Addressing Racial and Ethnic Microaggressions in Therapy Jesse Owen University of Louisville Karen W. Tao and Zac E. Imel University of Utah Bruce E. Wampold University of Wisconsin, Madison and Research Institute, Modum Bad Psychiatric Center, Vikersund, Norway Emil Rodolfa Alliant International University Racial and ethnic microaggressions in everyday life can negatively impact the well-being of racial and ethnic minorities (REM). When microaggressions are perceived in therapy they can interfere with therapeutic progress. However, little is known about whether microaggressions are addressed in therapy and if so, does addressing them impact the therapeutic relationship. REM clients from a university counseling center ( n 120) reported on their therapy experience. Approximately 53% of clients reported experiencing a microaggression from their therapist. Clients’ perceptions of micro- aggressions were negatively related to the working alliance, even after controlling for their current psychological well-being, number of sessions, and therapist racial and ethnic status. Of those clients who reported a microaggression, nearly 76% reported that the microaggression was not discussed. For those clients who experienced a microaggression and did not discuss it, alliance ratings were lower as compared with clients who did not experience a microaggression or who experienced a microaggression but discussed it. Keywords: alliance, cultural competence, microaggression, psychotherapy, rupture Overt forms of racism and prejudice have a long history in the United States and they still occur today. However, less overt forms of racism and prejudice are more ubiquitous (Sue et al., 2007). Pierce et al. (1978) identified an insidious form of cultural bias—racial microaggressions, which include the ex- perience of a variety of direct and indirect (conscious and unconscious) insults, slights, and discriminatory messages. Sue et al. (2007) explicated three specific types of microaggres- sions: microinvalidations (e.g., denying that racism exists), microassualts (e.g., more direct racism, but conducted in private or safe environments), and microinsults (e.g., believing the cultural norms of a group are pathological). A core character- istic of a microaggression is that it is ambiguous or subtle in nature, which is in contrast to an overt racial slur Microaggres- sions can target individuals’ cultural identities (e.g., gender, sexual orientation; Owen, Tao, & Rodolfa, 2010; Shelton & Delgado-Romero, 2011) and generally when they target indi- viduals’ race and ethnicity (e.g., subtle discriminatory mes- sages), as they are biased comments referring to nationality, values, cultural customs, and language and can also be based on physical appearance (Pollard & O’Hare, 1999). Ethnicity typi- cally refers to group membership related to a common cultural heritage, values, attitudes, and behaviors; whereas race is typ- ically defined by physical attributes (e.g., skin color) that is shared by a group of people (Cokley, 2007; Quintana, 2007). A hybrid definition that integrates both race and ethnicity can be useful in some cases (Cokley, 2007) and in particular for therapy studies as therapists are reacting to both the clients’ race and ethnicity during therapy. Racial and ethnic microaggressions have been associated with psychological distress, such as anger, anxiety, confusion, and contempt (e.g., Blume, Lovato, Thyken, & Denny, 2012; Mercer, Zeigler-Hill, Hayes, & Wallace, 2011; Schoulte, Schultz, & Alt- maier, 2011; Wang, Leu, & Shoda, 2011). Many racial and ethnic minority (REM) individuals experience microaggressions in ev- eryday life, and unfortunately they also can be experienced during J ESSE O WEN received his PhD in counseling psychology from the Uni- versity of Denver. He is currently an associate professor and training director at the University of Denver. His research interest includes multi- cultural processes in psychotherapy and psychotherapy outcomes and processes. K AREN W. T AO received her PhD in counseling psychology from Uni- versity of Wisconsin, Madison in Counseling Psychology. She is currently an assistant professor at the University of Utah. Her research interest includes cultural processes, psychotherapy outcomes, and processes. Z AC E. I MEL received his PhD in counseling psychology from University of Wisconsin, Madison. He is an assistant professor at the University of Utah. His research interest includes psychotherapy outcome and processes. B RUCE E. W AMPOLD received his PhD in counseling psychology from University of California, Santa Barbara. He is currently a professor at the University of Wisconsin, Madison and Modum Bad Psychiatric Center. His research interest includes psychotherapy outcomes and processes. E MIL R ODOLFA received his PhD in counseling psychology from Texas A&M University. He is currently a professor at the Alliant International University. His research interests include supervision, training, ethics, and boundaries. C ORRESPONDENCE CONCERNING THIS ARTICLE should be addressed to Jesse Owen, Psychology Department, College of Education, University of Lou- isville, Louisville, KY 40292. E-mail: jesse.owen@louisville.edu This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Professional Psychology: Research and Practice © 2014 American Psychological Association 2014, Vol. 45, No. 4, 283–290 0735-7028/14/$12.00 http://dx.doi.org/10.1037/a0037420 283 therapy. For instance, clients are likely to perceive microinvalida- tions and microinsults 1 from their therapists in the form of dismis- sive or negating messages about their cultural heritage or culturally inappropriate interventions (Burkard & Knox, 2004; Neville, Lilly, Duran, Lee & Browne, 2000; Salvatore & Shelton, 2007; Solór- zano, Ceja, & Yosso, 2000; Thompson & Jenal, 1994). Concep- tually, microaggressions could be experienced by REM clients (either tacitly or explicitly) as a recapitulation of previous cultural injustices, which when not properly handled could impede the therapy process. For example, a microinsult to an Asian American client (e.g., “I don’t know why you just don’t speak up for yourself and tell your parents how you feel”) potentially pathologizes the client’s cultural ways of relating to elders. In doing so, the thera- pist possibly jeopardizes the working alliance, as the client and therapist will likely disagree on standing up to the client’s parents as an appropriate method to alleviate the client’s distress with his family. Thus, therapists may need to be aware of the cultural dynamics in the therapy process and be able to successfully nav- igate these conversations (Cardemil & Battle, 2003; Maxie, Ar- nold, & Stephenson, 2006). To date, there are only three known studies that have examined racial and ethnic microaggressions in therapy with two samples of African American clients ( n 40, Constantine, 2007; n 19, Morton, 2011) and one sample was more diverse including ap- proximately 20% Asian American clients, 16% multiracial/ethnic clients, 9% Hispanic clients, 1% African American, and 50% Euro American clients ( n 232, Owen et al., 2011). Moreover, there are only two other known studies examining microaggressions against women and LGBQ clients in therapy (Owen et al., 2010; Shelton & Delgado-Romero, 2011). Despite the paucity of therapy studies, one emerging trend is becoming clear: the experience of microaggressions can lead to feelings of invalidation and/or dis- respect, thus resulting in a rupture wherein the emotional bond with a therapist is disrupted. For instance, in several analogue studies, therapists’ endorsement of a specific type of microaggres- sion— colorblind attitudes (Neville et al., 2000)— has been shown to impact aspects of the working alliance, such as their empathy toward an African American client (Burkard & Knox, 2004; also see Gushue, 2004; Neville et al., 2006). In actual therapy sessions, clients’ perceptions of racial and ethnic microaggressions were negatively associated with the quality of the working alliance (Constantine, 2007; Morton, 2011; Owen et al., 2011) as well as satisfaction with services and therapy outcomes (Constantine, 2007; Owen et al., 2011). The negative association between microaggressions and the working alliance is of particular importance. First, the alliance— or the agreement between the client and therapist on the goals for therapy and the methods to reach those goals as well as the relational bond between the client and therapist (Bordin, 1979)—is one of the most robust predictors of therapy outcomes, accounting for approximately 7% to 14% of the variance (Crits-Christoph et al., 2011; Horvath et al., 2011). Accordingly, it is of utmost importance for therapists to attend to microaggressions that impact the ability to form or maintain a therapeutic working alliance. Second, the alliance has been shown to be a mediator for the association between microaggression and therapy outcomes, sug- gesting those client–therapist dyads who are able to maintain high quality alliances are able to work through the rupture in a way that does not impede clients’ clinical goals (Constantine, 2007; Owen et al., 2011; Owen & Imel, 2010). Given the deleterious effects of clients’ perceptions of micro- aggressions on therapy processes, it is important to ascertain whether therapists and clients are discussing these potentially negative experiences. Therapists may not be fully aware of the specific statement or action that the client perceived as a micro- aggression, because microaggressions can be indirect, uncon- scious, and unintentional. Generally, therapists are not good pre- dictors of treatment process and outcome. For example, therapists tend to underestimate the number of clients who deteriorate during therapy (Chapman et al., 2012; Hannan et al., 2005). Moreover, therapists have been shown to be reluctant and uncomfortable when addressing issues of race and ethnicity (Knox, Burkard, Johnson, Suzuki, & Ponterotto, 2003). Nonetheless, therapists should strive to be aware of how the client reacts to what they say and any potential changes in the therapeutic relationship (Safran & Murran, 2000). Ideally, therapists who are aware of potential problems can directly process the specific microaggression, seek to understand why it occurred, clarify misunderstandings, and work to realign with the client (Nadal, 2009; Safran & Murran, 2000). In doing so, therapists should be able to repair the therapeutic rela- tionship. Currently, we are not aware of any data establishing whether therapists attempt to mend the therapeutic relationship with clients after the presence of microaggressions. We predicted that REM clients’ perceptions of microaggres- sions would be negatively associated with their ratings of the alliance (Hypothesis 1), after controlling for current psychological well-being, number of sessions, and therapist race and ethnic status. The use of current psychological well-being as a control variable should help reduce the likelihood that some of the vari- ance in the alliance are due to positive mental health or gains in therapy (cf. Barber, 2009). Additionally, we predicted that client– therapist dyads that addressed the microaggression (and were able to come to resolution) would (a) demonstrate similar levels of working alliance to clients who never perceived a microaggression (Hypothesis 2) and (b) have higher levels of alliance than clients whose therapist never discussed the microaggression or were un- able to resolve the issue (Hypothesis 3). Method Participants Clients included 88 females and 32 males with a median age of 22 years old ( range 18 to 51). Thirty-one percent (24.2%) of the clients were graduate students, 28.3% were seniors, 20.8% were juniors, 11.7% were sophomores, 13.3% were freshman, and 1.7% were nonstudents. Clients were asked to self-identify their race/ ethnicity; 1.7% of clients identified as African American, 42.5% identified as Asian American, 24.2% identified as Hispanic, 30.8% identified as multiethnic, and less than 1% did not identify their race or ethnicity. The small number of clients did not allow us to test for differences between racial and ethnic demographic groups. 1 Therapists are less likely to use microassualts in session with clients because this subtype of microaggression is more conscious and resembles overt racism. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 284 OWEN, TAO, IMEL, WAMPOLD, AND RODOLFA These figures are consistent with the overall university’s race and ethnic composition, wherein Asian Americans represent 41% of the student body, Hispanic/Latino(a) students represent 16%, and African American students represent 3% (note the university does not include multiethnic/racial as a category). Thirty-three therapists treated the 120 clients. The average num- ber of clients reporting for each therapist was 3.64 ( range 1 to 9). Ten of the therapists self-identified as REM and 23 self- identified as White. Therapists were not directly assessed in this study and subsequently further demographic or other therapeutic information was not gathered. In general, the therapists were predoctoral interns, postdoctoral fellows, staff psychologists, and staff therapists. There is no prescribed therapeutic approach for therapists. In prior assessments at this counseling center, where therapists were asked to indicate their theoretical orientation in an open-ended format, all therapists indicated that they practiced some form of integrative therapy (e.g., psychodynamic/cognitive– behavioral, relational/systems/cultural; see Owen et al., 2011). This counseling center generally provides brief therapy (6 – 10 sessions; the median number of sessions for this sample was four). It is common practice at this counseling center for the therapist who conducts the intake to continue to see the client for therapy. Measures Racial Microaggressions in Counseling Scale (RMCS; Con- stantine, 2007). The RMCS was used to assess clients’ per- ceptions of racial microaggressions during the course of ther- apy. It is a 10-item scale with items that assesses a client’s perception of counseling situations related to race or ethnicity. Example items include the following: My counselor minimized the importance of cultural issues in our session(s) , and My counselor sometimes was insensitive about my cultural group when trying to understand or treat my concerns or issues . We retained the wording for the items; however we altered the instructions and the rating scales. The original rating scale was a 0 ( this never happened to me ), 1 ( this happened, but it did not bother me ), and 2 ( this happened and I was bothered by it ). The existing rating scale can be characterized as a categorical rating system, reflecting two dimensions: (a) if the microaggression happened or not, and (b) if the microaggression bothered the client or not. Moreover, there might be reactions that could be delineated between the ratings of it did not bother me and I was bothered by it , (e.g., I was slightly bothered by it) that are conceptually meaningful. Statistically, rating scales with few response options can also impact the reliability of the measure, with rating scales that have 5 to 7 response options generally demonstrating better reliability estimates (Nunnally, 1978). As such, to assess frequency and impact of microaggressions, clients were instructed to First rate how often each situation occurred, then rate how the situation affected you. The rating scale to assess frequency ranged from 1 ( Never ) to 5 ( Always ). Next, clients were asked, in a separate question for each item, If this occurred, how much did it bother you ? to assess the impact of microaggression, which was rated on a 5-point scale ranging from 1 ( Not at all ) to 5 ( Very much ). The distribution of the scores was restricted for both sub- scales. The mean for the frequency ratings were M 1.22 ( SD 0.35) and for the impact ratings were M 1.20 ( SD 0.48). As such, we created two scores for the RMCS: (a) whether microaggression occurred at all (Yes 64 [53.3%]; No 56 [46.7%]), and (b) a total microaggression score. For the latter score, we included all clients including those who did not report any microaggressions. To do so, we treated ‘no reported microaggressions’ responses as the anchor point on the scale (i.e., 1) and then continued with the other rating points on the impact scale. For example, a client who experienced a microaggression but reported that it did not bother then at all would have a score of 2 on the impact of microaggression score. This scaling was done to include all clients. The total micro- aggression scale was consistent to creating a composite score by summing the frequency and impact subscales. The reliability and validity of RMCS has been tested in three separate samples (Constantine, 2007; Morton, 2011; Owen et al., 2011). To establish concurrent validity, Constantine (2007) found the clients’ scores on the RMCS were negatively corre- lated with their ratings of cross-cultural competency of White therapists. Furthermore, the RMCS was negatively related to client rated working alliance and satisfaction with therapy, suggesting that the measure captures meaningful aspects of the psychotherapy process (Constantine, 2007; Morton, 2011; Owen et al., 2011). Previous reliability estimates have ranged from .66 to .79 for REM clients. In the current study, the Cronbach’s alpha was .88. Discussion of microaggressions. For clients who reported that they experienced a microaggression, they were asked if the microaggression(s) was discussed in therapy. Specifically, they were asked the following: If any of the above occurred [microag- gressions from the RMCS], did you and your counselor discuss the issue(s)? and the response categories were Yes, and we were able to work it out ; Yes, and we were unable to work it out ; and No, we did not discuss the issue Working Alliance Inventory-Short Form (WAI-S, Tracey & Kokotovic, 1989). The WAI-S is a client rated measure of working alliance that consists of 12 items that assess goals and tasks for therapy as well as the relational bond between the client-therapist. These items were rated on a seven-point scale ranging from 1 ( Strongly disagree ) to 7 ( Strongly agree ) with higher scores indicating a better working alliance. The WAI-S is a commonly used measure of working alliance and the reliability and validity has been demonstrated in numerous studies comparing the WAI-S to other working alliance scales and therapy outcome (see Horvath et al., 2011). For the current study, the total scale score was used and the Cronbach’s alpha was .95. Schwartz Outcome Scale-10 (SOS-10; Blais et al., 1999). The SOS-10 is a 10-item scale designed to assess current psycho- logical well-being over the past week, which is rated on a seven- point scale ranging from 1 ( Never ) to 7 ( All the time or nearly all the time ). The reference samples ( n 10,000) for the SOS-10 were drawn from various clinical populations (e.g., inpatient, outpatient, college counseling centers) and nonclinical populations (e.g., adults from the community, college students; Owen & Imel, 2010). Across studies, the SOS-10’s reliability was: test/retest, r .88 and Cronbach’s alpha ( ) .91. Furthermore, the SOS-10 has strong convergent and divergent validity as it correlates in the predicted direction with a variety of clinical and psychological well-being scales and reliably discriminates between clinical and This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 285 RACIAL AND ETHNIC MICROAGGRESSIONS IN THERAPY nonclinical samples (see Owen & Imel, 2010 for review). The Cronbach’s alpha coefficient for this study was .94. Procedure During intake at a large West Coast university counseling center clients were asked, on their intake card, if they would be willing to receive a survey about their therapy experience. Clients who agreed were sent an email at the end of the academic quarter directing them to an anonymous online survey. Three-hundred fifty-seven individuals responded to the electronic survey (29% response rate); however, we only included clients who identified as an REM, participated in individual therapy, and identified their therapist. For purposes of the research, participants initially com- pleted an informed consent form and then were then asked to identify their therapist. Next, clients completed a series of instru- ment and other questions related to the functioning of the coun- seling center (not analyzed here). Clients were directed to identify and report on a specific therapist for the alliance and microaggres- sion measures. If clients identified multiple individual therapists they were excluded from the analyses. Furthermore, clients who did not identify any therapist were also excluded from the analy- ses. After completion of the study, clients were able to enter a raffle for $100.00 dollars. All procedures were approved by the university IRB. Data Analysis Considerations Psychotherapy data frequently violate a basic assumption of statistical testing—independence of observations—as multiple cli- ents are treated by the same therapist. Multilevel models can correct for the biases that occur when this assumption is violated. We conducted multilevel models with Bayesian estimation utiliz- ing the statistical package Mplus 7.0 (Muthén, & Muthén, 1998 – 2012). Bayesian models are not based on normality assumptions or asymptotic results and provide better estimations for unbalanced and small sample sizes (Hamaker & Klugkist, 2011), which is well suited for our study given that we had a moderately small number of therapists and an unbalanced number of clients treated by each therapist. In Bayesian models the results describe posterior distri- bution, which is the range of uncertainty left in the model after accounting data included in the model (Hamaker & Klugkist, 2011). Thus, larger posterior distributions mean that there is more uncertainty in the results. Credible intervals (CI), similar to con- fidence intervals, are utilized in Bayesian models to help describe the range of the posterior distribution. For more detailed overview of Bayesian analysis see Hamaker and Klugkist (2011). Results Preliminary Analyses Initially, we examined the degree to which therapists varied in their clients’ ratings of alliance and microaggressions. Therapists accounted for approximately 9.8% of the variance in their clients’ ratings of alliance (client level variance 1.19, p .001, 95% CI 0.94, 1.61; therapist level variance 0.13, p .001, 95% CI 0.004, 0.61) and they accounted for approximately 6% of the variance in their clients’ perceptions of microaggressions (client level variance 0.15, p .001, 95% CI 0.12, 0.20; therapist level variance 0.01, p .001, 95% CI 0.001, 0.03). These results suggest that therapists vary in their clients’ average ratings of the alliance and microaggressions. However, some caution should be taken with these results as the ratio of clients per therapist was small. Next, we screened for differences in clients’ ratings of mi- croaggressions based on therapists’ racial/ethnic status. The results were not statistically significant, b 2 0.04, SD .08, p .64, 95% CI 0.18, 0.12, suggesting that REM clients did not significantly differ in their report of microaggressions when their therapist was White or REM. Similarly, when we tested whether therapists’ racial/ethnic status would be associ- ated with clients’ perception of microaggressions occurring at all (yes/no), the results were not statistically significant, b 0.57, SD .30, p .086, 95% CI 0.11, 1.12. In total, 53.3% ( n 64) of REM clients reported experiencing a micro- aggression (68.4% of REM clients who were treated by REM therapists and 46.3% of REM clients who were treated by White therapists) 3 . The descriptive statistics and bivariate correlations for the variables in the study are presented in Table 1 and Table 2. Primary Analyses We tested whether clients’ perceptions of microaggressions would be negatively associated with their ratings of the working alliance (Hypothesis 1). To do so, we predicted alliance scores by clients’ perceptions of microaggressions (total score), and we controlled for number of sessions (level 1-grand mean centered), current well-being (level 1-grand mean centered), and therapists’ race and ethnicity (level 2-REM 1, White 0). The results supported our hypothesis, insofar as clients who reported stronger alliances reported fewer microaggressions, after controlling for the variability among therapists and the variance in the other control variables (see Table 3). Of the control variables, only clients’ current psychological well-being (SOS-10) was positively associ- ated with alliance scores. Of the 64 clients who reported experiencing a microaggression, 76% ( n 42) reported that the microaggression experience was not discussed, 24% ( n 13) reported the microaggression was discussed. Of these 13 clients, only one client reported that the discussion was not successful. REM therapists were less likely to discuss the microaggression experience as compared with White therapists (90.5% vs. 67.6%); however, after controlling for ther- apist effects, these differences were not statistically significant ( b 0.78, SD 0.37, p .06, 95% CI 0.04, 1.54). The lack of statistical significance is likely a result of the small sample size as there were only 10 REM therapists and 23 White therapists cou- pled with the fact that discussion of the microaggression experi- ence was a low base rate event. 2 For simplicity we will just report the b value here, but it reflects the median of the posterior distribution. 3 We also tested whether client reports of microaggressions differed among client racial/ethnic groups (e.g., African American, Asian Ameri- can, Hispanic, Multi-Racial/Ethnic). The results were not statistically sig- nificant ( p .05). However, our sample size per group was relatively small. Further, information about these differences can be requested from the first author. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 286 OWEN, TAO, IMEL, WAMPOLD, AND RODOLFA To address our second and third hypothesis, we examined three groups: (a) client and therapist dyads who successfully discussed the microaggression experience, (b) client and therapist dyads who did not discuss the microaggression, and (c) client and therapist dyads where there were no reported microaggressions. There was one client who discussed the microaggression experience but it was unresolved, thus this client was excluded from comparisons. Table 4 shows the means and standard deviations for the three groups. The results demonstrated that alliance scores were signif- icantly different among the three groups ( b 0.27, SD 0.08, p .001, 95% CI 0.44, 0.12). As seen in the table, client and therapist dyads who successfully discussed the microaggression experience had alliance scores comparable with clients who did not perceive any microaggressions ( p .05, d 0.15, 95% CI 0.40, 0.10; supporting Hypothesis 2). Client and therapist dyads who successfully discussed the microaggression experience had higher alliance scores as compared with client and therapist dyads who did not discuss the microaggression experience ( p .05, d 0.42, 95% CI 0.73, 0.12; supporting Hypothesis 3). Client and therapist dyads that did not discuss the microaggression ex- perience had significantly lower alliance scores as compared with client and therapist dyads where there were no reported microag- gressions ( p .05, d 0.56, 95% CI 0.79, 0.34). Discussion Many REM clients may have experienced microaggressions in daily life (e.g., Sue et al., 2007), and our results suggest that these societal experiences of discrimination do not stop at the therapy door. Indeed, 53% of REM clients in the current study reported that a microaggression occurred in their therapy (any score on any item above the rating of Never ). Moreover, clients’ perceptions of microaggressions were related to lower quality alliances with their therapists. Consequently, our findings add to the other three stud- ies that have demonstrated that clients’ perceptions of racial and ethnic microaggressions are negatively associated with the alli- ance—a central element in the therapeutic process and significant mediator between clients’ perceptions of microaggressions and therapy outcomes (Owen et al., 2010). Conceptually, microaggressions can be thought of as a special case of ruptures in therapy, wherein experiences of discrimination and oppression from the larger society are recapitulated, which places the therapeutic relationship under duress and strain. In the current study, only 24% of client and therapist dyads discussed the microaggression experience. Given that therapists are not gener- ally proficient at identifying clients who are deteriorating in ther- apy (Hannan et al., 2005), this finding might be a multicultural variant of a similar phenomenon. Indeed, the very nature of mi- croaggressions as subtle, unconscious, and indirect may prevent well-meaning therapists from recognizing that their clients are experiencing such offenses. Moreover, therapists have been shown to be reluctant and uncomfortable when addressing issues of race Table 1 Means and Standard Deviations for the Alliance, SOS-10, and Microaggressions REM clients with REM therapists REM clients with White therapists Scale information M ( SD ) M ( SD ) Range Alliance (WAI-S) 5.60 (1.17) 5.51 (1.09) 1–7 Psych well-being (SOS-10) 4.98 (1.25) 5.16 (1.09) 1–7 Microaggression (RMCS) 1.11 (0.25) 1.14 (0.43) 1–6 n (%) n (%) Total N Any Micro 26 (68.4%) 38 (46.3%) 38/82 No discuss micro 19 (90.5%) 23 (67.6%) 21/34 a Note REM clients treated by REM therapists, n 38; REM clients treated by White therapists, n 82. WAI-S Working alliance Inventory- Short Form; SOS-10 Schwartz Outcome Scale-10; RMCS Racial Microaggression in Counseling Scale. a There were nine missing cases for the discussion of microaggression item, thus the number of clients reporting a microaggression slightly exceeds the number of clients reporting on whether a discussion of the microaggression occurred. Table 2 Bivariate Correlations Between Working Alliance, SOS-10, and Microaggressions Any micro Total micro Alliance SOS-10 Any micro — Total micro .21 — Alliance .21 .28 — SOS-10 .17 .27 .41 — Note These correlations do not account for therapist effects. Any micro was coded 1 Yes , 0 No p .05. p .01. p .001. Table 3 Summary of Multilevel Model Predicting Alliance by Clients’ Perceptions of Microaggressions Effect b ( SD ) p value 95% CI Fixed effects Intercept-alliance 4.88 (0.36) .001 4.29, 5.69 Total microaggression 0.82 (0.21) .001 1.20, 0.38 Number of sessions 0.04 (0.03) .11 0.01, 0.10 SOS-10 0.24 (0.06) .001 0.12, 0.37 Therapist ethnicity 0.16 (0.30) .68 0.41, 0.70 Random effects Therapist variance 0.26 (0.19) .001 0.06, 0.75 Client variance 0.89 (0.13) .001 0.65, 1.15 Note Ethnicity was coded 1 for REM and 0 for White. Coefficients are unstandardized effects. CI credible intervals. Table 4 Means and Standard Deviations for Alliance by No Discussion of Microaggressions, Successful Discussion of Microaggressions, and No Microaggressions No discussion ( n 42) Successful discussion ( n 12) No micro ( n 51) Alliance 5.17 (1.21) 5.65 (0.91) 5.80 (1.06) Note One client reported that the discussion of microaggression was not successfully resolved, but his/her/their scores are not reported here. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 287 RACIAL AND ETHNIC MICROAGGRESSIONS IN THERAPY and ethnicity (Knox et al., 2003). Nonetheless, therapists are responsible for the well-being of their clients. Client and therapist dyads that were able to successfully discuss the microaggression experience had higher quality al- liances as compared with client and therapist dyads where the microaggression experience was not discussed. In fact, client and therapist dyads who successfully discussed the microag- gression experience had similar alliances as those client and therapist dyads where there was no perceived microaggression. These results illuminate the power of addressing the missteps that can occur in therapy (see Safran & Murran, 2000). More- over, they are consistent with the positive effects of resolutions to ruptures in nonmicroaggression studies (e.g., Muran et al., 2009). At present and given the cross-sectional nature of our study, it is unclear whether the initial alliance was stronger for client and therapist dyads who successfully addressed the mi- croaggression as compared with the dyads who did not address the microaggression. Limitations and Future Directions The merits of our results should be interpreted within the scope of the study’s methodological strengths and limitations. First, the retrospective nature of this study coupled with the electronic survey methodology raises several concerns: (a) inability to cap- ture session by session process of alliance and microaggressions, (b) our findings are correlational making the directionality of microaggressions and alliance correlations potentially bidirec- tional, and (c) the response rate (29%) was relatively low. All known microaggression studies have utilized the same methodol- ogy (Constantine, 2007; Morton, 2011; Owen et al., 2011), and in our opinion the lack of research examining racial/ethnic microag- gressions (only three studies) as well as therapists multicultural competencies in actual therapy settings is particularly hampering for the progress of our profession. There are many questions that can be untangled by assessing microaggressions in a prospective manner, including how the alliance changes after a microaggres- sion and how client and therapists are able to address the experi- ence. Potentially, therapists’ ability to establish a sound relational foundation early in therapy may enhance therapy process by pro- viding a safe environment, which may lend itself to having dis- cussions about microaggressions. There may be advantage to as- sess microaggressions via therapist rating to gauge the degree of overlap in perceptions. Second, we only assessed microaggressions related to clients’ racial and ethnic identity, whereas other types of cultural identities (e.g., gender, sexual orientation) were not explored. Future studies should continue to examine the potential intersection of gender and ethnicity, for example, in relation to the experience of microag- gressions. Moreover, between racial/ethnic group comparisons on their perceptions of microaggressions was not statistically signif- icant; however, we did not have sufficient sample size to fully test these differences. Consistently, the limitations of restricting a study to a single institution with disproportionality in racial and ethnic demographics compelled us to combine racial and ethnic minority clients into one group. We recognize this constrains interpretation regarding differences between how specific groups historically and currently experience racism or discrimination in the United States or how specific groups on this campus may be impacted by microaggressions based on the racial and ethnic breakdown. We also acknowledge this decision to aggregate the sample preferences nomothetic (e.g., Students of Color) over id- iosyncratic (e.g., Vietnamese American compared with Mexican American) information. Accordingly, our sample limited our abil- ity to test potentially important distinctions within and between racial/ethnic groups as well as differences for particular cross- racial/ethnic therapist-client dyad pairings. Future research may want to examine whether and how microaggressions are experi- enced differentially among the various racial and ethnic groups. Third, this study caries the typical strengths and limitations of data sets obtained in naturalistic settings. As such, we sacrificed more rigorous controls (e.g., prescreening, training therapists, monitoring interventions) for a larger, more diverse sample, and naturalistic treatments. Lastly, we adjusted rating scale for the RMCS to provide separate ratings for frequency and impact of the microaggression. The distinction between frequency and impact may prove to be important in understanding how microaggressions can influence the process of therapy. For instance, why are some clients who perceive microaggressions not as impacted? Unfortu- nately, our research design does not allow us to fully test these hypotheses. Moreover, the scores on the RMCS were restricted, suggesting that clients do not generally report that microaggres- sions occur frequently throughout treatment and they do not report that the microaggression experience bothered them a great deal—at least based on scores on the ratings scale. These findings parallel the frequency and impact of microaggressions in previous psychotherapy studies (Constantine, 2007; Morton, 2011; Owen et al., 2011). The restricted range on the measures does not neces- sarily minimize the importance of microaggressions; rather, it appears that even modest ratings of microaggressions can have a deleterious effect on the process and outcome of psychotherapy. Implications for Training and Practice Given the complexity of microaggressions it is likely that all therapists will have clients who experience them. The overwhelm- ing aspect of microaggressions for many clinicians and trainees is that they are often unconscious and unintentional. What we have learned about the microaggression research, however, is that no therapist is immune from unintentionally offending or invalidating their clients. Therapists must realize that developing a strong multicultural orientation requires a lifelong process of a willing- ness to examine biases, attitudes, and beliefs— ultimately reflected in a cultural humble stance with clients (Hook et al., 2013; Owen, Tao, Leach, & Rodolfa, 2011; Owen, 2013). Accordingly, students in counseling programs must be introduced to the concept and impact of microaggressions early on in their training and beyond the multicultural counseling class. Core courses such as counseling skills, ethics and professional issues, and practicum should also incorporate activities and readings that allow students to explore, both didactically and experientially, the ways in which microag- gressions manifest (e.g., implicit or unconscious biases) as well as how they contrast to more explicit or overt forms of racism. Moreover, the open endorsement of egalitarian or nonprejudiced views is not a reliable indicator of an individual’s awareness about their biases or stereotypes. In fact, evidence suggests those who consider themselves politically liberal are often at greatest risk of maintaining implicit bias or rationalizing t