The effect of childhood trauma, personal wellness, supervisory working alliance, and organizational factors on vicarious traumatization.

Scholars have identified vicarious traumatization (VT) as one of the most extreme effects of working with traumatized clients; however, not all mental health therapists develop VT (Pearlman & Saakvitne, 1995). Path analytic procedures were used to assess a comprehensive theoretical VT model based on constructivist self-development theory (CSDT). The model explained 46% of the variance in VT in mental health therapists surveyed.

**********

Because the incidence of trauma in the United States has become immense and pervasive, mental health counselors are inevitably exposed to demoralizing stories of trauma, disempowerment, and abuse. According to the National Crime Victimization Survey, in 2008 about 8.4% of Americans reported being victims of a violent personal crime (sexual assault, robbery, or theft; Bureau of Justice Statistics, 2008). Further, an estimated 16.7% to 25% of women will experience a violent sexual assault, 16.7% of women and 10% of men have experienced childhood sexual abuse, and approximately 32 million Americans are survivors of child sexual abuse (Healing Sexual Trauma, 2010; Heppner et al., 1995; Ratna & Mukergree, 1998). Many survivors of violent crime seek treatment for mental health disorders (National Center for Victims of Crime, 2010). As a result, an estimated 82% to 94% clients seeking treatment in community mental health centers have survived some type of traumatic event (Bride, 2004). Most therapists will therefore work with trauma survivors at some point (Bride, 2004; Trippany, White Kress, & Wilcoxon, 2004).

A counselor's empathic connection with the client is critical to the counseling relationship and therapeutic progress; however, empathy for clients leaves those therapists who work with traumatized clients affectively vulnerable (McCann & Pearlman, 1990b; Pearlman & Saakvitne, 1995). This vulnerability may cause them to experience symptoms similar to those of their survivor clients, such as feelings of fear, helplessness, and lack of control (McCann & Pearlman, 1990b). Because trauma work is intense, therapists are more vulnerable to developing such impairments as countertransference, burnout, compassion fatigue, and vicarious traumatization (Bride, 2004; Pearlman & Saakvitne, 1995).

Forms of impairment previously identified failed to encompass the unique manifestation of disruptive psychological effects on those working with traumatized clients (Pearlman & Saakvitne, 1995). To describe the impact, McCann and Pearlman (1990b) introduced the term vicarious traumatization (VT). Pearlman and Saakvitne described VT as the "transformation in the inner experience of the therapist that comes about as a result of empathic engagement with clients' trauma material" (1995, p. 31). Essentially, VT is a shift in the internal experience and psychological wellbeing of the therapist. It encompasses the negative impact of trauma work on the therapist's psychological functioning and describes changes in worldview, identity, values, philosophy of life, and sense of the world as a result of prolonged exposure to clients' experiences of trauma (Pearlman & Saakvitne, 1995; Way, VanDeusen, & Cottrell, 2007).

The impact of VT on a practitioner's psychological functioning and worldview results in both intra- and interpersonal difficulties (McCann & Pearlman, 1990b; Trippany et al., 2004). Practitioners affected experience a wide range of symptoms including loss of energy, difficulty in maintaining interpersonal relationships, cynicism, nightmares, feelings of hopelessness and despair, a disrupted flame of reference (change in worldview), disrupted psychological functioning, increased emotionality, emotional numbness, and even dissociation or depersonalization (Saakvitne & Pearlman, 1995). In sum, VT undermines a practitioner's sense of safety in the world and sense of control over life situations (Pearlman & Saakvitne, 1995). For example, affected practitioners may begin to have difficulty' trusting people they previously trusted and feeling safe in previously unthreatening situations or environments (McCann & Pearlman, 1990b; Trippany et al., 2004). The disrupted worldview caused by the negative impact of VT on the practitioner's sense of safety, and trust can cause interpersonal difficulties for clients, such as relational conflict and strained interpersonal relationships (Saakvitne & Pearlman, 1999).

Because VT affects a therapist's personhood and worldview, it also has a major impact on the therapeutic process. It can lead to compromised therapeutic boundaries, misdiagnosis, diminished ability to attend to client needs, and loss of energy, optimism, and commitment (Sexton, 1999; Trippany et al., 2004). To understand VT and its unique manifestation in practitioners, it is important to understand the constructivist self-development theory (CSDT), which is VT's theoretical basis (McCann & Pearlman, 1990a; Pearlman & Saakvitne, 1995).

CONSTRUCTIVIST SELF-DEVELOPMENT THEORY

CSDT is based on a constructivist foundation postulating that individuals construct personal realities based on the complex cognitive schemas used to interpret and make sense of life experiences (Pearlman & Saakvitne, 1995). In the process of self-development, individuals recreate and restructure their realities and perceptions based on those experiences (McCann & Pearlman, 1990a). Thus, counselors working with traumatized clients actively restructure and recreate their perceptions and realities based on the interaction between their own frames of reference and client stories of trauma (Pearlman & Saakvitne, 1995). After continued exposure to traumatic material, counselors adapt their belief systems and worldviews to extract meaning from these events (McCann & Pearlman, 1990a; Pearlman & Saakvitne, 1995). For example, when, through empathic engagement, practitioners identify with clients who have experienced trauma, they may have difficulty making sense of these experiences, which may lead them to feel vulnerable and unsafe in the world. Thus, therapists develop VT when they are unable to maintain a consistent sense of self and make sense of clients' traumatic stories based on their existing worldview and belief system (McCann & Pearlman, 1990b; Trippany et al., 2004). With its basis in CSDT, VT is a unique construct that differs conceptually from such forms of practitioner impairment as countertransference, burnout, and compassion fatigue (American Counseling Association [AGA], 2003; Figley, 1995; Pearlman & Saakvitne, 1995).

PREVALENCE OF VICARIOUS TRAUMATIZATION

Data on the exact number of practitioners impacted by VT is limited by confusion about what constitutes VT, the lack of specific diagnostic criteria for it, and the fact that many researchers have failed to distinguish VT from other forms of counselor impairment (e.g., Argentero & Setti, 2011 ; Nelson-Gardell & Harris, 2003; Sabin-Farrell & Turpin, 2003). However, the impact of VT on mental health counselors has been described as an occupational hazard and a normal counselor adaptation "to recurrent client-presented traumatic material" (Trippany et al., 2004, p. 32). Despite the lack of specific diagnostic criteria, scholars have identified the Trauma Attachment and Belief Scale (TABS: Pearlman, 2003) as the instrument of choice to assess VT (Traumatic Stress Institute [TSI], 2010; Trippany et al., 2004). One team of researchers using the TABS has reported that VT prevalence in counselors is 45.9% (Dunkley & Whelan, 2006); unfortunately, no other prevalence statistics have been reported.

A COMPREHENSIVE VT MODEL

Based on CSDT, Pearlman and Saakvitne (1995) proposed that therapist, work, and supportive factors contribute to the emergence of VT in mental health therapists; however, no studies to date have examined the combined influence of these variables. Researchers have explored the influence of childhood trauma (Dunkley & Whelan, 2006; Pearlman & Mac Ian, 1995; Schauben & Frazier, 1995; Way et al., 2007); clinical supervision (Bober & Regehr, 2005; Dunkley & Whelan, 2006; Hunter & Schofield, 2006); personal wellness (Brady, Guy, Polestra, & Brokaw., 1999; Bride, 2004; Schauben & Frazier, 1995): and organizational factors (Linley & Joseph, 2007; Pearlman & Mac Ian, 1995; Schauben & Frazier, 1995) on the development of VT, but researchers have not used a comprehensive CSDT-based model to examine the relationships between and the combined impact of these variables.

Childhood trauma: Although several researchers have studied the influence of a therapist's experience of childhood trauma on VT (see, e.g., Adams, Matto, & Harrington, 2001; Bride, 2004; Pearlman & Mac Ian, 1995; Schauben & Frazier, 1995), the results have been inconclusive. Some reported significant positive correlations between VF and childhood trauma (Bride, 2004; Pearlman & Mac Ian, 1995), and others found no relationship (Adams et al., 2001; Dunkley & Whelan, 2006; Schauben & Frazier, 1995). Although a history of childhood trauma does seem to contribute to the development of VT in some therapists, it does not alone explain its incidence.

Personal wellness: Theorists have proposed that personal wellness and self-care may mediate the relationship between childhood trauma history and VT (McCann & Pearlman, 1990b). Pearlman and Saakvitne (1995) proposed that a holistic wellness approach helps to prevent and alleviate symptoms of VT. Both qualitative and quantitative studies have found that consistent participation in wellness activities decreased vulnerability to VT (Bober, Regehr, & Zhou, 2006; Brady et al., 1999; Bride, 2004; Hunter & Schofield, 2006; Schauben & Frazier, 1995). Although researchers have found that self-care and wellness activities mediate the relationship between childhood trauma history and development of VT, the influence of a holistic approach to wellness on VT has not been researched.

Supervisory working alliance: McCann and Pearlman (1990b) suggested that clinical supervision could mediate the relationship between childhood trauma history and VT because it helps therapists to avoid professional isolation, normalizes their reactions to trauma work, and promotes self-awareness (Sommer, 2008). Both quantitative and qualitative studies have found that those who reported regular participation in clinical supervision and having a positive supervisory relationship were less vulnerable to developing VT (Bober & Regehr, 2005; Dunkley & Whelan, 2006; Hunter & Schofield, 2006; Pearlman & Mac Ian, 1995; Sommer & Cox, 2005).

Organizational factors: Organizational culture, organizational support, work environment, workload, and caseload are factors thought to impact the development of VT (Bell, Kulkarni, & Dalton, 2003; McCann & Pearlman, 1990b; Pearlman & Saakvitne, 1995). Organizational factors have been found to directly influence a therapist's vulnerability or resilience toward developing VT (Pearlman & Saakvitne, 1995). Although there is a gap in the literature examining the influence of organizational factors on VT, theoretical support for their influence is strong (Bell et al., 2003; Neumann & Gamble, 1995; Pearlman & Saakvitne, 1995).

According to CSDT, the development of VT is influenced by a combination of therapist, work, and supportive factors (Harrison & Westwood, 2009; Trippany et al., 2004). There is evidence that a therapist's experience of childhood trauma may directly influence the emergence of VT, but personal wellness and a positive supervisory working alliance may have mediating functions (Pearlman & Mac Ian, 1995; Pearlman & Saakvitne, 1995). There is also evidence that organizational factors may directly affect the development of VT (Brady et al., 1999; Linley & Joseph, 2007; Pearlman & Mac Ian, 1995; Schauben & Frazier, 1995). A supportive organizational culture, personal wellness, and a positive supervisory working alliance are predicted to increase resilience toward developing VT, and workload and childhood trauma history are predicted to increase vulnerability toward developing VT.

Purpose

The purpose of this study was to test a comprehensive model of factors contributing to the development of VT in mental health therapists. A path model based on CSDT was created a priori to describe relationships between variables (McCann & Pearlman, 1990b; Pearlman & Saakvitne, 1995). The model described the theoretical relationships between childhood trauma, personal wellness, supervisory working alliance, organizational culture, workload, and VT and predicted that childhood trauma and workload would have a positive direct effect on VT, organizational culture a negative direct effect, and personal wellness and supervisory working alliance partial negative mediating effects (Figure 1). Alternative path analysis models were considered, but the literature did not provide strong enough support for any alternative.

[FIGURE 1 OMITTED]

METHODS

Participants

The study population was comprised of mental health counselors working full-time in community mental health agencies because of their exposure to traumatized clients and experience within an organization (Bride, 2004; Pearlman & Saakvitne, 1995). In a community mental health agency a cluster of practitioners serve clients from the community. Pearlman and Saakvitne (1995) proposed that practitioners working in community agencies may be more likely to develop VT because they cannot control organizational factors and caseloads. Unlike practitioners working in private practice, they are more likely to be exposed to traumatized clients because they cannot screen out clients based on the nature or severity of their presenting problem, and an estimated 93% to 94% of clients in community mental health centers have experienced trauma (Bride, 2004). Participants in the study represented a stratified, random, cluster sample of mental health therapists in urban (population >100,000); suburban (population < 100,000 and within 20 miles of an urban center); and rural (population < less than 100,000 and more than 20 miles from an urban center) communities in the Rocky Mountain region of the United States (Colorado, Idaho, Montana, New Mexico, Utah, Wyoming). All participants were full-time counselors who spent at least 60% of their time in direct client service and received at least one hour of hierarchical supervision per month.

The community mental health centers in the region were divided into three strata according to the size of community. From a database provided by the Substance Abuse and Mental Health Services Administration (SAMHSA, 2009), researchers contacted 33 randomly selected community mental health centers (8 urban, 10 suburban, and 15 rural). The final sample was comprised of therapists from 13 community mental health centers (3 urban, 5 suburban, and 5 rural) and the overall response rate was 39.4%. At these centers, 134 mental health counselors volunteered to participate. Three survey packets were excluded from the analysis due to missing data, resulting in a final sample of 131 and a completion rate of 97.8%.

Of the 131 participants, 83 were female (63.4%) and 48 male (36.6%). Most were Caucasian (n = 106; 80.9%), with others being Hispanic (n = 19; 14.5%), American Indian (n = 1; 0.8%); or Multiethnic/Multiracial (n = 2; 1.5%); 3 participants (2.3%) indicated Other, without further specification. Participants ranged in age from 25 to 71 (M = 42.18; SD = 11.00). They were licensed as clinical social workers (n = 50; 42.7%); marriage and family therapists (17 = 11; 8.4%); professional counselors (n = 40; 30.5%); or psychologists (n = 7; 5.3%) or were unlicensed professionals (n = 17; 13.0%). Their clinical experience ranged from 2 to 33 years (M = 10.31; SD = 8.11). Practitioners' caseloads averaged 39.11 (SD = 25.87) and ranged from 8 to 126. Of clients on current caseloads, an average of 50.2% had experienced a significant trauma and had either a primary or secondary treatment goal of addressing its impact on their lives.

Instruments

Childhood trauma: The Childhood Trauma Questionnaire (CTQ; Bernstein & Fink, 1998) is a 28-item self-report survey designed to assess history of childhood abuse (emotional, physical, and sexual) or neglect (emotional and physical). It measures the incidence of childhood trauma using a five-point frequency scale ranging from 1 (never true) to 5 (very often true). The overall CTQ score is used to assess childhood trauma. Bernstein et al. (1994) reported that initial reliability of CTQ scores and the alpha coefficient for the total score was .95. A significant test-retest reliability coefficient for intervals ranging from 1.6 to 5.6 months was reported as .86 for the total score, suggesting respondent reports of trauma were consistent over time (Bernstein & Fink, 1998). Bernstein and Fink (1998) found content validity because the content domains were written to reflect the domains of childhood trauma described in the maltreatment literature. The internal consistency coefficient (Cronbach's alpha) for the present study was .91.

Personal wellness. The Five Factor Wellness Inventory--Form A (5F-Wel-A; Myers & Sweeney, 2005) was used to measure personal wellness. The 5F-Wel-A is a comprehensive 73-item measure of personal wellness based on a holistic Adlerian view of optimal health (Myers & Sweeney, 2005). Participants answered items on a 4-point Likert-type scale ranging from 1 (strongly agree) to 4 (strongly disagree). The total wellness score was used to measure personal wellness in the hypothesized model. The manual for the 5F-Wel-A reported the test is both reliable and valid (Myers & Sweeney, 2005). The 5F-Wel-A demonstrates internal consistency; based on a 5-year study of 2,093 participants, the authors reported an alpha coefficient of .98 for total wellness. Similarly, Hattie, Myers, and Sweeney (2004) reported an internal consistency score of .94 for total wellness. Myers and Sweeney (2005) also claimed convergent and divergent validity of the 5F-Wel-A factors (subscales) based on findings that each factor was able to "discriminate among a variety of populations related to these variables" (p. 16). In this study, the Cronbach's alpha reliability estimate was .93 for total wellness.

Supervisory working alliance: Researchers designed the Supervisory Working Alliance Inventory--Supervisee Form (SWAI; Efstation, Patton, & Kardash, 1990) to measure the quality of the counseling supervision relationship. The 19-item questionnaire is designed to measure the supervisee's perception of the quality of the supervisory relationship from responses to a 7-point, Likert-type scale ranging from 1 (almost never) to 7 (almost always). A factor analysis of the supervisee form identified two subscales, rapport and client focus. The rapport subscale score was used to measure supervisory working alliance in this study. To demonstrate the internal consistency of the scale, Efstation et al. (1990) reported a Cronbach's alpha coefficient of .90 for the rapport subscale, indicating adequate reliability. Compared to other established measures, the SWAI-Supervisee form demonstrated convergent and divergent validity. Its subscales correlated with similar subscales on the Supervisory Styles Inventory (SSI), indicating convergent validity (Efstation et al., 1990). For the present study, the Cronbach's alpha coefficient for the rapport subscale was .95.

Organizational culture: The Job Satisfaction Survey (Spector, 1985) is a 36-item self-report survey designed to assess employee altitudes about her or his job. Participants were required to rate their attitudes-related items on a 6-point Likert-type scale ranging from 1 (disagree very much) to 6 (agree very much). Higher scores indicate higher job satisfaction. Internal consistency reliability scores were reported based on a normative sample of 2,870 (Spector, 1985); the Cronbach's alpha for the total score was .91. To determine discriminant and convergent validity, Spector (1985) employed a multitrait-multimethod analysis of the Job Satisfaction Survey and the previously established Job Descriptive Index. Equivalent subscales of both measures had significant validity correlations, ranging from .61 to .80 (Spector, 1985). In the present study, the total Job Satisfaction Survey score was used to measure organizational culture, and the Cronbach's alpha reliability estimate was .92.

Workload: The Quantitative Workload Inventory (QWI; Spector & Jex, 1997) was used to measure workload from the therapists' perspective (Spector & Jex, p. 358). The 5-item QWI uses a 5-point Likert-type scale ranging from 1 (less than once per month or never) to 5 (several times per day); higher scores reflect higher perceived workload. Spector and Jex (1998) indicated internal consistency reliability for the QWI and reported an average coefficient alpha of .82 from 15 studies. However, they found that determining convergent and discriminant validity is difficult because no other instruments measure this construct (Spector & Jex, 1997). The Cronbach's alpha reliability estimate for the QWI in the present study was .80.

Vicarious traumatization: The Trauma and Attachment Belief Scale (TABS; Pearlman, 2003) is an 84-item self-report questionnaire based on CSDT. The TABS measures an individual's beliefs about self and others related to the five psychological needs (safety, intimacy, trust, control, and esteem) that are commonly altered by exposure to trauma (Pearlman, 2003). Participants answer items on a 6-point Likert-type scale ranging from 1 (disagree strongly) to 6 (agree strongly). TABS results consist of 10 subscale scores and a total score. Subscales represent disruptions in beliefs about self and others related to the five psychological needs, including self-safety ("I believe I am safe"); other-safely ("I can't stop worrying about others' safety"); self-trust ("I don't trust my instincts"); other-trust ("trusting people is not smart"); self-esteem ("I'm not worth much"); other-esteem (I often think the worst of others"); self-intimacy ("I feel hollow inside when I am alone"); other-intimacy ("I don't feel much love from anyone"); self-control ("I feel like I can't control myself"); and other-control ("I often feel people are trying to control me"). Higher subscale and total scores indicate greater disruption in beliefs about safety, trust, esteem, intimacy, and control and are therefore associated with high levels of VT.

Pearlman (2003) provided evidence for the face validity of the instrument in that items on it ask respondents directly about their beliefs related to the five psychological needs. She also provided an argument for its construct validity through the use of interscale correlations, factor analysis, and correlations with other instruments measuring similar constructs (e.g., the Trauma Symptom Inventory). Though reliability and validity research on the current version of TABS is limited, it is very similar to the 1994 version, the Traumatic Stress Institute Belief Scale-Revision L (TSI-BSL). TSI-BSL items were revised for readability to form the current TABS instrument. In examining the reliability and validity of the TSI-BSL, Jenkins and Baird (2002) reported a Cronbach's alpha of .95 for the total score and .62 to .83 for the 10 subscales. They also reported concurrent and discriminant validity for the measure when it was correlated with other measures. In the current study, the total TABS score was used to measure VT; the Cronbach's alpha reliability estimate for the total score was .95.

Procedure

After receiving institutional review board approval, the lead researcher contacted the selected community mental health centers by telephone to explain the research. When a center agreed to participate, the researcher made an appointment to go to the organization to administer the surveys and provide an in-service training. Clinicians from the center who agreed to participate signed a consent form, which was collected separately to ensure confidentiality, before participating. They then completed the surveys, which were administered in random order to help control for score differences based on testing fatigue. The researcher then provided a one-hour in-service on managing the impact of trauma work for all employees, whether or not they had participated in the survey.

RESULTS

This study used the weighted least squares estimation method to estimate path coefficients in the path model using LISREL 8.80, which is recommended over other estimation methods when variables are measured using ordinal data (i.e., Likert-type scales) rather than interval data (Joreskog, 2005; Kline, 2005). Solutions (path coefficients) were then standardized for ease of interpretation.

Descriptive statistics for each variable in the path model are described in Table 1. A correlation matrix of the variables was analyzed before the path analysis was conducted to assess for multicollinearity. Scores for the Pearson product-moment correlations are presented in a matrix in Table 2. Many of the relationships between variables in the model were statistically significant and are considered small ([greater than or equal to] 0.10) to medium ([greater than or equal to] 0.30) effect sizes in counseling research; however, statistical significance may be a function of the large sample size required for path analytic procedures (Granello, 2007; Kline, 2005). Although the variables are correlated, bivariate correlations among all variables are less than r = [absolute value of .85]; therefore, the correlations do not violate the path analytic assumption of multicollinearity (Kline, 2005; Weston & Gore, 2006).

The researchers used the following fit indices to determine how well the model fits the data: model chi square ([chi square]), Bentler comparative fit index (CFI), incremental fit index (IFI), standardized root mean square residual (SRMR), and Steiger-Lind root mean square error of approximation (RMSEA). The [chi square] was used to test the null hypothesis that the model fits the data well, and a statistically significant [chi square] resulted in rejection of the null hypothesis, indicating the model does not fit the data well (Kline, 2005; Thompson, 2000; Weston & Gore, 2006). The CFI and IFI were also used to test the goodness of fit of the model. Scores for the CFI range from 0 to 1.0, with values closer to 1.0 indicating that the model fits the data well (Hu & Bentler, 1999; Kline, 2005; Weston & Gore, 2006). The IFI is a non-normed fit index; therefore, scores can range from 0 to larger than 1.0, with values closer to 1.0 indicating the data fit the model well (Hu & Bentler, 1999). To determine model fit, researchers used the cutoff score recommended for samples of less than 500, which is [greater than or equal to] .90 for both the CFI and the IFI (Kline, 2005; Weston & Gore, 2006).

Researchers also used the SRMR and RMSEA to assess the badness of fit of the model (Kline, 2005). Scores for the SRMR and RMSEA range from 0 to 1.0 with scores closer to 0 indicating better model fit (Kline, 2005; Thompson, 2000). In this study, [less than or equal to] .10, the common cutoff criterion for samples of less than 500, was used for both the SRMR and RMSEA (Kline, 2005; Thompson, 2000; Weston & Gore, 2006). All of the fit statistics ([chi square], CFI, IFI, SRMR, and RMSEA) were used to determine the degree to which the data fit the hypothesized path model. Before data for each research question were analyzed, overall model fit was considered.

Overall Model Fit

To assess the overall fit of the path model, researchers examined multiple fit indices. The [chi square] was used to assess for model misspecification and was statistically significant ([chi square] = 26.41, df = 12, p <0.05), indicating that the model does not fit the data well, since [chi square] assesses badness of fit; however, because [chi square] is heavily influenced by sample size, it is not a good indicator of fit when data are ordinal (Kline, 2005) and must be examined in conjunction with other fit statistics (Martens, 2005). The SRMR and the Steiger-Lind RMSEA were also used to assess badness of fit. The SRMR was found to be 0.30 (greater than the recommended cutoff score of [less than or equal to] .10 for samples of less than 500), indicating poor model fit (Kline, 2005; Weston & Gore, 2006). Similarly, the RMSEA of 0.13 also suggested the model does not fit the data well. Thus, all three fit indices used to assess badness of fit found the model did not fit the data well.

The CFI, the goodness of fit index (GFI), and the IFI were goodness of fit statistics used in conjunction with the badness of fit statistics to assess the overall fit of the model. The CFI, which compares the model to a baseline or null model, was 1.0; since this is greater than the recommended cutoff score of 0.90, the model fit the data well (Martens, 2005; Weston & Gore, 2006). In the present study the GFI was .94, which also indicates good model fit (Martens, 2005). Filmily, the IFI assesses model fit while compensating for sample size; the IFI of 1.10 for this study was also greater than the recommended cutoff score, 0.90, which again indicates good model fit (Hu & Bentler, 1999). Therefore, the goodness of fit indices indicated that the model fits the data well. Examining the fit indices simultaneously suggests that certain aspects of the model fit the data well, and other aspects do not (Klein, 2000).

Research Question One

The hypothesis was that there would be a positive direct effect of childhood trauma and negative partial mediating effects of personal wellness and supervisory working alliance on VT. Results of the path analysis provided partial support for the hypothesis (see Figure 1). Although the direction of relationships between variables was consistent with the hypothesis, not all the relationships were statistically significant. The direct effect of childhood trauma (0.20) and the partial mediating effect of personal wellness on VT (-0.58) were statistically significant. Personal wellness partially mediated the relationship between childhood trauma and VT. The partial mediating effect of supervisory working alliance on VT (-0.06) was not statistically significant. These results suggest that the direct effect of childhood trauma and the partial mediating effect of personal wellness are aspects of the model that fit the data well, whereas the partial mediating effect of supervisory working alliance is an aspect of the model that does not fit the data well.

Research Question Two

The hypothesis that organizational culture would have a negative direct effect on VT was not supported. Although the direction of the effect was consistent with the hypothesis, the direct effect of organizational culture on VT (-0.13, p = .07) was not statistically significant. These results suggest the organizational culture is an aspect of the model that did not fit the data well (Kline, 2005; Weston & Gore, 2006).

Research Question Three

The hypothesis that workload would have a positive direct effect on VT was also not supported. Again, the direction of the relationship was consistent with the hypothesis, but the direct effect of workload on VT (0.08) was not statistically significant. Thus, workload is another aspect of the model that did not fit the data well (Weston & Gore, 2006).

Overall Variance Explained by the Model

Examining the squared multiple correlation coefficient ([R.sup.2]) for the endogenous variable indicates the proportion of total variance in each variable that is explained by the model (Kline, 2005). The path model accounted for 46% of the variance in VT, which in the fields of mental health and education is considered a large effect ([greater than or equal to] 35%) and practically significant (Granello, 2007; Thompson, 2002).

DISCUSSION

When examining model fit, it is important that all fit indices are examined simultaneously because each measures a different aspect of model fit (Klem, 2000; Kline, 2005). The results related to overall model fit were contradictory: the badness of fit indices ([chi square], SRMR, and RMSEA) suggested the model did not fit the data well but the goodness of fit indices (CFI, CFI, and IFI) suggested good model fit. The insignificant badness of fit indices suggested model misspecification or that some aspects of the model did not fit the data well, but the statistically significant goodness of fit indices indicated the tested model fit the data better than a null model. Simultaneous evaluation of the fit indices indicates that CSDT seems to have some relevance in explaining the development of VT in mental health counselors, but it does not provide a comprehensive explanation.

Relationships Among Variables

To find which aspects of the model fit the data well, researchers examined the path coefficients. Because the Cronbach's alpha coefficients of the six scales used to assess variables in the model are considered very good to excellent (.80 to .95), examining path coefficients to assist in interpretation is appropriate (Kaplan, 2000).

The model tested the CSDT claim that a history of childhood trauma affects the development of VT but may be partially mediated by personal wellness and a strong supervisory working alliance. The childhood trauma and personal wellness constructs behaved in ways that are consistent with theory, but supervisory working alliance did not. Therapists in the current study who reported more frequent experiences of childhood trauma were more likely to also experience symptoms of VT. Also, those who engaged in more personal wellness activities benefited from the partial mediating effect of wellness on VT when they had experienced childhood trauma. These results, which aligned with CSDT, indicated that clinicians who reported a history of childhood trauma also reported more cognitive distortions associated with VT (Pearlman & Mac Ian, 1995; Way et al., 2007). The findings are consistent with the assertions of Pearlman and Saakvitne (1995) that engagement in personal wellness practices decreases vulnerability to VT and partially mediates the effect of childhood trauma on VT.

The findings diverged from theory and previous research on the construct of supervisory working alliance, which did not significantly mediate cognitive distortions associated with VT in the presence of other model variables. Previous researchers found that a strong supervisory working alliance was important for decreasing therapist vulnerability to VT (e.g., Bober & Regehr, 2005; Hunter & Schofield, 2006; Pearlman & Mac Ian, 1995). The current study did not support this assertion because supervisory working alliance did not have a significant mediating effect on VT. This may be a result of the limited range of scores on the SWAI: Because most participants reported a strong supervisory working alliance, clinicians who experienced poor or inadequate relationships with supervisors were not represented by the sample.

Finally, the model tested the CSDT claim that a positive organizational culture decreases therapist vulnerability to VT, but high workload (collective work responsibilities) increases this vulnerability (Pearlman & Saakvitne, 1995; Trippany et al., 2004). This study, the first to examine the effect of organizational factors on VT, found that organizational culture and workload did not significantly affect development of symptoms of VT when other variables were present. Thus, the results did not align with CSDT, and the theory that organizational culture and workload directly influence a therapist's vulnerability to VT was not supported (e.g., Bell et al., 2003, Neumann & Gamble, 1995; Pearlman & Saakvitne, 1995; Sexton, 1999). When examining scores on the QWI, which was used to assess workload, it is important to note the limited variability in participant scores; most participants (n = 120, 91.6%) reported experiencing heavy workloads according to the scale's cutoff score (total scores [greater than or equal to] 15; Spector & Jex, 1998). The clustering of scores on this scale might explain why this variable was not significant. It may be necessary to study practitioners employed in a variety of mental health settings to better assess the significance of this variable, taking into account that counseling practitioners in community mental health settings consistently report heavier workloads (Dadich & Muir, 2009).

Overall Variance Explained by the Model

Besides examining the overall fit of the model, to determine its practical significance it is also important to evaluate the proportion of overall variance explained in the endogenous variables (Weston & Core, 2006). According to the results of the path analysis, CSDT explained 46% of the variance in VT. In the social sciences, this is considered a large effect (Granello, 2007); however, each variable must be examined in conjunction with the current literature because effect size alone does not determine practical significance (Granello, 2007; Thompson, 2002).

In this study CSDT failed to comprehensively explain VT in mental health counselors. Some aspects of the model (childhood trauma and personal wellness) fit the data well; others (supervisory working alliance, organizational culture, and workload) did not. Nevertheless, the results have practical implications for therapists, educators, and supervisors in terms of decreasing the impact of VT.

Implications

Mental health professionals. This study reaffirmed that a personal history of childhood trauma can have a direct effect on the development of VT. Theorists have attributed increased vulnerability toward developing cognitive distortions associated with VT to being reminded of one's own trauma history upon exposure to client stories of trauma (McCann & Pearlman, 1990b; Pearlman & Saakvitne; 1995). These findings have substantive implications for therapists who have experienced childhood trauma. First, it is important that they recognize their possible increased vulnerability to VT. Awareness of the potential impact of their own history of trauma will enable therapists to recognize when emotions and memories related to their own experiences have been triggered and seek help to manage shifting beliefs about the goodness of self, others, and the world associated with VT. A therapist who becomes aware of altered beliefs can engage in activities to mediate the development of VT. The American Mental Health Counselors Association (AMHCA, 2010) mandates that counselors "recognize that their effectiveness is dependent on their own mental and physical health" and seek help and supervision when their competence begins to be affected (p. 9). Thus, practitioners who become aware of altered belief systems related to VT should engage in activities to mediate the effects on their personal and professional functioning.

In the present study, personal wellness had a strong partial mediating effect on VT; counselors who reported a personal history of childhood trauma were able to partially mediate the development of VT by engaging in activities to promote personal wellness. Although it cannot be inferred directly from these results, engaging in personal wellness activities has also been found to decrease the impact of VT in counselors after it develops (Brady et al., 1999; Harrison & Westwood, 2009; Saakvitne, 2002; Schauben & Frazier, 1995). Because the study measured the partial mediating effect of a holistic approach to wellness (Harrison & Westwood, 2009; Myers & Sweeney, 2005), counselors are encouraged to engage in a variety of practices to promote physical, emotional, cognitive, spiritual, and interpersonal wellness.

Regular participation in wellness activities had a significant negative correlation to VT, which suggests that counselors who engaged in wellness practices more often experienced fewer cognitive distortions related to VT. Besides engaging in wellness activities to mediate the effect of childhood trauma, practitioners should work to decrease their vulnerability to VT by creating a wellness plan to follow regularly. Such a plan should include activities to address physical wellness (e.g., eating healthily, getting adequate sleep, exercising regularly); emotional wellness (e.g., maintaining a sense of humor, engaging in activities to promote self-awareness); cognitive wellness (e.g., engaging one's imagination, building problem-solving skills); spiritual wellness (e.g., attending religious services, meditating); and interpersonal wellness (e.g., fostering interpersonal relationships). Since this was the most significant effect in the model and accounted for the most variance in VT, engaging in personal wellness activities seems to be essential to decreasing counselors' vulnerability to VT.

Although supervisory working alliance did not partially mediate the effect of childhood trauma on VT, the two variables were negatively correlated, indicating a possible relationship between the constructs that the present model does not account for. Developing a strong supervisory working alliance may help practitioners decrease their vulnerability to VT, although it is not able to partially mediate the effect of childhood trauma. Thus, practitioners should seek to form a close working alliance with supervisors to provide a context in which they can explore their reactions to client trauma material and avoid professional isolation (Knight, 2004; Pearlman & Mac Ian, 1995; Sommer, 2008).

Finally, organizational culture and workload did not have direct effects on VT, but in this study, VT was significantly positively correlated with workload (r = .22, p < 0.05), indicating that practitioners reporting a greater workload also reported higher levels of VT. Also, VT was negatively correlated with organizational culture (r = -.19, p < 0.05), which suggests that therapists who are more satisfied with their jobs as a result of a positive organizational culture also reported lower levels of VT. Based on these findings, therapists might advocate for changes in workload and organizational culture, which may decrease their vulnerability to VT. For example, those working in community agencies can advocate for a more manageable workload, increased opportunities for promotion, administrative support, contingent rewards, increased communication, and more opportunities to participate in activities to develop relationships with colleagues (Bell et al., 2003; Trippany et al., 2004).

Educators: University faculty are in a unique position to introduce therapists-in-training to the risk of professional impairment. Many accreditation standards (e.g., American Psychological Association: APA, 2007; Council for the Accreditation of Counseling and Related Educational Programs: CACREP, 2009) and professional practice standards (American Counseling Association: ACA, 2005; AMHCA, 2010; APA, 2010) emphasize the necessity of educating therapists-in-training about the potential for professional impairment and promoting personal wellness. Recommendations for introducing students to impairment and wellness issues include leading discussions about the possibility that a personal history of trauma will make them more vulnerable to VT and designing activities to help them identify tools to help mediate the potentially negative impact of working with traumatized clients (i.e., personal wellness, self-care).

Because personal wellness had a strong partial mediating effect on VT, it is essential for educators to promote wellness in students throughout their training. Modeling wellness, constructing a wellness course, and requiring students to develop holistic wellness plans have been found to increase wellness in counseling students (e.g., Myers, Mobley, & Booth, 2003; Roach & Young, 2007). Educators can require students to draft a personal wellness plan during an orientation course and evaluate and revise the plan at other critical points in the program (i.e., practicum and internship).

As gatekeepers for the mental health profession, faculty are responsible for recognizing impairment in trainees and taking steps to mediate it (ACA, 2005). Impairment can be recognized during practicum and internship experiences. Early recognition of impairment coupled with discussions of impairment and wellness throughout the training program can help therapists-in-training to build the skills necessary to mediate the potentially negative impact of trauma work on new professionals.

Supervisors. Like educators, supervisors can promote wellness and self-care in their supervisees. Specifically, they can encourage supervisees to draft and consistently follow a comprehensive wellness plan. To emphasize the importance of personal wellness, supervisors can solicit information about supervisee struggles with maintaining their wellness plans and also share their own struggles with maintaining personal wellness. In general, supervisors can promote wellness and self-care by modeling them, encouraging supervisees to adhere to a holistic wellness plan, and providing a means of accountability for therapists to follow through on self-care activities (Sommer, 2008).

Limitations

Although precautions were taken to minimize limitations, the following may be considered threats to internal validity:

As is common in social science research, the study used self-report instruments to measure constructs in the proposed model. These measures are susceptible to social desirability bias, especially when used to gather data on belief systems, attitudes, or objective measures of behavior (Gall, Gall, & Borg, 2007). Also, the use of Likert-type scales to measure constructs may be considered a limitation because participants may have different interpretations of points on the scale (Gall et al., 2007).

Sample size is another possible limitation; the 131 participants met the 10:1 rule of thumb (10 participants per free model parameter) and were considered a medium sample; however, sample sizes larger than 200 are preferable when using path analytic procedures (Kline, 2005). Additionally, mixed or contradictory results among fit indices are more likely with smaller sample sizes; increasing sample size may produce more definitive results (Kline, 2005; Weston & Gore, 2006).

Finally, the response rate of community mental health centers contacted to participate in this study was 39.4%, which may be a limitation because practitioners employed in centers that participated may differ from those employed in centers that did not.

DIRECTIONS FOR FUTURE RESEARCH

The results and the limitations of this study provide several directions for research, including assessing the relevance of CSDT to the development of VT by testing other models. First, assessing a similar CSDT model but using a larger sample size of therapists working in community mental health centers may provide more conclusive results (Klein, 2000; Weston & Gore, 2006). Second, reorganizing the model using similar constructs might be warranted. For example, it might be more appropriate to include supervisory working alliance in the model as an exogenous rather than an endogenous or partial mediating variable to examine its direct effect on VT. Because more parsimonious models are preferable when using path analytic procedures, removing from the model variables that have small effects (workload or organizational culture) might also provide an avenue for future research (Kline, 2005). Third, comparing the current model to alternative models with samples that have more diverse workloads might provide additional perspective on the validity of the model and CSDT.

CONCLUSION

This study examined a comprehensive theoretical model for the development of VT based on CSDT. While CSDT failed to provide a comprehensive framework for VT, the results explained 46% of the variance in the development of VT in mental health therapists surveyed. Childhood trauma and personal wellness had significant effects on VT, but the effects of supervisory working alliance, organizational culture, and workload were not statistically significant. Examining these results in the context of the literature elicited practical implications for therapists, educators, and supervisors for decreasing the impact of VT in community mental health therapists.

REFERENCES

Adams, K. B., Matto, H. C., & Harrington, D. (2001). The Traumatic Stress Institute Belief Scale as a measure of vicarious trauma in a national sample of clinical social workers. Families in Society, 82, 363-371.

American Counseling Association. (2003). ACA's taskforce on counselor wellness and impairment. Retrieved from http://www.counseling.org/wellness_taskforce/tf_definitions.htm

American Counseling Association (2005). ACA code of ethics. Retrieved from http://www.counseling.org

American Mental Health Counselors Association (2010). Principles for AMHCA code of ethics. Retrieved from www.amhca.org/assets/news/AMHCA_Code_of_Ethics_2010_w_pagination_ cxd_51110.pdf

American Psychological Association (2007). Guidelines and principles for accreditation of programs in professional psychology. Retrieved from www.apa.org/ed/accreditation

American Psychological Association (2010). Ethical principles of psychologists and code of conduct. Retrieved from http://www.apa.org/ethics/code/index.aspx

Argentero, P., & Setti, I. (2011). Engagement and vicarious traumatization in rescue workers. International Archives of Occupational & Environmental Health, 84, 67-75.

Bell, H., Kulkarni, S., & Dalton, L. (2003). Organizational prevention of vicarious trauma. Families in Society, 84, 463-470.

Bernstein, D. P., & Fink, L. (1998). Childhood trauma questionnaire. San Antonio, TX: Psychological Corporation.

Bober, T., & Regehr, C. (2005). Strategies for reducing secondary or vicarious trauma: Do they work. Brief Treatment and Crisis Intervention, 6, 1-9.

Bober, T., Regehr, C., & Zhou, Y. R. (2006). Development of the coping strategies inventory for trauma counselors. Journal of Loss and Trauma, 11, 71-83.

Brady, J. L., Guy, J. D., Polestra, P. L., & Brokaw, B. F. (1999). Vicarious traumatization, spirituality, and the treatment of sexual abuse survivors: A national survey of women psychotherapists. Professional Psychology: Research and Practice, 30, 386-393.

Bride, B. (2004). The impact of providing psychosocial services to traumatized populations. Stress, Trauma, and Crisis, 7, 29-46.

Bureau of Justice Statistics (2008). National Crime Victimization Survey 2008. Retrieved from http://bjs.ojp.usdoj.gov/index.cfm?ty=pbdetail&iid=2218

Council for Accreditation of Counseling and Related Educational Programs. (2009). 2009 standards. Retrieved January 11, 2009 from http://www.cacrep.org/2009standards.pdf.

Dadich, A., & Muir, K. (2009) Tricks of the trade in community mental health research: Working with mental health services and clients. Evaluation and the Health Professions, 32, 38-58.

Dunkley, J., & Whelan, T. A. (2006). Vicarious traumatization in telephone counselors: Internal and external influences. British Journal of Guidance & Counselling, 34, 451-469.

Efstation, J. F., Patton, M. J., & Kardash, C. M. (1990). Measuring the working alliance in counselor supervision. Journal of Counseling Psychology, 37, 322-329.

Figley, C. (Ed.). (1995). Compassion fatigue: Coping with secondary traumatic stress disorder in those who treat the traumatized. New York, NY: Brunner/Mazel, Publishers.

Gall, M. D., Gall, J. P., & Borg, W. R. (2007). Educational research: An introduction (8th ed.). Boston, MA: Pearson.

Granello, D. H. (2007). Publishing quantitative manuscripts in counselor education and supervision: General guidelines and expectations. Counselor Education and Supervision, 47, 66-75.

Harrison, R. L., & Westwood, M. J. (2009). Preventing vicarious traumatization of mental health therapists: Identifying protective practices. Psychotherapy Theory, Research, Practice, Training, 46, 203-219.

Healing Sexual Trauma (2010). Sexual assault statistics. Retrieved from http://www.healing sexualtrauma.org/resources/statistics/

Heppner, M. J., Good, G. E., Hillenbrand-Gunn, T. L., Hawkins, A. K., Hacquard, L. L., Nichols, R. K., ... Brock, K. J. (1995). Examining sex differences in altering attitudes about rape: A test of the elaboration likelihood model. Journal of Counseling and Development, 73, 640-747.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indices in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55.

Hunter, S. V., & Schofield, M. J. (2006). How counselors cope with traumatized clients: Personal, professional and organizational strategies. International Journal for the Advancement of Counselling, 28, 121-138.

Joreskog, K. G. (2005). Structural equation modeling with ordinal variables using LISREL. Retrieved April 17, 2009 from http://www.ssicentral.com/lisrel/techdocs/ordinal.pdf

Kaplan, D. (2000). Structural equation modeling foundations and extensions. Thousand Oaks, CA: Sage Publications, Inc.

Klem, L. (2000). Structural equation modeling. In L. G. Grimm, & P. R. Yarnold (Eds.), Reading and understanding more multivariate statistics (pp. 227-260). Washington, DC: American Psychological Association.

Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). New York, NY: The Guilford Press.

Knight, C. (2004). Working with survivors of childhood trauma: Implications for clinical supervision. The Clinical Supervisor, 23, 81-105.

Linley, P. A., & Joseph, S. (2007). Therapy work and therapists' positive and negative well-being. Journal of Social and Clinical Psychology, 26, 385-403.

Martens, M. P. (2005). The use of structural equation modeling in counseling psychology research. The Counseling Psychologist, 33, 269-298.

McCann, I. L., & Pearlman, L. A. (1990a). Psychological trauma and the adult survivor: Theory, therapy, and transformation. New York, NY: Brunner/Mazel.

McCann, I. L., & Pearlman, L. A. (1990b). Vicarious traumatization: A framework for understanding the psychological effects of working with victims. Journal of Traumatic Stress, 3, 131-149.

Myers, J. E., Mobley, A. K., & Booth, C. S. (2003). Wellness of counseling students: Practicing what we preach. Counselor Education & Supervision, 42, 264-274.

Myers, J. E., & Sweeney, T. J. (2005). Five Factor Wellness Inventory: Adult, teenage, and elementary school versions. Menlo Park, CA: Mind Garden, Inc.

National Center for Victims of Crime (2010). Crime victimization in the United States: Statistical overviews. Retrieved from http://ovc.ncjrs.gov/ncvrw2010/pdf/6_StatisticalOverviews.pdf

Nelson-Gardell, D., & Harris, D. (2003). Child abuse history, secondary traumatic stress, and child welfare workers. Child Welfare, 82, 5-26.

Neumann, D. A., & Gamble, S. J. (1995). Issues in the professional development of psychotherapists: Countertransference and vicarious traumatization in the new trauma therapist. Psychotherapy, 32, 341-347.

Pearlman, L. A. (2003). Trauma and attachment belief scale. Los Angeles, CA: Western Psychological Services.

Pearlman, L. A., & Mac Ian, P. S. (1995). Vicarious traumatization: An empirical study of the effects of trauma work on trauma therapists. Professional Psychology: Research and Practice, 26, 558-565.

Pearlman, L. A., & Saakvitne, K. W. (1995). Trauma and the therapist: Countertransference and vicarious traumatization in psychotherapy with incest survivors. New York, NY: Norton.

Ratna, L., & Mukergree, S. (1998). The long-term effects of childhood sexual abuse: Rationale for and experience of pharmacotherapy with nefazodone. International Journal of Psychiatry in Clinical Practice, 2, 83-95.

Roach, L. F., & Young, M. E. (2007). Do counselor education programs promote wellness in their students? Counselor Education & Supervision, 47, 29-45.

Saakvitne, K. W. (2002). Shared trauma: The therapist's increased vulnerability. Psychoanalytic Dialogues, 12, 443-449.

Sabin-Farrell, R., & Turpin, G. (2003). Vicarious traumatization: Implications for the mental health of health workers. Clinical Psychology Review, 23, 449-480.

Schauben, L. J., & Frazier, P. A. (1995). Vicarious trauma: The effects on female counselors of working with sexual violence survivors. Psychology of Women Quarterly, 19, 49-64.

Sexton, L. (1999). Vicarious traumatization of counselors and effects on their workplaces. British Journal of Guidance & Counseling, 27, 393-403.

Sommer, C. A. (2008). Vicarious traumatization, trauma-sensitive supervision, and counselor preparation. Counselor Education & Supervision, 48, 61-71.

Sommer, C. A., & Cox, J. A. (2005). Elements of supervision in sexual violence counselors' narratives: A qualitative analysis. Counselor Education and Supervision, 45, 119-134.

Spector, P. E. (1985). Measurement of human service staff satisfaction: Development of the job satisfaction survey. American Journal of Community Psychology, 13, 693-713.

Spector, P. E., & Jex, S. M. (1998). Development of font self-report measures of job stressors and strain: Interpersonal conflict at work scale, organizational constraints scale, quantitative workload inventory, and physical symptoms inventory. Journal of Occupational Health Psychology, 3, 356-367.

Substance Abuse and Mental Health Services Administration (2009). Mental health center locator. Retrieved May 26, 2009, from http://mentalhealth.samhsa.gov/databases/.

Thompson, B. (2002). 'Statistical,' 'practical,' and 'clinical': How many kinds of significance do counselors need to consider. Journal of Counseling and Development, 80, 64-71.

Traumatic Stress Institute (2010). Research. Retrieved from http://www.tsicaap.com/research.htm

Trippany, R., White Kress, V., & Wilcoxon, A. (2004). Preventing vicarious trauma: What counselors should know when working with trauma survivors. Journal of Counseling and Development, 82, 31-37.

Way, I., VanDeusen, K., & Cottrell, T. (2007). Vicarious trauma: Predictors of clinicians' disrupted cognitions about self-esteem and self-intimacy. Journal of Child Sexual Abuse, 16, 81-98.

Weston, R., & Gore, P. A. (2006). A brief guide to structural equation modeling. The Counseling Psychologist, 34, 719-751.

Amy M. Williams is affiliated with AspenPointe Health Services, Heather M. Helm and Elysia V Clemens are affiliated with the University of Northern Colorado. Correspondence concerning this article should be sent to Dr. Amy M. Williams, AspenPointe Health Services, 179 Parkside Drive, Colorado Springs, CO 80831. E-mail: [e-mail].

Table 1. Descriptive Statistics for Variables Included in the
Hypothesized Model

                                            Supervisory
                     Childhood   Personal     Working
                      Trauma     Wellness    Alliance

Mean                   43.00      81.02        66.83
Standard deviation     14.54       7.42        13.44
Minimum                 25         58.9         23
Maximum                 86         97.3         84
Range                   61         38.4         61
Skewness                97         -.37        -1.12
Kurtosis                .38        -.15        1.07
Likert scale            1-5        1-4          1-7

                     Organizational                Vicarious
                        Culture       Workload   Traumatization

Mean                     142.15        19.75         175.02
Standard deviation       24.28          3.91         36.97
Minimum                    67            9            113
Maximum                   200            25           327
Range                     133            16           214
Skewness                  -.30          -.57          .91
Kurtosis                  .12           -.32          1.32
Likert scale              1-6           1-5           1-6

Note. n = 131. For all scales the standard error for skewness was
.21 and the standard error for kurtosis was .42.

Table 2. Correlation Matrix for Variables in the Hypothesized Model

                                   1         2         3

1  Childhood trauma                1.0
2. Personal wellness               -.16      1.0
3. Supervisory working alliance    -.23 **    .22 *    1.0
4. Organizational culture          -.16       .28 **    .49 **
5. Workload                         .27 **   -.25 **   -.04
6. Vicarious traumatization         .36 **   -.63 **   -.26 **

   4         5        6

1
2.
3.
4. 1.0
5. -.24 **   1.0
6. -.19 *     .22 *   1.0

Note. N = 131. * indicates correlation is significant at p [less than or
equal to] 0.05 level. ** indicates correlation is significant at p [less
than or equal to] 0.01 level.