Mental Illness and Criminal Justice

Criminal Justice Policy Review 2016, Vol. 27(1) 46 –75

© 2014 SAGE Publications Reprints and permissions:

sagepub.com/journalsPermissions.nav DOI: 10.1177/0887403414554997

cjp.sagepub.com

Article

A Criminal Justice System–Wide Response to Mental Illness: Evaluating the Effectiveness of the Memphis Crisis Intervention Team Training Curriculum Among Law Enforcement and Correctional Officers

Megan L. Davidson1

Abstract The present study sought to evaluate the effectiveness of the Memphis Crisis Intervention Team (CIT) training curriculum among law enforcement and correctional officers. A panel research design was used in which a sample of 179 law enforcement officers and 100 correctional officers in nine Florida counties were surveyed on the first day of training (pretest), the last day of training (posttest), and 1 month following their completion of CIT training (follow-up). These surveys measured the extent to which CIT training achieved several officer-level objectives, including increased knowledge of mental illness, improved self-efficacy when responding to mental health crises, and enhanced perceptions of verbal de-escalation skills. The results of these surveys revealed that the training effectively achieved these objectives in the short term, but a measurable decay was found in the follow-up time frame with regard to self-efficacy and perceptions of verbal de-escalation.

Keywords program evaluation, mental illness, crisis intervention introduction

1East Carolina University, Greenville, NC, USA

Corresponding Author: Megan L. Davidson, Assistant Professor, Department of Criminal Justice, College of Human Ecology, East Carolina University, East Fifth Street, Greenville, NC 27858-4353, USA. Email: [email protected]

554997CJPXXX10.1177/0887403414554997Criminal Justice Policy ReviewDavidson research-article2014

Davidson 47

Mental illness is a public policy concern that pervades every facet of the public sector in communities around the United States. Individuals with a mental illness often cycle through various systems of care throughout their lifetime. The responsibilities of inter- vening, managing, and treating this population are shared by numerous mechanisms of formal social control, including the criminal justice system. Understanding how the criminal justice system responds to persons with a mental illness is paramount to developing the most effective and appropriate intervention strategies. The purpose of the present study is to evaluate the effectiveness of a formal criminal justice response to mental illness across nine counties in Florida. The Memphis Crisis Intervention Team (CIT) model has been widely adopted across the country and around the world. Therefore, gaining an understanding of the true effectiveness of this model contributes to the literature surrounding this widespread criminal justice program.

While the criminal justice system was originally developed solely to enforce the law and punish wrongdoers, the responsibilities of this system have expanded over time. Due to dramatic policy changes in the mental health field over the last few decades, agents of the criminal justice system are now frequently faced with the chal- lenge of intervening and managing situations involving persons with a mental illness. Specifically, deinstitutionalization, the tightening of civil commitment laws, and inad- equate community treatment alternatives have collectively increased the visibility of persons with a mental illness on urban streets around the country since the 1960s and 1970s (Erickson & Erickson, 2008; Teplin, 1984). This trend has not been accompa- nied by a greater tolerance of this population by the general public. The dangerousness stigma attached to the erratic and disruptive behavior that is often associated with mental illness frequently compels citizens to invoke the criminal justice system to manage incidents involving this population that occur in the community (Angermeyer, Cooper, & Link, 1998; Erickson & Erickson, 2008; Horwitz, 2002; Phelan & Link, 1998; Schnittker, 2000; Teplin, 1984).

According to the Council of State Governments (2002) in the Criminal Justice Mental Health Consensus Project, law enforcement officers typically encounter per- sons with a serious mental illness in one of the following scenarios: (a) as a victim of a crime, (b) as a witness to a crime, (c) as the subject of a call for assistance, (d) as a suspected offender, and (e) as a danger to themselves or others. An estimated 7% to 10% of all police contacts involve a person with a mental illness (Borum, Deanne, Steadman, & Morissey, 1998; Wells & Schafer, 2006). Results from a survey of law enforcement officers from three different agencies indicated that approximately 92% reported having responded to at least one mental health crisis in the month prior to the survey, with 84% reportedly responding to more than one of these incidents during the same time frame (Borum et al., 1998). Likewise, people with a mental illness often report coming into contact with law enforcement, with many of them having been arrested at least once (Borum, 2000).

Many communities lack dispositional alternatives to arrest for persons with a men- tal illness engaging in maladaptive, criminal behavior. While this behavior may be a manifestation of the individual’s mental illness, law enforcement officers must fre- quently resort to arrest when responding to these calls for service. As a result, the

48 Criminal Justice Policy Review 27(1)

burden of caring for and managing this population has been largely transferred to cor- rectional facilities. When estimating the prevalence of mental illness found among the currently incarcerated population, prior studies have found rates ranging from as low as 14% to as high as 64% (Ditton, 1999; James & Glaze, 2006; Wilper et al., 2009). The estimates vary dramatically based on the inmates sampled, measures utilized, and methodologies used, making it difficult to obtain a precise prevalence figure. With regard to serious mental illness, Aderibigbe (1997) asserts that the rate of schizophre- nia and bipolar disorder among jail and prison inmates is nearly three times greater than the general population.

In Florida, the statistics follow the nationwide trend for the rates of arrest and incar- ceration found among persons with a mental illness. According to a report presented by the Florida Supreme Court (2007), approximately 125,000 individuals booked into Florida jails each year have a diagnosed mental illness. This report also states that roughly 23% of county jail inmates and 17% of prison inmates in the State of Florida have a serious mental illness. This translates to approximately 16,000 prison inmates and 15,000 local jail inmates with a serious mental illness on any given day in Florida (Florida Supreme Court, 2007).

These figures enumerate the breadth of mental illness as a social problem that per- vades the criminal justice system throughout the country and in Florida specifically. The disproportionate rates of arrest and incarceration found among persons with a mental illness have compelled members of the criminal justice system to collaborate with mental health advocates and practitioners to develop diversionary strategies to mitigate this issue. The focus of the present study is to conduct a program evaluation of the Memphis CIT model, a diversionary program that was developed to improve the criminal justice response to persons with a mental illness.

The Memphis CIT Model

The Memphis CIT model was established in the late 1980s following the fatal police shooting of an individual with a history of mental illness. In September of 1987, law enforcement officers in Memphis responded to a call for service involving an indi- vidual with a history of mental illness. This individual was wielding a knife and threat- ening to harm himself. The officers demanded that he drop the knife, at which time he ran toward the officers with the knife in-hand. In response to this threat, the officers fired several shots that resulted in the death of the individual. Following this incident, mental health advocates and the general public expressed tremendous concern sur- rounding the manner in which the police were managing mental health crises in the community. To quell their concerns, the Memphis Mayor established a task force com- prised of representatives from law enforcement agencies, the mental health system, advocates, and academic contributors. This collaborative task force sought to develop a program that would effectively reduce the likelihood of injury to the officer and the person in crisis. The additional goals of criminal justice diversion and increasing access to mental health treatment were also on the agenda (Compton, Broussard, et al., 2011).

Davidson 49

The Memphis CIT model emerged as a comprehensive strategy to enhance the criminal justice response to mental health crises. The two defining elements of this program are a 40-hr specialized police-based training curriculum and a community- wide partnership between the criminal justice and mental health systems (Compton, Bahora, Watson, & Oliva, 2008; Wells & Schafer, 2006). There are currently more than 400 jurisdictions employing the CIT model throughout the country and approxi- mately 1,500 CIT programs in existence around the world (Compton, Broussard, et al., 2011). These figures enumerate the widespread adoption and diffusion of this strategy in the field of criminal justice.

The voluntary 40-hr Memphis CIT training program is designed to provide law enforcement officers with the knowledge and skills needed to recognize and manage mental health crises (Borum, 2000; Compton, Broussard, et al., 2011; Florida CIT Coalition, 2005). This training includes presentations from local mental health provid- ers, legal experts, advocacy organizations, and consumers of mental health services. Officers are also given the opportunity to tour local mental health facilities to gain insight into the functioning of the mental health system in their communities. One additional component of the training involves learning and practicing verbal de-esca- lation skills that are deemed useful when responding to situations involving persons with a mental illness. The training component of the Memphis CIT model empowers first-responders to resolve mental health crises effectively by diverting individuals into a treatment setting as opposed to the traditional criminal justice setting, when appropriate (Borum, 2000; Compton, Broussard, et al., 2011; Wells & Schafer, 2006).

While the training program was originally intended solely for street-level law enforcement officers, correctional officers and other first-responders are now being integrated into the training classes. Of particular importance for the present study is the decision to provide CIT training to correctional officers to address the continual flow of inmates with mental illnesses and the accompanying prevalence of mental health crises in these institutions. With the expansion to the correctional domain, CIT can now be conceptualized as a tool that is being utilized throughout the criminal jus- tice system to improve the system-wide response to persons with serious mental illnesses.

Existing Literature on the Memphis CIT Model

Prior research surrounding the Memphis CIT model has focused almost exclusively on the effectiveness of this program within the law enforcement setting. According to Compton et al. (2011), the existing literature can be consolidated into three main cat- egories: (a) officer-level outcomes, (b) disposition of CIT calls, and (c) mental health referral characteristics and outcomes. The officer-level studies have evaluated the extent to which the program is effectively achieving the objectives of the 40-hr train- ing curriculum. According to the Florida CIT Coalition (2005), the first goal of CIT training is to increase officers’ knowledge of mental illnesses and available commu- nity resources. An additional goal of the training is to increase officers’ confidence in managing incidents involving a person with a mental illness. Third, the training is

50 Criminal Justice Policy Review 27(1)

intended to provide officers with verbal de-escalation skills that they can use in the future to effectively diffuse a mental health crisis. Reduced use of force and decreased incidence of officer and subject injuries are also desired outcomes of CIT training (Borum, 2000; Compton, Neubert et al., 2011).

One of the most important goals of the Memphis CIT model is related to the pre- ferred disposition of calls for service involving persons with a mental illness. The criminal justice diversionary element of the CIT program seeks to increase officers’ utilization of the psychiatric referral process and decrease their tendency to arrest per- sons with a mental illness (Franz & Borum, 2011; Teller, Munetz, Gil, & Ritter, 2006). In addition, a long-term objective of the Memphis CIT model is improved mental health outcomes for persons with a mental illness encountering law enforcement (Broussard, McGriff, Neubert, D’Orio, & Compton, 2010; Compton, Broussard et al., 2011; Strauss et al., 2005). However, tracking these individuals through the mental health system and measuring mental health outcomes is a daunting task. Thus, researchers evaluating this objective of the CIT model have focused on comparing the characteristics of mental health referrals initiated by CIT officers to those initiated by other sources to identify the extent to which these officers are appropriately referring individuals with a mental illness to treatment.

To summarize the officer-level studies, the Memphis CIT model has been linked to reduced stigma associated with mental illness and improved self-efficacy among offi- cers when handling mental health crises (Bahora, Hanafi, Chien, & Compton, 2008; Compton, Esterberg, McGee, Kotwicki, & Oliva, 2006; Hanafi, Bahora, Demir, & Compton, 2008; Teller et al., 2006). Specifically, a study conducted by Bahora et al. (2008) explored the relationship between CIT training and reduced mental illness stigma among officers. They evaluated this relationship by measuring the social dis- tance reported by officers in their interactions with persons with schizophrenia. Social distance is a form of stigma that measures one’s comfort level in terms of how close they are willing to be to a person with a mental illness. Bahora et al. (2008) used vignettes to compare the responses of 40 officers prior to and after receiving CIT train- ing to 34 non-CIT-trained officers in terms of their reported social distance when responding to scenarios involving a person with a mental illness. They found that social distance decreased after officers received CIT training. Decreased social dis- tance and reduced stigma were also supported by other studies examining the officer- level effects of CIT training (Compton et al., 2006; Hanafi et al., 2008; Teller et al., 2006).

To assess whether CIT training achieved the intended improved self-efficacy out- come, Wells and Schafer (2006) distributed surveys to law enforcement officers before and after they participated in the training. The results of their study indicated a statisti- cally significant increase in officers’ perceptions of their ability to manage situations involving persons with a mental illness following their completion of CIT training. In addition, there was a slight increase in officers’ perceptions of their ability to commu- nicate with persons with a mental illness and their family members (Wells & Schafer, 2006). Similar findings were presented by Bahora et al. (2008) and Hanafi et al. (2008) in their respective studies of CIT training effectiveness.

Davidson 51

An additional intended objective of CIT training is improved officer knowledge of mental illness and an enhanced ability to recognize and resolve issues involving a person experiencing a mental health crisis following the training (Compton & Chien, 2008; Compton et al., 2006; Hanafi et al., 2008; Wells & Schafer, 2006). Several stud- ies have utilized basic knowledge tests derived from the standard CIT curriculum to assess knowledge gain and retention among officers participating in the training. Results from these studies indicate that CIT training improves knowledge of certain mental health conditions and the mental health referral process (Compton & Chien, 2008; Compton et al., 2006; Wells & Schafer, 2006). Furthermore, Compton and Chien (2008) administered a knowledge test to officers at the conclusion of CIT train- ing and 1 year following their completion of CIT training in an effort to identify fac- tors associated with knowledge retention. While they found a slight decrease in knowledge between these two points in time, the only factor associated with the knowledge decline was officer years of service, indicating less experienced officers had lower follow-up scores when compared with more experienced officers (Compton & Chien, 2008).

The CIT model has also been shown to effectively reduce the use of force and the incidence of officer and offender injury in mental health crises (Borum, 2000; Compton, Neubert, et al., 2011; Morabito et al., 2010; Skeem & Bibeau, 2008). A study conducted by Morabito, et al. (2010) explored the relationship between CIT training and use of force among law enforcement officers in Chicago. Morabito et al. (2010) conducted interviews with 216 officers in several Chicago districts, in which they asked several questions pertaining to the level of force used in their most recent encounter with an individual with a mental illness. One particular research question of interest was the role of CIT training in the decision to use force. They found that CIT- trained officers were much less likely to use force as a person became more resistant when compared with non-CIT-trained officers. This finding suggests that CIT-trained officers may be more patient and tolerant of noncompliant behavior as a result of their training. An additional outcome of CIT training that is related to use of force is the implementation of CIT in certain jurisdictions has decreased the utilization of high- intensity police units (SWAT [special weapons and tactics]; Bower & Pettit, 2001; Dupont & Cochran, 2000).

The second category of existing CIT research concerns the disposition of calls for service involving a person with a mental illness. With regard to the disposition of mental health disturbance calls, research suggests that CIT-trained officers are signifi- cantly more likely than non-CIT-trained officers to initiate a mental health referral (Teller et al., 2006; Watson, Morabito, Draine, & Otatti, 2008; Watson et al., 2010). However, there is conflicting evidence surrounding whether CIT actually reduces rates of arrest. While Franz and Borum (2011) posit that the implementation of CIT reduces the rate of arrest among persons with a mental illness, other researchers suggest that there is very little difference in arrest rates before and after CIT implementation (Teller et al., 2006). In addition, existing research indicates that any differential arrest rates among CIT and non-CIT-trained officers are miniscule (Teller et al., 2006; Watson et al., 2008; Watson et al., 2010). However, when compared with other law

52 Criminal Justice Policy Review 27(1)

enforcement diversionary models, CIT has demonstrated the lowest arrest rate for per- sons with a mental illness (Steadman, Deane, Borum, & Morrissey, 2000).

The final category of the extant literature evaluating the Memphis CIT model includes studies examining the characteristics and outcomes for psychiatric referrals initiated by CIT-trained officers. Studies evaluating this component of the CIT model have thoroughly reviewed medical charts from psychiatric facilities and emergency rooms to determine whether psychiatric referral was an appropriate disposition for these individuals. Findings from studies examining the mental health characteristics and outcomes of those diverted through the Memphis CIT model indicate that CIT- trained officers are utilizing the psychiatric referral process appropriately (Broussard et al., 2010; Strauss et al., 2005). In addition, these studies have indicated that CIT- trained officers are correctly identifying the calls for service involving a person with a mental illness that need to be resolved through a mental health referral.

CIT in Corrections

As mentioned previously, existing research has focused primarily on the effectiveness of the CIT model within the law enforcement setting. However, in 2004, the National Alliance on Mental Illness (NAMI) partnered with the Maine Health Access Foundation and local officials in Androscoggin County, Maine, to launch a pilot study to examine the effectiveness of the CIT model in the local jail (Public Health Research Institute, 2005). The goal was to determine whether the benefits derived from this training pro- gram for police officers could also benefit correctional officers in their daily interac- tions with inmates with mental illnesses. The researchers utilized official reports and focus group data to conduct a process and outcome evaluation of CIT in the Androscoggin County Jail.

The findings from this study provided empirical support for the implementation of this program in jails. Correctional officers expressed an appreciation for the new skills and knowledge acquired through the training. They also indicated that they felt better prepared to handle a mental health crisis within the jail following the training. In addi- tion, the correctional officers reported an increased use of verbal de-escalation and a decreased use of physical or chemical force when responding to incidents involving inmates with a mental illness following CIT training (Public Health Research Institute, 2005). Due to the demonstrated success of this study, an expansion project was sup- ported that entailed the implementation of CIT in eight county jails across the state of Maine. The findings of this project provided further support for the utilization of CITs in jails, as they replicated the findings stemming from the Androscoggin County Jail study (University of New England Center for Health Policy, Planning and Research, 2007).

There are several limitations to the existing literature surrounding the CIT model. First, most of the studies just cited examined relatively small sample sizes covering a narrow geographical area, thus limiting the generalizability of the findings. Second, only a select few incorporated a control group or employed a pre-/posttest research design, both of which are considered best practices for a program evaluation study. In addition, the limited

Davidson 53

testing of potential confounding variables further constricts the quality of the findings gen- erated from these studies. Finally, only two empirical studies to date have examined the effectiveness of the Memphis CIT model in the correctional setting, both of which were conducted in the same geographical area. Therefore, additional research is needed to explore the relative effectiveness of this model in other jails around the country.

The present study seeks to address several of the shortcomings outlined in this review of the literature. First, this study evaluates the effectiveness of the training among a sample of law enforcement and correctional officers across a broad geographi- cal region, which has not been done previously in the same study setting. This unique component of the study fosters a deeper understanding of the comparative effectiveness of CIT training across these two groups of first-responders in the criminal justice sys- tem. Second, the methodological strategies previously used meet the minimal standard for evidence-based practices set forth by Taxman and Belenko (2012). The panel research design utilized in the present study seeks to enhance the evidence base sur- rounding this topic by assessing change on several measures of training effectiveness over time. This evaluation approach is useful for capturing knowledge and skill improvement, as well as retention. Finally, this study contributes to the literature by examining the relationship between officers’ demographic characteristics and training effects, an important consideration cited by Watson et al. (2008) in the CIT context.

Prior research suggests that the incorporation of demographic variables is essential to controlling for possible mediating or confounding variables when evaluating the effectiveness of a training program (Blume, Ford, Baldwin, & Huang, 2010; Ozturk, 2011; Severiens & Ten Dam, 1994). In their meta-analysis summarizing the transfer of training literature, Blume et al. (2010) noted that demographic characteristics of train- ees influence knowledge retention and skill application in the occupational training. In the context of policing, officer characteristics such as gender, age, and years of service have been identified as factors associated with self-efficacy, decision making, and receptivity to training (Brandi & Stroshine, 2013; Garner, Maxwell, & Heraux, 2002; Kakar, 2002; Ozturk, 2011). According to Ozturk (2011), demographic characteristics are indicators of differential background experiences and learning styles that influence the effectiveness of police training. By incorporating a panel research design, includ- ing correctional officers in the sample, and assessing the influence of individual offi- cer characteristics on training outcomes, the present study involves a more extensive evaluation of the CIT training program than attempted previously and provides a unique contribution to the CIT literature.

Method

Research Questions and Hypotheses

The program evaluation conducted in this study measured the extent to which the CIT training curriculum is achieving the intended officer-level objectives. In addition, this study seeks to examine the relative effectiveness of the training among a sample of correctional officers when compared with law enforcement officers. The following

54 Criminal Justice Policy Review 27(1)

hypotheses that were tested in this study reflect the training objectives laid forth by the Florida CIT Coalition1:

Hypothesis 1 (H1): CIT training will increase officers’ knowledge of mental illness. Hypothesis 2 (H2): Officers will experience an increase in their perceived level of self-efficacy when managing mental health crises upon completion of CIT training. Hypothesis 3 (H3): Officers’ perceptions of verbal de-escalation will be enhanced as a result of CIT training

Research Design

This study utilized a panel research design with three data collection points: (a) pretest (first day of the training), (b) posttest (last day of training), and (c) follow-up (1 month fol- lowing completion of the training). According to Kirkpatrick (1976), the before-and-after approach to program evaluation is the most effective method for gaining insight into the true effectiveness of a training program. This research design sheds light on the extent to which the training is achieving its intended objectives by comparing the baseline scores on the variables of interest at the pretest to scores on the posttest and follow-up questionnaires to measure any changes in responses that can be attributed to the training program.

For the pre- and posttest data collection points, paper questionnaires (PAPI—Pencil and Paper Interview) were distributed in-person to all of the officers attending the training sessions on the first and last days of the training. The initial survey included four main sections: (a) demographics, (b) basic knowledge of mental illness, (c) per- ceptions of self-efficacy, and (d) perceptions of verbal de-escalation. This survey also contained questions pertaining to how they first learned about CIT, what prompted them to attend the class, their prior exposure to mental illness, and previous mental health training. The survey administered at the conclusion of the training included sec- tions two, three, and four from the initial survey.

A follow-up survey was constructed using the online survey (CASI—Computer- Assisted Self-Interview) development software Qualtrics. A link to this survey was distributed via email to officers who responded to the previous surveys 1 month after they completed CIT training. This questionnaire included Sections 2, 3, and 4 from the initial survey. As mentioned previously, measuring scores on the variables of interest at the beginning and end of the training provides an accurate picture of the extent to which the training is achieving its initial objectives (Kirkpatrick, 1970). This study also tests whether there is any decline between the posttest and follow-up data collec- tion points on these variables of interest.

Measures

Knowledge of mental illness (H1). The assumption underlying the importance of knowledge acquisition in a training program is that enhanced knowledge will improve on-the-job

Davidson 55

performance among those receiving the training. An effective training program provides trainees with knowledge that enhances their cognitive response to on-the-job situations (Kraiger, Ford, & Salas, 1993). The CIT training program intends to provide law enforce- ment and correctional officers with knowledge about mental illness (H1) to improve their response to mental health crises.

Eight fact-based, True/False questions derived from CIT training curricula were included to measure knowledge of mental illness across the three data collection points (see Table 1). The responses to these knowledge-based questions were coded as fol- lows: 1 = correct, 0 = either incorrect or don’t know. A summed variable comprised of the aggregated responses to these eight questions was created to identify changes in knowledge between the pretest, posttest, and follow-up data collection points. Therefore, in the analytical procedures, “Knowledge of Mental Illness” is represented by a summed variable with a range of 0 to 8, with 0 being no correct answers and 8 being all correct answers. The purpose of creating this summed variable is to differentiate

Table 1. Eight Fact-Based Questions Measuring “Knowledge of Mental Illness.”

Statement True False Don’t know

When someone has a mental illness, their brain is impaired in a way that affects their behavior and emotions.

X

An individual with a developmental disorder is likely to respond to a command differently than an individual with a mood disorder.

X

Schizophrenia is a mental illness that is often accompanied by hallucinations.

X

An individual with bipolar disorder is sometimes unpredictable because his or her mood fluctuates between depression and mania.

X

One of the main causes of mental illness is a lack of self-discipline and will-power.

X

A person with a mental illness is more dangerous than a person without a mental illness.

X

The best way to deal with a person displaying signs of a mental illness is to set firm limits and make it clear that the officers are in charge.

X

When an individual is paranoid and believes that everyone is out to get them, it is best to play along with him or her to get them to do what you want.

X

If a person is threatening to harm himself or herself or others, it is best to approach him or her with an aggressive response.

X

Note. The correct answers are marked with an X, based on CIT curriculum. CIT = Crisis Intervention Team.

56 Criminal Justice Policy Review 27(1)

between those that scored lower and those that scored higher on the knowledge-based questions at all three time points.

Self-efficacy and the management of mental health crises (H2). Self-efficacy is defined as “a personal judgment of how well one can execute a course of action required to deal with prospective situations” (Stajkovic & Luthans, 1998, p. 240). Research suggests that self-efficacy is strongly related to on-the-job performance in the organizational setting, meaning that an individual’s perception of their ability to do their job is directly related to their actual ability to carry out job-related activities (Stajkovic & Luthans, 1998). As this relates to training, enhancing trainees’ self-efficacy serves the important purpose of improving their job performance. One objective of CIT training is to increase officers’ self-efficacy with regard to the intervention and management of mental health crises, thereby increasing their ability to manage these situations. To examine the extent to which the training achieves this objective, six perception-based questions were included on the pretest, posttest, and follow-up questionnaires. Their responses were based on a 5-point Likert-type scale (0 = strongly disagree, 1 = dis- agree, 2 = neutral, 3 = agree, 4 = strongly agree). These questions tapped into the officers’ perceived ability to intervene and manage a situation involving a person with a mental illness.

A composite measure comprised of the aggregated responses to the six perception- based questions was created to represent self-efficacy at each data collection point. The internal consistency among these questions was high (Cronbach’s α = .878). When the responses were summed for the “Self-Efficacy” composite variable, the scores ranged between 0 and 24. The summation of the responses to these individual ques- tions allows the distinction to be made between officers with lower overall perceptions of self-efficacy and those with higher perceptions of self-efficacy across the three data collection points.

Perceptions of verbal de-escalation (H3). Perceptions and attitudes are important indica- tors of decision making and behavioral change. The theory of planned behavior asserts that an individual’s behavior is driven in part by their attitudes, beliefs, and percep- tions of that behavior (Ajzen, 1991). The more favorable the attitude toward the behavior, the more likely the individual is to engage in that behavior. The present study seeks to examine officers’ perceptions of verbal de-escalation techniques (H3). CIT training intends to improve their perceptions of verbal de-escalation based on the pre- sumption that improved perceptions will lead to an increase in their utilization of these skills. Three questions were included on the pretest, posttest, and follow-up question- naires that asked officers to indicate whether they believe verbal de-escalation is an advantageous tool, whether they are comfortable using verbal de-escalation, and whether they feel as though the type of intervention skills used can impact the outcome of incidents.

The responses to these three questions (Cronbach’s α = .826) were summed to cre- ate a composite variable that represented “Perceptions of Verbal De-escalation” for the three data collection points. The responses were coded in the same fashion as the other

Davidson 57

Likert-type questions utilized in the study, with strongly disagree being “0,” and strongly agree being equal to “4.” Therefore, the range of possible scores on the com- posite measure “Perceptions of Verbal De-escalation” ranged between “0” and “12.” Again, the purpose of aggregating these questions to create a summed composite vari- able is to discriminate between those with lower perceptions of verbal de-escalation and those with higher perceptions of verbal de-escalation across the three time points.

Independent variables. Prior research has not incorporated officer-level independent variables when examining the effectiveness of CIT training. These factors are impor- tant because they represent possible mediators of measurable training effects. The officer-level demographic characteristics that are utilized as independent variables in this study include race (non-White = 0, White = 1), sex (female = 0, male = 1), and age. The occupational characteristics of officers included in this survey are officer type (correctional = 0, law enforcement = 1), years of service, and rank (patrol/line officer = 0, supervisory officer = 1). Other officer-level variables that were examined as they relate to potential training effects include previous mental health training (0 = no, 1 = yes), prior exposure to mental illness (0 = no, 1 = yes), and whether the officer volun- teered for the training (0 = no, 1 = yes). The voluntary status of the officer is an impor- tant variable to consider in the present study because the Memphis CIT training curriculum was originally designed to be provided only to officers who volunteered for the training. However, agencies are increasingly seeking full implementation with the desire to have their entire agency trained in CIT. Therefore, gaining an understand- ing as to whether officers’ voluntary status really matters in terms of training effective- ness has valuable practical implications. In addition to examining officer-level characteristics, this study also explored county-level differences among officers in terms of program effectiveness.

Sampling

Out of the 67 counties in Florida, 30 have at least one criminal justice agency that has adopted the CIT model. Based on communication with CIT Coordinators from the agencies in the 14 counties that train both law enforcement and correctional officers in CIT, a convenience sample was formulated representing those counties willing to par- ticipate in the study. The 9 counties included in this study comprise approximately 32% of Florida’s overall population and 30% of the State’s average daily jail popula- tion. In addition, an estimated 34% of all arrests that occurred in Florida in 2011 took place in these 9 counties (Florida Department of Law Enforcement, 2012). Therefore, it is fair to assert that the counties included in this study represent about a third of the State’s law enforcement and local correctional agencies.

In total, surveys were distributed to officers in 11 CIT classes in 9 Florida counties (one class in 9 counties, and two classes in 2 counties) between July and December of 2012. Of the 300 total officers who received the pretest survey, 294 completed and returned the surveys, including 103 correctional officers and 191 law enforcement officers. The posttest survey was administered to the 294 officers who completed the

58 Criminal Justice Policy Review 27(1)

pretest. Of the 279 officers who completed the posttest, 100 were correctional officers and 179 were law enforcement officers. An attrition rate of 12% brought the possible sample size for the follow-up survey down to 215. Of the 215 officers who received the follow-up survey, 117 total officers completed the survey, comprised of 43 cor- rectional officers and 74 law enforcement officers. The response rates for the officer surveys were 98% for the pretest, 95% for the posttest, and 42% for the follow-up survey.

The low response rate for the follow-up survey is likely attributable to the web- based mode of distribution. While there are numerous advantages to online surveys, including cost-savings and ease of access, the response rates for these types of surveys are notably lower than other modes of survey research (Dillman, Smyth, & Christian, 2009; Marsden & Wright, 2010; Nulty, 2008). As noted by Marsden and Wright (2010) and Nulty (2008), the average response rate for online surveys is approximately 50%, which is just slightly higher than the response rate for the follow-up survey in this study.

Results

Officer-Level Characteristics

In addition to officer type, the pretest captured several demographic and occupational characteristics of respondents: age, sex, race, ethnicity, years of service, and rank (see Table 2). The descriptive statistics for these variables are reported for the 279 officers who completed the pretest and posttest surveys. The minimum age for the respondents was 20 and the maximum age was 60, with a mean age of 36 years. Nearly 80% of the officers who responded to the pre- and posttest surveys were male. In addition, approx- imately 75% of the respondents were White, while 17% reported Black as their race, and the remaining 8% were American Indian, Asian, or “Other.” For analytical pur- poses, this categorical variable was collapsed into White or non-White officers. Among the respondents, 16% reported their ethnicity as Hispanic. The minimum year(s) of service was 0 because there were several new recruits in the CIT training classes. The maximum year(s) of service was 32, and the mean was 8 years of service. The officers were asked to indicate their rank in an open-ended question. This variable was later collapsed into line officer (patrol or detention deputy), or supervisory rank (lieutenant, captain, sergeant, etc.). Roughly 78% were line officers, with the remaining 22% fall- ing into the supervisory rank category.

The pretest survey included several additional questions that were utilized as inde- pendent variables when examining the outcome measures associated with training effectiveness (see Table 2). The first question asked the officers to indicate whether they volunteered for CIT training. In this sample, 62% volunteered for the training, while 37% indicated they did not volunteer for the training, and the remaining 1% failed to answer the question. The goal of including this question as an independent variable is to determine whether an officer’s volunteer status has any impact on the effectiveness of the curriculum.

Davidson 59

Two additional independent variables that are included in this study reflect questions in the pretest that address prior exposure to mental illness. The first question asked offi- cers to indicate whether they received mental health training in their training academy, and the second question asked the officers to indicate whether they know someone per- sonally with a mental illness (i.e., family member, friend, coworker, etc.). Approximately 44% of the respondents reportedly received mental health training in the academy and 32% reported knowing someone with a mental illness. Variables reflecting these officer- level characteristics were tested as possible mediating variables that could contribute to any changes that may have occurred between the pretest and posttest, as well as the fol- low-up data collection point with regard to the key measures of training effectiveness.2

Table 2. Officer-Level Characteristics (n = 279).

Variable n % Variable Years

Officer type Age Law enforcement 179 64 Minimum 20 Correctional 100 36 Maximum 60 White M 36 White 208 75 SD 9 Non-White 71 25 Hispanic Years of service Hispanic 45 16 Minimum 0 Non-Hispanic 232 83 Maximum 32 Missing 2 1 M 8 Sex SD 7 Male 220 79 Female 59 21 Rank Line officer 218 78 Supervisor 59 21 Missing 2 1 Volunteer status Volunteer 172 62 Nonvolunteer 103 37 Missing 4 1 Prior MH training Yes 150 54 No 123 44 Missing 6 2 Know someone with MI Yes 89 32 No or “don’t know” 187 67 Missing 3 1

Note. MI = mental illness; MH = mental health.

60 Criminal Justice Policy Review 27(1)

Immediate Training Effects: Pretest/Posttest Results

To test the previously outlined program evaluation hypotheses, three key outcome mea- sures associated with training effectiveness were captured on the pretest and posttest sur- veys: (a) basic knowledge of mental illness, (b) perceptions of self-efficacy, and (c) perceptions of verbal de-escalation. The first step in the program evaluation analytical pro- cess involved examining the immediate training effects, by conducting a series of paired samples t tests to measure the mean changes in scores on the key outcome measures between the pretest and posttest. The results of these analyses are presented in Table 3.

To identify changes in “Knowledge of Mental Illness,” a paired samples t test was conducted to compare the mean number of correct responses on the pretest to the mean number of correct responses on the posttest. As illustrated in Table 3, this analysis revealed a statistically significant increase in the mean number of correct responses from the pretest to the posttest, which suggests CIT training does effectively improve officers’ knowledge of mental illness on average.

For the “Self-Efficacy” outcome, responses to six questions were summed to create a composite variable for the pretest and a separate composite measure comprised of the same questions was created using the posttest responses. The paired samples t test comparing the pretest mean score with the posttest mean score was also significant on this measure, suggesting that on average, CIT training significantly increased officers’ self-efficacy with regard to responding to mental health crises. Similarly, two separate composite variables consisting of the aggregated responses to three questions were created to represent “Perceptions of Verbal De-escalation” at the pretest and posttest data collection points. The results of a paired samples t test comparing the pretest and posttest means on this measure indicated that CIT training significantly increased offi- cers’ perceptions of verbal de-escalation, on average.

Table 3. Immediate Training Effects (n = 279).

Pretest Posttest SD t df

H1: Knowledge of mental illness

5.85 (out of 8) 6.67 1.375 9.836*** 271

H2: Self-efficacy 15.71 (out of 24) 19.62 4.55 14.09*** 268 H3: Perceptions of

verbal de-escalation 9.89 (out of 12) 10.65 2.52 4.980*** 271

H4: Perceptions of MH services and referral process

6.21 (out of 12) 8.10 2.79 10.998*** 265

H5a: Knowledge of MH referral process

2.40 (out of 4) 3.37 1.043 15.162*** 267

H5b: Knowledge of MH services

2.18 (out of 4) 3.31 1.047 17.871*** 270

Note. MH = mental health. ***p < .001.

Davidson 61

The results of the paired samples t tests measuring the immediate training effects indicate that on every measure of training effectiveness there was a statistically sig- nificant increase between the pretest and posttest. The most pronounced growth was found on the “perceptions of self-efficacy” measure, which increased by approxi- mately 16% between the pretest and posttest data collection points. However, the “Knowledge of Mental Illness” measure increased by 10% and the “Perceptions of Verbal De-escalation” measure improved by 6%. Any growth on these measures is considered an indication of training effectiveness.

To explore the immediate training effects further, the relationships between the three key outcome measures and the officer characteristics previously outlined were tested using a series of bivariate analytical procedures. The following independent variables were included in this analytical process: officer type, sex, race, rank, years of service, volunteer status, prior exposure to mental illness, and previous mental health training. The main purpose of these analyses was to identify any possible mediating relationships between the independent variables and the outcome measures that may explain the growth noted previously on the measures of training effectiveness. These analytical procedures served the additional purpose of identifying whether officers with certain characteristics benefited more than others from the training. Specifically, this study sought to determine whether the training was equally effective for correc- tional and law enforcement officers.

Prior to conducting these analyses, a change variable was created for each outcome measure, which was calculated by subtracting the pretest mean from the posttest mean. A series of independent samples t tests were conducted to test the relationships between the categorical independent variables and the change variables that represent the mea- sures of training effectiveness. In addition, Pearson’s correlation coefficients were uti- lized to examine the relationships between years of service and the immediate training effects. The Pearson’s correlation coefficients revealed no statistically significant rela- tionships, indicating that officers’ level of experience does not play a role in their receptivity to CIT training.

The results of the independent samples t tests involving officer type (correctional or law enforcement), rank, and prior exposure to mental illness indicated that these vari- ables are not related to the officer-level CIT training objectives. These findings sug- gest the immediate training effects are not directly attributable to officers’ years of service, rank, prior exposure to mental illness, or whether the officer is in law enforce- ment or corrections. In addition, no statistically significant relationships were identi- fied between the independent variables and mean changes on the “Perceptions of Verbal De-escalation” measure.

However, a significant relationship was identified between officer sex and change on the “Self-Efficacy” measure. Females demonstrated a 22% increase (change = 5.25, SD = 5.02) in “Self-Efficacy,” whereas males experienced a 15% mean increase (change = 3.52, SD = 4.36). However, female officers scored significantly lower than males on the pretest surveys, indicating they had more to gain from the training at the outset in terms of self-confidence. Also worth noting, males gained more than females on the other measures, although these findings did not reach statistical significance.

62 Criminal Justice Policy Review 27(1)

With regard to the “Knowledge of Mental Illness” measure, nonvolunteers gained significantly more (change = 1.05, SD = 1.4) than volunteers (change = 0.70, SD = 1.35). The growth translates to a 13% increase for nonvolunteers and a 9% improve- ment rate for volunteers. In addition, officers without prior mental health training gained significantly more (change = 1.06, SD = 1.45) than officers with prior mental health training (change = 0.61, SD = 1.26) on the “Knowledge of Mental Illness” mea- sure. Officers with prior mental health training had a higher mean score on this mea- sure at the pretest, indicating they had a higher baseline knowledge of mental illness than officers with no prior mental health training.

When examining the impact of the independent variables on the three key measures of CIT training effectiveness, it appears several variables may play a mediating role in predicting immediate training effects. Specifically, officer volunteer status and prior mental health training are significantly related to mean changes on the “Knowledge of Mental Illness” measure. In addition, a significant relationship was identified between officer sex and mean changes on the “Self-Efficacy” measure. Although some groups demonstrated greater improvements on certain objectives when compared with others, all groups increased on every training objective, indicating no detectable declines on these measures.3,4

Intermediate Training Effects

The analytical procedures used to examine the intermediate training effects included only the sample of respondents who completed all three surveys (n = 117). Recognizing that attrition can potentially result in nonresponse bias that diminishes the generaliz- ability of the findings, a series of bivariate analyses were conducted to identify differ- ences between the group of officers who responded to the follow-up survey and the group of officers who did not respond to this survey. These analyses revealed females were more likely to respond than males, officers who reported having previously received mental health training were more likely to respond than officers with no prior mental health training, officers who volunteered for CIT training were more likely to respond than officers who did not volunteer for the training, and officers who knew someone with a mental illness were less likely to respond to the follow-up survey than officers who did not report knowing someone with a mental illness. Respondents also possessed less favorable attitudes about verbal de-escalation than nonrespondents at the beginning of CIT training. Worth noting, this difference between respondents and nonrespondents in terms of baseline perceptions of verbal de-escalation diminished to a nonsignificant level at the posttest data collection point, suggesting this may be an artifact of sample distribution.

It can be determined from this preliminary analysis that the missing data (nonre- sponse) are not at random, which means the differences between the respondents and nonrespondents in terms of sex, prior mental health training, volunteer status, and knowing someone with a mental illness suggest potential sources of bias. Thus, the findings pertaining to these officer-level characteristics and the intermediate training effects may not be generalizable to the entire sample or a broader population with a

Davidson 63

similar distribution. In addition, it would be difficult to extrapolate the true meaning of findings related to these officer characteristics and the intermediate training effects because this subsample does not proportionately represent the original sample. However, additional analyses involving the follow-up sample serves the important purpose of exploring the retention and usefulness of CIT training.

To examine the intermediate effectiveness of CIT training, a series of paired sam- ples t tests were performed to assess the mean change between the posttest and follow- up surveys on the three measures of training effectiveness. As evidenced in Table 4, no substantial change was identified between the posttest mean and follow-up mean on the “Knowledge of Mental Illness” measure. However, the mean scores on the other two measures significantly declined between the two data collection points. On aver- age, officers experienced an estimated 22% decline on the “Self-Efficacy” measure and a 28% decrease on the “Perceptions of Verbal De-escalation” measure between the posttest and follow-up data collection points. The significant declines on these mea- sures represent a diminishing effect, or decay, of the training over time.

Separate change variables were created by subtracting the posttest mean from the follow-up mean to represent the changes that occurred between the posttest and fol- low-up data collection points on the “Knowledge of Mental Illness,” “Self-Efficacy,” and “Perceptions of Verbal De-escalation” measures. Independent samples t tests were performed to examine the relationships between the original nine independent vari- ables and these change variables to determine if officer characteristics played a role in the change that occurred between these two time points. The nine independent vari- ables that were tested include officer type, sex, race, rank, previous mental health training, volunteer status, prior exposure to mental illness, and years of service. None of the independent variables were significantly related to the “Knowledge of Mental Illness” or “Self-Efficacy” change variables. However, officer rank and race were sig- nificantly related to the “Perceptions of Verbal De-escalation” change variable. Unranked officers lost more (change = −3.62, SD = 2.55) than ranked officers (change = −2.44, SD = 2.62) on this measure, t(98) = −2.024, p < . 05). In addition, non-White officers experienced a greater deterioration (change = −4.59, SD = 2.09) than White officers (change = −3.04, SD = 2.64) between the posttest and follow-up survey data collection points in terms of their “Perceptions of Verbal De-escalation,” t(98) = −2.280, p < . 05).5

Table 4. Intermediate Training Effects (n = 117).

Variable Posttest Follow-up SD t df

H1: Knowledge of mental illness

6.69 (out of 8) 6.67 0.95 0.22 95

H2: Self-efficacy 20.12 (out of 24) 14.61 3.90 14.07*** 98 H3: Perceptions of

verbal de-escalation 11.03 (out of 12) 7.73 2.61 12.64*** 99

***p < .001.

64 Criminal Justice Policy Review 27(1)

To further explore the relationship between time and training effectiveness, several linear growth curve models were created using pretest, posttest, and follow-up scores for the three measures of training effectiveness: “Knowledge of Mental Illness,” “Self- Efficacy,” and “Perceptions of Verbal De-escalation.” The key independent variable for each of these models is Time (pretest, posttest, and follow-up). Each growth curve model also incorporated the independent variables previously identified as predictors of change on the individual outcome measures. As noted previously, prior mental health training and volunteer status were significantly related to change on the “Knowledge of Mental Illness” measure, sex was the only officer characteristic associ- ated with change on the “Self-Efficacy” measure, while officer rank and race were associated with change on the “Perceptions of Verbal De-escalation” measure.

The results of the growth curve models are presented in Table 5. As illustrated, Time was a significant predictor of growth in all three models. This finding suggests that officers’ scores pertaining to “Knowledge of Mental Illness,” “Self-Efficacy,” and “Perceptions of Verbal De-escalation” were significantly affected by Time. The refer- ence category for Time in the growth curve models was Time 3 (follow-up survey), meaning officers’ scores at Time 1 (pretest) and Time 2 (posttest) were compared with their scores at Time 3 (follow-up survey). As indicated in Table 5, officers’ scores on the “Knowledge of Mental Illness” measure at Time 1 were significantly lower (9%) than their scores at Time 3. However, officers’ “Knowledge of Mental Illness” at Time 2 was only slightly greater (less than 1%) than Time 3. The results of the growth curve

Table 5. Growth Curve Models.

Effect β SE

Model 1: Knowledge of mental illness Time 1 (Pretest) −0.694** 0.128 Time 2 (Posttest) 0.031 0.122 Prior mental health training −0.064 0.120 Volunteer status 0.074 0.126 Model 2: Self-efficacy Time 1 (Pretest) 1.27** 0.451 Time 2 (Posttest) 5.39** 0.390 Sex (Female) −1.25 0.536 Model 3: Perceptions of verbal de-escalation Time 1 (Pretest) 2.545** 0.244 Time 2 (Posttest) 3.227** 0.228 Race (Non-White) 0.234 0.282 Rank (Patrol) −0.074 0.250

Note. Reference categories for variables: Time 3 (Follow-Up), Sex (Male), Prior mental health (Yes), Volunteer status (Volunteer), Race (White), Rank (Supervisor). Likelihood Ratio Tests: Knowledge of Mental Illness (5, n = 110) = 70.95, p < .001; Self-Efficacy (3, n = 117) = 154.65, p < .001; Perceptions of Verbal De-escalation (4, n = 117) = 150.26, p < .001. **p < .01.

Davidson 65

model with “Knowledge of Mental Illness” as the dependent variable align with the findings presented previously regarding the changes that occurred on this measure between the pretest and posttest, and posttest and follow-up data collection points. However, the independent variables previously identified as predictors of change on the “Knowledge of Mental Illness” measure were not significant in the growth curve model.

With regard to the “Self-Efficacy” growth curve model, both Time 1 and Time 2 were significantly different from Time 3. In this model, officers’ “Self-Efficacy” at Time 1 was 5% higher than Time 3, and their “Self-Efficacy” at Time 2 was 22% higher than Time 3. These findings support the results previously provided related to changes on the “Self-Efficacy” measure over time, with one exception. While sex was associated with change on the “Self-Efficacy” measure in the bivariate analysis, this finding was not corroborated in the growth curve model.

In the final growth curve model, significant differences were noted between Time 1 and Time 3, as well as Time 2 and Time 3 on the “Perceptions of Verbal De-escalation” measure. Officers’ scores at Time 3 were 21% lower than Time 1 and 27% lower than Time 2. The results of this growth curve model confirm the findings previously out- lined pertaining to changes over time on the “Perceptions of Verbal De-escalation” measure. However, similar to the other growth curve models, the effects of the inde- pendent variables on officers’ “Perceptions of Verbal De-escalation” that were previ- ously significant decreased to a nonsignificant level in this model. Overall, the growth curve models further indicate that the “Knowledge of Mental Illness” gained through the training was retained by officers at the follow-up data collection point. However, officers’ “Self-Efficacy” and “Perceptions of Verbal De-escalation” declined signifi- cantly between the posttest and follow-up time frame, making their follow-up scores lower than their baseline scores. Potential explanations for these declines will be pos- ited in the upcoming discussion.

Discussion

The purpose of this study was to evaluate the effectiveness of the Memphis CIT train- ing curriculum across several counties in Florida. An additional objective entailed exploring the relative effectiveness of the training among law enforcement and cor- rectional officers. Central to this study was identifying whether the training was simi- larly effective for correctional officers, since this training has recently been expanded to include this set of participants. When examining the immediate effectiveness of the training, analyses revealed a significant increase on all three measures of training effectiveness between the pretest and posttest data collection periods. These analyses indicate that the training does effectively achieve the intended officer-level objectives in the short-term time frame among both law enforcement and correctional officers.

Further analyses were conducted to determine whether certain officer characteris- tics mediate the immediate effectiveness of the training. These analyses revealed that officer sex was significantly related to growth on the “Self-Efficacy” measure, indicat- ing that females gained more than males on this measure. In addition, nonvolunteers

66 Criminal Justice Policy Review 27(1)

and officers without prior mental health training experienced a greater increase on the “Knowledge of Mental Illness” measure than their counterparts. In an effort to deter- mine whether the differences identified could be attributable to pretraining dissimilari- ties rather than susceptibility to the training, a series of independent samples t tests were conducted to compare the baseline (pretest) scores among the groups just described on the measures where a significant difference was identified. Volunteers and nonvolunteers entered the training with roughly the same mean scores on this measure, which indicates the training truly did have a greater impact on nonvolunteers with regard to improving their “Knowledge of Mental Illness.”

However, females started with a lower baseline score than males on the “Self- Efficacy” measure and officers with no prior mental health training began with lower baseline scores on the “Knowledge of Mental Illness” measure than officers with prior mental health training, which suggests females and officers with no prior mental health training had more to gain from the training than their counterparts. These findings point to the presence of a ceiling effect, in which there is only so much to be gained from a training program and the ceiling is considered the maximum effectiveness of training (Lewis-Beck, Bryman, & Futing Liao, 2003). Thus, the groups of officers with lower pretest mean scores on the measures of training effectiveness had further to climb to reach the ceiling, which could explain the differential growth rates.

Further testing to examine possible explanations for the gender difference in the effectiveness of the training ruled out prior exposure to mental illness as an explana- tion. Existing research is mixed on gender differences among law enforcement officers in terms of perceived self-efficacy, with findings from certain studies indicating that male officers are more self-confident than female officers, and other studies indicating no significant difference between male and female officers in terms of self-efficacy (Kakar, 2002; Love & Singer, 1988). This finding is particularly interesting in the context of the present study because most of the officers in charge of the CIT program within these agencies were female. As stated by Kakar (2002),

It has been argued that women bring an “ethic of care” to law enforcement. This ethic is seen to translate into policewomen’s possessing greater communication skills and more calming demeanors and being more empathetic with and better equipped to interact with citizens. (p. 242)

In addition, the cognitive learning styles of men and women differ greatly, which may explain disparate gains in various areas of the training (Severiens & Ten Dam, 1994). Future research is needed to explore gender differences in terms of amenability to training and self-efficacy in the field of law enforcement, particularly in relation to crisis intervention.

When examining the intermediate effectiveness of the training, the results of the statistical tests revealed that officers experienced very little change on the “Knowledge of Mental Illness” measure between the posttest and follow-up data collection points. This finding indicates that the knowledge acquired during CIT training was largely retained during the 1-month follow-up period. However, significant reductions were

Davidson 67

identified when examining the follow-up mean scores for the “Self-Efficacy” and “Perceptions of Verbal De-escalation” measures. These results suggest officers expe- rienced a significant decline, or decay, on these measures when they returned to the field following their completion of CIT training. These declines took their mean scores to a lower level than documented on the pretest survey.

To determine whether these declines may be attributable to officers not having the opportunity to use the skills in the month following the training, a secondary analysis was conducted using a question on the follow-up survey that asked officers to indicate the number of times they have used verbal de-escalation since completing CIT train- ing. This analysis revealed that 30% of the follow-up sample reported not having the opportunity use the verbal de-escalation skills since completing the training. To assess whether this might explain the decline, an independent samples t test was conducted to compare the mean scores of officers who reported using verbal de-escalation skills since the training with those who have not had the opportunity to use them on the post- test follow-up change variables for the “Perceptions of Verbal De-Escalation” and “Self-Efficacy” measures. No statistically significant differences between these two groups emerged, although a greater decline was found among the group of officers who reported not using verbal de-escalation since the training on the “Self-Efficacy” change variable. Furthermore, when looking at differences between these two groups on just the follow-up “Self-Efficacy” scores, officers who reported using verbal de- escalation since the training scored significantly higher than the group of officers who reported not using verbal de-escalation since the training. These analyses indicate future research should extend the follow-up period to identify whether the decay iden- tified in this study shrinks over time once more officers have the opportunity to use the learned skills. Moreover, further research is needed to explore the interrelatedness between perceptions of verbal de-escalation and self-efficacy in the CIT context.

Because the secondary analyses just presented did not provide a finite explanation of the noted decay, this issue warrants further discussion. As mentioned previously, “Self-Efficacy” refers to one’s perceptions of one’s ability. Within the context of the present study, officers were asked several questions at each time point pertaining to their level of confidence when intervening and managing situations involving persons with a mental illness. While officers experienced a significant increase on this measure between the pretest and posttest surveys, the significant decline identified on the fol- low-up survey suggests their self-confidence decreased over time. One potential explanation for this deterioration that could be empirically tested in the future might be the acquisition of knowledge and skills in the training leads to an increased aware- ness of shortcomings in these areas once officers return to the field. Law enforcement and correctional officers may leave CIT training with improved self-efficacy sur- rounding their ability to manage situations involving persons with a mental illness, but when given the opportunity to use the knowledge and skills they obtained in the train- ing after returning to duty, they could become acutely aware of inadequacies they did not know existed prior to the training.

Similarly, the findings related to the decline on the “Verbal De-escalation” measure suggest CIT training may have a boomerang effect in the intermediate time frame by

68 Criminal Justice Policy Review 27(1)

lowering officers’ perceptions of the de-escalation tools they learned once given the opportunity to exercise these skills in the field. Although CIT training is intended to improve officers’ perceptions and understanding of verbal de-escalation by providing them with additional de-escalation skills and exercises, officers may not find these tools as useful as hoped when encountering persons with a mental illness once back in the field. Future qualitative research is needed to explore this further, in which officers are interviewed to gain a better understanding of why officers’ perceptions of self- efficacy and verbal de-escalation declined so quickly after completing CIT training. In addition, an extended follow-up period would be beneficial to determine whether this deterioration is temporary or long-term.

When examining the relationships between the independent variables and the dependent change variables for the three measures of training effectiveness, the only significant relationships identified involved the “Perceptions of Verbal De-escalation” measure. On this measure, unranked officers experienced a greater decline than ranked officers, and non-White officers demonstrated a greater decrease than White officers. According to data from the follow-up survey, fewer ranked officers reported having the opportunity to utilize the de-escalation skills in the month following their comple- tion of the training, and thus were less likely to change their perceptions of these skills in the follow-up time period.

One possible explanation for the relationship between officer race and change in “Perceptions of Verbal De-escalation” could be attributable to differences that existed between White and non-White officers on this measure prior to the follow-up data col- lection point. Although White and non-White officers began the training with nearly the same mean score on this measure, non-White officers had a higher mean score on the posttest, indicating they gained more from the training on this measure. Therefore, they had more to lose going into the follow-up data collection point. In addition, a dif- ference existed between White and non-White officers in terms of prior exposure to mental illness that may provide an explanation for this finding. Nearly 35% of White officers reported knowing someone with a mental illness at the beginning of CIT train- ing, compared with 24% of non-White officers. While the chi-square test did not prove significant, this difference is potentially substantial enough to provide an explanation for the differences identified between these two groups on the “Self-Efficacy” mea- sure. Prior exposure to mental illness may be a salient factor that contributes to sus- tained “Self-Efficacy” following the completion of CIT training, although not measurable in this study. No significant differences between White and non-White officers were identified in relation to the use of verbal de-escalation skills at the fol- low-up data collection point.

Several recommendations for improvement of CIT training that can be derived from this study pertain specifically to the notable training decay found among the officers who completed the follow-up survey. The decay on the Self-Efficacy and Perceptions of Verbal De-escalation measures suggests greater emphasis should be placed on active learning exercises (i.e., role-playing), as opposed to relying so heav- ily on lecture-based content. In the CIT training curriculum, typically one full day is devoted to role-playing, with each individual officer getting the opportunity to

Davidson 69

practice the learned verbal de-escalation techniques in only one scenario. According to a synthesis of training literature provided by Burke and Hutchins (2007), behavioral modeling and active learning exercises through which trainees can practice skills and receive feedback are essential training strategies to promote the retention and on-the- job application of skills acquired through training (i.e., transfer of training). In addi- tion, the decay noted in this study indicates a greater need for refresher courses, or follow-up training, to reinforce the skills and knowledge gained from CIT training. An additional reinforcement strategy that might effectively improve the long-term reten- tion of CIT knowledge and skills would be assigning newly trained CIT officers to veteran CIT officers for a brief period of field training following their completion of the course. This would allow newly trained CIT officers to learn from more experi- enced CIT-trained officers as to how to apply the skills they learned in CIT training in real-world scenarios.

With regard to policy, one particularly important finding of this study is that CIT training appears to be equally effective for both law enforcement and correctional officers. Officer type was not significantly related to the training objectives, meaning the training is similarly effective for both types of officers. Since many jurisdictions are beginning to send correctional officers, probation officers, and other first-respond- ers to CIT training, it is critical to understand whether the training has the desired effects on these other groups of participants. The results of this study support prior research that indicates this type of training is beneficial for correctional officers to attend. Based on the findings from this study, more correctional agencies should con- sider enrolling their officers in CIT programs. In addition, although the standard CIT curriculum was effective for correctional officers, the development of a corrections- specific CIT program may be even more beneficial by catering specifically to the unique challenges faced by correctional officers when responding to inmates experi- encing mental health crises.

In addition, while CIT training was originally intended solely for volunteers, many agencies are now considering the full implementation of CIT, meaning the goal is to have 100% of their officers trained in CIT. This is a controversial practice in the CIT community as many stakeholders feel as though some officers may not possess the temperament necessary to be receptive to the training. The difference in approaches (partial vs. full implementation) is contingent upon jurisdictional needs and objec- tives, with some agencies treating CIT as a specialized unit of officers and others treat- ing CIT as “just another training.” Therefore, the decision to train nonvolunteers is discretionary and agency-specific. Nonetheless, the findings from this study indicate that CIT training is effective and beneficial for both volunteers and nonvolunteers.

Study Limitations and Future Research

This study has several limitations that should be noted. First, this study utilized a con- venience sample, which reduces the generalizability of the findings. Second, the true impact of CIT training on law enforcement and correctional officers is difficult to estimate because the study did not include a control group. Third, because

70 Criminal Justice Policy Review 27(1)

many officers volunteered for the training, selection bias may pose a threat to external validity. In addition, the short follow-up period does not allow the measurement of any long-term attitudinal or behavioral changes experienced by CIT-trained law enforce- ment and correctional officers. Furthermore, the low response rate for the follow-up survey diminished the quality of the implications that could be derived from the study, particularly with regard to nonresponse bias. Finally, the officer surveys did not include a measure of social desirability, making it difficult to determine whether the officers answered the questions in a false manner that projected a greater improvement across the time points than actually occurred.

This study brought to light several ideas for future studies involving the CIT model. First, duplicating this study with a larger sample size covering a broader geographical area would enhance the generalizability of these findings. Second, extending the fol- low-up period to a minimum of 6 months following the completion of the training would provide a more accurate picture of the long-term effectiveness of CIT training. In addition, future studies should incorporate official CIT reports completed by CIT- trained officers when they respond to an incident involving a person with a mental illness to gain a better understanding of the behavioral impact of the training. Furthermore, more research is needed surrounding the implementation and effective- ness of the CIT model in rural areas. Finally, the ideal assessment of training program effectiveness would entail the development of a randomized control research design in which officers were randomly assigned to a treatment (training) group and control group. This type of research design would permit the comparison of CIT-trained offi- cers with non-CIT-trained officers in terms of their performance on the measures of training effectiveness.

Conclusion

The findings from this study indicate CIT training effectively achieved the immediate officer-level objectives among both law enforcement and correctional officers. However, data from the follow-up data collection point suggested a noticeable deterio- ration or decay in terms of intermediate training effectiveness. While the knowledge attained through CIT training was retained throughout the follow-up period, officers’ “Self-Efficacy” and “Perceptions of Verbal De-escalation” declined significantly in the month following the completion of CIT training. Future research is needed to iden- tify explanations for these declines to determine whether improvements in training curriculum or refresher courses could address decay. This study also revealed that officer characteristics play a role in CIT training effectiveness, suggesting that officers with certain characteristics may be more susceptible to the training than others.

Of critical importance to this study is the finding that this training is similarly effec- tive for both law enforcement and correctional officers, indicating that CIT training is appropriate to apply in the correctional setting. In addition, this study revealed that officers who do not volunteer for the training may benefit just as much, if not more, than officers who volunteer for the training. Thus, voluntariness may not be a neces- sary criterion for enrollment in the training.

Davidson 71

The criminal justice involvement of persons with a mental illness is a critical public policy concern. As such, the time is ripe for more research examining existing and innovative solutions to this issue. Program evaluation research is critical to under- standing the true effectiveness of these solutions, which not only aids in building an evidence base for criminal justice programs but also promotes continuous quality improvement. This study adds to the evidence base surrounding the Memphis CIT model, and provides guidance for future research on this topic.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

1. The Florida Crisis Intervention Team (CIT) Coalition is a conglomerate of criminal jus- tice and mental health representatives responsible for the promotion of the Memphis CIT model and providing technical assistance for Florida counties considering the implementa- tion of the CIT program.

2. A series of Pearson’s correlation coefficients and chi-square tests were created to examine the relationships between the officer-level independent variables using the data collected for the 279 officers who completed the pretest and posttest surveys. Although several pos- sible sources of multicollinearity were identified, most of these variables warrant consid- eration in future analyses due to their theoretical importance. However, officer ethnicity and race tap into similar constructs, in the same way officer age and years of service are essentially measuring the same thing. Therefore, to reduce the risk of multicollinearity, the variables representing officer ethnicity and officer age were excluded from future analyti- cal procedures.

3. Recognizing the interactions that existed among the independent variables, a series of two- way between-group ANOVA models were conducted to determine whether the associated officer-level characteristics translated into interaction effects that could possibly explain the observed immediate training effects. These models were not significant, indicating no need to further explore these interrelationships.

4. Because officers were surveyed in multiple counties, a categorical variable repre- senting each of the counties in the study was treated as a random effect in a series of mixed linear models to determine whether county of training mediated the identified relationships between the independent variables and the immediate training effects. County of training was not a significant predictor, nor did it change the findings presented.

5. The ANOVA models and the mixed linear models with county of training were repeated here. No significant findings were revealed, thus precluding the need to further explore interactions or the relationship between county of training and the key measures of training effectiveness.

72 Criminal Justice Policy Review 27(1)

References

Aderibigbe, Y. (1997). Deinstitutionalization and criminalization: Tinkering in the interstices. Forensic Science International, 85, 127-134.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179-211.

Angermeyer, A., Cooper, B., & Link, B. (1998). Mental disorder and violence: Results of epi- demiological studies in the era of de-institutionalization. Social Psychiatry and Psychiatric Epidemiology, 33, S1-S6.

Bahora, M., Hanafi, S., Chien, V., & Compton, M. (2008). Preliminary evidence of effects of Crisis Intervention Team training on self-efficacy and social distance. Journal of Administration and Policy in Mental Health, 35, 159-167.

Blume, B. D., Ford, J. K., Baldwin, T. T., & Huang, J. L. (2010). Transfer of training: A meta- analytic review. Journal of Management, 36, 1065-1105.

Borum, R. (2000). Improving high risk encounters between people with mental illness and the police. Journal of the American Academy of Psychiatry and the Law, 28, 332-337.

Borum, R., Deanne, M., Steadman, H., & Morissey, J. (1998). Police perspectives on responding to mentally ill people in crisis: Perceptions of program effectiveness. Behavioral Sciences & the Law, 16, 393-405.

Bower, D., & Pettit, E. (2001, February). The Albuquerque police department’s Crisis Intervention Team: A report card. FBI Law Enforcement Bulletin, 1-5. Retrieved from http://www.au.af.mil/au/awc/awcgate/fbi/crisis_interven.pdf

Brandi, S. G., & Stroshine, M. S. (2013). The role of officer attributes, job characteristics, and arrest activity in explaining officer use of force. Criminal Justice Policy Review, 24, 551-572.

Broussard, B., McGriff, J., Neubert, D., D’Orio, B., & Compton, M. (2010). Characteristics of patients referred to psychiatric emergency services by Crisis Intervention Team police officers. Community Mental Health Journal, 46, 579-584.

Burke, L. A., & Hutchins, H. M. (2007). Training transfer: An integrative literature review. Human Resource Development Review, 5, 263-296.

Compton, M. T., Bahora, M., Watson, A., & Oliva, J. (2008). A comprehensive review of extant research on Crisis Intervention Team (CIT) programs. Journal of the American Academy of Psychiatry and the Law, 36, 47-55.

Compton, M. T., Broussard, B., Munetz, M., Oliva, J., & Watson, A. (2011). The Crisis Intervention Team model of collaboration between law enforcement and mental health. New York, NY: Nova Science Publishers.

Compton, M. T., & Chien, V. (2008). Factors related to knowledge retention after Crisis Intervention Team training for police officers. Psychiatric Services, 59, 1049-1051.

Compton, M. T., Esterberg, M., McGee, R., Kotwicki, R., & Oliva, J. (2006). Brief reports: Crisis Intervention Team training: Changes in knowledge, attitudes, and stigma related to schizophrenia. Psychiatric Services, 57, 1199-1202.

Compton, M. T., Neubert, B. N. D., Broussard, B., McGriff, J. A., Morgan, R., & Oliva, J. R. (2011). Use of force preferences and perceived effectiveness of actions among Crisis Intervention Team (CIT) police officers and non-CIT officers in an escalating psychiatric crisis involving a subject with schizophrenia. Schizophrenia Bulletin, 37, 737-745.

Council of State Governments. (2002). Criminal Justice/Mental Health Consensus Project. Retrieved from https://www.ncjrs.gov/pdffiles1/nij/grants/197103.pdf

Dillman, D. A., Smyth, J. D., & Christian, L. M. (2009). Internet, mail, and mixed-mode sur- veys: The tailored design method (3rd ed.). Hoboken, NJ: John Wiley.

Davidson 73

Ditton, P. (1999). Special report: Mental health and treatment of inmates and probationers. Washington, DC: Bureau of Justice Statistics, U.S. Department of Justice.

Dupont, R., & Cochran, S. (2000). Police response to mental health emergencies: Barriers to change. Journal of the American Academy of Psychiatry and Law, 28, 338-344.

Erickson, P., & Erickson, S. (2008). Crime, punishment, and mental illness: Law and the behav- ioral sciences in conflict. New Brunswick, NJ: Rutgers University Press.

Florida CIT Coalition. (2005). The Florida Crisis Intervention Team (CIT) Program. Retrieved from http://www.nami.org/Content/ContentGroups/Policy/CIT/Florida_CIT_program_model.pdf

Florida Department of Law Enforcement. (2012). Crime in Florida, 2011 Florida uniform crime report [Computer program]. Tallahassee, FL: Author.

Florida Supreme Court. (2007). Mental health: Transforming Florida’s mental health system. Retrieved from http://www.dcf.state.fl.us/programs/samh/publications/judgeleitrpt.pdf

Franz, S., & Borum, R. (2011). Crisis Intervention Teams may prevent arrests of people with mental illnesses. Police Practice & Research, 12, 265-272.

Garner, J. H., Maxwell, C. D., & Heraux, C. G. (2002). Characteristics associated with the prevalence and severity of force used by the police. Justice Quarterly, 19, 705-746.

Hanafi, S., Bahora, M., Demir, B., & Compton, M. (2008). Incorporating Crisis Intervention Team (CIT) knowledge and skills into the daily work of police officers: A focus group study. Community Mental Health Journal, 44, 427-432.

Horwitz, A. (2002). The social control of mental illness. Clinton Corners, NY: Percheron Press. James, D., & Glaze, L. (2006). Mental health problems of prison and jail inmates. Bureau of

Justice Statistics. Retrieved from http://bjs.ojp.usdoj.gov/content/pub/pdf/mhppji.pdf Kakar, S. (2002). Gender and police officers’ perceptions of their job performance: An analysis

of the relationship between gender and perceptions of job performance. Criminal Justice Policy Review, 13, 238-256.

Kirkpatrick, D.L. (1976). Evaluation of training. In R. Craig & L. Bittel (Eds.), Training and development handbook: A guide to human resource development (pp. 87-112). New York, NY: McGraw-Hill.

Kraiger, K., Ford, K., & Salas, E. (1993). Application of cognitive, skill-based, and affective theories of leaning outcomes to new methods of training evaluation. Journal of Applied Psychology, 78, 311-328.

Lewis-Beck, M., Bryman, A., & Futing Liao, T. (2003). The SAGE encyclopedia of social sci- ence research methods. London, England: SAGE.

Love, K., & Singer, M. (1988). Self-efficacy, psychological well-being, job satisfaction and job involvement: A comparison of male and female police officers. Police Studies, 11, 98-102.

Marsden, P. V., & Wright, J. D. (2010). Handbook of survey research (2nd ed.). Bingley, UK: Emerald.

Morabito, M. S., Kerr, A. N., Draine, J., Ottati, V., & Angell, B. (2010). Crisis Intervention Teams and people with mental illness: Exploring the factors that influence the use of force. Crime & Delinquency, 20, 1-21.

Nulty, D. D. (2008). The adequacy of response rates to online and paper surveys: What can be done? Assessment & Evaluation in Higher Education, 3, 301-314.

Ozturk, M. (2011). Teaching style preferences of trainees at police in-service training sessions and differences among demographic groups (Order No. 3482178, Spalding University). Available from ProQuest Dissertations and Theses. Retrieved from http://search.proquest. com.jproxy.lib.ecu.edu/docview/909969177/abstract?accountid=10639

74 Criminal Justice Policy Review 27(1)

Phelan, J., & Link, B. (1998). The growing belief that people with mental illness are violent: The role of the dangerousness criterion for civil commitment. Social Psychiatry and Psychiatric Epidemiology, 33, S7-S12.

Public Health Research Institute. (2005). Evaluation report of NAMI CIT implementa- tion at the Androscoggin County jail. Retrieved from http://www.nami.org/Content/ M i c r o s i t e s 1 8 6 / N A M I _ M a i n e / H o m e 1 7 4 / F A M I L Y _ N e w s l e t t e r - _ W i n t e r _ 2 0 0 6 / CITFinalReportDRAFT_12.31.05.pdf

Schnittker, J. (2000). Gender and reactions to psychological problems: An examination of social tolerance and perceived dangerousness. Journal of Health and Social Behavior, 41, 224- 240.

Severiens, S. E., & Ten Dam, G. T. (1994). Gender differences in learning styles: A narrative review and quantitative meta-analysis. Higher Education, 27, 487-501.

Skeem, J., & Bibeau, L. (2008). How does violence potential relate to Crisis Intervention Team response to emergencies. Psychiatric Services, 59, 201-204.

Stajkovic, A., & Luthans, F. (1998). Self-efficacy and work-related performance A meta-analy- sis. Psychological Bulletin, 124, 240-261.

Steadman, H., Deane, M., Borum, R., & Morrissey, J. (2000). Comparing outcomes of major models of police responses to mental health emergencies. Psychiatric Services, 51, 645- 649.

Strauss, G., Glenn, M., Reddi, P., Afaq, I., Podolskaya, A., Rybakova, T., . . .El-Mallakh, R. S. (2005). Psychiatric disposition of patients brought in by Crisis Intervention Team police officers. Community Mental Health Journal, 41, 223-228.

Taxman, F. S., & Belenko, S. (2012). Implementing evidence-based practices in community corrections and addiction treatment. New York, NY: Springer.

Teller, J., Munetz, M., Gil, K., & Ritter, C. (2006). Crisis Intervention Team training for police officers responding to mental disturbance calls. Psychiatric Services, 57, 232-237.

Teplin, L. (1984). Criminalizing mental disorder: The comparative arrest rate of the mentally ill. American Psychologist, 39, 794-803.

University of New England Center for Health Policy, Planning and Research. (2007). Crisis Intervention Team (CIT) training for correctional officers: An evaluation of NAMI Maine’s 2005-2007 expansion program. Retrieved from http://www.nami.org/Content/Microsites186/ NAMI_Maine/Home174/Criminal_Justice2/CITevaluationReport122107.pdf

Watson, A., Morabito, M., Draine, J., & Otatti, V. (2008). Improving police response to persons with mental illness: A multilevel conceptualization of CIT. International Journal of Law and Psychiatry, 31, 359-368.

Watson, A., Ottati, V., Morabito, M., Draine, J., Kerr, A. N., & Angell, B. (2010). Outcomes of police contacts with persons with mental illness: The impact of CIT. Administration and Policy in Mental Health, 37, 302-317.

Wells, W., & Schafer, J. (2006). Officer perceptions of police responses to persons with a mental illness. Policing: An International Journal of Police Strategies & Management, 29, 578-601.

Wilper, A. P., Woolhandler, S., Boyd, J. W., Lasser, K. E., McCormick, D., Bor, D. H., & Himmelstein, D. U. (2009). The health and health care of US prisoners: Results of a nation- wide survey. American Journal of Public Health, 99, 666-672.

Author Biography

Megan L. Davidson is an assistant professor in the Department of Criminal Justice at East Carolina University. Her research focuses on examining criminal justice responses to special

Davidson 75

populations of offenders, such as those suffering with mental illnesses, substance abuse issues, and/or homelessness. She is particularly interested in evaluating the effectiveness of jail diver- sion programs, corrections-based rehabilitation programs, and reentry initiatives targeting these special populations of offenders. She completed her PhD in Public Affairs with a Criminal Justice concentration at the University of Central Florida. She obtained her BA in psychology and BS/MS in justice administration from the University of Louisville.