455 Sarıkaya, Y., & İlhan, T. (2022). Supervisory roles styles and counseling self-efficacy: Mediating role of the supervisory working alliance. International Online Journal of Education and Teaching (IOJET), 10(1). 455-473. Received : 19.08.2022 Revised version received : 03.12.2022 Accepted : 05.12.2022 SUPERVISORY ROLES STYLES AND COUNSELING SELF-EFFICACY: MEDIATING ROLE OF THE SUPERVISORY WORKING ALLIANCE Research article Yusuf Sarıkaya (0000-0003-3057-3718) Tokat Gaziosmanpaşa University yusufsarikaya@gmail.com Tahsin İlhan (0000-0002-5007-5022) Tokat Gaziosmanpaşa University tahsinilhan73@gmail.com Correspondence: yusufsarikaya@gmail.com Yusuf Sarıkaya is an assistant professor of Guidance and Psychological Counseling at the department of Educational Sciences, Faculty of Education, Tokat Gaziosmanpaşa University, Turkey. His research interests are counseling education, death anxiety, family counseling. Tahsin İlhan is a professor of Guidance and Psychological Counseling at the department of Educational Sciences, Faculty of Education, Tokat Gaziosmanpaşa University, Turkey. His research interests are positive psychology, counseling education, cognitive behavioral therapy, family counseling, Copyright © 2014 by International Online Journal of Education and Teaching (IOJET). ISSN: 2148-225X. Material published and so copyrighted may not be published elsewhere without written permission of IOJET. mailto:yusufsarikaya@gmail.com mailto:tahsinilhan73@gmail.com mailto:yusufsarikaya@gmail.com Sarıkaya& İlhan 456 SUPERVISORY ROLES STYLES AND COUNSELING SELF- EFFICACY: MEDIATING ROLE OF THE SUPERVISORY WORKING ALLIANCE Yusuf Sarıkaya yusufsarikaya@gmail.com Tahsin İlhan tahsinilhan73@gmail.com Abstract This research aimed to examine the relationship between the supervisory roles and styles with counseling self-efficacy as mediated by supervisory working alliance. The research was carried out with 848 volunteer participants studying in their final year of the guidance and counseling undergraduate program in 15 different Turkish state universities. Counselor Activity Self-Efficacy Scale, Supervisor Roles Scale, Supervisory Styles Inventory - supervisee form and Supervisory Working Alliance Inventory - supervisee form were used for data collection. The maximum likelihood method was used in the structural equation modeling performed with latent variables. The significance of the mediation effect was tested by the bootstrapping procedure. According to the results of the analysis, the task-oriented style and guidance role affected the counseling self-efficacy significantly and positively. In addition, the supervisory working alliance had a full mediating role in the relationship between task-oriented style and adviser role with counseling self-efficacy. Findings showed the effect of task-oriented style and guidance role on counseling self-efficacy beliefs, but also supported that the supervision working alliance is a fundamental component of the supervision process. Keywords: counseling self-efficacy, supervisor roles, supervisory styles, working alliance 1. Introduction Guidance and counseling education began to develop in Türkiye from the middle of the 1950s and has undergone continuous development and change to date (Aladağ & Kemer, 2016; Doğan, 1996, 2000). At undergraduate, masters and doctoral levels, supervision is a part of the development process within these programs. Especially in recent years, it appears much research related to supervision has been performed in Türkiye. Counseling supervision is defined as a relational and case-focused process where a competent and senior expert in the field monitors and audits the practice of a novice inexperienced in the field of counseling (Bernard & Goodyear, 2019; Holloway, 1995; Milne, 2007). One of the important elements in a counselor offering effective support to a client is stated to be their counseling self-efficacy (CSE) beliefs (Jaafar et al., 2009; Larson, 1998; McCarthy, 2012). Supervision offers significant opportunities to strengthen the self-efficacy beliefs of counselors in practice. Research completed in Türkiye appears not to have sufficiently researched the relationship between CSE with supervision in the context of variables. When the differences between practices in Türkiye compared with Western samples especially are evaluated, research about self-efficacy appears important. Thus, in this study, the impacts of supervisory roles, supervisory styles and supervisory working alliance (SWA) on CSE were investigated within a model framework. mailto:yusufsarikaya@gmail.com mailto:tahsinilhan73@gmail.com International Online Journal of Education and Teaching (IOJET) 2022, 10(1), 455-473. 457 1.1. Counseling Self-efficacy (CSE) Self-efficacy is one of the most central concepts in the social cognitive theory (Bandura, 1977, 1986). Bandura (1977) proposed that there was an important differentiation between efficacy expectations and outcome expectations emerging from behavior. Outcome expectations are a prediction related to the result that will emerge from a certain behavior, while efficacy expectations are beliefs related to whether the behavior that will lead to the desired outcome can be performed or not. Bandura (1977) stated that self-efficacy beliefs had greater impact on behavioral change. According to Bandura, “among the mechanisms of agency, none is more central or pervasive than people's beliefs of personal efficacy” (Bandura, 1995, s. 2). Social persuasion, performance accomplishment, vicarious learning and physiological/emotional states are four important resources feeding self-efficacy beliefs (Bandura, 1977, 1986). Based on Bandura’s theory, CSE is defined as “one’s beliefs or judgments about her or his capabilities to effectively counsel a client in the near future” (Larson & Daniels, 1998, p. 180). In research, variables like the personality traits (Larson et al., 1992), age (Tang et al., 2004), and gender (Daniels, 1997; Larson et al., 1992) of the counselor were found to have insignificant or very weak levels of correlation with CSE. Contrary to this, variables like feedback and anxiety (Daniels, 1997; Larson et al., 1992), educational experiences (Tang et al., 2004), and mastery level (Melchert et al., 1996) were found to have stronger correlations with self-efficacy. The supervision process is an important educational experience affecting the self-efficacy beliefs of counselors. For example, studies comparing counselors who received or did not receive supervision showed receiving supervision positively affected self-efficacy (Al- Darmaki, 2004; Cashwell & Dooley, 2001; Schwartz, 2016). However, research investigating the associations between different supervision approaches and methods with self-efficacy revealed that different methods did not significantly change self-efficacy beliefs (Koç, 2013; Meydan, 2015; Prindle, 2012). In short, findings of previous research showed that receiving supervision positively impacted CSE; however, the supervision method and approach did not contribute to this positive change. Thus, variables like supervision experiences, relationship with the supervisor, and supervisor behavior appear to have stronger correlations with self- efficacy (Mesrie et al., 2018; Tugendrajch et al., 2021). 1.2. Supervisory Styles and Roles Supervisory roles and styles are important variables affecting the relationship between the supervisor and supervisee. When literature related to the topic is investigated, firstly supervisory roles were defined, while supervisory style appears to be a newer concept. Supervisory roles are defined as the form or approach used when dealing with teaching material offered by the supervisor to the supervisee (Bernard, 1979). These roles are related to the professional roles undertaken and mastered within the professional development process by the supervisor and these professional roles affect their forms of behavior, functions and expectations within the supervision process (Bernard & Goodyear, 2019; Hess, 2008). It appears a variety of supervisory roles, such as teacher, consultant, counselor and evaluator, were included within several supervisory approaches and models, led by the discrimination model (Bernard, 1979) and then the systems approach model (Holloway, 1995). In short, if these roles are considered, in the teacher role the supervisor focuses on elements like teaching the desired knowledge and skills to the supervisee, while in the counselor role the supervisor is interested in topics like the anxiety, fears and previous experiences of the supervisee (Bernard, 1979). In the consultant role, the focus is expected to be on the relationship between the supervisor and supervisee (Bernard, 1979), while in the evaluator role, the supervisor is Sarıkaya& İlhan 458 expected to monitor and assess the degree to which the supervisee has learned the expected knowledge and skills. The concept of supervisory styles is a concept described to a large extent in the study by Friedlander and Ward (1984). Defining supervisory styles as clearly discriminated approach styles toward the supervision practice and supervisee, Friedlander and Ward (1984) proposed that no matter how close the concepts of supervisory style and role are to each other, the supervisory role (e.g., teacher) is based on the social context (social life) rather than the supervision process and they chose the concept of style. During development studies for an adjective-based scale, they determined three supervisory styles as attractive, interpersonal sensitive and task oriented. The attractive style refers to a warm, empathic and supportive supervisor; the interpersonal sensitive style indicates creative supervisors with strong instincts directing the relationship with the supervisee; and the task-oriented style describes supervisors who are target oriented, focusing on teaching and evaluation tasks (Friedlander & Ward, 1984). Studies observed the supervisory styles and roles were associated with both the SWA and CSE. For example, a study by Efstation et al. (1990) found significant and positive correlations between supervisory styles with SWA and CSE. However, Hanson (2006) found moderate, significant and positive correlations between the attractive and interpersonal sensitive styles with SWA and CSE, but did not find a significant correlation between the task-oriented style with SWA and CSE. Fernando and Hulse-Killacky (2005) obtained different findings. Within the scope of research they identified a model comprising all supervisory styles that explained 13% of the variance in CSE and found that the task-oriented style was the only style significantly contributing to the model. Finally, another study by Robinson (2001) of supervisors concluded that participants who stated they mainly undertook a consultant role had significantly higher SWA points. All these research findings show the relationships between supervisory styles and roles with SWA and CSE are very complex. 1.3. Supervisory Working Alliance (SWA) as a Mediator The supervision process involves a three-way relationship (i.e. supervisor, supervisee, client). The SWA is a concept emphasizing the quality of the relationship between the supervisor and supervisee. Different conceptualizations related to the SWA are known. For example, Bordin (1983) proposed that the SWA comprised the components of goal, task and emotional bond, while Efstation et al. (1990) mentioned the dimensions of client focus and rapport. Studies found that the SWA was not associated with variables like age, sex and theoretical orientation (Baker 1990; Ladany & Friedlander, 1995); however, the SWA was shown to have correlations with variables like development level, positive or negative supervision experiences and supervision relationship (Ladany & Lehrman-Waterman, 1999; Ramos-Sánchez et al., 2002). A study conducted by Humeidan (2002) revealed that SWA explained 22% of the variance in CSE. Another study by Hanson (2006) observed that among many variables related to the supervision process, SWA was the variable that explained CSE most at rates of 31%. Other studies found similar findings about the relationship between SWA and CSE (e.g., Trangucci, 2013; Tsai, 2015). All these findings show that SWA is a basic element of the supervision relationship and that self-efficacy beliefs are strengthened as the quality of the relationship with the supervisor increases. Correlations between SWA and CSE can be said to show the power-based and hierarchical nature of the supervision relationship (Morgan & Sprenkle, 2007). Supervisors act as gatekeepers in the supervision process (Bernard & Goodyear, 2019). They assess the supervisee during supervision. For this reason, they are placed above the supervisee in International Online Journal of Education and Teaching (IOJET) 2022, 10(1), 455-473. 459 hierarchical terms. As a result, it is not wrong to state that the supervision relationship is more affected by supervisor behavior and forms of approach. Thus, supervisory roles and styles are encountered as important variables affecting the SWA. Many studies in the literature found significant correlations between supervisory styles and roles with the SWA (Efstation et al., 1990; Hanson, 2006; Lorenz, 2009; Spelliscy 2007; Rarick 2001). The research findings summarized here indicate that supervisory roles and styles positively affect the SWA and thus supervision within the environment of a quality relationship is assessed as strengthening the self-efficacy beliefs of the supervisee. Within this assessment framework, the decision was made to test the model given in Figure 1 within the scope of this research. Note: CASES: The Counselor Activity Self-Efficacy Scale total score; SWAI: The Supervisory Working Alliance Inventory total score Figure 1. Proposed model 2. Method 2.1. Participants The research was carried out with volunteer participants attending the final year of guidance and counseling undergraduate programs who had completed the supervision process. Within this scope, students attending 15 state universities located in different regions of Türkiye who had completed at least 5 sessions of counseling were included in the study. Firstly data were collected from a 952-person research group. However, preliminary analyses examined to test the fit of the structural equation modeling (i.e. missing data, outliers, normality) excluded some data from the research. Thus, analyses were conducted with an 848-person research group. Hair, Black, Babin and Anderson (2010) stated that structural equation modelling studies generally require large samples. For example, they recommended a sample group of at least 500 people for models containing more than seven constructs. Similarly, Kline (2011) stated rules such as 20:1 and 10:1 for the proportion of sample size to observed variables in the model. Within this scope, the 848-person working group of the study was assessed as sufficient. Sarıkaya& İlhan 460 The majority of participants were female (67.45%) and mean age was 22.26 years. The majority had received 6-10 weeks of supervision (58.37%) with 78.55% of participants stating they were supervised for 1-3 hours per week. The proportion of participants stating they only received individual supervision as the supervision method was 16.51%, while 42.99% of participants stated they only received supervision in a group. The majority of participants defined the approaches they used during counseling as cognitive behavioral and eclectic. Information related to participants is presented in Table 1. Table 1. Sample Characteristics n (%) 𝒙 sd Age 843 22.26 1.17 Gender Male 276 (% 32.55) Female 572 (% 67.45) Number of supervision sesssions 6-10 495 (% 58.37) 9.67 4.42 11 and above 339 (% 41.63) Supervision hours for per week 1-3 652 (% 78.55) 2.57 1.42 4-6 178 (% 21.45) Number of cases (client) 1 case 283 (% 33.97) 2.43 1.74 2 case 286 (% 34.33) 3 + case 264 (% 31.70 Number of counseling sessions with real clients 5-10 548 (% 66.10) 9.30 5.54 11-15 169 (% 20.39) 16 and above 112 (% 13.51) Supervision method Individual 139 (% 16.51) Group 362 (% 42.99) Individual and group 341 (% 40.50) Counseling approach Cognitive- Behavioral 269 (% 32.06) Eclectic 282 (% 33.61) Person centered 130 (% 15.49) Solution focused 88 (% 10.49) Existential 16 (% 1.91) Gestalt 14 (% 1.67) Adlerian 13 (% 1.55) Psychoanalysis 20 (% 2.38) Reality therapy 7 (% .84) Total 848 2.2. Instruments 2.2.1. Counselor Activity Self-Efficacy Scale (CASES) The CASES was developed by Lent et al. (2003) with the aim of measuring the self-efficacy perceptions of counselors (i.e. supervisees) related to counseling. The scale comprises 3 factors of helping skill self-efficacy, session management self-efficacy and counseling challenges self- efficacy. High points obtained from the CASES indicate high levels of self-efficacy. The alpha coefficient for the whole scale was found to be .97. The Turkish adaptation study for the scale was carried out by Pamukçu and Demir (2013) and the McDonald’s omega coefficient was .98 for the whole scale. The Cronbach’s alpha reliability coefficient for the scale was recalculated in this research and found to be .95. International Online Journal of Education and Teaching (IOJET) 2022, 10(1), 455-473. 461 2.2.2. Supervisor Roles Scale (SRS) The SRS, based on the perceptions of supervisees about the roles undertaken in the supervision process by supervisors, was developed by İlhan et al. (2015). The scale comprises three subscales of teacher role, counselor role and adviser role. The authors stated that the structure emerging in scale development studies confirmed Bernard’s (1979) discriminant model and hence the subscales were equivalent to the roles included in this model (i.e. teacher, counselor, consultant). High points obtained from the subscales mean that the role is displayed more. The Cronbach’s alpha coefficients for the subscales were reported to vary from .97 to .91 (İlhan et al., 2015). The Cronbach’s alpha reliability coefficients for the SRS were recalculated within the scope of this research and were observed to be between .88 and .96. 2.2.3. Supervisory Styles Inventory (SSI) The SSI was developed by Friedlander and Ward (1984) with the aim of measuring supervisory styles by the supervisor toward the supervisee. It was developed with two parallel forms for the supervisee (e.g., counselor) and supervisor. In this research, only the form developed for supervisee was used. The scale comprises three dimensions called attractive, interpersonal sensitive and task oriented (Friedlander & Ward, 1984). The internal consistency Cronbach’s alpha coefficients for the subscales were reported to vary from .93 to .84 (Friedlander & Ward, 1984). The adaptation of the SSI to Turkish was carried out by Atik (2017). The adaptation study stated the Cronbach’s alpha reliability coefficients varied from .94 to .95 for the subscales. The Cronbach’s alpha coefficients for the subscales were recalculated in this research and found to vary from .92 to .93. 2.2.4. Supervisory Working Alliance Inventory (SWAI) The SWAI was developed by Efstation et al. (1990) based on opinions related to the working alliance of Greenson (1965/2008) and Bordin (1983). It was developed as two separate forms for the supervisor and supervisee and in this research only the supervisee form was used. This form of the SWAI comprises two subscales named rapport and client focus. The Cronbach’s alpha coefficients for these subscales were found to be .90 and .77, respectively (Efstation et al., 1990). The Turkish adaptation study for the SWAI supervisee form was carried out by Sarıkaya (2017). The adaptation study reported the Cronbach’s alpha coefficients for rapport and client focus were .95 and .93, respectively. The Cronbach’s alpha coefficients were recalculated with the data in the present research and found to be .94 and .93. 2.3. Data Analysis Within the scope of the research, first the mean, standard deviation, kurtosis, skewness and correlation values related to the independent, dependent and mediator variables were investigated. With the aim of making the dataset suitable for structural equation modeling, missing data, normality, linearity, outliers and multicollinearity analyses were performed. The predictive and mediating relationships between variables were analyzed using structural equation modelling. Maximum likelihood was used as the estimation method. Evaluation of the model firstly investigated the chi-square and chi-square/df ratio. Additionally, the comparative fit index (CFI), Tucker-Lewis index (TLI), goodness of fit index (GFI), standardized root mean square residual (SRMR) and root mean square error of approximation (RMSEA) values were reported. Acceptable fit values used the criteria of χ2/df < 5, CFI, TLI, GFI > .90, SRMR < .05, RMSEA < .08 (Browne & Cudeck, 1993; Byrne, 2001; Hair et al., 2010; Hu & Bentler 1999; Tabachnick & Fidell, 2007). Sarıkaya& İlhan 462 In order to reduce the number of observed variables and control the measurement error in structural equation modelling of latent variables, the parceling method was used (Nasser-Abu Alhija & Wisenbaker, 2006). In the parceling method, items are randomly assigned to the parcels (Little et al., 2002). While 3 parcels each were formed for the attractive style, interpersonal sensitive style, task-oriented style, adviser role, counselor role and session management self-efficacy variables, 4 parcels each were used for the teacher role, helping skill self-efficacy, and counseling challenges self-efficacy variables. The parcels were added to the models as observed variables of the latent structures. In models using total CSE scores, the three subscales for self-efficacy were each added as observed variables. Similarly, the two subscales for the variable of SWA were added to the model as observed variables. The significance of indirect relationships in mediation models was investigated with the bootstrap method as recommended by the literature (Shrout & Bolger, 2002). Within this scope, a 5000-person bootstrap sample was created in the 95% confidence interval. As the confidence interval did not include zero, the mediation was interpreted to be significant. 3. Results 3.1. Descriptive Statistics and Zero-Order Correlations The mean, standard deviation and correlation coefficients for variables considered within the scope of the research are presented in Table 2. When correlation coefficients are investigated, there appeared to be positive significant correlations between the supervisory roles (i.e. teacher, counselor, adviser) and styles (i.e. interpersonal sensitive, attractive, task oriented) with CSE. The correlation coefficients between supervisory roles and styles with self- efficacy varied from .29 to .34 (p < .01). Similarly, the coefficients between supervisory roles and styles with SWA varied from .75 to .87 (p < .01). Finally, the relationship between SWA and self-efficacy was positive and significant (r = .36, p < .01). 463 Table 2. Descriptive statistics and zero order correlations Descriptive Zero-order correlations M SD CASES HSSE SMSE CCSE SWAI Rapport Client focus Teacher Counselor Adviser Attractive Interpersonal sensitive CASES 237.63 45.63 - HSSE 94.80 17.83 .87** - SMSE 63.68 13.08 .88** .80** - CCSE 79.16 22.17 .84** .51** .55** - SWAI 5.03 1.28 .36** .36** .34** .24** - Rapport 5.09 1.31 .35** .36** .34** .23** .97** - Client focus 4.92 1.38 .32** .33** .31** .22** .92** .82** - Teacher 60.89 16.57 .31** .28** .29** .24** .80** .74** .80** - Counselor 48.04 14.65 .29** .28** .27** .21** .87** .86** .79** .85** - Adviser 31.04 7.22 .34** .33** .33** .23** .76** .73** .73** .83** .79** - Attractive 5.42 1.33 .30** .30** .29** .20** .82** .83** .70** .70** .79** .68** - Interpersonal sensitive 5.40 1.21 .30** .31** .32** .18** .79** .75** .76** .77** .75** .74** .82** - Task oriented 5.35 1.09 .31** .33** .33** .18** .75** .69** .75** .76** .71** .75** .73** .89** Note: CASES: Counselor Activity Self-Efficacy Scale total score; HSSE: helping skill self-efficacy; SMSE: session management self-efficacy; CCSE: counseling challenges self-efficacy; SWAI: Supervisory Working Alliance Inventory total score; Rapport, Client focus: Subscales of the SWAI; Teacher, Counselor, Adviser: Supervisor roles; Attractive, Interpersonal sensitive, Task oriented: Supervisory Styles. **p < .01 464 3.2. Measurement Model The measurement model was tested for whether it showed good fit or not. The measurement model included 8 latent variables, with 3 for supervisory roles, 3 for supervisory styles, SWA and CSE. The parcels related to supervisory roles and styles variables were added to the model as observed variable. Subscale scores for CSE and SWA variables were added as observed variables. Thus, the measurement model included 8 latent constructs represented by 24 observed variables. As a result of analysis carried out with the maximum likelihood method, all indexes revealed there was good fit between the model and the research data (χ2(224) = 943.16, p < .001, χ2/df = 4.21, RMSEA = .06, CFI = .97, SRMR = .03, TLI = .96, GFI = .89). The factor loadings included in the measurement model were all positive and significant (range .60-.95, p < .001). 3.3. Structural Model While the independent variables of the structural model were all supervisor roles and styles, the SWA added as the mediator variable in the model. CSE was the dependent variable in the model (see Figure 1 for proposed model). Analyses performed with the maximum likelihood method firstly tested the full mediation model and then tested the partial mediation model. When the full mediation model is tested with structural equation modelling, values indicating weak fit were obtained (χ2/df = 5.37, RMSEA = .08). Additionally, it appeared that the paths between some supervisory roles and styles to SWA were insignificant. In the next stage, model variations with different numbers of independent variables were separately tested and some independent variables without significant path coefficients were removed from the model. In conclusion, the model including only task-oriented style and adviser role as independent variables (Model 1) was found to have acceptable goodness of fit values (see Table 3). All paths in the model were significant (p < .001). The standardized path coefficients for the model are presented in Figure 2. A similar process was followed when testing partial mediation. Task-oriented style and adviser role remained as independent variables in the model (Model 1a). In this model. The paths between adviser role and task-oriented style to self-efficacy were not significant (p > .05). The other paths were significant ( p < .001). The path coefficients between variables in model 1a are shown in Figure 2. This model was found to have acceptable goodness of fit values (see Table 3). However, the decision was made to accept model 1 (full mediation model) based on the lack of significance for direct paths between adviser role and task-oriented style with self-efficacy and the lack of significant increase in model fit for these new paths added to the model (Δχ2 = 3.96, df = 2, p > .05). Model 1 Model 1a Note: CASES: Counselor Activity Self-Efficacy Scale total score; SWAI: Supervisory Working Alliance Inventory total score **p < .01 Figure 2. Full and partial mediated structural models International Online Journal of Education and Teaching (IOJET) 2022, 10(1), 455-473. 465 Table 3. Fit indexes of structural and alternative models Models χ2 df χ2/df RMSEA CFI SRMR TLI GFI Model 1 129.23 40 3.23 .05 .99 .02 .98 .97 Model 1a 125.27 38 3.42 .05 .99 .02 .98 .97 Model 2 134.02 50 2.68 .04 .99 .02 .99 .97 Model 3 123.44 40 3.09 .05 .99 .02 .99 .97 Note: For all chi-square p < .001 3.4. Bootstrapping The significance of mediation analysis was tested with the bootstrapping procedure. The mediation effect was tested with 5000 bootstrap resamples in the 95% confidence interval. As a result of the analysis, the indirect effects were found to be significant. SWA mediated the relationship between task-oriented style and adviser role with self-efficacy. The lower and upper coefficients for direct and indirect effects are given in Table 4. Table 4. Direct and indirect effects for models Model Pathways Effect 95% CI Lower bonds 95% CI Upper bonds Model1 Direct effect Task oriented →SWAI .39 .28 .50 Adviser →SWAI .54 .42 .65 SWAI→CASES .42 .35 .49 Indirect effect Task oriented →SWAI→CASES .16 .12 .22 Adviser →SWAI→CASES .23 .17 .30 Model 2 Direct effect Task oriented →SWAI .39 .28 .50 Adviser →SWAI .54 .42 .65 SWAI→HSSE .40 .33 .47 indirect effect Task oriented →SWAI→ HSSE .16 .11 .21 Adviser →SWAI→ HSSE .22 .16 .28 Model 3 Direct effect Task oriented →SWAI .39 .28 .50 Adviser →SWAI .54 .42 .65 SWAI→SMSE .38 .31 .45 Indirect effect Task oriented →SWAI→ SMSE .15 .10 .21 Adviser →SWAI→ SMSE .21 .15 .27 Note: CI: Confidence interval; CASES: Counselor Activity Self-Efficacy Scale total score; HSSE: helping skill self-efficacy; SMSE: session management self-efficacy; SWAI: Supervisory Working Alliance Inventory total score Sarıkaya& İlhan 466 3.5. Alternative Models Some alternative models were developed and tested in accordance with the cross-sectional structure and exploratory nature of the research. The alternative models dealt with each of the subdimensions of CSE as dependent variables. Thus, the direct and indirect effects of supervisory roles and styles and the mediation effect of SWA were investigated in these alternative models. The procedures followed in previous models were performed for analysis of alternative models. For ease of reading, the process related to the alternative models is not reported in detail, but information related to models determined to be more acceptable is presented. In the first alternative model, the dependent variable was helping skill self-efficacy (Model 2). All supervisory roles and styles were independent variables in the model, while SWA was the mediating variable. Structural equation modeling performed with the maximum likelihood method firstly tested the full mediation model and values indicating weak fit were obtained. As with previous models, a variety of model variations were trialed with some insignificant paths deleted in order. Finally, the model in Figure 3 was found to have acceptable goodness of fit values (see Table 3). As the partial mediation model tested for model 2 did not have better fit, the full mediation model was determined to be more acceptable. All path coefficients in the full mediation model were significant (see Figure 3). After the bootstrap procedure, the indirect effects of the full mediation model were observed to be significant (see Table 4). These results show that task-oriented style and adviser role contribute to the variance related to helping skills self-efficacy mediated by SWA. Another alternative model that was tested dealt with session management self-efficacy as the independent variable (Model 3). The first analyses of this model did not have good fit and the model given in Figure 3 was reached by removing some variables. The goodness of fit values related to model 3 are given in Table 3. The partial mediation model was tested; however, the full mediation model was determined to be more appropriate. All path coefficients and indirect effects in the model were significant. The path coefficients are given in Figure 3, while the values related to indirect effects are shown in Table 4. The findings related to model 3 indicate that SWA has a full mediating role in the relationship between task-oriented style and adviser role with session management self-efficacy. Finally, the model with counseling challenges self-efficacy as dependent variable was tested in terms of full mediation and partial mediation effects. In trials of many models with the maximum likelihood method, no acceptable goodness of fit values were found. For this reason, alternative models with counseling challenges self-efficacy as dependent variable could not be tested. Model 2 Model 3 Note: HSSE: helping skill self-efficacy; SMSE: session management self-efficacy; SWAI: Supervisory Working Alliance Inventory total score **p < .01 Figure 3. Alternative models and standardize path coefficients International Online Journal of Education and Teaching (IOJET) 2022, 10(1), 455-473. 467 4. Discussion When the models tested within the scope of the research are investigated, it appeared the SWA played a full mediating role in the effect of task-oriented style and adviser role on CSE. In other words, when the SWA variable is controlled, the effects of this style and role on self- efficacy become insignificant. These findings show that SWA is an important element in the impacts of task-oriented style and adviser role on CSE. Furthermore, the findings indicate the importance of adviser role and task-oriented style in terms of the supervisory relationship and CSE. The supervision process with task-oriented style indicates focus on targets, planned, practical and concrete supervisors (Friedlander and Ward, 1984). The findings obtained in the research show the task-oriented style does not just predict total scores for CSE but also significantly and positively predicts the helping skill self-efficacy and session management self-efficacy mediated by the SWA. The SWA makes it mandatory for the supervisor and supervisee to reach a consensus about supervision goals and tasks (Bordin, 1983). Furthermore, if rapport cannot be ensured with the supervisee, it was stated that the SWA cannot be formed (Efstation et al., 1990). Supervisors approaching the supervisee with the task-oriented style, offering a structured training environment, may be considered to be able to determine common goals and tasks about supervision more easily. In the literature, studies showing that supervision with task-oriented style has more association with the tasks, goals and client focus subdimensions of SWA support this idea (e.g., Efstation et al., 1990; Ladany et al., 2001). A supervisee concretely learning the tasks required to complete the supervision process and expending effort to meet the agreed-upon common goals may have increased commitment to the supervisor and supervisory process. In the literature, a variety of studies showing positive significant correlations between task-oriented style with SWA support this view (e.g., Ladany et al., 2001; Lorenz, 2009; Rarick, 2001). Findings related to mediation analysis revealed that the task-oriented style displayed without SWA will not have an effect on CSE. These findings are thought to be consistent with previous studies. For example, a study performed by Fernando and Hulse-Killacky (2005) revealed that task-oriented style was the only style predicting CSE. A similar finding was reached by Meissner (2012) who found that among supervisory styles, the task-oriented style was the only style affecting the competence perception of the supervisee. A study conducted by Efstation et al. (1990) investigated the relationship between supervisory styles and CSE and the highest correlation belonged to the task-oriented style. In the literature, no study investigating the mediating role of the supervisory work alliance in the relationship between task-oriented style with CSE has been found. Though counselor candidates have received theoretical lessons previously, they complete their first counseling practice under supervision. Approaching these very inexperienced counselor candidates with the task-oriented style is assessed to have positive contribution to their self-efficacy perception. It is more probable that counselor candidates will achieve performance accomplishments by working with a task-oriented supervisor. Rønnestad and Skovholt (1993) stated that supervision practice completed with counselors at starting level and psychotherapy students were typically very structured, didactic oriented and more technically focused. Additionally, students learning many models and techniques only from a conceptual perspective in previous times may experience great anxiety in putting these into practice and they stated the need to perform structured supervision including positive feedback, support and encouragement to reduce this anxiety (Rønnestad & Skovholt, 1993; Skovholt & Rønnestad, 1992). For this reason, the task-oriented style may be said to provide the opportunity for educational experiences responding to the professional developmental needs of counselor candidates. In light of all these assessments and findings, it is not wrong to say that the SWA is a very important element for the impact of task-oriented style on CSE. Sarıkaya& İlhan 468 When the research findings are investigated, the SWA had a full mediating role in the impact of the adviser role on CSE total points and the subdimensions of helping skill self-efficacy and session management self-efficacy. Stated more openly, supervision with an adviser role significantly contributes to creating the SWA, and when the effect of SWA is controlled, the impact of adviser role on CSE becomes insignificant. Also, in all models tested within the scope of the research and with good fit, the only supervisor role predicting counselor self- efficacy via SWA was the adviser role. These findings reveal that the supervisory role of adviser may be relatively more important compared to the other roles. The adviser role, equivalent to the consultant role in the discrimination model developed by Bernard (1979), requires formation of an egalitarian relationship with the supervisee. Supervisors using this role act like colleagues of the counselor candidate and guide them in decision-making processes (Bernard, 1997). Furthermore, they believe in the power of the counselor candidate to solve problems related to the client (Bernard & Goodyear, 2019). As stated previously, the SWA requires the supervisor and supervisee to reach consensus about supervision goals and tasks (Bordin, 1983). Additionally, if a strong bond is not formed between the supervisor and supervisee and weak rapport develops, it was stated the SWA may not be adequately founded (Bordin, 1983; Efstation et al., 1990). The egalitarian nature of the adviser role is thought to provide significant opportunities in terms of these dimensions of the SWA. A supervisee feeling much closer in hierarchical terms to a supervisor displaying the adviser role may expend more effort in determining common goals and tasks and may state their desires and expectations without fear of judgement or rejection. This type of relationship may be expected to increase the rapport between the supervisor and supervisee. A variety of studies conducted within the scope of Bernard’s (1979) discrimination model revealed the teacher and counselor role were open and clear factors, while they stated the consultant role was a fuzzier factor (Stenack & Dye, 1982). Similar findings were reached in the study by İlhan et al. (2015), revealing that the adviser role was the most unclear factor. This fuzziness of the adviser role is related to including various aspects of both the teacher role and the counselor role (Bernard & Goodyear, 2019). It is thought that this fuzziness may play a role in the positive effect of the adviser role on CSE. For example, holding discussions about which method the supervisee will use is a task in the adviser role (Bernard & Goodyear, 2019; Stenack & Dye, 1982). The supervisee may learn a variety of information during these discussions. However, this information is mostly acquired by the supervisee using their own resources. For this reason, rather than the supervisor being just a teacher directly teaching information, they guide the supervisee in reaching the information they need themselves. Similarly, a supervisor adopting the adviser role and believing in the power of the supervisee may be expected to observe their preparation and anxiety and act as a counselor, though only partially. The characteristics related to the adviser role appear to be consistent with self- efficacy resources. In hierarchical terms, it is more probable that a supervisee who feels close to their supervisor will experience less anxiety and thus achieve diverse performance accomplishments when completing their first practice. This role offers the opportunity for counselor candidates to discover their own resources by focusing on revealing the power of the supervisee. When the literature is investigated, research related to supervisory roles appears to be very limited. However, there are still some studies supporting the findings of this research. A study by Robinson (2001) reported that counselor candidates reporting their supervisors took consultant roles had significantly higher points related to the client focus subdimension for the SWA. The same study found that for supervisor role perceptions of counselor candidates, the consultant role was in first place with the teacher role in last place. A study carried out by International Online Journal of Education and Teaching (IOJET) 2022, 10(1), 455-473. 469 Clemente (2006) found that participants stating that supervisors mostly took the consultant role had higher self-efficacy points. This finding shows that the consultant role may have an important place in counselor education. For this reason, the positive effect of the adviser role, equivalent to the consultant role, on self-efficacy appears to be consistent with the literature 4.1. Limitations and Recommendations The attractive style, interpersonal sensitive style, teacher role and counselor role were not included in the models tested with good fit within the scope of the research. The probable reason for this situation may be related to the nature of these styles and roles. Supervision practice completed in Türkiye is relatively short. When Table 1 is investigated, the majority of participants received nearly 10 sessions of supervision. The same table shows that supervision practice was mainly completed in groups. In relatively short-term supervision practice performed with groups, the attractive style, interpersonal sensitive style and counselor roles may not be displayed sufficiently. Furthermore, though some of these roles and styles (i.e. attractive style, teacher role, counselor role) are displayed, they may strengthen the hierarchical relationship pattern between the supervisor and supervisee. Thus, these styles and roles may negatively affect the supervisory relationship and self-efficacy. When the literature is investigated, though there are studies showing positive and significant correlations of these roles and styles (e.g., Hanson 2006; Ladany et al., 200; Lorenz, 2009), there are also studies showing no relationship or negative relationships (e.g., Atik, 2017, Meissner, 2012; Meydan 2021). All these findings show that there is a need for more studies to explain the functions of supervisory roles and styles. Some alternative models were tested involving the CSE subdimensions within the scope of the research. However, the model considering the counseling challenges self-efficacy subdimension as dependent variable could not be tested as it did not have good fit. Thus, within the scope of this model, the direct and indirect effects of supervisory role and style and the mediating role of SWA could not be investigated. Counseling challenges self-efficacy encompasses self-efficacy about coping with difficult cases like clients at risk of suicide, with severe cognitive disorders and with previous traumatic experiences (Lent et al., 2003). It is assessed that the chances of the supervisee meeting these types of cases in their first practice during the supervision process is very low. This is considered to be the probable reason for the lack of good fit for the model about counseling challenges. Findings obtained in the research reflect the perceptions of the supervisee. Research to reflect the perceptions of the supervisors may be carried out. Supervision is not just performed with counselors, but also with other professional groups like psychiatrists, clinical psychologists and social service workers. This element should be noted when generalizing the findings of this research completed with counseling students. Finally, this study was completed with counselors at undergraduate level. It should not be forgotten that supervision at masters and doctoral level may involve different dynamics. Sarıkaya& İlhan 470 References Aladağ, M., & Kemer, G. (2016). Clinical supervision: An emerging counseling specialty in Turkey. The Clinical Supervisor, 35(2), 175-191. Al-Darmaki, F. R. (2004). Counselor training, anxiety, and counseling self-efficacy: Implications for training psychology students from the United Arab Emirates University. Social Behavior & Personality: An International Journal, 32(5), 429-439. Atik, Z. (2017). Psikolojik danışman adaylarının bireyle psikolojik danışma uygulaması ve süpervizyonuna ilişkin değerlendirmeleri. Unpulished doctoral dissertation, Hacettepe Üniversitesi, Ankara, Türkiye. Baker, D. E. (1990). The relationship of the supervisory working alliance to supervisor and supervisee narcissism, gender, and theroetical orientation. Unpulished doctoral dissertation, University of Southern California, USA. Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84(2), 191-215. Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall. Bandura, A. (1995). Self-efficacy in changing societies. New York: Cambridge University. Bernard, J. M. (1979). Supervisor training: A discrimination model. Counselor Education and Supervision, 19, 60-68. Bernard, J. M., & Goodyear, R. K. (2019). Fundamentals of clinical supervision. Pearson. Bordin, E. S. (1983). A working alliance based model of supervision. The Counseling Psychologist, 11(1), 35-42. Browne, M. & Cudeck, R. (1993). Alternative ways of assessing model fit. In: K. A. Bollen, & J. S. Long (Eds.), Testing structural equation models. (pp. 136-162). Newbury Park, CA: Sage Byrne, B.M. (2001). Structural equation modeling with Amos: Basic concepts, applications, and programming. New York: Springer - Verlag. Cashwell, T. H., & Dooley, K. (2001). The impact of supervision on counselor self- efficacy. The Clinical Supervisor, 20(1), 39-47. Clemente, C. L. (2006). The relationship between perceived supervisory roles, styles, and working alliances and students' self-efficacy in speech-language pathology practicum experiences. Unpulished doctoral dissertation, Touro University International, USA Daniels, J. A. (1997). The influence of performance feedback and causal attributions upon ratings of counseling self-efficacy. Unpulished doctoral dissertation, University of Nebraska, USA. Doğan, S. (1996). Türkiye’de psikolojik danışma ve rehberlik alanında meslek kimliğinin gelişimi ve bazı sorunlar. Türk Psikolojik Danışma ve Rehberlik Dergisi, 2(7), 32-44. Doğan, S. (2000). Türkiye'de psikolojik danışma ve rehberliğin durumu ve geleceğe ilişkin yönelimler. Eğitim ve Bilim, 25(118), 3-8. Efstation, J. F., Patton, M. J., & Kardash, C. M. (1990). Measuring the working alliance in counselor supervision. .Journal of Counseling Psychology, 37(3), 322-329. International Online Journal of Education and Teaching (IOJET) 2022, 10(1), 455-473. 471 Fernando, D. M., & Hulse-Killacky, D. (2005). The relationship of supervisory styles to satisfaction with supervision and the perceived self-efficacy of master's-level counseling students. Counselor Education and Supervision, 44(4), 293-304. Friedlander, M. L., & Ward, L. G. (1984). Development and validation of the Supervisory Styles Inventory. Journal of Counseling Psychology, 31(4), 541-557. Greenson, R. (2008). The working alliance and the transference neurosis. The Psychoanalytic Quarterly, 77(1), 77-102 (Reprint from Psychanalytic Quarterly, 34, s. 155-181, 1965). Hair Jr, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate data analysis. Pearson Education. Hanson, M. G. (2006). Counselor self-efficacy: Supervision contributions, impact on performance, and mediation of the relationship between supervision and performance. Unpulished doctoral dissertation, Southern Illinois University at Carbondale, USA. Hess, A. K. (2008). Psychotherapy supervision: A conceptual review. In Psychotherapy supervision: Theory, research, and practice (A. K. Hess, K. D. Hess, & T. H. Hess (Eds.), (pp. 3-24). John Wiley & Sons. Holloway, E. (1995). Clinical supervision: A systems approach. Thousand Oaks, CA: Sage Publication. Hu, L.T. & Bentler, P.M. (1999). Cut-off criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. Humeidan, M. A. (2002). Counseling self-efficacy, supervisory working alliance, and social influence in supervision. Unpulished doctoral dissertation, Ball State University, USA. İlhan, T., Sarıkaya, Y., Başkal, A., & Yöntem, M. K. (2015, October 7-9). Süpervizör Rolleri Ölçeği’nin geliştirilmesi: İlk psikometrik bulgular [Conference presentation]. 13rd National Counseling and Guidance Congress, Mersin, Türkiye. Jaafar, W. W., Mohamed, O., Bakar, A. R., & Tarmizi, R. A. (2009). The Influence of counseling self-efficacy towards trainee counselor performance. International Journal of Learning, 16(8), 247-260. Kline, R. B. (2011). Principles and practice of structural equation modeling. New York: The Guilford Press. Koç, İ. (2013). Kişiler arası süreci hatırlama tekniğine dayalı süpervizyonun psikolojik danışman adaylarının psikolojik danışma becerilerine, özyeterlik ve kaygı düzeylerine etkisi. Unpulished doctoral dissertation. Ege Üniversitesi, İzmir. Ladany, N., & Friedlander, M.L. (1995). The relationship between the supervisory working alliance and trainees’ experience of role conflict and role ambiguity. Counselor Education and Supervision, 34(3), 220-231. Ladany, N., & Lehrman-Waterman, D. (1999). The content and frequency of supervisor self- disclosures and their relationship to supervisor style and the supervisory working alliance. Counselor Education and Supervision, 38(3), 143-160. Ladany, N., Walker, J. A., & Melincoff, D. S. (2001). Supervisory style: Its relation to the supervisory working alliance and supervisor self-disclosure. Counselor Education & Supervision, 40(4), 263-275. Sarıkaya& İlhan 472 Larson, L. M. (1998). The social cognitive model of counselor training. The Counseling Psychologist, 26, 219-273. Larson, L. M., & Daniels, J. A. (1998). Review of counseling self-efficacy literature. The Counseling Psychologist, 26, 179-218. Larson, L. M., Suzuki, L. A., Gillespie, K. N., Potenza, M. T., Bechtel, M. A., & Toulouse, A. (1992). Development and validation of the counseling self-estimate inventory. Journal of Counseling Psychology, 39, 105-120. Lent, R. W., Hill, C. E., & Hoffman, M. A. (2003). Development and validation of the counselor activity self-efficacy scales. Journal of Counseling Psychology, 50, 97-108. Little, T. D., Cunningham, W. A., Shahar, G., & Widaman, K. F. (2002). To parcel or not to parcel: Exploring the question, weighing the merits. Structural equation modeling, 9(2), 151-173. Lorenz, D. C. (2009). Counseling self-efficacy in practicum students: Contributions of supervision. Unpulished doctoral dissertation, The Pennsylvania State University, USA. McCarthy, A. K. (2012). Relationships between supervisory working alliance, counselor self- efficacy, and successful client outcomes, and in state/federal rehabilitation counseling system. Unpulished doctoral dissertation, Northern Illinois University, USA. Melchert, T. P., Hays, V. L., Wiljanen, L. M., & Kolocek, A. K. (1996). Testing models of counselor development with a measure of counseling self-efficacy. Journal of Counseling & Development, 74(6), 640-644. Mesrie, V., Diener, M. J., & Clark, A. (2018). Trainee attachment to supervisor and perceptions of novice psychotherapist counseling self-efficacy: The moderating role of level of experience. Psychotherapy, 55(3), 216-221. Meydan, B. (2015). Bireyle psikolojik danışma uygulamasında mikro beceri süpervizyon modelinin etkililiğinin incelenmesi. Türk Psikolojik Danışma ve Rehberlik Dergisi, 5(43), 55-68. Meydan, B. (2021). Turkish First-Time Supervisees' Counseling Self-Efficacy. Eurasian Journal of Educational Research, 92, 61-78. Milne, D. (2007). An empirical definition of clinical supervision. British Journal of Clinical Psychology, 46(4), 437-447 Morgan, M. M., & Sprenkle, D. H. (2007). Toward a common-factors approach to supervision. Journal of Marital and Family Therapy, 33(1), 1-17. Nasser-Abu Alhija, F., & Wisenbaker, J. (2006). A Monte Carlo study investigating the impact of item parceling strategies on parameter estimates and their standard errors in CFA. Structural Equation Modeling, 13(2), 204-228. Pamukçu, B., & Demir, A. (2013). Psikolojik danışma öz-yeterlik ölçeği Türkçe formunun geçerlik ve güvenirlik çalışması. Türk Psikolojik Danışma ve Rehberlik Dergisi, 5(40), 212-221. Prindle, J. E. C. (2012). Narrative reflecting team supervision: A literature review and quantitative examination of narrative reflecting team supervisions' impact on self- efficacy of counseling trainees. Unpulished doctoral dissertation, Oregon State University, USA. International Online Journal of Education and Teaching (IOJET) 2022, 10(1), 455-473. 473 Ramos-Sánchez, L., Esnil, E., Goodwin, A., Riggs, S., Touster, L. O., Wright, L. K., ... Rodolfa, E. (2002). Negative supervisory events: Effects on supervision and supervisory alliance. Professional Psychology: Research and Practice, 33(2), 197-202. Rarick, S. L. (2001). The relationship of supervisor and trainee gender match and gender attitude match to supervisory style and the supervisory working alliance. Unpulished doctoral dissertation, Lehigh University, USA. Robinson, J. K. G. (2001). Relationships between perceived supervisory styles and counseling students' self-efficacy in practicum. Unpulished doctoral dissertation, University of South Carolina, USA. Rønnestad, M. H., & Skovholt, T. M. (1993). Supervision of beginning and advanced graduate students of counseling and psychotherapy. Journal of Counseling & Development, 71(4), 396-405. Sarıkaya, Y. (2017). Süpervizör rolleri, tarzları ve süpervizyon terapötik ittifakının psikolojik danışma öz yeterliği ile ilişkisi. Unpulished doctoral dissertation, Tokat Gaziosmanpaşa Üniversitesi, Tokat, Türkiye. Schwartz, G. L. (2016). The relationship of supervision variables to counseling self-efficacy and outcome expectations among practicing psychologists. Unpulished doctoral dissertation, Andrews University, USA. Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7, 422– 445. Skovholt, T. M., & Rønnestad, M. H. (1992). Themes in therapist and counselor development. Journal of Counseling and Development, 70, 505-515. DOI: 10.1002/j.1556-6676.1992.tb01646.x Spelliscy, D. (2007). Predicting supervisee role conflict/ambiguity: A model of attachment, supervisory style, and supervisory working alliance. Unpulished doctoral dissertation, Fordham University, USA. Stenack, R. J., & Dye, H. A. (1982). Behavioral descriptions of counseling supervision roles. Counselor Education and Supervision, 21(3), 295-304. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics. Boston: Pearson Education. Tang, M., Addison, K. D., LaSure-Bryant, D., Norman, R., O’Connell, W., & Stewart-Sicking, J. A. (2004). Factors that influence self-efficacy of counseling students: An exploratory study. Counselor Education and Supervision, 44(1), 70-80. Trangucci, K. A. (2013). The supervisory working alliance and self-efficacy of school psychology graduate interns. Unpulished doctoral dissertation, Fordham University of South Carolina, USA. Tsai, P. (2015). Trainee's anxiety and counseling self-efficacy in counseling sessions. Unpulished doctoral dissertation, Iowa State University, USA. Tugendrajch, S. K., Sheerin, K. M., Andrews, J. H., Reimers, R., Marriott, B. R., Cho, E., & Hawley, K. M. (2021). What is the evidence for supervision best practices?. The Clinical Supervisor, 40(1), 68-87.