

The challenge of co-production: How equal partnerships 

1119-1138. 

Boyle, D. & Harris, M. (2009). 

. NESTA. 

Brix, J., Krogstrup, H. K., & Mortensen, N. M. (2020). Evaluating the outcomes of co-
(2), 169-185. 

production in local government. 

Local Government Studies, 46

Brown, L. & Osborne, S. P. (2013). Risk and innovation: Towards a framework for risk 

Public Management Review, 15

governance in public services. 
Public Administration Review, 43

(2), 186-208. 

Brudney, J. L. & England, R. E. (1983). Toward a de�inition of the coproduction concept. 

(1), 59-65. 

Public 

Bryson, J. M., Crosby, B. C., & Bloomberg, L. (2014). Public value governance: Moving beyond 

Administration Review, 74
traditional public administration and the new public management. 
(4), 445-456. 

Buchanan, M. J. & Meller, A. (2019, Oct. 1). Brett Kavanaugh: A representation of the 

damaged U. S. judiciary. https://www.americanprogress.org/article/brett-
kavanaugh-representation-damaged-u-s-judiciary/ 

Cheng, Y. (2019). Exploring the role of nonpro�its in public service provision: Moving from 

coproduction to cogovernance. 

(2), 203-214. 

Public Administration Review, 79

67 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Journal of Applied Psychology, 92

De Cremer, D. & Tyler, T. R. (2007). The effects of trust in authority and procedural fairness 

on cooperation. 

(3), 639-649. 

Public Administration Review, 60

Denhardt, R. B. & Denhardt, J. V. (2000). The new public service: Serving rather than 

steering. 

(6), 549-559. 

Diaz, J. (2023, June 21). Jan. 6 rioter who used a stun gun on Of�icer Michael Fanone 

rioter-sentenced-of�icer-michael-fanone 
sentenced to prison. 

NPR. https://www.npr.org/2023/06/21/1183558868/jan-6-

Environmental Science & Policy, 73, 

Diver, S. (2017). Negotiating Indigenous knowledge at the science-policy interface: Insights 

from the Xáxli’p Community Forest. 
http://dx.doi.org/10.1016/j.envsci.2017.03.001 

1-11. 

British Journal of Criminology, 60, 
Doherty, E. E. & Bersani, B. E. (2020). What protects those at high risk from criminal justice 

contact despite the odds? A negative case analysis. 
1627-1647. 

Evidence & Policy, 13

Durose, C., Needham, C., Mangan, C., & Rees, J. (2017). Generating “good enough” evidence 

for co-production. 

The process is the punishment: Handling cases in a lower criminal court. 

(1), 135-151. 

Feeley, M. M. (1979). 

Russell Sage Foundation. 

Fledderus, J., Brandsen, T., & Honingh, M. (2014). Restoring trust through the co-production 

Public Management Review, 16

of public services: A theoretical elaboration. 
443. 

(3), 424-

Public Administration Review, 56
Frederickson, H. G. (1996). Comparing the reinventing government movement with the new 

public administration. 

(3), 263-270. 

Graham, J. W., Olchowski, A. E., & Gilreath, T. D. (2007). How many imputations are really 

Science, 8
needed? Some practical clari�ications of multiple imputation theory. 

Prevention 

(3), 206-213. 

The Journal of Private 
Haeffele, S. & Storr, V. H. (2019). Hierarchical management structures and housing the poor: 

Enterprise, 34
An analysis of Habitat for Humanity in Birmingham, Alabama. 

(1), 15-37. 

2023 state-of-the-states survey of jury improvement 

efforts

Hannaford-Agor, P. & Moffett, M. (2024). 

. National Center for State Courts. 

Public Administration, 69, 

Hood, C. (1991). A public management for all seasons? 

3-19. 

68 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Impute missing data values (multiple imputation). 

IBM (2024). 

https://www.ibm.com/docs/en/spss-statistics/29.0.0?topic=imputation-impute-
missing-data-values-multiple 

Public Management Review, 21

Kang, S. & Van Ryzin, G. G. (2019). Coproduction and trust in government: Evidence from 

survey experiments. 

(11), 1646-1664. 

Kim, Y. (2021). Searching for newness in management paradigms: An analysis of intellectual 

51
history in U.S. public administration. 

American Review of Public Administration, 

(2), 79-106. 

Co-production of public services and outcomes

Loef�ler, E. (2021). 

Palgrave MacMillan. 

. Cham, Switzerland: 

International 

Loef�ler, E. & Bovaird, T. (2020). Assessing the impact of co-production on pathways to 

Public Management Journal, 23
outcomes in pubic services: The case of policing and criminal justice. 

(2), 205-223. 

Loef�ler, E. & Timm-Arnold, P. (2020). Comparing user and community co-production 

Public Policy and Administration, 36

approaches in local “welfare” and “law and order” services: Does the governance 
mode matter? 
https://www.nytimes.com/2023/06/16/us/defund-police-minneapolis.html 

The New York Times. 

(1), 115-137. 

Londoñ o, E. (2023, June 16). How ‘defund the police’ failed. 

Longazel, J. G., Parker, L. S., & Sun, I. Y. (2011). Experiencing court, experiencing race: 

Perceived procedural injustice among court users. 

(2), 202-227. 

Race and Justice, 1

Loper Bright Enterprises v. Raimondo, 603 U.S. ___ (2024). 

https://www.supremecourt.gov/opinions/23pdf/22-451_7m58.pdf 

Lynn, L. E. Jr. (2001). The myth of the bureaucratic paradigm: What traditional public 
(2), 144-160. 

administration really stood for. 

Public Administration Review, 61

Manly, C. A. & Wells, R. S. (2015). Reporting the use of multiple imputation for missing data 

Research in Higher Education, 56, 

in higher education research. 

Academy of Management Review, 20

397-409. 

Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organizational 

trust. 

(3), 709-734. 

Criminology & Criminal Justice, 16

McCulloch, T. & Members of Positive Prison? Positive Futures (PPPF) (2016). Co-producing 

justice sanctions? Citizen perspectives. 
451. 
Criminology and Criminal Justice, 9

(4), 431-

McIvor, G. (2009). Therapeutic jurisprudence and procedural justice in Scottish drug courts. 
(1), 29-49. 

69 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
McLennan, B. J. (2020). Conditions for effective coproduction in community-led disaster 

Voluntas, 31

risk management. 
9957-2 
Political Research Quarterly, 61

, 316-332. https://doi.org/10.1007/s11266-018-

Meinke, S. R. (2008). Institutional change and the electoral connection in the Senate. 

(3), 445-457. 

Mitchell, M. M., Fahmy, C., Clark, K. J., & Pyrooz, D. C. (2022). Non-random study attrition: 

Journal of Quantitative Criminology, 38, 

Assessing correction techniques and the magnitude of bias in a longitudinal study of 
reentry from prison. 
Creating public value: Strategic management in government. 

755-790. 

Moore, M. (1995). 

Harvard University Press. 

Boston: 

Motley, R. O. Jr., Chen, Y., Johnson, C., & Joe, S. (2020). Exposure to community-based 

Social Work Research, 44

violence on social media among black male emerging adults involved with the 
criminal justice system. 

(2), 87-97. 

Nabatchi, T., Sancino, A., & Sicilia, M. (2017). Varieties of participation in public services: 

The who, when, and what of coproduction. 
776. 
the innocent and makes America more unequal

(5), 766-
Punishment without crime: How our massive misdemeanor system traps 

Public Administration Review, 77

Natapoff, A. (2018). 

National Center for State Courts (NCSC) (2024). 

https://www.ncsc.org/consulting-and-research/areas-of-expertise/racial-
justice/resources/community-engagement-initiative/public-engagement-pilot-
projects 

Public Engagement Pilot Projects. 

. Basic Books. 

Nederhand, J., Klijn, E. H.,  Van der Steen, M., & Van Twist, M. (2019). The governance of self-

Policy Sciences, 52, 
organization: Which governance strategy do policy of�icials and citizens prefer? 

233-253. 

Social Policy & Society, 7

Needham, C. (2008). Realising the potential of co-production: Negotiating improvements in 

public services. 

(2), 221-231. 

Nelson, M. J. (2014). Responsive justice? Retention elections, prosecutors, and public 

Journal of Law and Courts, 2

https://doi.org/10.1086/674527 

opinion. 

(1), 117-152. 

Osborne, S. P., Radnor, Z., & Strokosch, K. (2016). Co-production and the co-creation of value 

Public Management Review, 18

in public services: A suitable case for treatment? 
639-653. 
Science Quarterly, 53

(5), 

Social 

Ostrom, E. (1972). Metropolitan reform: Propositions derived from two traditions. 

(3), 474-493. 

70 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
World Development, 24

Ostrom, E. (1996). Crossing the Great Divide: Co-production, synergy, and development. 

(6), 1073-1087. 

Journal of Economic 

Perspectives, 14

Ostrom, E. (2000). Collective action and the evolution of social norms. 

(3), 137-158. 

Ostrom, E., Parks, R. B., Whitaker, G. P., & Percy, S. L. (1978). The public service production 

process: A framework for analyzing police services. 
389. 

381-

Policy Studies Journal, 7, 

Parks, R. B., Baker, P. C., Kiser, L., Oakerson, R., Ostrom, E., Ostrom, V., Percy, S. L., Vandivort, 
M. B., Whitaker, G. P., & Wilson, R. (1981). Consumers as coproducers of public 
9
services: Some economic and institutional considerations. 

Policy Studies Journal, 

(7), 1001-1011. 

Robertson, G. D. (2024, July 25). North Carolina review say nonpro�it led by lieutenant 

Associated Press

governor’s wife ‘seriously de�icient.’ 
https://apnews.com/article/north-carolina-lieutenant-governor-spouse-nonpro�it-
review-2b2d2303df7165d129675c3143de7237# 
Public Policy, 6

Criminology & 

. 

Rottman, D. B. (2007). Adhere to procedural fairness in the justice system. 

(4), 835-842. 

Rousseau, D. M., Sitkin, S. B., Burt, R. S., & Camerer, C. (1998). Not so different after all: A 

Academy of Management Review, 23

cross-discipline view of trust. 
missing data 

Rubin, D. B. (2018). Multiple imputation. In S. van Buuren (Ed.), 

nd

(2

 ed., pp. 29-62). CRC Press. 

(3), 393-404. 
Flexible imputation of 

Scott, J. D. (2002). Assessing the relationship between police-community coproduction and 

Journal of Contemporary Criminal Justice, 18

neighborhood-level social capital. 
147-166. 
127

The North American Review, 

(2), 

Seymour, H. (1878). The government of the United States. 

(265), 359-374. 

Shaw, K., Baude, W., & Vladeck, S. I. (2024, July 11). ‘The justices dropped this bomb’: Three 

The New York Times

legal experts on a shocking Supreme Court term. 
https://www.nytimes.com/2024/07/11/opinion/supreme-court-term-
immunity.html 
Private wants and public needs

. 

Stigler, G. J. (1962). The tenable range of functions of local government. In E. S. Phelps (Ed.), 

 (p. 146). W. W. Norton. 

Vanderbilt Law Review, 74

Sudeall, L. & Pasciuti, D. (2021). Praxis and paradox: Inside the black box of eviction court. 

(5), 1365-1434. 

71 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
Tor�ing, J., Sørensen, E., & Røiseland, A. (2019). Transforming the public sector into an arena 

Society, 51
for co-creation: Barriers, drivers, bene�its, and ways forward. 

Administration & 

(5), 795-825. 

Journal of Criminal Justice, 34

Sun, I. Y. & Wu, Y. (2006). Citizens’ perceptions of the courts: The impact of race, gender, and 

recent experience. 

Procedural justice: A psychological analysis. 

, 457-467. 

Thibaut, J. & Walker, L. (1975). 

Erlbaum. 

Lawrence 

Tyler, T. R. (1988). What is procedural justice? Criteria used by citizens to assess the 

fairness of legal procedures. 

Why people obey the law. 

(1), 103-135. 

Law and Society Review, 22

Tyler, T. R. (2006). 
Press. 

Princeton, New Jersey: Princeton University 

Ugwudike, P. (2017). Understanding compliance dynamics in community justice settings: 

Review, 27
The relevance of Bourdieu’s habitus, �ield, and capital. 

International Criminal Justice 

(1), 40-59. 

Public Administration Review, 78
Uzochukwu, K. & Thomas, J. C. (2018). Who engages in the coproduction of local public 

services and why? The case of Atlanta, Georgia. 
514-526. 

(4), 

International Review of Administrative Sciences, 
Van Eijk, C. & Steen, T. (2016). Why engage in co-production of public services? Mixing 

82
theory and empirical evidence. 

(1), 28-46. 

Van Gils, M., Baardman, F., & Langbroek, P. (2021). Feedback for professionals: Co-

Justice System Journal, 42

production of court services by mirrormeeting-focusgroups for the judiciary in the 
Netherlands. 

(2), 164-179. 

Wadsworth, T. & Roberts, J. M. Jr. (2008). When missing data are not missing: A new 
Criminology, 46
approach to evaluating supplemental homicide report imputation strategies. 

(4), 841-870. 

Weaver, B. (2011). Co-producing community justice: The transformative potential of 

personalization for penal sanctions. 

1038-1057. 

British Journal of Social Work, 41, 

Social Policy & Administration, 53

, 

Weaver, B. (2019). Co-production, governance and practice: The dynamics and effects of 
(2)
The New York Times

user voice prison councils. 

249-264. 

Wegman, J. (2024, July 12). The Supreme Court is gaslighting us all. 

. 
https://www.nytimes.com/2024/07/12/opinion/supreme-court-psychological-
manipulation.html 

72 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Public Administration 

Review, 40

Whitaker, G. P. (1980). Citizen participation in service delivery. 

(3), 240-246. 

Statistics in Medicine, 30

White, I. R., Royston, P., & Wood, A. M. (2011). Multiple imputation using chained equations: 

Issues and guidance for practice. 
Risk Analysis, 39

(4), 377-399. 

Wilson, R. S., Zwickle, A., & Walpole, H. (2019). Developing a broadly applicable measure of 
(4), 777-791. 

The little book of restorative justice, revised and updated

risk perception. 

Zehr, H. (2015). 

Books. 

. New York: Good 

73 

 
 
 
 
 
 
 
APPENDIX A: IMPUTATIONS 

Table A.1. OLS coefficients for each of the imputed data sets: Model 5. 

Procedural fairness 
Original OLS Model 5 (n=122) 
Trust 
Risk 

Procedural fairness 
Imputation 1—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 2—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 3—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 4—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 5—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 6—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 7—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 8—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 9—Model 5 
Trust 
Risk 

B 
-.036 
.389 
.092 

-.026 
.292 
.075 

.014 
.269 
.078 

-.016 
.203 
.094 

-.026 
.290 
-.028 

-.100 
.386 
.186 

-.063 
.301 
-.039 

-.037 
.333 
.105 

-.075 
.336 
.001 

-.137 
.399 
.067 

t 
-.229 
2.365 
.862 

-.214 
2.237 
.873 

.115 
2.002 
.884 

-.139 
1.629 
1.137 

-.220 
2.311 
-.344 

-.867 
3.088 
2.265 

-.516 
2.309 
-.461 

-.319 
2.703 
1.294 

-.635 
2.668 
.011 

-1.215 
3.292 
.842 

p 

.819 
.020* 
.391 

.831 
.027* 
.384 

.909 
.047* 
.378 

.889 
.105 
.257 

.827 
.022* 
.731 

.387 
.002** 
.025* 

.607 
.022* 
.645 

.750 
.008** 
.198 

.526 
.008** 
.991 

.226 
.001** 
.401 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

Overall 

1.734 
.029* 
.322 

2.912 
<.001** 
.304 

2.789 
<.001** 
.295 

2.984 
<.001** 
.310 

2.219 
<.001** 
.250 

2.538 
<.001** 
.276 

3.156 
<.001** 
.322 

2.105 
.003** 
.240 

3.463 
<.001** 
.342 

3.262 
<.001** 
.329 

74 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Table A.1 (cont’d) 

Procedural fairness 
Imputation 10—Model 5 
Trust 
Risk 

Procedural fairness 
Pooled model 
Trust 
Risk 
p 

p 

**

< .01; *

< .05

F 
Sig. 
R2 

3.535 
<.001** 
.347 

.084 
.176 
-.020 

-.054 
.390 
.097 

.728 
1.402 
-.245 

-.337 
2.085 
1.041 

.468 
.163 
.806 

.711 
.056 
.310 

75 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
APPENDIX B: ALTERNATIVE MEASURES OF KEY CONCEPTS 

Table B.1. All OLS regression models with single-item measures. 

Procedural fairness 
Model 1 (n=208) 
Trust 
Risk 

Procedural fairness  
Model 2 (n=208) 
Trust 
Risk 
Data collection site: 

Procedural fairness 
Model 3 (n=197) 
Trust 
Risk 
Experience: 

Midwest state 
Southern state 
East Coast state 
Great Lakes state 

Work with the courts 
Community leadership 
Served on a jury 
Defendant in civil or criminal case 
Witness 
Plaintiff (brought case to court) 
Juvenile justice or child welfare case 
Probationer (on probation) 
Engaged as member of the public 
Other 

Procedural fairness 
Model 4 (n=198) 
Trust 
Risk 
Demographics: 

Age 
American Indian or Alaska Native 
Asian 
Black or African American 
Spanish, Hispanic, Latina/o/x, Puerto Rican 
Race, other 

B 
.062 
.118 
.085 

.063 
.093 
.089 

-.168 
.192 
.216 
.401 

.041 
.149 
.086 

-.302 
.147 
-.164 
-.109 
-.052 
.033 
-.078 
-.100 
.323 
-.197 

.070 
.104 
.091 

.004 
-.050 
.361 
.068 
.335 
-.900 

t 
1.477 
2.417 
1.161 

1.440 
1.808 
1.170 

-.823 
1.085 
.916 
.782 

.929 
2.825 
1.044 

-1.572 
.902 
-1.038 
-.573 
-.305 
.158 
-.469 
-.449 
1.896 
-.821 

1.569 
1.928 
1.195 

.773 
-.282 
.712 
.335 
1.713 
-2.016 

p 

.141 
.017* 
.247 

.151 
.072 
.243 

.411 
.279 
.361 
.435 

.354 
.005** 
.298 

.118 
.368 
.301 
.567 
.761 
.874 
.639 
.654 
.060 
.412 

.118 
.055 
.234 

.440 
.778 
.477 
.738 
.088 
.045* 

F 
Sig. 
R2 

Overall 

4.700 
.003** 
.065 

F 
Sig. 
R2 

2.564 
.015* 
.082 

F 
Sig. 
R2 

1.936 
.029* 
.121 

F 
Sig. 
R2 

2.781 
.002** 
.153 

76 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
F 
Sig. 
R2 

2.242 
.001** 
.267 

Ideology (liberal is high)
Education level 
Gender 

Table B.1 (cont’d) 

Procedural fairness 
Model 5 (n=187) 
Trust 
Risk 
Demographics: 

Experience: 

Age 
American Indian or Alaska Native 
Asian 
Black or African American 
Spanish, Hispanic, Latina/o/x, Puerto Rican 
Race, other 
Ideology (liberal is high) 
Education level 
Gender 

Work with the courts 
Community leadership 
Served on a jury 
Defendant in civil or criminal case 
Witness 
Plaintiff (brought case to court) 
Juvenile justice or child welfare case 
Probationer (on probation) 
Engaged as a member of the public 
Other 

Data collection site: 

p 

p 

**

< .01; *

< .05 

Midwest state 
Southern state 
East Coast state 
Great Lakes state 

-.017 
-.123 
.054 

B 
.055 
.121 
.108 

.007 
.042 
.598 
.023 
.455 
-1.342 
-.063 
-.157 
.106 

-.261 
.336 
-.055 
-.045 
.066 
.043 
-.012 
-.247 
.308 
-.389 

-.131 
.068 
.444 
.743 

-.215 
-3.096 
.356 

t 
1.121 
2.122 
1.244 

1.268 
.157 
1.131 
.090 
1.957 
-2.802 
-.744 
-3.600 
.644 

-1.341 
1.948 
-.337 
-.232 
.385 
.206 
-.071 
-1.087 
1.784 
-1.619 

-.388 
.216 
1.305 
1.071 

.830 
.002** 
p 
.722 

.264 
.035* 
.215 

.207 
.875 
.260 
.928 
.052 
.006** 
.458 
<.001** 
.520 

.182 
.053 
.737 
.816 
.701 
.837 
.944 
.278 
.076 
.108 

.698 
.829 
.194 
.286 

77 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Table B.2. OLS coefficients for imputed data sets with single-item measures: Model 5. 

Procedural fairness 
OLS Model 5 (n=187) 
Trust 
Risk 

Procedural fairness 
Imputation 1—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 2—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 3—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 4—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 5—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 6—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 7—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 8—Model 5 
Trust 
Risk 

Procedural fairness 
Imputation 9—Model 5 
Trust 
Risk 

B 
.055 
.121 
.108 

-.031 
.131 
-.066 

.041 
.165 
.100 

.060 
.102 
.110 

.035 
.105 
.117 

.034 
.074 
.035 

.025 
.131 
.038 

.098 
.068 
.134 

.023 
.081 
.015 

.039 
.126 
.055 

t 
1.121 
2.122 
1.244 

-.925 
3.298 
-1.144 

1.209 
4.059 
1.833 

1.770 
2.591 
1.956 

1.035 
2.660 
2.164 

1.022 
1.918 
.659 

.770 
3.375 
.727 

2.755 
1.648 
2.353 

.687 
1.999 
.266 

1.172 
3.349 
1.027 

p 

.264 
.035* 
.215 

.356 
.001** 
.253 

.227 
<.001** 
.068 

.078 
.010** 
.051 

.301 
.008** 
.031* 

.307 
.056 
.510 

.442 
<.001** 
.468 

.006** 
.100 
.019* 

.493 
.046* 
.790 

.242 
<.001** 
.305 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

F 
Sig. 
R2 

Overall 

2.242 
.001** 
.267 

5.871 
<.001** 
.280 

5.958 
<.001** 
.283 

5.628 
<.001** 
.272 

4.942 
<.001** 
.247 

6.266 
<.001** 
.294 

4.965 
<.001** 
.248 

4.605 
<.001** 
.234 

4.303 
<.001** 
.222 

6.676 
<.001** 
.307 

78 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
F 
Sig. 
R2 

6.221 
<.001** 
.292 

Table B.2 (cont’d) 

Procedural fairness 
Imputation 10—Model 5 
Trust 
Risk 

Procedural fairness 
Pooled model (n=419) 
Trust 
Risk 
Demographics: 

Age 
American Indian or Alaska Native 
Asian 
Black or African American 
Spanish, Hispanic, Latina/o/x, Puerto Rican 
Race, other 
Ideology (liberal is high) 
Education level 
Gender 

Experience: 

Work with the courts 
Community leadership 
Served on a jury 
Defendant in civil or criminal case 
Witness 
Plaintiff (brought case to court) 
Juvenile justice or child welfare case 
Probationer (on probation) 
Engaged as a member of the public 
Other 

Data collection site: 

Midwest state 
Southern state 
East Coast state 
Great Lakes state 

p 

p 

**

< .01; *

< .05

.052 
.110 
.030 

.038 
.109 
.057 

.009 
.043 
.452 
.183 
.245 
-1.230 
-.137 
-.110 
.101 

-.211 
.180 
-.052 
.078 
-.007 
.096 
.142 
-.288 
.310 
-.351 

-.303 
.208 
.614 
.495 

1.537 
2.792 
.556 

.785 
2.164 
.680 

2.403 
.180 
.901 
.918 
1.256 
-2.464 
-2.265 
-3.572 
.732 

-1.093 
1.365 
-.333 
.446 
-.040 
.469 
1.087 
-1.335 
2.389 
-1.521 

-1.053 
.793 
2.067 
.815 

.125 
.005** 
.579 

.438 
.034* 
.502 

.017* 
.858 
.369 
.363 
.215 
.018* 
.025* 
<.001** 
.469 

.282 
.177 
.741 
.657 
.969 
.643 
.278 
.187 
.019* 
.136 

.299 
.429 
.046* 
.421 

79