Supervisor Cultural Humility and Supervisee Nondisclosure: The Supervisory Working Alliance Matters

Abstract

Given that half or more of supervisees (therapist trainees) never have their clinical work monitored or observed, supervisees who withhold salient information in clinical supervision compromise supervisors’ ability to monitor client welfare and promote supervisees’ professional development. Attempting to further understand the factors explaining supervisee nondisclosure, we tested the supervisory working alliance as a mediator of the hypothesized inverse relations of cultural humility and collaborative supervision with supervisee nondisclosure (supervision-related and clinically-related nondisclosure) among a diverse sample of 214 supervisees in applied psychology and allied mental health programs. Results supported the hypotheses that (1) descriptively, supervision-related nondisclosure was more prominent than clinically-related nondisclosure, (2) cultural humility substantially inversely predicted supervisee nondisclosure, and (3) the supervisory working alliance fully mediated the inverse relations of cultural humility and collaborative supervision with supervisee nondisclosure. Understanding the mechanisms underlying supervisee nondisclosure have broad implications for clinicians and researchers alike.

Clinical supervision has been described as a primary vehicle for training therapists to practice competently (Borders et al., 2014). Clinical supervision is associated with improved job satisfaction (O’Donovan & Kavanagh, 2014), enhanced treatment outcomes (Wrape et al., 2015), reduced emotional exhaustion among therapists (Knudsen et al., 2013), and fewer intentions to quit the field (Fukui et al., 2019). Supervision can be viewed as necessary to protect the public by ensuring that therapists enter the field prepared to practice ethically and capably (Bernard & Goodyear, 2019; Falender & Shafranske, 2022). For supervision to operate effectively (Ladany et al., 2013), supervisees need to feel safe enough within the hierarchical, evaluative relationship to disclose sensitive and risky clinical and supervision-related material (Staples-Bradley et al., 2019). Thus, supervisee disclosure is essential to the process of supervision (Knox, 2015). For example, supervisees’ negative reactions to clients, other supervisees, or the supervisor can interfere with the working alliance and client outcome (Falender & Shafranske, 2022). These reactions should be disclosed, dealt with, and addressed in supervision to ensure client welfare.

Despite the importance of supervisee disclosure in supervision, many supervisees choose not to disclose information they know is important to their supervisors (Ladany et al., 2013). Although rates of nondisclosure in past research range widely from 6.2% (Siembor & Ellis, 2012) to 84.3% (Mehr et al., 2010), nondisclosure is conceived of as somewhat common (cf. Hutman & Ellis, 2020). Supervisees often report that they tend not to disclose specific types of information to their supervisors, including negative reactions to supervision, personal issues, clinical mistakes, evaluation concerns, general client observations, and negative or critical reactions to the client or supervisor (Ladany et al., 2013; Mehr et al., 2015; Staples-Bradley et al., 2019). Supervisee nondisclosure thereby delimits opportunities for supervisees to develop their competencies regarding the clinically- and supervision-related content that is not disclosed and compromises supervisors’ abilities to protect client welfare (Knox, 2015). Clinically-related content has been operationalized as knowledge or information of mistakes, interactions with clients, or aspects of the therapy content and process that are not disclosed in supervision, whereas nondisclosure of supervision-related content pertains to the supervisees’ feelings, reactions, or concerns about the supervisory process, the supervisor, or the clinical setting that are not shared in supervision (Ellis & Colvin, 2016). The consequences of withholding clinically-related content in supervision are especially concerning since most supervisors do not review or monitor supervisees’ clinical work (Rodriguez-Menendez et al., 2017).

A Framework of Supervisee Nondisclosure

Why do supervisees withhold important information in supervision? In the context of an evaluative, hierarchical supervisory alliance, supervisees may choose not to disclose for myriad reasons (e.g., Knox, 2015; Ladany et al., 2013). For example, fear of the consequences of disclosing information that could indicate a lack of competencies or dissatisfaction with supervision are often reasons cited for not disclosing in supervision (e.g., dismissal from the program, offending supervisors, or embarrassing themselves; Gunn & Pistole, 2012; Knox, 2015). A developing framework of supervisee nondisclosure has been put forth to investigate factors that promote or inhibit nondisclosure (Ellis & Colvin, 2016; Gibson et al., 2019; Hutman & Ellis, 2020; Mehr et al., 2015; Siembor & Ellis, 2012). Studies based on this framework have emphasized factors implicated in the quality of supervision, the supervisees’ experience in supervision, characteristics of the supervisee (e.g., anxiety, self-efficacy), and supervisees’ perceptions of their supervisor (e.g., multicultural competence, cultural humility). Of the few factors tested empirically (e.g., Cook & Welfare, 2018; Gibson et al., 2019; Gunn & Pistole, 2012; Jakob et al., 2014; McKibben et al., 2019), one consistent predictor of supervisee nondisclosure is the supervisory working alliance (e.g., Gibson et al., 2019; Hutman & Ellis, 2020). Supervisees’ perceptions are particularly important to study since their perceptions of their supervisors’ demeanor and presentation during supervision may influence their perceptions of the supervisory working alliance and in turn, influence what is not disclosed in supervision.

Supervisory Working Alliance Theory

The supervisory working alliance is defined as agreements on goals, agreements on tasks, and the emotional bond between the supervisee and supervisor (Bordin, 1983). Supervisory working alliance theory (Bordin, 1983) posits that a strong alliance between supervisor and supervisee promotes positive supervision outcomes for supervisees (e.g., supervisee growth, self-efficacy, satisfaction with supervision) and clinical care outcomes for clients (Enlow et al., 2019; Park et al., 2019). Hence, the quality of the supervisory working alliance is a critical element in the process of supervision and is one of the single most important predictors of positive outcomes in supervision (Bernard & Goodyear, 2019). Given that supervisee nondisclosure has been characterized as a critically important supervision outcome of interest with potential negative consequences (e.g., Knox, 2015), and guided by theory positing that the supervisory working alliance would be linked with (and may mediate) critical supervision outcomes (e.g., Hutman & Ellis, 2020; Vandament et al., 2021; Zhang et al., 2021), this study examined the supervisory working alliance as a mediator in relation to supervisee nondisclosure.

To wit, supervisees experiencing a positive supervisory working alliance report more disclosure of information in supervision (e.g., Cook & Welfare, 2018; Gunn & Pistole, 2012; McKibben et al., 2019; Mehr et al., 2010, 2015). If the supervisory working alliance is poor, however, disclosure about the supervisees’ experience of supervision can be particularly challenging (Ladany et al., 2013). Yet, other highly relevant predictors of supervisee nondisclosure largely remain untested or are not well established (e.g., collaborative supervision, Gibson et al., 2019), nor are the mechanisms underlying supervisee nondisclosure. Some evidence suggests the supervisory working alliance may be one such mechanism that explains (i.e., mediates) relations between other clinical predictors and outcomes (Crockett & Hays, 2015; Hook et al., 2013; Li et al., 2021).

For example, one study found that the supervisory working alliance mediated the association between a feminist supervisory approach and supervisee nondisclosure (McKibben et al., 2019), suggesting the supervisory working alliance may be key in understanding supervisee nondisclosure and may explain the mechanisms underlying supervisee nondisclosure. Thus, we theorized that the supervisory working alliance may be a mechanism through which other variables, such as collaborative supervision (Rousmaniere & Ellis, 2013) and supervisor cultural humility (Watkins et al., 2018), affect supervisee disclosure in supervision. The purpose of our study was to build on recent empirical work (Gibson et al., 2019) to further understand the factors explaining supervisee nondisclosure by testing the supervisory working alliance as a mediator of two predictors (i.e., supervisor cultural humility, collaborative supervision) with supervisee nondisclosure. Because past research has found that supervisee nondisclosure appears to consist of two types—supervision-related nondisclosure and clinically-related nondisclosure (Gibson et al., 2019; Jakob et al., 2014)—we also sought to investigate how supervisor cultural humility, collaborative supervision, and supervisory working alliance might be differentially linked with both supervision-related and clinically-related nondisclosure.

Collaborative Supervision

A separate but related concept to the supervisory working alliance is collaborative supervision (Rousmaniere & Ellis, 2013). Collaborative supervision is defined “in the context of the supervisory alliance as the extent to which the supervisor and supervisee(s) mutually agree and work together on the processes and activities of clinical supervision” (Rousmaniere & Ellis, 2013, p. 300). Prior research that empirically examined collaborative supervision and the supervisory working alliance challenged the assumption that collaboration is inherent in or essential to the alliance, finding that collaborative supervision and the supervisory working alliance are directly and moderately linked, but not identical constructs (Rousmaniere & Ellis, 2013). Notably, the construct of collaborative supervision, as conceived by Rousmaniere and Ellis (2013), measures the frequency in which supervisors and supervisees engage in open discussion about the process of supervision. Hence, although theoretically and empirically linked with the supervisory working alliance (Gibson et al., 2019), collaborative supervision distinctly emphasizes the process related to a collaborative, cooperative supervision.

Collaborative supervision was theorized and found to be related to supervisee nondisclosure (Gibson et al., 2019). Supervisors who explicitly (or implicitly) empower and encourage supervisees to collaborate evidenced more positive supervision outcomes, greater satisfaction, and lower supervisee role conflict than non-collaborative supervisors (Zlatev, 2015). However, the indirect effect of collaborative supervision on supervisee nondisclosure through the supervisory working alliance heretofore has not been tested empirically. We expected collaborative supervision to be inversely linked with supervisee nondisclosure and that this association would be mediated by the supervisory working alliance.

Supervisor Cultural Humility

Another factor posited to be associated with supervisee nondisclosure is supervisor cultural humility. Cultural humility is defined “as having an interpersonal stance that is other-oriented rather than self-focused, characterized by respect and lack of superiority toward an individual’s cultural background and experience” (Hook et al., 2013, p. 353). Supervisor cultural humility is considered to be a contributing factor to best practices in supervision through enhancing multicultural competence, improving the supervisory alliance, increasing supervisees’ responsibility to feedback, and fostering engagement in supervision (Upshaw et al., 2020; Watkins et al., 2018). Supervisors who are viewed as more open and genuine about their own cultural background tend to make supervisees feel comfortable in disclosing more, which is important for facilitating self-awareness and critical consciousness among supervisees (Soheilian et al., 2014). Accordingly, supervisees may disclose less when supervisors demonstrate less willingness to explore multicultural identities and cultural issues in supervision (Cook et al., 2020; Hutman & Ellis, 2020; Jernigan et al., 2010). Consistent with this theorizing, supervisor cultural humility was found to be inversely related to supervisee nondisclosure in a sample of 101 post-master’s counselors (Cook et al., 2020). However, these findings merit further research attention, as supervisee nondisclosure was assessed using measures that lacked psychometric validity data and did not take into account the two types of supervisee nondisclosure (e.g., Jakob et al., 2014), supervision-related nondisclosure and clinically-related nondisclosure (cf. Gibson et al., 2019). As such, it remains untested as to how cultural humility is distinctly linked with supervision-related nondisclosure and clinically-related nondisclosure.

Although tentative evidence suggested that supervisor cultural humility may be inversely associated with supervisee nondisclosure (Cook et al., 2020), there is a dearth of research on potential mediators or explanatory mechanisms of this relation. The only potential mediator suggested by a recent large-scale review of the literature on cultural humility in supervision was the supervisory working alliance (Zhang et al., 2021). Specifically, the supervisory working alliance was found to explain the positive impact of cultural humility on supervisory outcomes (Vandament et al., 2021). Similarly, the supervisory working alliance was found to mediate the association between supervisee nondisclosure and the related construct of supervisor multicultural competence (Hutman & Ellis, 2020). Based on supervisory working alliance theory, the developing framework of supervisee nondisclosure, and in continuing the line of programmatic research on the working alliance as a potential mediator of critical supervision outcomes (Crockett & Hays, 2015; Hutman & Ellis, 2020; Li et al., 2021; McKibben et al., 2019; Vandament et al., 2021; Zhang et al., 2021), this study sought to explore if the supervisory working alliance would be a potential mediator of supervisee nondisclosure. Examining the factors of supervisor cultural humility, collaborative supervision, the supervisory working alliance, and supervisee nondisclosure jointly through mediational analyses may lead to an important discovery as to key factors in facilitating supervisee disclosure.

If mediation is supported based on the results of this study, results may inform the training of clinical supervisors. Standardized training programs for clinical supervisors are lacking in the literature (Borders et al., 2014), and as such, the findings could inform efforts to ensure that supervisors are trained to be culturally humble, collaborative, and able to develop and maintain a positive supervisory working alliance (Gregus et al., 2020). The findings may support joint efforts to develop supervisor cultural humility in tandem with efforts to build the supervisory working alliance among clinical supervisors (Hook et al., 2013). Finally, a secondary aim of this study was to provide a nuanced descriptive understanding of which clinical and supervisory issues are frequently and infrequently disclosed in supervision, results of which will inform supervision training efforts and the work of supervisors, who may attend more to specific clinical and supervision areas knowing they are particularly challenging for disclosure.

Research Hypotheses

To replicate previous studies, we hypothesized (Hypothesis: H1) collaborative supervision to be inversely related to supervisee nondisclosure. In terms of the novel contributions of the current research to the clinical supervision literature, we hypothesized (Hypothesis: H2) that supervisor cultural humility would inversely, significantly predict supervisee nondisclosure (supervision-related nondisclosure and clinically-related nondisclosure as a multivariate set). To test the theorized mediation, we hypothesized (Hypothesis 3: H3) that the supervisory working alliance would mediate the inverse relations of supervisor cultural humility and collaborative supervision with supervisee nondisclosure, such that the supervisory working alliance would inversely, significantly predict supervisee nondisclosure (See Figure 1). Lacking necessary theorizing for hypotheses of differential effects, we only did follow-up tests on supervision-related nondisclosure and clinically-related nondisclosure for significant supervisee nondisclosure results.

Figure 1. Mediated Model for Supervisee Nondisclosure.

Note. Paths are positive relations except as noted. Path coefficients are shrunken multivariate eta squared coefficients (${\widehat{\rho }}_{MV}^2$; i.e., proportion of explained variance). c is the path coefficient when SWA is not included; c’ is the coefficient when it is. Paths are significant at p < .0001, unless noted otherwise. SND = Supervisee Nondisclosure (SRND and CRND) as a multivariate set; CH = Cultural Humility; CSB = Collaborative Supervision Behavior; SWA = Supervisory Working Alliance.

A secondary aim was to shed light on descriptive findings about what supervision-related and clinically-related topics are not disclosed by supervisees in supervision. Results may provide important implications for areas that would benefit from enhanced attention in supervision, improve our understanding of topics of particular difficulty for supervisees to disclose, and spur pathways for future research to address such critical areas. Regarding descriptive findings on nondisclosure, we expected that nondisclosure would be normative in regard to particularly sensitive topics (e.g., negative reactions to supervisors). Although vastly different rates of nondisclosure have been found in past research (i.e., 6.2% to 84.3%; Mehr et al., 2010; Siembor & Ellis, 2012)—likely based on differing study methodology (e.g., mixed methods vs. quantitative methods, nondisclosure in a single-supervision session vs. nondisclosure across all supervision sessions with the supervisor)—we expected overall rates might be relatively low, similar to other recent research on supervisee nondisclosure that quantitatively assessed nondisclosure in supervision throughout the duration of supervision with their current supervisor (e.g., under 10%; Hutman & Ellis, 2020; Siembor & Ellis, 2012).

Method

Participants

Based on multivariate multiple regression with three independent variables and two dependent variables, an a priori power analysis calculated with power (1 – β) set at .95, the experiment-wise Type I error rate set at α = .05, and an estimated population effect size of ${\widehat{\rho }}^2$ = .05 based on past research (Hutman & Ellis, 2020) and a medium effect size in counseling psychology (Haase et al., 2005), this study required a target sample of 212 participants. Inclusion criteria for eligibility included: (a) being pre-licensure master’s, doctoral, or post-doctoral supervisees who were (b) training in counseling or clinical programs, (c) engaged in clinical work, and (d) receiving clinical supervision. Approximately 274 individuals accessed the online survey to participate, and 214 completed the study, representing a 78% completion rate. Participants who did not complete the study were excluded from analyses.

The 214 participants were between the ages of 21 and 56 (M = 27.89; SD = 5.10). On average, participants had 2.39 years of supervised clinical experience (SD = 1.90). Participants reported an average of 5.22 clinical supervisors across their supervised clinical experiences (SD = 3.88). Most participants received formal training in clinical supervision (n = 122; 57.8%). Participants reported working with their current supervisors for approximately 5 months on average (M = 0.42 years, SD = 0.49 years). See Table 1 for demographic information of study participants and their supervisors, which was self-reported by participants.

Table 1.

Sample Demographics Among Participating Supervisees (N = 214)

Supervisee Demographic	Response	Frequency (%)	N
Race/Ethnicity	White or European American	73.7	154
	Latinx or Hispanic American	9.1	19
	Asian or Asian American/Pacific Islander	7.7	16
	Black or African American	4.3	9
	Biracial or Multiracial	4.3	9
	Self-Identified Race or Ethnicity	1.0	2
Gender Identity	Cisgender woman	85.3	180
	Cisgender man	13.7	29
	Transgender, agender, or non-binary	0.9	2
Sexual Orientation	Heterosexual	72.0	152
	Bisexual	10.4	22
	Queer	8.1	17
	Gay	3.8	8
	Lesbian	2.8	6
	Self-Identified Other Sexual Orientation	2.8	6
Ability Status	No disability	94.7	198
	Disability or impairment	5.1	11
Socioeconomic Status	Lower class	5.3	11
	Working class	13.0	27
	Lower middle class	19.7	41
	Middle class	31.3	65
	Upper middle class	24.5	51
	Lower upper class	3.8	8
	Upper class	2.4	5
Student Status	International	3.3	7
	Domestic (U.S.)	96.7	204
Education Achieved	Bachelor’s degree	29.9	63
	Master’s degree	62.1	131
	Doctoral degree	6.2	13
	Other professional degree	1.4	3
Field of Training	Clinical Psychology	50.7	107
	Counseling Psychology	40.3	85
	Mental Health Counseling	4.3	9
	Counselor Education	1.4	3
	Rehabilitation Counseling	0.5	1
	School Counseling	0.5	1
	Social Work	0.5	1
	Other Field	1.9	4
Stage of Training	Pre-practicum	3.3	7
	Practicum	71.4	150
	Internship	15.7	33
	Post-Doctoral Clinical Training	3.3	7
	Pre-Licensure Employment	3.3	7
	Other (e.g., combination of aforementioned stages)	2.9	6
Clinical Setting	University-based training clinic	29.0	61
	College counseling center	21.0	44
	Community mental health center	20.1	43
	Private practice	3.8	8
	Private psychiatric hospital	3.8	8
	Primary/secondary school	3.3	7
	State psychiatric hospital	2.9	6
	Substance use/rehabilitation facility	1.4	3
	Veterans Affairs hospital	2.9	6
	Forensic site/prison	1.4	3
	Other setting	10.0	21
Supervisor Demographic	Response	Frequency (%)	N
Race/Ethnicity	White or European American	84.1	175
	Latinx or Hispanic American	3.8	8
	Asian or Asian American/Pacific Islander	4.8	10
	Black or African American	3.4	7
	Biracial or Multiracial	2.9	6
	Self-Identified Race or Ethnicity	1.0	2
Gender	Cisgender woman	68.6	144
	Cisgender man	30.5	64
	Transgender, agender, or non-binary	1.0	2
Licensure	Licensed	92.9	196
	Unlicensed	6.2	13
Field	Clinical Psychology	64.5	136
	Counseling Psychology	25.1	53
	Mental Health Counseling	4.3	9
	Counselor Education	0.9	2
	Rehabilitation Counseling	0.9	2
	School Counseling	0.5	1
	Social Work	0.5	1
	Marriage and family therapy	0.5	1
	Other Field	3.0	6

Note. Supervisees self-reported their supervisor’s demographic data.

Measures

As suggested by others (e.g., Shadish et al., 2002), we tested proxy comparison groups (α_pc = .01) to increase the scientific rigor of the one-group ex post facto research design; that is, scores on the major variables were compared to their respective norms. These tests served four purposes: (a) most of the measures used here were developed recently and are not well known, hence, there is little evidence to understand what a given score means; (b) to circumscribe the study’s sample and inferences from it; (c) to provide a context to interpret the subjects’ scores descriptively; and (d) to assist in interpreting the major findings.

Supervisee Nondisclosure Scales (SNDS)

The SNDS (Ellis & Colvin, 2016; Siembor & Ellis, 2012) measured the extent to which supervisees choose to withhold information from their supervisors. For inferential analyses, this study used the 11-item version of the SNDS that Ellis and Colvin (2016) derived and tested via graded response item response theory (IRTPRO; Cai et al., 2011) from the larger 30-item measure (Siembor & Ellis, 2012), although the full measure was used to characterize the sample in terms of clinically-related and supervision-related nondisclosure at the item level. Participants rated items on a 7-point Likert-type scale (1 = Fully disclosed, 7 = Decided to not disclose), with a Not Applicable (NA) option for descriptive purposes. A rating scale utility analysis (Linacre, 1999) found that a 3-point scale (1 = NA and Fully disclosed; 2 = 2–6, Somewhat disclosed; 3 = 7, Decided to not disclose) was optimal (vs. a 7-point scale), and this 3-point scale was used in characterizing the rates of disclosure across items in descriptive results. Using an item-level fit statistic, S-χ² (Kang & Chen, 2011) and the standardized local dependence χ² fit statistic (Liu & Thissen, 2014), items with better fit indices were retained yielding two SND scales: Supervision-Related Nondisclosure (SRND; 4 items) and Clinically-Related Nondisclosure (CRND; 7 items). The goodness of fit indices suggested acceptable psychometric properties, SRND: M₂(20) = 446.34, p < .001, RMSEA = .20, “marginal reliability,” similar to Cronbach’s alpha, = .77; CRND: M₂(77) = 442.36, p < .001, RMSEA = .07; marginal reliability = .83. IRT scores range from 1 to 10, where higher scores indicate greater nondisclosure. Participants’ scores (see Table 2) were significantly lower than the original CRND scores (M = 4.22), t(213) = −3.68, p = .001, ${\widehat{\rho }}^2$ = 0.055, 95% CI [0.02, 0.12], but did not differ from original SRND scores, M = 4.11, t(213) = −0.5, p = .91, ${\widehat{\rho }}^2$ = 0.0, CI [0.0, 0.0].

Table 2.

Means, Standard Deviations, and Correlations Among the Major Variables.

Variable	M	SD	SRND	CRND	SWA	CSB
SRND	4.10	2.93
CRND	3.81	1.63	.43
SWA	196.08	38.99	−.71	−.55
CSBS	12.02	4.35	−.48	−.42	.66
CH	46.06	10.96	−.54	−.44	.72	.54

Note. All correlations p < .0001. SRND = Supervision-Related Nondisclosure; CRND = Clinically-Related Nondisclosure; SWA = Supervisory Working Alliance -Trainee; CH= Cultural Humility; CSB = Collaborative Supervision Behaviors.

Collaborative Supervision

The Collaborative Supervision Behaviors Scale (CSBS; Rousmaniere & Ellis, 2013) was used to assess supervisees’ perceptions of the extent to which they and their supervisors collaborated in supervision. The CSBS is a four-item measure rated on a 5-point Likert-type scale ranging from 1 (Never) to 5 (Always). Summed total scores range from 4 to 20, with higher scores reflecting more perceived collaborative supervision. Rousmaniere and Ellis (2013) developed and tested the CSBS using confirmatory factor analyses, reporting solid convergent and divergent validity of the CSBS. In this study, Cronbach’s alpha was α = .91, consistent with other studies (Gibson et al., 2019; Rousmaniere & Ellis, 2013). Participants’ CSBS scores were significantly higher than Rousmaniere and Ellis’s development sample, M = 10.52, t(213) = 5.04, p < .0001, ${\widehat{\rho }}^2$ =.10, CI [0.05, 0.18].

Supervisor Cultural Humility

The 12-item Cultural Humility Scale (CHS; Hook et al., 2013) was used to assess participants’ perceptions of their supervisors’ multicultural orientation and values, or the extent to which their supervisor had an interpersonal stance that was respectful toward others’ cultural backgrounds and experiences. Items are rated on a 5-point Likert-type scale (1 = Strongly disagree, 5 = Strongly agree), with 5 items being reverse scored. Summed total scores range from 12 to 60; higher scores reflect more cultural humility. Strong support was found for internal consistency reliability, as well as convergent validity of the CHS with multicultural competencies (Hook et al., 2013). Reliability in this sample was α = .92. Participants’ CHS scores were significantly lower than Hook et al.’s (2013, Study 2) sample of 134 adults’ ratings of their therapist (M = 20.15), t(213) = −5.46, p < .0001, ${\widehat{\rho }}^2$ =.12, CI [0.06, 0.20].

Supervisory Working Alliance

The supervisory working alliance was assessed using the Supervision Working Alliance Inventory-Trainee Version (SWAI-T; Bahrick, 1989). The SWAI-T is a self-report measure of supervisees’ perceptions of the supervisory working alliance based on the conceptualization of Bordin (1983). The SWAI-T contains 36 items that are rated on a 7-point Likert-type scale from 1 (Never) to 7 (Always), and a total score is indicated due to higher intercorrelations among the three subscales of agreement on goals, agreement on tasks, and emotional bond (Ladany et al., 1999). Fourteen items are reverse scored. Total scores range from 36 to 252. Higher ratings indicate a stronger alliance. Past research supports evidence of convergent and divergent validity and internal consistency reliability of the SWAI-T (Ladany et al., 1999). Cronbach’s alpha for the SWAI-T in this study was α = .97. No significant differences were found between the present participants’ SWAI-T scores (M = 196.08, SD = 38.99) or those in Ellis et al.’s (2003) sample, M = 197.06, t(213) = −0.37, p = 0.71, ${\widehat{\rho }}^2$ = 0.00, 95% CI [0.00, 0.02].

Procedure

Upon Institutional Review Board Approval from a midsize Northeastern public university, a call to participate was advertised via professional psychology list servs (e.g., Student Affiliates of Seventeen, National Latinx Psychological Association) and sent to directors of training of applied psychology doctoral programs for circulation to potential participants. Supervisees were invited to participate in a study to investigate supervisees’ experiences in clinical supervision with their primary supervisors. Those interested could choose to access the link included in the call for participants, which also included a brief description of the study. All participants were volunteers who completed this online study that used Qualtrics survey software. Main study measures were counterbalanced and administered in random order to avoid potential sequencing effects. Participants who completed the survey were entered into a raffle to win 1 of 20 $10 Amazon gift cards. Because recruitment emails were sent to professional listservs and directors of training, and confirmation cannot be obtained on how many trainees viewed the email or how many directors of training shared the call to participate with their trainees, the response rate is unknown and cannot be calculated.

Results

Preliminary Analyses

Initially, 278 people responded to the study, of which 64 were excluded due to more than 5% missing data; participants with less than 5% missing data (n = 4) were retained for analysis, and scale means were calculated with available items only. Tests of the statistical assumptions (i.e., violations of independence of data, linearity, normality, homoscedasticity, multicollinearity, or leverage, influential cases, multivariate outliers, or measure administration sequence effects) found no violations, suggesting the data were appropriate for the major analyses. Data were examined for missingness, and missing values were examined for patterned occurrences. Little’s Missing Completely at Random (MCAR; Little, 1988) test revealed nonsignificant findings, which indicated a failure to reject the null hypothesis and suggested that data were missing completely at random, χ² = 2,956.90 (2,905), p = .246. Because parameters derived from MCAR data under list-wise deletion have been found to be equivalent to those derived from complete data (Schlomer et al., 2010), this study proceeded with complete case analysis by excluding participants with more than 5% missing data (n = 64).

Descriptive Findings of Supervisee Nondisclosure

To characterize the rates of disclosure and non-disclosure of clinically-related and supervision-related content among supervisees in this sample, supervisees in the present study rated items of the Supervisee Nondisclosure Scales (Ellis & Colvin, 2016; Siembor & Ellis, 2012). Results indicated the clinical topics and issues they frequently to infrequently disclosed in supervision, as well as which topics were not applicable to their supervision experiences. Participants ratings of SNDS items on a 3-point scale (1 = Fully disclosed; 2 = Somewhat disclosed; 3 = Decided to not disclose) are reported in the following section to provide granular detail about disclosure and nondisclosure in this sample.

Supervision-Related Issues Frequently Not Disclosed in Supervision

For supervisees indicating that the issues arose in their supervision, supervision-related issues that supervisees decided not to disclose included negative reactions about (a) supervisor’s competence (61.5%; n = 40/65), (b) feeling unsafe in supervision (45.9%, n = 17/37), (c) supervisor’s stereotypes or biases (43.2%, n = 38/88), (d) not getting enough out of supervision (41.7%; n = 43/103), (e) the process of supervision (e.g., how supervision flows and proceeds, such as who speaks and when) (38.8%; n = 40/103), (f) supervisor’s behavior or attitudes (e.g., something the supervisor says or does in supervision) (36.9%, n = 48/130), (g) concerns about the power differential in supervision (34.6%; n = 37/107), (h) the focus in supervision (e.g., whether and what the supervisee and supervisor agree to spend time on) (34.1%; n = 31/91), and (i) the format of supervision (e.g., how the supervision session is structured) (31.1%; n = 33/106). In contrast, 33.0% (n = 69/209) of supervisees fully disclosed positive reactions to their supervisors when they occurred.

Common Clinically-Related Issues Disclosed in Supervision

Regarding clinically-related issues, supervisees (a) fully disclosed (82.6%, n = 166) or partially disclosed (16.9%, n = 34) important client information, (b) fully disclosed (49.7%, n = 99) or partially disclosed (48.2%, n = 96) positive non-romantic reactions to a client, (c) fully disclosed (44.3%, n = 86) or partially disclosed (55.2%, n = 107) negative reactions to a client, (d) fully disclosed (43.6%, n = 88) or partially disclosed (55.0%, n = 111) clinical mistakes, (e) fully disclosed (43.0%, n = 89) or partially disclosed (54.1%, n = 112) mistakes they were worried they might make, (f) fully disclosed (38.4%, n = 71) or partially disclosed (56.2%, n = 104) personally identifying with clients issues (countertransference), (g) fully disclosed (36.9%, n = 65) or partially disclosed (52.3%, n = 92) personal issues relevant to working with clients, (h) fully disclosed (29.8%, n = 62) or partially disclosed (64.9%, n = 135) feeling inadequate, and (i) fully disclosed (40.4%, n = 21) or partially disclosed (46.2%, n = 24) client’s attraction to them (the supervisee). When these clinically-related issues arose, no more than 11 (5.4%) supervisees chose to not disclose, except for personal issues relevant to working with clients (10.8%, n = 19).

Less Commonly Occurring Issues

Some supervision-related issues were frequently not applicable, including feeling unsafe in supervision (82.7%; n = 177), negative reactions to the supervisor’s competence (69.6%, n = 149), negative reactions to their supervisor’s stereotypes or biases (58.9%; n = 126), and negative reactions to the focus of supervision (57.5%, n = 123). Similarly, some clinically-related issues were frequently not applicable; for example, feelings of romantic attraction to a client (84.1%; n = 180), client’s apparent attraction to them (75.7%; n = 162), and behavior with a client that might have been unprofessional (63.1%, n = 135).

Major Inferential Analyses

See Table 2 for descriptive statistics for the major variables and Table 3 for major analyses results. Following Hayes’ (2018) procedures for testing mediation, we performed a series of simultaneous multivariate multiple regression analyses (Cohen et al., 2003) to test the hypotheses on supervisee nondisclosure as a multivariate construct—the multivariate composite or set of supervision-related nondisclosure and clinically-related nondisclosure analyzed jointly. Multivariate tests were conducted to control for the confounding (correlation) between the two dependent variables (supervision-related nondisclosure and clinically-related nondisclosure; e.g., Haase & Ellis, 1987). We followed up significant multivariate effects with univariate F-tests plus the standardized discriminant function coefficients (sdfc; Haase & Ellis, 1987). Tests of unique effects (i.e., tests of one predictor controlling for the other predictor variables) were performed to control for the correlations among the predictor variables (see Table 2). A significant test of unique effects suggests that the given variable predicts the dependent variable(s) over and above the influence of the other predictors and confounding in the statistical model. With a study-wise Type I error rate of α_EW = .05, the per comparison alpha was set at α_PC = .01. Predictors were considered substantive contributors if they explained ≥ 5% of the variance (${\widehat{\rho }}_{MV}^2$ ≥ .05), a medium effect size for counseling psychology (Ellis & Ladany, 1997; Haase et al., 2005).

Table 3.

Multivariate Multiple Regression Analyses Between the Predictors and SND (SRND & CRND).

Statistical Model	Pillai’s V	F	p	${\widehat{\rho }}_{MV}^2$	Dependent Variable	F	p	${\widehat{\rho }}^2$	95% CI	sdcf
Single Predictor Tests (PV – DV)
SWA – SND	.57	141.60	.0001	.57	SRND	210.12	.0001	.49	0.34, 0.48	.82
					CRND	92.82	.0001	.30	0.13, 0.26	.51
CH – SND	.35	55.51	.0001	.34	SRND	86.39	.0001	.29	0.20, 0.37	.76
					CRND	52.08	.0001	.19	0.12, 0.28	.49
CSB – SND	.28	41.33	.0001	.28	SRND	61.76	.0001	.23	0.14, 0.31	.71
					CRND	44.92	.0001	.17	0.09, 0.25	.53
Test of Full Model
	.58	28.75	.0001	.27	SRND	69.90	.0001	.49	0.41, 0.57	.82
					CRND	31.88	.0001	.30	0.21, 0.39	.52
Tests of Unique Effects
SWA – SND	.29	42.36	.0001	.28	SRND	68.14	.0001	.24	0.16, 0.33	.87
					CRND	21.69	.0001	.09	0.04, 0.16	.45
CH – SND	.008	.84	.43	0.0					0.00, 0.02
CSB – SND	.006	.59	.56	0.0		.			0.00, 0.01

Note: ${\widehat{\rho }}_{MV}^2$ is the estimated population multivariate effect size; sdfc = standardized discriminant function coefficient; PV = Predictor variable; DV = Dependent variable; Tests of Unique Effects = Tests of one predictor variable controlling for the influence of all other predictor variables; SRND = Supervision-Related Nondisclosure; CRND = Clinically-Related Nondisclosure; SND = Supervisee Nondisclosure (SRND and CRND) as a multivariate set; CH = Cultural Humility; CSB = Collaborative Supervision Behaviors; SWA = Supervisory Working Alliance.

Per H1, collaborative supervision (with no other predictors in the analysis) significantly inversely predicted supervisee nondisclosure (as a multivariate set), F(2, 211) = 41.33, p < .0001, ${\widehat{\rho }}_{MV}^2$ = 0.28. Follow up tests found that collaborative supervision was significantly inversely related to supervision-related nondisclosure, F(1, 212) = 61.76, p < .0001, ${\widehat{\rho }}^2$ = 0.23, sdfc = 0.71, and clinically-related nondisclosure, F(1, 212) = 44.92, p < .0001, ${\widehat{\rho }}^2$ = 0.17, sdfc = 0.53. Thus, H1 was supported.

As predicted for H2, supervisor cultural humility (as a single predictor) was significantly inversely associated with supervisee nondisclosure (as a multivariate set), F(2, 211) = 55.51, p < .0001, ${\widehat{\rho }}_{MV}^2$ = 0.34. The follow up tests found that cultural humility was significantly inversely related to supervision-related nondisclosure, F(1, 212) = 86.39, p < .0001, ${\widehat{\rho }}^2$ = 0.29, sdfc = 0.76, and clinically-related nondisclosure, F(1, 212) = 52.08, p < .0001, ${\widehat{\rho }}^2$ = 0.19, sdfc = 0.49. Thus, H2 was supported.

Testing H3 (supervisory working alliance as a mediator) involved a series of multivariate tests per Hayes (2018): (a) test of the full model of the three predictors and the two dependent variables as a multivariate set, (b) tests of each predictor separately with supervisee nondisclosure as a multivariate set, (c) tests of the unique effects of each predictor on supervisee nondisclosure as a multivariate set controlling for the other predictors. As expected, the test of the full model (supervisory working alliance, cultural humility, and collaborative supervision) and supervisee nondisclosure as a multivariate set was significant (p < .0001, ${\widehat{\rho }}_{MV}^2$ = .27), as were the follow up tests for supervision-related nondisclosure (p < .0001, ${\widehat{\rho }}^2$ = 0.49,) and for clinically-related nondisclosure (p < .0001, ${\widehat{\rho }}^2$ = 0.30). As can be seen in Table 3, each predictor separately (single predictor tests) was strongly associated with supervisee nondisclosure as a set (${\widehat{\rho }}_{MV}^2$s = .28 - .57) and with supervision-related nondisclosure (${\widehat{\rho }}^2$s = .23 - .49) and clinically-related nondisclosure (${\widehat{\rho }}^2$s = .17 - .30). Finally, only the supervisory working alliance was a unique predictor of (a) the multivariate composite of supervisee nondisclosure controlling for the other predictors (p < .0001, ${\widehat{\rho }}_{MV}^2$ = .28), and (b) supervision-related nondisclosure (p < .0001, ${\widehat{\rho }}^2$ = 0.24, sdfc = 0.87) and clinically-related nondisclosure (p < .0001, ${\widehat{\rho }}^2$ = 0.09, sdfc = 0.45). As predicted for mediation, neither cultural humility nor collaborative supervision were uniquely associated with supervisee nondisclosure (ps > .43, ${\widehat{\rho }}_{MV}^2$s = 0.0) when controlling for the supervisory working alliance in the statistical model, yet both were individually (as single predictors) strongly associated with the supervisory working alliance (ps < .0001, ${\widehat{\rho }}^2$s > .44) and multivariate supervisee nondisclosure (ps < .0001, ${\widehat{\rho }}_{MV}^2$s > .28). That is, the substantive inverse relations of cultural humility and collaborative supervision with supervisee nondisclosure were significantly attenuated when the supervisory working alliance was included in the statistical model. Collectively, these data support the supervisory working alliance mediating the relations of cultural humility and collaborative supervision with supervisee nondisclosure. Thus, H3 was supported. Figure 1 depicts the paths using the estimated population effect sizes (i.e., shrunken squared effect size coefficients; e.g., ${\widehat{\rho }}_{MV}^2$; Fairchild et al., 2009) of the relations for the two predictors and the mediator (supervisory working alliance) from Table 3.

Discussion

The present study was the first to test whether supervisor cultural humility would be inversely related to supervision-related and clinically-related supervisee nondisclosure among supervisees, as well as whether the supervisory working alliance (assessed according to Bordin’s [1983] conceptualization based on the supervisees’ self-reported perceptions of agreement on goals, agreement on tasks, and emotional bond) would mediate hypothesized links between supervisor cultural humility, collaborative supervision, and supervisee nondisclosure. Of note, supervisees reported the race of their supervisor, indicating that 36.0% (n = 77) of supervisory dyads were cross-racial, or dyads comprised of at least one person of Color; the rest were White-identified dyads (64.0%; n = 137). As such, the results should be viewed within this context. Future researchers are encouraged to investigate cross-racial supervisory experiences—an understudied but increasingly common and important supervisory context. Findings underscored the importance of perceived supervisor cultural humility and the role of the supervisory working alliance in influencing supervisees’ willingness to engage in disclosure in supervision. Beyond contributing to the empirical literature on supervisee nondisclosure, results support supervisory working alliance theory and the developing framework of supervisee nondisclosure and provide potential implications for training and practice in counseling psychology and other clinical fields.

In addition to testing a theorized mediation model of supervisee nondisclosure, this study also revealed important descriptive findings about what supervision-related and clinically-related topics are critical in supervisee nondisclosure. These results provided important implications for the practice of supervision and future research in this area.

Descriptive Findings of Supervisee Nondisclosure

Several supervision-related issues were challenging for supervisees to disclose in supervision. For example, supervisees struggled to disclose their negative reactions about the perceived safety of supervision; concerns about the supervisory power differential; beliefs that they were not getting enough out of supervision; the behavior or attitudes, competence, and stereotypes or biases of the supervisor; as well as their concerns about the process, focus, or format of supervision sessions. On the other hand, about one-third of supervisees fully disclosed positive reactions to their supervisors when they occurred. Findings underscored supervisees’ reticence to disclose negative reactions to supervisors or aspects of the supervision, perhaps due to concerns about safety, power, and evaluation (Bernard & Goodyear, 2019).

Conversely, supervisees in this study reported many clinical issues that were frequently encountered and fully disclosed in supervision (versus not disclosed). The large majority of supervisee in this study reported fully or partially disclosing important client information, positive non-romantic reactions to a client, negative reactions to a client, clinical mistakes, mistakes they were worried they might make, their countertransference, personal issues relevant to working with clients, feelings of inadequacy, and perceptions of client’s attraction to them. Thus, these findings accentuated the notion that supervisees tend to disclose clinically-related content at relatively higher rates than supervision-related content, paralleling and building on findings of previous studies that documented low rates of clinically-related nondisclosure (e.g., Gibson et al., 2019; Gunn & Pistole, 2012; Hutman & Ellis, 2020; Jakob et al., 2014).

The descriptive results suggested that much of the nondisclosed material was supervision-related in nature and not clinically-related material. This pattern of findings was consistent with the results of Gibson et al. (2019) and others, who found that disclosure of perceptions of the supervision process occurred mainly in the context of a high quality supervisory working alliance, whereas most supervisees disclosed clinically-related content irrespective of working alliance quality (Gunn & Pistole, 2012; Hutman & Ellis, 2020; Jakob et al., 2014). Nevertheless, nondisclosure in these areas may hinder clinical outcomes to the extent that it interferes with the supervisory process and undermines supervisee growth and development. An implication is that strengthening the supervisory working alliance (via increased agreement on goals and tasks and improved emotional bond between supervisee-supervisory) may be linked with decreased nondisclosure, particularly related to supervision-related nondisclosure. As this study sample was largely comprised of White supervisees who reported having a White supervisor, more research is needed on rates of nondisclosure among supervisees of Color or supervisees working in cross-racial supervisory dyads.

Inferential Findings of Supervisee Nondisclosure

Regarding inferential findings, in accordance with the hypotheses, supervisor cultural humility was inversely associated with supervisee nondisclosure as a multivariate construct as well as with supervision-related nondisclosure and clinically-related nondisclosure. Thus, supervisees in the present study reported less nondisclosure in supervision when they also perceived that their supervisor was more culturally humble. Findings suggested that increased cultural humility among supervisors may foster a supervisory context that is perceived to be safe for disclosure, whether supervision-related or clinically-related. These results extended previous research (Gibson et al., 2019) and supported the notion that supervisors’ ability to develop and maintain an open, other-oriented approach to supervision where cultural context can be explored and discussed is a best practice in supervision that may potentially be capable of facilitating positive supervisory outcomes, including an improved supervisory working alliance and lower supervisee nondisclosure (Upshaw et al., 2020; Watkins et al., 2018). Particularly in times that challenge efforts for social justice and equity, the capacity for supervisors and supervisees to acknowledge and discuss sociocultural contexts that influence the personal and professional is increasingly important (see Mosley et al., 2020 for a comprehensive, culturally relevant psychological framework).

Notably, the supervisory working alliance fully mediated the inverse associations between (a) supervisor cultural humility and supervisee nondisclosure and (b) collaborative supervision and supervisee nondisclosure. The influence of supervisees’ perceptions of their supervisors’ cultural humility and collaboration on their level of nondisclosure in supervision appeared to depend on the quality of the supervisory alliance. For instance, the non-experimental data observed here suggested that supervisees experiencing a stronger supervisory alliance tended to overshadow their perceptions of their supervisors’ cultural humility and collaboration, as supervisory alliance predicted less nondisclosure in terms of both supervision-related and clinically-related nondisclosure. Thus, the supervisory working alliance—or the agreement on goals, agreement on tasks, and emotional bond of the supervisor and supervisee—may be the mechanism through which the inverse relation between supervisors’ collaboration and cultural humility and supervisee nondisclosure can be explained.

It is important to ground these findings—which underscore the importance of supervisees’ perceptions of their supervisors’ cultural humility, collaboration, and the working alliance in relation to supervisees’ comfort with disclosure in supervision—in the context of the current sample. These inferential findings provide insight into factors linked with reduced supervisee nondisclosure in supervision. Yet, descriptive findings illuminated some concerning areas where supervisees felt less empowered to make a disclosure. As noted previously, approximately 64.0% of this sample were White supervisees paired with a White clinical supervisor, and descriptive findings revealed that approximately 43.2% of supervisees reported that they decided not to disclose negative reactions they had about their supervisors’ stereotypes or biases evident in supervision when such a negative reaction occurred. Thus, descriptive findings suggest that White supervisees may struggle to enact cultural humility through discussing and addressing stigmatizing or discriminatory attitudes, beliefs, and behaviors of the supervisor in supervision in the context of the supervisory power dynamic. These are concerning trends that point to the increasingly critical need to address whiteness in training and clinical supervision (Baima & Sude, 2020). Mental health and counseling training programs must attend more fully to race as a significant factor in the organization of U.S. society and culture that permeates all spaces, whether acknowledged or not. Given that a colorblind approach to counseling and supervision risks harming clients (Neville et al., 2013), supervisees and supervisors must become aware of, acknowledge, and interrogate stereotypes and biases and use supervision as a tool for this critical work regarding personal and professional growth.

Taken together, results of this study build on recent findings that the supervisory working alliance may be a mediator in clinical supervision (e.g., Hutman & Ellis, 2020; McKibben et al., 2019) and support the supervisory working alliance as a potential integral explanatory factor in understanding and predicting supervisee nondisclosure. Thus, findings underscored prior theory and literature on how a strong and positive supervisory working alliance may facilitate positive clinical outcomes. Accordingly, previous scholarship has highlighted how a strong and positive supervisory working alliance is linked with more positive outcomes in supervision, including increased supervisee safety, growth, and development, as well as less nondisclosure (e.g., Bernard & Goodyear, 2019; Cook & Welfare, 2018; Gunn & Pistole, 2012; Ladany et al., 1999; Mehr et al., 2010, 2015). This study extended prior research on cultural humility and the supervisory working alliance, delineating how supervisor cultural humility and collaboration are connected to the alliance, and which in turn, was inversely related to supervisee nondisclosure. A key takeaway for clinical supervisors is that fostering a strong supervisory working alliance may be instrumental to minimizing supervisee nondisclosure and promoting growth of the supervisee in supervision. Findings demonstrate that when supervisors are collaborative and demonstrate cultural humility, supervisees may be more likely to disclose supervision-related and clinically-related content because there is a stronger supervisor-supervisee working alliance. Supervisors should also monitor the alliance to assess for indications to be more collaborative and humble.

Limitations

This study has several limitations. Nearly 25% of individuals who accessed the study ultimately did not complete it. Most of these non-completers (n = 39 of 64 total; 60.9%) did not provide any response, but instead navigated away after accessing the study landing page. Although the study non-completion rate was in line with typical online research completion rates (LaRose & Tsai, 2014), this remains a notable limitation since those who completed the study may have differed from non-completers in unknown (but possibly important) ways (e.g., level of motivation to complete a study on supervisee nondisclosure). This is particularly relevant because convenience sampling (i.e., non-random sampling) was used to recruit participants. For that reason, we cannot assume that this sample of supervisees is representative of the broader population of clinical supervisees, and thus care must be taken in generalizing findings.

Descriptively, the present sample of supervisees is relatively similar to prior studies of supervisee nondisclosure in terms of representation (and lack thereof) for race, ethnicity, and gender identity, while being notably more diverse in terms of participants’ sexual orientations (e.g., Gibson et al., 2019; Hutman & Ellis, 2020). Yet, this sample does not adequately represent supervisees of specific race and ethnicity, gender identity, sexual orientation, and ability identities. Specifically, most participants were White cisgender women who identified as heterosexual and able-bodied. As such, there is an immense need for more targeted studies of diversity in terms of all intersecting social identities in clinical supervision research. Additionally, in assessing race and ethnicity of participants, we did not disaggregate racial and ethnic groups beyond terms that gloss over critical within-group heterogeneity based on sociohistorical and political realities. For example, participants were asked whether they identified as White/European American, Latinx/Hispanic American, Asian/Asian American/Pacific Islander, Black/African American, Biracial or Multiracial, or another race or ethnicity they could self-identify. Accordingly, we could not disaggregate Asian, Asian American, and Pacific Islander participants; Indigenous participants; or individuals with Biracial or Multiracial identities. We acknowledge this ethnic gloss as problematic and in future research, recommend comprehensive assessment of supervisees’ intersecting identities with particular attention to avoiding the use of ethnic gloss. Furthermore, future researchers are encouraged to focus on the experiences of supervisees who hold marginalized identities to better understand supervisee nondisclosure in relation to the societal dynamics of power, privilege, and oppression as they function in the supervisory context (e.g., supervisees who identify as nonbinary or gender fluid, and supervisees across the spectrum of sexual orientations, supervisees with disabilities).

As this cross-sectional study administered measures at one time point, we were unable to determine temporal precedence among constructs in the theorized mediation model. Until the mediation process is tested more thoroughly, inferences about mediation need to be used with caution. Additionally, all study measures were self-report and assessed supervisees’ perceptions of their supervision, which introduces potential bias into the data (Shadish et al., 2002). Although the study design was not dyadic and did not assess supervisors’ perceptions of the supervision, there is a critical importance in studying supervisees’ perceptions of their supervisors and of the supervision given that supervisees’ perceptions, whether accurate or inaccurate, will shape their behavior and the supervision process (e.g., Cook et al., 2018; McKibben et al., 2019). Nevertheless, each of the aforementioned limitations diminishes the validity of this study.

The strengths of the study include that it was an initial test of a hypothesized mediation model of supervisee nondisclosure, explicated falsifiable hypotheses, used an a priori power analysis, controlled Type I and II error rates, used psychometrically viable measures, and tested multivariate mediation following current recommendations (Hayes, 2018). Future research should build on the growing literature related to supervisee nondisclosure by examining other constructs that may impact nondisclosure, including contextual factors not assessed in this current study, such as supervisor and supervisee clinical experiences, supervisors’ perceptions about nondisclosure, and supervisory style. A theoretically-based, conceptual model of supervisee nondisclosure should be developed to account for the many supervisee and supervisor factors that may influence the contextual process of supervisee nondisclosure. Perhaps future research could examine other constructs that may potentially impact supervisee nondisclosure, such as stage of identity development specific to same-racial and cross-racial supervisory dyads. Such endeavors may illuminate if and how these processes are operating in supervision. For example, researchers could examine how whiteness operates in supervision (Baima & Sude, 2020). Finally, no studies to our knowledge have examined how supervisees’ own self-reported cultural humility may be related to their nondisclosure in supervision, which may be a fruitful area for investigation.

Implications for Practice, Advocacy, Education/Training, and Research

This study has notable implications for practice, advocacy, education/training, and research in counseling psychology. Regarding future research, it is important to note that the present cross-sectional study tested the associations of cultural humility and collaborative supervision with supervisee nondisclosure mediated through the supervisory working alliance in a sample comprised of mostly White women supervisees. Although findings suggested mediation of the supervisory working alliance on the outcome of supervisee nondisclosure, future research could examine mediation longitudinally to assess for temporal precedence among these associations. For example, although current theory and research in clinical supervision would suggest that supervisee nondisclosure is an outcome influenced by supervisor cultural humility, the supervisory working alliance, and collaborative supervision, perhaps these theorized associations are bidirectional, or causality flows in the opposite direction. Ultimately, it is an empirical question as to whether supervisee nondisclosure could affect supervisor cultural humility, how collaborative the supervision is, and the quality of the supervisory working alliance. Moreover, this study assessed supervisees’ perceptions of their supervisors’ level of cultural humility and collaborative style, but future dyadic studies would help evaluate whether and how supervisees’ and supervisors’ ratings of cultural humility, collaborative supervision, and the supervisory working alliance may facilitate other outcomes beyond reduced nondisclosure.

The results suggested the distinct importance of cultural humility, collaborative supervision, and a strong supervisory alliance as factors that may minimize supervisee nondisclosure, especially supervision-related nondisclosure, suggesting that these factors may be essential ingredients to positive outcomes in supervision. Although few standardized training programs for clinical supervisors exist (Borders et al., 2014), collectively, the findings here and elsewhere (e.g., Gibson et al., 2019; Hutman & Ellis, 2020; Jakob et al., 2014; Watkins et al., 2018) suggested the importance of training clinical supervisors to be culturally humble, collaborative, and able to foster effectively strong working alliances with supervisees. Likewise, supervisees could be encouraged to disclose relevant supervision-related and clinically-related content with supervisors so as to maximize the effectiveness of supervision and enhance the clinical process. In the practice of supervision, supervisors can be encouraged to make space for supervisees to bring up supervision-related and clinically-related concerns, with particular attention to power dynamics and supervisee safety, since supervision-related negative reactions may be difficult for supervisees to feel comfortable disclosing (Bernard & Goodyear, 2019).

For advocacy in the realm of clinical supervision, culturally humble supervisors could develop an anti-racist lens, employ culturally-based supervisory frameworks, empower their supervisees, and enhance supervisees’ critical consciousness (e.g., Arczynski & Morrow, 2017), tasks aimed at instilling, infusing, and modeling advocacy principles in supervision. As cultural humility inversely predicted supervisee nondisclosure through a more positive supervisory working alliance in this study, supervisors could model for supervisees how to engage with others (e.g., clients, supervisees, colleagues, staff, community members) in a humble manner (e.g., openness to explore others’ sociocultural backgrounds, asking questions when uncertain, expressing curiosity about and interest in others’ worldviews), which may benefit supervision, clinical work, and interactions with others (Hook et al., 2013). Since supervisees may have more training in multicultural supervision than their supervisors (Wong & Wong, 2020), training supervisors in microaggressions and microinterventions may also be beneficial (Sue et al., 2019).

Conclusion

Findings suggest supervision-related nondisclosure as more prevalent and more important than clinically-related nondisclosure (e.g., Gibson et al., 2019) and highlight the importance of the supervisory working alliance in supervisees’ disclosure of salient information in supervision (Bernard & Goodyear, 2019). This study contributes to the literature on clinical supervision by demonstrating the supervisory working alliance as a potential explanatory factor underlying inverse associations between supervisor cultural humility and supervisee nondisclosure as well as collaborative supervision and supervisee nondisclosure. Findings inform efforts to decrease the incidence of supervisee nondisclosure, which limits the effectiveness of supervision.

Public Significance Statement.

Findings of this study revealed that supervisor cultural humility and a collaborative supervisory style inversely predict supervisee nondisclosure through the influence of the strength of the supervisory working alliance. Results highlight supervisor cultural humility and collaborative supervision as possible facilitators of an open, safe, and supportive supervisory alliance that may increase supervisees’ disclosure of important supervision-related and clinically-related information