Since the call for the recruitment of more multiculturally diverse counselling students (e.g., speak with an accent, older, of colour) in the Association for Counselor Education and Supervision conference in 1988 (Walz, Gazda, & Shertzer, Reference Walz, Gazda and Shertzer1991), the number of international counselling students (ICSs) has increased in the United States (U.S.; Haktanir, Reference Haktanir2018; Reid & Dixon, Reference Reid and Dixon2012). ICSs studying in the U.S. come from various countries in which the counselling profession is at different developmental stages, such as similar to that of the U.S., or where counselling is not recognised as a profession. ICSs diversify the class environment and may help foster their domestic classmates’ multicultural competencies (Ng & Smith, Reference Ng and Smith2009) at various levels (e.g., attitudes and beliefs, knowledge). Additionally, ICSs can have a pivotal role in creating a unified definition of counselling around the world by spreading their knowledge and experience should they return to their country of origin.

With the increasing number of ICSs studying in the U.S. (Karaman & Watson, Reference Karaman and Watson2017), the experiences of ICSs studying there are well-documented (e.g., Nayar-Bhalerao, Reference Nayar-Bhalerao2013; Ng & Smith, Reference Ng and Smith2012). ICSs have reported academic, personal and cultural challenges, such as issues related to adaptation to and understanding of American culture, language competency, and social inclusion (Nayar-Bhalerao, Reference Nayar-Bhalerao2013; Nilsson & Anderson, Reference Nilsson and Anderson2004). Considering the significance of language competency and cultural understanding in the counselling process, several researchers have examined the relationship between ICSs’ acculturation levels, which comprises both language competency and understanding and practising the mainstream culture, and the extent to which they feel comfortable in providing effective counselling services to clients (e.g., Li, Lu, Bernstein, & Bang, Reference Li, Lu, Bernstein and Bang2018; Ng & Smith, Reference Ng and Smith2012; Nilsson & Anderson, Reference Nilsson and Anderson2004). The findings of these studies are inconsistent as some researchers have suggested that overall acculturation is related to the extent to which individuals believe they can carry out counselling services effectively (Haktanir, Reference Haktanir2018; Nilsson & Anderson, Reference Nilsson and Anderson2004) while others have noted a nonsignificant relationship (Li et al., Reference Li, Lu, Bernstein and Bang2018; Ng & Smith, Reference Ng and Smith2012).

Counselling self-efficacy (CSE) is conceptualised in the literature as the concept of feeling comfortable with one’s skills while providing counselling services (e.g., Larson & Daniels, Reference Larson and Daniels1998). Counselling self-efficacy, in general, has been found to be positively associated with clinical experience (Haktanir, Reference Haktanir2018; Watson, Reference Watson1992), counselling-related coursework (Al-Darmaki, Reference Al-Darmaki2004; Meyer, Reference Meyer2015), and counsellor performance (Hanson, Reference Hanson2006), while clinical experience and counselling-related coursework together explain more than 30% of the variance in CSE scores. Additionally, CSE has been found to be negatively correlated with anxiety (Larson & Daniels, Reference Larson and Daniels1998; Tsai, Reference Tsai2015). Although studies focusing on CSE generally did not examine ICSs and domestic counselling students (DCSs) separately, several studies have focused on CSE among ICSs. However, there is a lack of studies exploring CSE among domestic students. Both domestic and international counselling students report anxiety about seeing clients (Al-Darmaki, Reference Al-Darmaki2004; Tsai, Reference Tsai2015), yet it is unclear to date how these groups’ CSE compares. Although it is portrayed in the literature that ICSs experience more stressors than DCSs that may be detrimental to the development of CSE (e.g., language proficiency issues, cultural understanding), there is a lack of research comparing the two groups (i.e., ICSs and DCSs) in terms of their levels of CSE.

Moreover, the Council for Accreditation of Counselling and Related Educational Programs (CACREP) is an accreditation body for master’s- and doctoral-level counselling programs in the U.S. and across the world. To obtain CACREP-accreditation, counselling programs must meet certain requirements, such as full-time faculty-to-student ratio, number of credit hours and courses offered. Accordingly, the CACREP proposed accreditation standards and current 2016 standards are being used. The purpose of these standards is to safeguard equitable education by standardisation across programs (Wilcoxon, Reference Wilcoxon1990). However, it is unclear whether the CACREP standards (2016) are developed with ICSs in mind and can address these students’ training needs. Additionally, little has been investigated regarding whether the pedagogical approach employed in counselling programs in the U.S. is as effective for ICSs as it is for DCSs. Although CSE is a perception and not equivalent to counselling competency, counsellors reporting higher CSE tend to provide a higher quality of services to their clients (Hanson, Reference Hanson2006; Jaafer, Mohamed, Bakar, & Tarmizi, Reference Jaafar, Mohamed, Bakar and Tarmizi2009). In fact, Bernard and Goodyear (2008) suggested promoting CSE of students to be one of the major goals of counselling programs. Additionally, researchers asserted CSE to be a key outcome measure in the evaluation of counsellor education (Goreczny, Hamilton, Lubinski, & Pasquinelli, Reference Goreczny, Hamilton, Lubinski and Pasquinelli2015). Consequently, a deeper understanding of CSE of students may help counselling programs to evaluate their work in preparing successful future counsellors. Sending out counsellors to the workforce who are competent and confident in their skills is not only a desired outcome of counselling programs, but also an ethical necessity for a counselling faculty. If counsellors in training who do not feel confident in their counselling skills join the workforce, then they may be less likely to be of help to clients, which can raise an ethical and gatekeeping issue.

To this end, the main purpose of this study is to compare the CSE of DCSs and ICS after controlling for clinical experience and counselling-related coursework. Since it is documented in the literature that counselling students who acquire more counselling experience and take more counselling-related classes generally report higher levels of CSE (e.g., Haktanir, Reference Haktanir2018; Watson, Reference Watson1992), these two variables are accounted for in the analysis. Furthermore, this study examines the role of clinical experience and graduate-level, counselling-related coursework in explaining CSE for DCSs and ICSs separately, and compares the two regression models to ascertain whether they differ in explaining CSE. Finally, this study investigates whether participants’ CSE scores vary across participants’ degree level (i.e., master’s level, doctoral level) or specialty area (e.g., School Counselling, Counsellor Education).

With these purposes in mind, this study will address the following research questions:

1. 1. Is there a significant difference between counselling self-efficacy of ICSs and DCSs after controlling for advancement in counselling program, as indicated by the amount of clinical experience and counselling-related coursework?

2. 2. Are clinical experience and amount of counselling training significant predictors of CSE for ICSs and DCSs separately?

3. 3. Do the regression models for DCSs and ICSs differ significantly?

4. 4. Are there significant differences in CSE scores across specialty areas and degree levels?

Method

Participants

International counselling students (ICSs) Please check that the hierarchy of section headings is set correctly.enrolled in U.S. institutions were recruited as a part of a dissertation study, and the researcher collected data separately from DCSs for the current study. An ICS was described as a counselling student who was born and raised outside of the U.S., regardless of their citizenship status. One hundred and twenty-three DCSs (i.e., American students enrolled in counselling programs in the U.S.) and 114 ICSs were recruited. Any cases with excessive missing data points or violating test assumption(s) were removed. Consequently, the researcher ran the analysis with a total of 179 participants, of whom 93 were DCSs (52%) and 86 were ICSs (48%). The mean age for DCSs was 32.89 (SD = 9.25) while it was 31.36 for ICSs (SD = 6.59), yielding an average of 32.16 (SD = 8.10) for the entire group. Overall, both groups appeared to be rather similar in the gender composition, with each group including 18 men, and 75 and 68 women in the DCS group and in the ICS groups respectively. Of the participants, 20.10% (n = 36) were men and 79.90% (n = 143) were women.

Just over half of DCSs (n = 48; 51.61%) identified as Caucasian or non-Hispanic White, 19 identified as Black or African American (20.43%), 12 identified as Hispanic (14.90%), 8 identified as either biracial or multiracial (8.60%), 3 identified as Asian American (3.23%), and 3 identified as Other (3.23%). Approximately 75% of ICSs identified as either Asian (n = 33; 38.37%) or non-Hispanic White (n = 31; 36.05 %), 11 identified as Black (12.79%), 10 identified as Hispanic (11.63%), and 1 identified as biracial (1.16%). When participants’ specialty areas were examined, participants reported to be in one of the seven specialty areas: Addiction Counselling (n = 5), Clinical Mental Health Counselling (n = 64), Counselling Psychology (n = 3), Counsellor Education (n = 44), Marriage and Family Counselling (n = 4), School Counselling (n = 55), and others (n = 4); see Table 1 for participant characteristics.

Table 1.

Essential participant characteristics

Instruments

Demographic questionnaire

Participants completed a demographic questionnaire that contained items related to age, gender, ethnicity, accreditation status of the participant’s program, participant’s specialty area, the number of counselling-related credits participant had completed, and clinical (counselling) experience in hours.

Counsellor Activity Self-Efficacy Scales

The Counsellor Activity Self-Efficacy Scale (CASES) was developed by Lent, Hill, and Hoffman, (Reference Lent, Hill and Hoffman2003) to measure the degree to which counsellors believe they possess basic counselling skills and can help clients with specific issues in the near future (i.e., over the next week). The CASES is comprised of three subscales: (a) basic counselling skills, (b) session management skills, and (c) helping clients with certain issues (e.g., depressed, extremely anxious, suicidal). This scale has 41 items, each ranging from 0 (no confidence) to 9 (complete confidence), with higher total scores reflecting higher counselling self-efficacy. Sample items include ‘How confident are you that you could repeat or rephrase what the client has said, in a way that is succinct, concrete, and clear?’, ‘How confident are you that you know what to do or say next after your client talks?’ and ‘How confident are you that you could work effectively with a client who is clinically depressed?’ The CASES showed with excellent overall Cronbach’s alpha value, .97, with subscales ranging from .79 to .92 (Lent et al., Reference Lent, Hill and Hoffman2003). Furthermore, the scale demonstrated high correlation (r = .76) with the Counselling Self-Estimate Inventory (COSE; Larson et al., Reference Larson, Suzuki, Gillespie, Potenza, Bechtel and Toulouse1992), indicative of strong convergent validity. In the sample of the current study, an overall Cronbach’s alpha value of .98 for CASES, and .93, .97 and .96 for the basic counselling skills subscale, session management skills subscale and specific counselling topics subscale respectively were found.

Procedures

After receiving Institutional Review Board (IRB) approval, the researcher disseminated the study survey to DCSs and ICSs enrolled in U.S. institutions. The recruitment procedures for the groups were different. To collect data from domestic students, the study link was shared through a counselling list-serv (CESNET-L). Additionally, counsellor educators working in different departments across the U.S. were contacted to share the study information with their domestic students. Due to the limited number of ICSs studying in the U.S., a snowball sampling method was employed to collect data from the ICSs. The researcher contacted ICSs among a circle of acquaintances through email and invited them to participate in the study and share study information with other ICSs. Additionally, 20 counsellor educators across the U.S. were invited to share the study information with their ICSs, if any existed. Moreover, the researcher recruited ICSs through counselling conferences. Due to the inability to acquire the precise number of potential participants intended for recruitment, no response rate was calculated (Granello & Wheaton, Reference Granello and Wheaton2004).

Data screening and analysis

Power analysis

The researcher conducted an a priori G*Power analysis to ascertain the minimum number of participants required to detect a medium effect size given an alpha level of .05 and power of .80. For the analysis of covariance (ANCOVA) test (with two groups), a minimum of 158 participants were required to detect a significant difference if a medium effect size existed (f = .25), while a minimum of 68 participants were required to detect a medium effect size (f ² = .15) for the multiple regression analysis. Finally, a minimum number of 216 participants were required to find a significant difference, when present, for the analysis of variance (ANOVA) test (with seven groups). Given that 179 participants were recruited in the present study, it can be inferred that this study demonstrated sufficient power for the ANCOVA and multiple regression tests, yet lacked statistical power for the ANOVA test.

Preliminary analysis

Prior to carrying out the primary analysis, data were inspected, the ratio of missing values to all values was calculated (less than .002), and missing values were replaced with the series mean function in SPSS. Then the researcher tested the assumptions for the ANCOVA and multiple regression analysis, including independence of observations, normality, homoscedasticity, homogeneity of variance, normality, linearity and homogeneity of regression slopes, and noncollinearity of predictor variables (Field, Reference Field2013; Tabachnick & Fidell, Reference Tabachnick and Fidell2015). Six cases were removed due to a high influential effect on the regression line. Clinical experience violated the assumption of homoscedasticity; thus, it was transformed using the log10 function in SPSS. After removing the high influential data points and transforming the data, homoscedasticity was met (i.e., no more funnelling z graph).

Results

Research question 1

The first research question explored whether or not a significant difference existed after controlling for participants’ advancement (as measured by the amount of clinical experience and counselling-related coursework), results of which can be seen in Table 2. The researcher analysed the data using SPSS (Version 22) and conducted an ANCOVA to test whether there was a statistically significant difference between DCSs and ICSs’ counselling self-efficacy after adjusting for how advanced the students were in their program. To determine students’ advancement in the program, the researcher utilised their clinical experience (CE) and counselling-related coursework (CRC), as more advanced counselling students would accumulate more clinical hours and take more counselling-related courses. Using principle component analysis, these two variables (i.e., CE and CRC) were combined and reduced to one factor, which was then used as the covariate. After adjusting for advancement in a counselling program (i.e., clinical experience and counselling-related coursework), there was no statistically significant difference in CSE scores between DCSs and ICSs, F(1, 176) = 2.48, p = .12, partial η² = .02.

Table 2.

ANCOVA analysis results

Note: ANCOVA, analysis of covariance; SS, Type III sum of squares.

Research question 2

The second question was related to whether or not clinical experience and amount of counselling training significantly explained the variance in CSE scores. A multiple regression analysis was conducted to predict CSE from CE and CRC among DCSs and ICSs separately. The multiple regression model for the DCSs was statistically significantly, F(2, 91) = 21.92, p < .001, R ² = .34. A multiple regression analysis for the ICSs also revealed a statistically significant model, F(2, 83) = 24.13, p < .001, R ² = .37, both suggesting a large effect size (Field, Reference Field2013); see Table 3. In other words, CE and CRC accounted for 34% and 37% of the variance in CSE scores among DCSs and ICSs respectively. For the DCS group, while the clinical experience uniquely accounted for 24% of the variation in CSE scores, the amount of CRC uniquely explained approximately 1% of the variation in the CSE scores. For the ICS group, clinical experience uniquely explained 12% of the variance in CSE scores while the amount of CRC uniquely explained approximately 3% of the variance in CSE scores. Consequently, the remaining 9% for the DCSs and 22% for ICSs were the shared variance between CE and CRC.

Table 3.

Multiple regression analysis results for ICSs and DCSs

Note: Dependent = Counselling self-efficacy

Research question 3

The third question was about whether or not the regression models for DCSs and ICSs varied significantly. The researcher conducted a Fisher’s z transformation to compare correlation coefficients associated with the two regression models (i.e., multiple regression analysis models). A comparison of the regression models for DCSs and ICSs suggested no significant difference, Fisher z = .27, p > .05, as the z score was below the cutoff score of 1.98.

Research question 4

The fourth research question focused on whether or not CSE scores varied across the participants’ specialty area (e.g., school counselling, mental health counselling) and degree levels (e.g., master’s level and doctoral level). The researcher examined whether CSE scores differed across participants’ degree level and specialty areas. An independent-samples t test was carried out to determine whether significant differences existed in CSE between doctoral-level students and master’s-level students. Doctoral-level students (M = 335.79, SD = 41.61) reported significantly higher CSE than master’s-level students (M = 288.03, SD = 60.22), t(177) = −4.89, p < .001. When the same analysis was conducted for each group separately, similar differences were detected in domestic students group, t(91) = −3.49, p = .001, and international students group, t(84) = −3.31, p = .001; see Table 4.

Table 4.

Degree level and specialty area comparison results

A one-way ANOVA was run to determine whether the CSE level was different for different counselling specialty areas. CSE scores were statistically significantly different among specialty areas, F(6,172) = 6.23, p < .001. Tukey post hoc analysis showed that CSE scores of counsellor education students significantly differed from those of school counselling and clinical mental health counselling students (p < .001), yet no differences between master’s-level specialty areas were statistically significant; see Table 4 for more details.