Intimate partner violence (IPV) is the largest cause of injury to women in the US, with 22% of women experiencing a physical assault by a male intimate partner during their lifetime (Reference Campbell, Webster, Koziol-McLain, Block, Campbell, Curry, Gary, Glass, McFarlane, Sachs, Sharps, Ulrich, Wilt, Manganello, Xu, Schollenberger, Frye and LaughonCampbell et al. 2003; Reference Tjaden and ThoennesTjaden & Thoenes 2000). In response to this violence, many victims turn to legal measures to protect themselves and their children. Civil protection orders, often referred to as restraining orders, are among the most widely used legal interventions for IPV (Reference Ballou, Tabol, Liriano, Vazquez-Nuttall, Butler, Boorstein and McGovernBallou et al. 2007; Reference DeJong and Burgess-ProctorDeJong & Burgess-Proctor 2006; Reference JordanJordan 2004; Reference Logan, Shannon and WalkerLogan et al. 2005b). US courts issue roughly 1 million civil protection orders every year (Reference Fleury-Steiner, Miller, Maloney and PostelFleury-Steiner et al. 2016).
A civil protection order can mean the difference between life and death, but not every victim who files a petition is granted a protection order (Reference DurfeeDurfee 2009). Judges hold a great deal of discretion in civil protection order cases, and success in securing orders varies greatly (Reference McFarlane, Malecha, Gist, Watson, Batten, Hall and SmithMcFarlane et al. 2004; Reference Zoellner, Feeny, Alvarez, Watlington, O'Neill, Zager and FoaZoellner et al. 2000). For instance, Ptacek's formative study highlighted the range of demeanors judges used interacting with female victims and male defendants in domestic violence protection order cases, and how judges viewed their role in these cases, and how their demeanor impacted the quality of Reference Ptacekcourtroom interactions (1999).
Judges may deny orders because cases lack evidence to support victims' claims of violence, stalking, or sexual offenses. But empirical research suggests that extralegal factors not prescribed in the legal statutes, such as the victim's gender or location, can also affect outcomes. Understanding which factors impact judicial decisionmaking in protection order cases carries important implications for victims' safety and for potential inequalities in the justice system.
To identify what factors determine whether judges grant victims' requests for a protection order, this study draws on an original data set of over 1,000 civil protection order cases from Nebraska. Specifically, I examine whether factors such as judges' and victims' sociodemographic characteristics and location influence case outcomes while controlling for objective legal factors such as recency of abuse. By exploring the role of key demographic, abuse, and geographic factors in civil protection order cases, this study speaks to a burgeoning area of research on judicial decisionmaking, expands the empirical basis for theorizing differences in judicial outcomes, and points to new directions for investigating disparities in access to justice for victims of IPV.
1. Background
1.1 Civil Protection Orders
A civil protection order is a legally binding court order that can prohibit abusers from contacting, threatening, or harassing victims. Civil protection orders make violators who breach orders subject to criminal and civil penalties, including fines, contempt penalties, and criminal charges (Klein & Reference Klein and OrloffOrloff 1992; Reference McFarlane, Malecha, Gist, Watson, Batten, Hall and SmithMcFarlane et al. 2004; Reference Orloff, Noche, Olavarria, Martinez-McIntosh, Rose, Baran, Orloff and SullivanOrloff et al. 2004). Broadly speaking, protection orders can require abusers to stay away from locations such as a victim's workplace, residence, or school (Capshew & Reference Capshew and Aaron McNeeceMcNeece 2000; Reference Carlson, Harris and HoldenCarlson et al. 1999) and require the removal of weapons such as firearms from perpetrators' possession (Reference Zeoli, Frattaroli, Roskam and HerreraZeoli et al. 2017). Jurisdictions vary on who qualifies for orders, the protections they offer, and the consequences abusers face if they violate the orders (Reference Eigenberg, McGuffee, Berry and HallEigenberg et al. 2003; U.S. Department of Justice 2002).
The use of civil protection orders originates from reforms in the 1970s that amended the justice system's response to domestic violence. These reforms called for the treatment of domestic violence, or IPV, as legal issues rather than a private family matter. Until the federal Violence Against Women Act (VAWA) of 1994, protection orders were only available to victims through pending divorce cases (Reference ErezErez 2002). The VAWA expanded these protections to unmarried women and men, and strengthened victims' legal recourse by mandating interstate enforcement of protection orders.
Civil protection orders have become one of the most common legal responses to IPV (Reference DeJong and Burgess-ProctorDeJong & Burgess-Proctor 2006; Reference Logan, Shannon and WalkerLogan et al. 2005b). Civil orders are often easier to obtain than criminal convictions (Reference Logan, Shannon, Walker and FaragherLogan et al. 2006). Victims can initiate civil orders pro se, meaning that they can begin the process without an attorney. Civil hearings do not require the approval of a state prosecutor (Johnson & Reference Johnson and DawsonDawson 2010) and are often scheduled within a week of victims' initial filing (Buzawa & Reference Buzawa and BuzawaBuzawa 2003). Civil protection orders also require a lower burden of proof than criminal cases, needing only a “preponderance of the evidence” rather than “proof beyond a reasonable doubt” to proceed (Reference Finn and SteinmanFinn 1991; Reference Lucken, Rosky and WatkinsLucken et al. 2015; Reference WeissmanWan 2000; Weissman 2007). This lower standard makes civil protection orders more available to victims who have not sought formal medical or police confirmation of their abuse, or who have fewer confidants who can corroborate their accounts of abuse (Reference CahnCahn 1991; Reference Durfee, Garcia and CliffordDurfee 2010). In many states, lower standards also apply to enforcement of civil protection orders, allowing police to make arrests for violation of protection orders without meeting any other probable cause standard (Reference Logan, Shannon, Walker and FaragherLogan et al. 2006).
Civil protection orders also empower victims to pursue responses that they see as appropriate to their situations (Fischer & Reference Fischer and RoseRose 1995). For example, victims may file for civil protection orders because they believe that their abusers are less likely to retaliate for a civil protection order than for criminal proceedings (Wallace & Reference Wallace and RobersonRoberson 1996). Victims control the initial filing (Johnson & Reference Johnson and DawsonDawson 2010) and can request specific provisions from the court when filing their petitions, allowing them to tailor their case to their specific concerns. For example, judges can prohibit abusers from contacting or harassing victims, or require abusers to stay away from a specific location, such as the victim's residence (Reference Orloff, Noche, Olavarria, Martinez-McIntosh, Rose, Baran, Orloff and SullivanOrloff et al. 2004). Civil protection orders can also include specific requirements for child support (Reference Logan, Shannon, Walker and FaragherLogan et al. 2006), provisions for temporary protection of minor children (Reference Finn and SteinmanFinn 1991; Reference HartHart 1992; National Council of Juvenile and Family Court Judges 2010) and stipulations that victims must have access to financial resources and transportation (Reference Eigenberg, McGuffee, Berry and HallEigenberg et al. 2003).
1.2 Explaining Judicial Decisionmaking
By providing both short-term and long-term safety measures for victims and their families, granted or rejected civil protection orders can be a life-or-death issue (Reference DurfeeDurfee 2009). But, in contrast to criminal proceedings in which many different attorneys and intermediaries may be involved, this issue depends on direct, discretionary judicial decisionmaking. In the civil protection order process, a victim files a petition, and a judge decides what to do.
Although the specific procedures for filing vary by state, in most cases the victim submits a petition by filing court forms in which they provide written descriptions of the most recent incidents of abuse. In response, the judge may deny the request, grant an ex parte order, or schedule a show cause hearing.Footnote 1 An ex parte protection order, which is a temporary protection order that recognizes an “immediate and present danger” for an individual or household members, does not require the alleged abuser to be present at court (Reference Eigenberg, McGuffee, Berry and HallEigenberg et al. 2003; Reference Logan, Shannon, Walker and FaragherLogan et al. 2006).Footnote 2 Reference Richards, Tudor and GoverRichard et al. (2018) provide an updated replication of Reference DeJong and Burgess-ProctorDeJong and Burgess-Proctor's (2006) research on the victim-friendliness of protection order statutes using 2014 statutes for each state. In their discussion, Richard and co-authors reference that Nebraska was the only state to forbid civil protection order filing assistance by noting that the court “clerk and his or her employees shall not provide assistance in completing the forms” (Nebraska Penal Code §42-924.02).
Each state establishes evidentiary standards for issuing an ex parte order, but Nebraskan judges have discretion to grant an order based on the finding it “reasonably appears from the specific facts included in the affidavit that the petitioner will be in immediate danger of abuse before the matter can be heard on notice” (Nebraska Revised Statute §42-925). In Nebraska, an ex parte protection order or show cause order lasts until the victim goes to court. In this study, a show cause order or an ex parte order are both treated as temporary orders since these outcomes reflect short-term protection, before a final hearing or show cause hearing takes place. This study examines the issuance of temporary orders rather than the outcomes of a final hearing at which time both parties can provide evidence for whether a longer-term order should be denied or continued for a year.
Because they are rich in narrative data and arrive at clearly identifiable outcomes, petitions for civil protection orders provide a unique opportunity to study key questions about decisionmaking and justice. While some scholars in the legal formalist tradition have viewed judicial decisionmaking as a rational, deliberate process of applying the content of the law to the facts of the case (Reference Leiter, Golding and EdmundsonLeiter 2005), more recent critical legal scholarship has highlighted the importance of attention to psychological, political, and social factors in judicial decisionmaking (Reference Danziger, Levav and Avnaim-PessoDanziger, Levav, & Avnaim-Pesso 2011; Reference Sisk, Heise and MorrissSisk, Heise & Morriss 1998).
Previous research has shown that the race and gender of petitioners and respondents (Reference BasileBasile 2005; Reference Muller, Desmarais and HamelMuller, Desmarais, & Hamel 2009), the presence of children, the type of relationship between the parties (married, divorced, dating, etc.), references to a firearm (Reference Jordan, Pritchard, Wilcox and PritchardJordan et al. 2008; Reference Lucken, Rosky and WatkinsLucken, Rosky, & Watkins 2015; Reference YearwoodYearwood 2005), and the organization and quality of victim narratives (Reference DurfeeDurfee 2009; Reference Fitzgerald and DouglasFitzgerald and Douglas 2020) affect civil protection order outcomes. For example, Reference Vittes and SorensonVittes and Sorenson's (2006) analysis of petitions filed at a Los Angeles County domestic violence clinic found that the mention of sexual assault in a petition, but not mention of a firearm, was associated with receiving a temporary protection order.
Other scholarship has examined the role of “place” in the incidence and prosecution of IPV (Reference PruittPruitt 2008; Reference SandbergSandberg 2013; Reference WendtWendt 2009) and highlighted identifiable differences in the utilization of and barriers to services for victims by urban and rural communities (Reference Logan, Evans, Stevenson and JordanLogan et al. 2005a). Reference Logan, Shannon and WalkerLogan et al. (2005b) found that women in three rural Kentucky counties faced more barriers to obtaining and enforcing protection orders than victims living in an urban county. But until now, studies often have been restricted to a sample from a single court, county, city, or organization (Reference McFarlane, Malecha, Gist, Watson, Batten, Hall and SmithMcFarlane et al. 2004; Reference Mears, Carlson, Holden and HarrisMears et al. 2001). In turn, previous research was often restricted to smaller sample sizes and limited control variables (Kingsnorth, Wolcott & Reference Kingsnorth, Wolcott and LonnquistLonnquist 2013).
This study overcomes these limitations by analyzing a sample of over 1,000 civil protection order cases from across an entire US state. Drawing on previous scholarship to generate six hypotheses about judicial decisionmaking, I examine the effects of violence, gender, personal relationships, and place on the granting of civil protection orders. Because the sample scope includes metropolitan and nonmetropolitan counties in the state, I also conduct a novel analysis of “place” as an important factor in judicial decisionmaking. I identify the effect of “place,” determined through residing in a metropolitan or nonmetropolitan county, and victims' sociodemographic characteristics on issuance of temporary protection orders. This study extends the study of civil protection orders in scale and in scope, and points to new directions in the scholarship of judicial decisionmaking that should spotlight factors such as place or marital status of victims.
[Correction added on March 22, 2021, after first online publication: correct probabilities of the gender differences are added after original publication.]
2. Hypotheses
2.1 Gender of Victim
Previous research suggests that judges are more likely to grant a protection order requested by women than by men (Reference BasileBasile 2005; Reference Muller, Desmarais and HamelMuller, Desmarais, and Hamel 2009), a finding consistent with social psychological research on how people employ gender-specific evaluations of violent behavior (Reference FeatherFeather 1996; Reference GerberGerber 1991; Reference Willis, Hallinan and MelbyWillis et al. 1996). Assuming a control for the level of violence and threats referenced in cases, I hypothesize:
H1 Judges are more likely to issue temporary protection orders for female victims than for male victims.
2.2 Relationship Status
Beyond gender, there are variations in rates of victimization by relationship status. The frequency and severity of IPV is higher among cohabiting couples than their married counterparts (Brown & Reference Brown and BulandaBulanda 2008; Reference BrownridgeBrownridge 2008; Reference Caetano, Vaeth and Ramisetty-MiklerCaetano et al. 2008; Reference Kenney and McLanahanKenney & McLanahan 2006). Spousal abuse was both a socially and legally condoned practice in the US for many years (Reference AndersonAnderson 2002; Reference SiegelSiegel 1995; Reference SewellSewell 1989; Reference Straus, Gelles and SteinmetzStraus et al. 2017). Victims' relationship status, specifically whether victims are married or living together, may also affect civil protection order outcomes (Reference Shannon, Logan and ColeShannon et al. 2007). The following hypotheses on relationship status are making comparisons between victims whose relationship would adhere to the relationship criteria outlined in Nebraska Revised Statutes (§42-903):
Family or household members includes spouses or former spouses, children, persons who are presently residing together or who have resided together in the past, persons who have a child in common whether or not they have been married or have lived together at any time, other persons related by consanguinity or affinity, and persons who are presently involved in a dating relationship with each other or who have been involved in a dating relationship with each other.
Controlling for the level of violence and threats referenced in cases, do married victims fare better or worse than their unmarried counterparts? Studies of attitudes and responses to violence show a preference for keeping married couples together, even in cases of spousal abuse (Reference DroutDrout 1997; Reference Ross and GlissonRoss & Glisson 1991; Reference Shotland and StrawShotland & Straw 1976). Since judges may be less likely to grant orders to married victims because of a cultural ideology in which the institution of marriage takes precedence over victims' safety, I hypothesize:
H2 Judges are less likely to issue temporary protection orders for married victims than for unmarried victims who are cohabiting with their abusers.
A victim may also have children with the abuser. Previous scholarship has found that judges are concerned about addressing custody and visitation through protection orders (Reference Agnew-Brune, Moracco, Person and Michael BowlingAgnew-Brune et al. 2017), which could make them concerned with granting orders to victims who have children with the respondent. In addition, this reluctance may be linked to negative stereotypes that individuals, especially women, may file for protection orders in order to gain leverage or advantage in their divorce or custody cases (Reference Muller, Desmarais and HamelMuller et al. 2009). I hypothesize:
H3 Judges are less likely to issue temporary protection orders for victims who have children with their abusers.
2.3 Place
Comparisons of rural and urban areas have found differences in victim characteristics and outcomes, rates of IPV, and attitudes toward IPV (Reference EdwardsEdwards 2015). For instance, rural respondents have been found to blame IPV victims (Eastman & Reference Eastman and BunchBunch 2007), and rural legislators were less likely than their nonrural counterparts to support a statewide IPV policy (Hamm 1989). In contrast to urban communities, rural communities hold cultural norms of keeping family matters private (Billings & Reference Billings and BleeBlee 2000). I hypothesize:
H4 Judges in metropolitan counties are more likely to issue temporary protection orders than judges in nonmetropolitan counties.
2.4 Judges
Critical legal scholarship has found that judges' demographic characteristics affect outcomes in criminal cases (Boyd & Reference Boyd and NelsonNelson 2017; Reference Steffensmeier and BrittSteffensmeier & Britt 2001; Reference Steffensmeier and HebertSteffensmeier & Hebert 1999; Reference Welch, Combs and GruhlWelch et al. 1988). These findings are consistent with social psychological research showing that women are more likely than men to view domestic violence as unacceptable and that women are more sympathetic than men towards victims of domestic violence (Bryant & Reference Bryant and SpencerSpencer 2003; Reference Locke and RichmanLocke & Richman 1999; Reference West and WandreiWest & Wandrei 2002). In hypothetical scenarios involving rape, men are less harsh in their sentencing of male perpetrators (Basow & Reference Basow and MinieriMinieri 2011; Reference Hammond, Berry and RodriguezHammond, Berry, & Rodriguez 2011) and less likely to deliver severe sentencing to convicted husbands (Kern, Libkuman, & Reference Kern, Libkuman and TempleTemple 2007). I hypothesize:
H5 Female judges are more likely to issue temporary protection orders than male judges.
While research into the effects of a judge's race on civil protection orders is lacking, research examining the differences in laypeople's perceptions of IPV has found that white students are less accepting or have less tolerant attitude of violence toward intimate partners than ethnic or racial minorities (Reference CowanCowan 2000; Reference Lee, Pomeroy, Yoo and RheinboldtLett et al. 2005; Reference Locke and RichmanLocke & Richman 1999; Reference Simon, Anderson, Thompson, Crosby, Shelley and SacksSimon et al. 2001). However, experimental research also suggests that black or African American respondents more accurately perceive others' emotions compared to white respondents (Reference Gitter, Black and MostofskyGitter et al. 1972), and that people with low-status positions (e.g., female judges or racial minority judges) are better able to understand the perspective of others as accurate role-takers than individuals in high status positions (Reference Love and DavisLove & Davis, 2014). I hypothesize:
H6 Racial minority judges are more likely to issue temporary protection orders than white judges.
3. Data and Methods
To examine these hypotheses, I analyze the accounts of victims as provided in both petitions and judges' decisions sampled from cases filed in all counties in Nebraska in 2015. Victims record these accounts when they initiate a petition. The court form prompts victims to provide facts of the most recent incidents of domestic abuse towards themselves or their children. Victims are instructed to “please write a brief description of each incident including shoves, kicks or blows inflicted, weapons used, threats made, injuries sustained, medical or hospital treatment, if any.” Over 8,000 pages of court documents were manually coded to create variables on the characteristics of victims, their abusers, the judges who heard their cases, and the case outcomes. Prior to data collection, the university's institutional review board approved the study protocol.
3.1 Sample
A sample of 1,426 domestic abuse civil protection order cases was retrieved from Nebraska's online JUSTICE registry. Cases were restricted to protection order domestic abuse filings from 2015 and randomly selected from each county in the state. If rural counties had fewer than 30 cases filed in 2015, cases were sampled from filings available. Cases that did not provide the court's decision were excluded from the sample and were replaced with another case extracted from the county. After an initial sample of 30 cases from 80 counties, cases were then selected from each county until a sample of over 1,400 cases was collected. To address the potential limitation of not using judges in the selection criteria, the models used in this study include random effects for judges.
The sample was then restricted to IPV for heterosexual relationships. This was accomplished by dropping cases involving relatives, “persons related by consanguinity or affinity,” which is an additional relationship outlined in Nebraska Revised Statutes (§42-903). Other cases in which the relationship would not pass Nebraska's specific “qualifying relationship” for a domestic violence protection order, such as a neighbor, filing against a neighbor were also dropped. The final sample consisted of 1,044 civil protection order cases, including petitions and initial decisions.
3.2 Analytic Approach
Using Stata 13.0, I ran a mixed-effects logistic regression model that accounts for individual idiosyncrasies of judges through random effects for their gender and political affiliation. A second mixed-effect logistic regression model was used to examine effects of the petitioner (victim) having children with their respondent (abuser) by using a binary variable of “shared child” in lieu of the “relationship” variable. However, other models were tested and compared for optimal fit.Footnote 3 The independent variables in the models represent characteristics of the victims, judges, petition content, and the county in which the case was filed.
For ease of interpretation, estimates from these mixed-effects logistic regressions are reported as average marginal effects of each predictor such as abuse factors or whether victims filed in a metropolitan county, holding any other predictors in the model at their actual value for each observation. Average marginal effects help to give a sense of the magnitude of effects as compared to odds ratios and are easier to interpret than logistic regression coefficients (Reference WilliamsWilliams 2012). Average marginal effects of abuse variables factors indicate the average change in probability of the receiving a temporary protection order with a one-unit change in the independent variable.
I also used the Stata command margins to calculate predictive margins (Long & Reference Long and FreeseFreese 2006). This allowed me to specify the values of independent variables in the model, such as being a female victim who described experiencing physical abuse in the petition, and then compute the predicted probability of receiving a temporary protection order for an individual who has those values. Because a majority of the sample consists of cases heard by white male judges, I set the values accordingly, and report the predicted probabilities for cases heard by white male judges. I then tested whether differences in the probability of receiving a temporary protection order between groups were statistically significant, employing the mlincom command in the spost13 package in Stata (Long & Reference Long and FreeseFreese 2006).
3.3 Dependent Variables
The case outcome of interest, whether the victim received a temporary protection order through either an ex parte order or a show cause order, was coded as a binary variable (“1” for yes, “0” for no). Because the questions of interest concern judicial decisionmaking, this study focuses only on this initial decision by a judge to deny or proceed. Outcomes of later stages are not solely attributable to judges, as victims may request the dismissal of their initial petitions or fail to attend their final hearings (Harrell & Reference Harrell, Smith, Buzawa and BuzawaSmith 1996; Reference Zoellner, Feeny, Alvarez, Watlington, O'Neill, Zager and FoaZoellner et al. 2000).
3.4 Independent Variables
3.4.1 Petition Content
Several key variables created using petition content reflect variables that correspond to the legal statute that allows judges to take action if a victim “will be in immediate danger.” To create variables that correspond to this statute, I coded for the types of abuses referenced in victims' petitions. Based on the Nebraska statute (Nebraska Revised Statute §42-903), defining domestic violence as “Attempting to cause or intentionally and knowingly causing bodily injury with or without a dangerous instrument; placing, by physical menace, another person in fear of imminent bodily injury; or engaging in sexual contact or sexual penetration without consent,” I created variables for physical abuse, verbal threats, and sexual abuse.
Since judicial officers may also take into account nonphysical acts of abuse when assessing the need for protection, a number of other legally relevant variables were constructed. For instance, I coded for stalking from victims' own narratives. Stalking is defined as to “detain, stalk, or harass them or impose any restraint on their personal liberty and which will not prohibit constitutionally protected activities” (Nebraska Revised Statute §28-311.02). However, stalking is covered under a Nebraska Harassment order and not a Domestic Abuse order. Stalking was coded for in this study because victims stalked by a former intimate partner often experience coercive control and report more severe partner violence victimization (Reference BrewsterBrewster 2003; Reference Logan, Shannon and WalkerLogan et al. 2007b). I created other abuse variables, including variables for financial abuse, destruction of property, emotional abuse, harassment (frequent calls, texts, or emails), abuse of pets, and the threat of revenge porn (the posting of explicit photos of the victim on social media or sharing them by other means).
Four variables captured the situational context of the abuse, such as the abuse of children, children witnessing the abuse, and references to the use of alcohol or drugs in connection to abuse. While these variables cannot fully encapsulate every legal factor, such as case law, I sought to examine whether extraneous variables, such as the county in which victims file, impacts the issuance of temporary protection orders after controlling for the legally relevant factors provided through victims' petitions.
Table 2 provides descriptions of the coding criteria. The variables for petition content began as count variables. For example, I constructed a variable for the number of physical abuse incidents described in victims' petitions. From these variables, I then created binary variables, so that cases that included at least one reference to physical abuse were assigned a value of “1” and cases that included no incidents of physical abuse were assigned a value of “0.” Two binary variables were also created to assess the severity of physical injury. “Minor injury” includes references to bruises or scratches; “severe injury” includes broken bones, being burned on purpose, and choking. Because more recent incidents of abuse would likely affect whether judges decided that “the petitioner will be in immediate danger of abuse” (Nebraska Revised Statute §42-925), a binary variable was created to track whether the incidents described in the petition had occurred in the three months preceding the date the victim filed the petition (“1” at least one incident of abuse in the last three months, “0” for no recent incidents of abuse). I also created a control variable for the number of incidents described in victims' petitions: one, two, three, or more than three incidents.
To investigate the effects of sociodemographic factors such as the victim's gender, I control for recency of abuse, the number of incidents described in the victim's petition, and the parties' relationship status. A relationship variable was constructed using the information provided on the court form where they checked boxes: (1) spouse; (2) former spouse; (3) children; (4) a person I am currently living with; (5) a person I have lived with in the past; (6) the father/mother of one or more of my children; (7) a person I am presently involved with in a dating relationship; (8) a person I was involved with in a dating relationship; (9) other. From this information I created a categorical relationship variable with married as the reference category, and the remaining statuses as former spouse, dating, previously dated, living together, and previously lived together. To track whether the petitioner and respondent had a child together, I examined information from the relationship boxes on the court form and another section of the form that provided a space for the petitioner (victim) to record the names of any minor children they had with the respondent (abuser). Utilizing information provided from the entire form, a “shared child” binary variable was created across relationships to indicate when the petitioner and respondent had children together.
3.4.2 Sociodemographic Factors
Ideally, variables such as a victim's gender or the county in which they file should have no bearing on legal decisions since they do not correspond to provisions of legal statutes governing civil protection orders. Specifically, I examine the role of sociodemographic characteristics of judges and victims, relationship status,Footnote 4 and place in judicial decisions for civil protection orders. Petitions were coded for victims' gender (coded 1 = male, 0 = female). However, petitions do not contain information on the victim's race or socioeconomic class.
Petitions do specify the type of relationship between the victim and abuser. I used the margins command in Stata to create comparisons if the victim was married to their abuser and if the victim was unmarried but cohabited with their abuser. This variable allowed for comparisons between married petitioners and their unmarried counterparts who reside with the respondent while controlling for legally relevant factors, such as abuse. As outlined above, I used another binary variable of whether the victim had minor children with the abuser.
I also created variables for each judge's gender (male or female), race (white or nonwhite), and political affiliation. The race variable was constructed using information from judges' profiles on the official justice department Web site. Judges' racial classifications were estimated using their profile image. Given that the sample of judges was predominantly white, I combined judges classified as black or Hispanic/Latinx into a single nonwhite category (0 = nonwhite, 1 = white).
3.4.3 County
I also created a variable for the county in which the case was filed and classified whether the case was filed in a metropolitan or nonmetropolitan county. These classifications were based upon metropolitan and nonmetropolitan definitions released by the federal Office of Management and Budget (0 = nonmetropolitan, 1 = metropolitan) (Drozd 2013).
3.5 Control Variables
A control variable was created to account for the population of each county. As outlined below, I used mixed-effects logistic regression models which allowed me to add an idiosyncratic error term to the model for each judge.
4. Results
4.1 Descriptive Statistics of Cases
Table 1 presents the descriptive statistics of civil protection order cases. Most victims in the sample were female (85%), and over 54% of the victims were currently dating, married to, or living with the respondent at the time they filed their petitions for civil protection orders. These findings are similar to those in other studies (Harrell & Reference Harrell, Smith, Buzawa and BuzawaSmith 1996; Reference Harrell, Smith and NewmarkHarrell et al. 1993; Reference Lucken, Rosky and WatkinsLucken et al. 2015).
Table 1. Descriptive Statistics of Cases (N = 1,044)
Over half the sample (51%) of cases involved filings against respondents with whom the victim had children. This also aligns with previous studies of civil protection orders (Reference Gondolf, McWilliams, Hart and StuehlingGondolf et al. 1994; Reference Logan, Shannon, Walker and FaragherLogan et al. 2006; Reference Logan, Shannon and Cole2007a; Reference Wolf, Holt, Kernic and RivaraWolf et al. 2000). The majority (65.4%) of victims received ex parte orders, 14.4% of victims received show cause orders, and 20.2% of cases in the statewide sample had their petitions for protection orders denied.
The sample included orders by 111 judges, who on average heard ten cases each. Judges were predominantly male (76%) and white (97%). Approximately 64% of cases were filed in nonmetropolitan counties.
Table 2 reports rates for specific acts of abuse described in petitions. Over 59% of cases reported incidents of physical abuse, such as shoving, kicking, or pushing, and approximately 16% of cases referenced severe injury, such as physical violence that led to broken bones or strangulation. Approximately 40% of cases referenced minor children witnessing incidents of abuse. Over 43% of cases described at least three incidents of abuse, and 12% of cases described four or more incidents of abuse. On average, petitions included two incidents of abuse.
Table 2. Petition Content (N = 1,044)
Table 3 reports the results of a t-test of means comparing abuse rates for male and female victims. Several trends stand out. First, female victims were more likely than male victims to report severe injury (15% vs. 5%, p < 0.001) and to report minor injury such as bruises (18% vs. 10%, p < 0.001). Second, female petitioners reported significantly higher rates of sexual abuse, threats, and stalking. However, there were no significant differences in male and female petitioners being married to the respondent or having children with the respondent.
Table 3. Differences of Means of Abuse and Relationship by Male and Female Victim
Note: N = 1044 T-tests, standard errors are in parentheses.
*p < 0.05; **p < 0.01; ***p < 0.001.
To further examine the role of the gender of victim, abuse variables and relationship between victim and abuser were included in the analyses. Table 4 displays the average marginal effects from a mixed-effect logistic regression model. This approach allowed me to examine differences in civil protection order case outcomes by victim demographic factors, relationship status, judge demographics, the county in which victims file, and a range of legal factors. Specifically, the model shows the effects of factors such as a victim's gender, filing in a metropolitan county compared to a nonmetropolitan, and specific types of abuse.
Table 4. Average Marginal Effects from Mixed-Effects Logistic Regression Predicting Being Granted Temporary Protection Order
Note: N = 1044. Coefficient estimates are presented as average marginal effects calculated while holding any other predictors at their observed values; standard errors are in parentheses.
*p < 0.05; **p < 0.01; ***p < 0.001.
a Relationship reference category is married. Controls for county population and random effects for judges were included in the mixed-effects logistic regression model.
4.2 Abuse
Table 4 shows that cases which referenced physical abuse, severe injury, sexual abuse, or threats were significantly more likely to receive temporary protection orders. These are all factors that reflect the legal statute for protection orders. Cases with at least one recent incident of abuse (within three months) in the petition were significantly more likely to be granted than petitions that reported older incidents of abuse.
Judges were significantly more likely to issue temporary orders for petitions that include reference to severe injury, physical abuse, sexual abuse, threats, or recent incidents of abuse. On average, cases that referenced physical abuse increased the probability of receiving a temporary protection order by 21% (p < 0.001) and cases that included at least one incident of sexual abuse increased the probability of receiving a temporary protection order by 9% (p < 0.01).
Similarly, cases that referenced threats or a recent incident of abuse had a significantly higher predicted probability of receiving a temporary protection order than those that did not reference threats, sexual abuse, or recent abuse. On average, victims who mentioned experiencing threats in their petitions had an increased probability of being granted a temporary protection order of 9% (p < 0.001). Similarly, on average, cases that reference severe injury such as broking bones or choking increase the probability of receiving a temporary order by 10% (p < 0.01). Lastly, on average, victim narratives with at least one incident that had occurred in the past three months had an increased probability of receiving a temporary protection order by 16% (p < 0.01).
One key takeaway is that physical abuse was the most important factor of abuse for judicial decisionmaking. Physical abuse had the largest impact on case outcomes, with both male and female victims being significantly more likely to receive temporary orders when their petition referenced physical abuse (p < 0.001). Additional analyses were run using the count variable for physical abuse. For female petitioners, having one incident of physical abuse compared to no references to physical abuse had a 17% higher probability of receiving a temporary order (p < 0.001) and female petitioners that referenced two incidents of physical abuse compared to one incident of physical abuse had a 5% higher probability of receiving a temporary order (p < 0.05).
4.3 Gender of Victim
Table 4 shows that male victims are significantly less likely to receive a temporary protection order, controlling for abuse and relationship status (p < 0.001). On average, female petitioners had a 13% higher predicted probability of receiving a temporary protection order than male petitioners (p < 0.001).
Table 5 displays the predicted probability of receiving a temporary protection order for male and female petitioners when values of independent variables such as physical abuse are specified in the models. Table 5 shows that across all categories, male victims have a significantly lower predicted probability of receiving temporary protection orders. Specifically, female victims who filed in a nonmetropolitan county and referenced physical abuse were 12% more likely to receive a temporary order than their male counterparts who do the same (p < 0.01). A similar gender pattern existed across other abuse categories as well (e.g., severe injury, sexual abuse, and threats). Male victims had a significantly lower predicted probability of a receiving temporary order than female victims, even when both cases included severe injury, sexual abuse, or threats. This shows that whether the petition included sexual abuse, severe injury, or threats, gender of the victim was a significant determinant for whether a temporary order was issued. Therefore, I find support for H1 that judges are more likely to issue temporary orders for female victims than for male victims.
Table 5. Predicted Probability of Receiving a Temporary Protection Order by Victim Gender and County (Mixed-effects logistic regression)
Note: N = 1044; standard errors in parentheses. Predicted probabilities calculated for cases heard by White male judges.
*p < 0.05; **p < 0.01; ***p < 0.001.
a Predicted probability of receiving a temporary order by shared children was calculated using the mixed effects logistic regression model in which the shared children variable was used in place of the relationship variable.
4.5 Relationship Status of Victim
The second hypothesis stated that judges are more likely to issue temporary orders for unmarried victims who cohabit with their abusers than married victims. Table 4 shows that unmarried petitioners (victims) living with the respondents (abuser) fare significantly better than their married counterparts, controlling for types of abuse and recency of incidents. On average, unmarried petitioners who cohabit with the respondent had a 15% increased predicted probability of receiving a temporary protection order compared to married victims (p < 0.001). Table 5 shows that unmarried female victims in nonmetropolitan counties had a 17% increased predicted probability of receiving a temporary order compared to married female victims in nonmetropolitan counties (p < 0.001). I therefore find support for H2, with judges being less likely to grant orders to married individuals than to victims cohabiting with their abusers. One possible explanation for this pattern is that many judges seek to preserve existing marriages and are therefore more reluctant to grant protection orders for married victims than their unmarried counterparts, even after accounting for the recency and severity of abuse.
4.6 Shared Minor Children
To examine effects of the petitioner (victim) having children with their respondent (abuser) I calculated the average marginal effects from the second mixed-effect logistic regression model with a binary variable of “shared child” in lieu of the “relationship” variable. This binary variable of “shared child” was created across relationships to indicate when the petitioner and respondent had children together. I predicted the average probabilities of receiving temporary orders by whether the victim shared a child with the respondent (e.g., no shared child = 0, shared child = 1), specifying that the physical abuse was referenced in the petition for cases heard by white male judges. Specifying the values of key predictors such as physical abuse helps to further highlight the effect of “shared parent” on the probability of receiving an order. I find that both female and male victims who had a minor child with the respondent were significantly less likely to receive a temporary protection order (p < 0.01). Therefore, I find support for H3. Judges are less likely to issue protection orders for victims who have children with the respondent, even when they include references to physical abuse in their petitions. This finding is particularly meaningful given that many cases referenced children in the household being abused or witnessing the abuse of others.
Place and Gender
Despite the fact that each judge in each county relies on the same statewide legal statutes, I found significant effects of place on case outcomes, even after controlling for a number of legal factors. Table 4 shows that victims who file their petitions in metropolitan counties were significantly more likely to be granted temporary protection orders compared to those in nonmetropolitan ones. On average, victims who filed in a metropolitan county have 9% increased probability of receiving a temporary protection order than victims who filed in a nonmetropolitan county (p < 0.05).
To better examine whether the relationship case outcomes differ across metropolitan and nonmetropolitan counties, I tested the differences in the probability of receiving temporary orders by the gender of victims across county types. Table 5 shows that on average, a female victim who referenced physical abuse and filed in a metropolitan county had a 6% higher probability of receiving a temporary order than a similar female victim who had filed in a nonmetropolitan county (p < 0.05). Similarly, on average, a male victim who filed in a metropolitan county and referenced physical abuse in their petitions had a 11% higher probability of receiving a temporary order compared to his nonmetropolitan counterpart (p < 0.05). Married victims, unmarried cohabiting victims, and victims with shared children in nonmetropolitan counties were significantly less likely to receive temporary protection orders compared to victims who filed in metropolitan counties (p < 0.05). In other words, both male and female victims who file in metropolitan counties are more likely to receive temporary protection orders than victims in more rural areas, even controlling for the abuses described in their petition and for their relationship to their abuser. These results provide support for H4. Judges in metropolitan counties are more likely to issue protection orders than judges in nonmetropolitan counties, especially for male petitioners.
I did not initially hypothesize that place would have effects on gender-specific probabilities of being granted temporary protection orders. However, in the course of the analysis, I found that the overall gender difference was largest in nonmetropolitan areas. Table 5 shows that in metropolitan counties, female victims had a 11% higher probability of a protection order than males (p < 0.01). Female victims in nonmetropolitan counties had a 15% higher probability of receiving a temporary order than male victims in nonmetropolitan counties (p < 0.001).
4.7 Judge Characteristics
One potential explanation for the impact of place is that judges in nonmetropolitan counties differ significantly from judges working in metropolitan counties. I found a significant difference between the kind of judges working in nonmetropolitan and metropolitan counties. Using t-tests, I found that male judges were significantly more likely to work in nonmetropolitan counties (79% vs 71%, p < 0.001), and black or Hispanic/Latinx judges did not hear cases in nonmetropolitan counties. In addition, nonmetropolitan counties were more likely to have cases heard by Republican judges compared to metropolitan counties (77% vs. 64%, p < 0.001). However, since neither the gender nor race of judges had a statistically significant relationship with their decisions in the mixed-effects logistic models, H5 and H6 were not supported. Additional analyses were run with interaction terms between political affiliation and judges' demographic characteristics, but these coefficients were only marginally significant. This may reflect a limitation of the sample, in which judges were predominantly white, male, and Republican and this issue is discussed in the following section.
5. Limitations
This study has several important limitations. First, because the data consist of civil protection order cases from Nebraska, generalizability is limited. Many legal statutes regarding protection orders vary by state, and cultural factors affecting judges' opinions may vary as well.
By analyzing court documents, this study captures a specific sample of victims who seek formal legal intervention by applying for civil protection orders. Such victims may not be representative of all victims. For example, victims who file petitions for civil protections may be more likely to experience chronic or severe violence, which in turn could bias the data collected through court filings to overestimate the effect of physical abuse. Previous research has found that abused women who sought legal services were more likely to be employed full-time and were more highly educated than women who did not seek these services (Reference Duterte, Bonomi, Kernic, Schiff, Thompson and RivaraDuterte et al. 2008).
This study was also restricted to heterosexual intimate partner relationships. Cases with same-sex relationships or filings between family members such as brothers or parents were excluded from the analyses due to sampling. Future scholarship should examine judicial decisionmaking in civil protection order cases for same-sex relationships, as more researchers have come to examine perceptions of IPV in same-sex relationships (Reference Baker, Buick, Kim, Moniz and NavaBaker et al. 2013; Reference Poorman, Seelau and SeelauPoorman et al. 2003; Sorenson & Thomas 2009).
I was also unable to determine whether petitioners and respondents in the sample had previous petitions for civil protection orders against each other or pending divorce cases. In order to establish the effect of cross-filings (e.g., mutual restraining orders), previously dismissed protection order cases, or other filings such as custody or divorces cases on outcomes, a systematic check would need to be conducted on each petitioner and respondent in the Nebraska's online JUSTICE registry. Whether the parties had other court filings or whether the parties previously appeared before the judge may affect judicial decisionmaking. However, an experimental design would be better suited to examine this effect since it would require tracking judges' recall and knowledge of prior cases. While these factors likely shape judicial decisionmaking and are an important area of research, they are beyond the scope of this study.
Another limitation was that the judges studied in Nebraska were predominantly white and male. Future research should include a greater number of judges of other races and genders. In addition, estimating judges' racial classifications using profile images is another concern. Ideally, judges demographic characteristics would be self-reported and a larger sample size would be necessary to fully explore differences in judicial decisionmaking by judge's gender, race, and political affiliation. Another limitation is that coding victims' petitions for types of abuse and threats cannot fully account for judges' personal experiences or additional training on IPV they might have received. While Nebraska outlines formal rational statutes for protection orders, judicial interpretation of the state statutes are likely influenced by social factors such judges' understanding or misconceptions of the dynamics of IPV (Stark & Reference Stark and ChoplinChoplin 2017).
There were also a number of other variables that were unavailable in my data, such as the race, ethnicity, social class, age of the victim, or whether the victim had legal assistance when filing their petition. Since the same petition form was used across the state, no information was provided about whether the victim had assistance from an advocate or attorney when filing. While later court proceedings did indicate whether an attorney was present at the hearing, no such information was provided about the initial decision. This limitation is important given that Reference DurfeeDurfee (2009) highlighted advantages of legal representation, with narratives written by lawyers being more likely to include specific acts of violence with dates compared to narratives written by petitioners without assistance. But despite this limitation, this study examines predictors of for temporary protection orders in order to expand understanding of judicial decisionmaking.
6. Discussion
Overall, my findings support what critical legal scholars have long claimed: extralegal factors influence judicial decisionmaking. This is not to say that legal factors such as references to severe injury or sexual abuse do not also affect judicial outcomes. Both legal factors and extralegal factors, such as the county victims file in, play a role. I have shown how deeper consideration of factors beyond criteria outlined in statutes can expand our understanding of judicial decisionmaking in practice. Three key finding emerged that highlight how victim gender, shared children, and place influence whether victims receive temporary protection orders.
Physical abuse was a key predictor of temporary orders being granted, yet comparisons between cases that included physical abuse showed significant gender- and place-related effects for whether victims receive temporary protection orders. I found that male victims were significantly less likely to receive temporary orders than female victims, even after controlling for the recency or severity of injury. This significant difference suggests that a victim's gender can impact her or his access to justice, with male victims potentially being disregarded.
One potential explanation is that judges have access to additional information, such as whether the parties have any existing filings and believe that men in this sample are filing vexatious or unfounded petitions. For instance, judges may be aware of or have received training on abusers' coercive control. Reference StarkStark (2007) highlights how abusers control or compel obedience through instances of intimidation, isolation, depriving victims of their financial independence or material possessions and regulating their behavior. Judges may be identifying another aspect of coercive control, “paper abuse,” where batterers file frivolous petitions and seek protection order hearings as a way to continue victimization (Miller & Reference Miller and SmolterSmolter 2011). In this sample, judges may have been less likely to grant temporary orders to certain victims if they believed the petitions were filed as a means of manipulation and extension of coercive control.
But if this gender effect is explained by judges' awareness of coercive control we would expect other significant risk indicators to also affect judicial decisionmaking. But this is not the case. For instance, stalking was not significantly associated with temporary protection order outcomes. While stalking is not outlined in the domestic violence protection statutes, previous scholarship has shown that female victims who have experienced stalking also experience other forms of violence and have post traumatic stress disorder and anxiety symptoms, even after controlling for severe threats, severe physical violence, and sexual violence (Reference BrewsterBrewster 2002; Reference Logan, Shannon and WalkerLogan et al. 2007b; Reference Mechanic, Uhlmansiek, Weaver and ResickMechanic et al. 2000). While judges may be less likely to grant orders to men because they believe male petitioners are using court proceedings to further victimize their current or previous partner, this is not the only possible explanation.
Judicial decisions are not made in a vacuum but reflect a complex medley of interactions and heuristic shortcuts. A bounded rational model would view judges as burdened with a large caseload of imperfect victim narratives or insufficient case details and have finite amount of time and resources to make a court decision. The content, structure, and form of victims' narratives vary greatly and their petitions often provide the only evidence judges use to make a quick determination whether an ex parte order should be granted (Reference Durfee, Garcia and CliffordDurfee 2010). In response, judges use heuristics or cognitive shortcuts (Reference SimonSimon 1957) that allow individuals to efficiently process information and make a quick decision (Reference Guthrie, Rachlinski and WistrichGuthrie et al. 2000; Reference Agnew-Brune, Moracco, Person and Michael BowlingKahneman et al. 1982). This leads judges to routinize their legal decision-making process by simplifying their conceptions of victims. Other scholarship has examined how state statutes and previous judgments guide judicial decisionmaking and the use of heuristics as quick cognitive shortcuts also influences this process (Agnew-Brune et al. 2017; Reference Guthrie and GeorgeGuthrie & George 2005; Reference Guthrie, Rachlinski and WistrichGuthrie, Rachlinski, and Wistrich 2000). Consistent with broader gender cultural stereotypes (Reference GerberGerber 1991; Reference Willis, Hallinan and MelbyWillis et al. 1996), and common framing of IPV issues in news media (Reference Carlyle, Scarduzio and SlaterCarlyle et al. 2014), judges may rely on a cognitive gender bias with men presumed to be the perpetrators and not victims. In this case, the use of heuristics may create a bias against heterosexual male victims who do not conform to judges' preexisting conceptions of victims and are less likely to be granted temporary orders.
My findings suggest that judges may face cultural pressures in cases when children are involved or the parties are married. Judges were significantly less to grant orders to married individuals than to victims cohabiting with their abusers. One possible explanation is that many judges seek to preserve existing marriages and are therefore more reluctant to grant protection orders for married victims than their unmarried counterparts.
Similarly, judges were significantly less likely to grant temporary protection orders in cases in which the victim and the abuser had minor children together. Given that children's experiences with domestic abuse affect their mental health (Reference Bogat, DeJonghe, Levendosky, Davidson and von EyeBogat et al. 2006; Reference Kitzmann, Gaylord, Holt and KennyKitzmann et al. 2003; Reference Meltzer, Doos, Vostanis, Ford and GoodmanMeltzer et al. 2009), their physical health (Bair-Merritt, Blackstone, & Reference Bair-Merritt, Blackstone and FeudtnerFeudtner 2006), and their own future intimate relationships (Black, Sussman, & Reference Black, Sussman and UngerUnger 2010; Reference Cui, Ueno, Gordon and FinchamCui et al. 2013), these disparities in case outcomes may translate into disparities in life outcomes for children.
Judges may believe that issues of protection can be addressed in divorce or custody proceedings and fear that false complaints in domestic abuse cases were filed to obtain sole custody of a child and bypass family court (Reference Agnew-Brune, Moracco, Person and Michael BowlingAgnew-Brune et al. 2017). Judges may be less likely to grant orders to married victims, or victims with children, because of traditional cultural beliefs about preserving marriage or families. Judges in rural counties may be especially likely to hold such views because rural communities are often more socially and politically conservative and are more likely to hold adhere to traditional views of gender roles (Bush & Reference Bush, Lash, Abramson and HaskellLash 2006; Reference PruittPruitt 2008).
Expanding the geographical range of the study to include both nonmetropolitan and metropolitan areas also revealed the importance of place to judicial decisionmaking. Results from this study show that both male and female victims who file in nonmetropolitan counties are significantly less likely to receive orders than their metropolitan counterparts. Understanding rates of protection order issuance by place could help agencies serving victims to target areas where legal representation might be most beneficial. Future research should build on the findings in this study not only to expand our understanding of judicial decisionmaking, but also highlight sociospatial and socioeconomic characteristics that influence this process to address barriers to justice for victims.
Judges in both metropolitan and nonmetropolitan counties may also hold the common belief that individuals manipulate the justice system by filing for protection orders to seek revenge, to retaliate in a dispute, or to get the upper hand in custody or divorce proceedings (Reference Logan, Shannon and WalkerLogan et al. 2005b; Reference Muller, Desmarais and HamelMuller et al. 2009). Given the consequences for victims and families, each of these possibilities merits further investigation. These findings carry implications for future research on victims' access to justice and IPV-related services. Training could educate judges and other legal actors on common biases within the justice system, and on the negative effects of children witnessing incidents of abuse.
Extralegal factors, such as the gender of the victim, or whether the parties shared minor children, become stronger in nonmetropolitan counties. For example, the “parent penalty” for victims who had minor children with the respondent is larger in nonmetropolitan counties. Similarly, while male victims were less likely overall to receive temporary orders compared to female victims, this was especially true in nonmetropolitan counties.
These findings about place expand the scope of future research into judicial decisionmaking, and point to new directions for research on access to justice. Why do victims in nonmetropolitan counties, and especially male victims, fare worse than their metropolitan counterparts? As referenced above, one possible explanation involves local county culture with judges in rural communities holding more socially and politically conservative attitudes. Judges in nonmetropolitan counties in this sample may hold traditional gender stereotypes that lead to preferential treatment of women's civil protection order requests. Or rather, judges in metropolitan counties may have a local county culture that is more educated about issues surrounding IPV. Future scholarship should examine how resources, as well as training or attitudes, impact judicial decisionmaking.
7. Conclusion
A civil protection order is a primary legal recourse for preventing further abuse, but its implementation depends on judicial decisionmaking. Given the life-or-death consequences for victims of interpersonal violence, it is important to understand which factors affect judicial decisionmaking for these cases. In this study, both legal and extralegal factors played a role in predicting civil protection order case outcomes. I have found that it is not merely the content of petitions that matters. Victims' legal outcomes are also shaped by where they live and who was victimized. Male victims in nonmetropolitan counties were least likely to receive a temporary order of protection. Judges in this study appear biased toward preserving existing marriage and family statuses, with orders being less likely to be granted in nonmetropolitan counties, despite the serious risks to victims.
Because demographic and geographic factors are numerous and complicated, with effects that are difficult to predict in advance, they present a “gray area” of law and its inequities. Addressing these issues requires large-scale empirical research that identifies which factors affect victims' access to justice. This study begins this process by showing that there are statistically significant differences in the granting of protection orders by victim's gender, marital status, shared minor children, and the county in which they file.
