The Influence of Client Socioeconomic Status on Psychotherapists’ Attributional Biases and Countertransference Reactions

Introduction

It has been noted that the cultural significance of socioeconomic status (SES) and its influence on the psychotherapy process has not been given the attention that it deserves in mental health literature (Lam & Sue, 2001; Liu, 2001). Prior research examined some forms of clinical judgment in relation to SES (Abramowitz & Dokecki, 1977), but little empirical research has examined how SES impacts the therapeutic exchange between psychotherapist and client (Liu, Soleck, Hopps, Dunston, & Pickett, 2004). In particular, little has been written to help psychotherapists identify and confront their personal reactions and attitudes to differences in client SES. The present study attempts to illustrate how psychotherapists respond to clients of different SES through the examination of attributional biases and countertransference reactions.

Many authors suggest that SES is part of an individual’s identity (Liu, Soleck, et al., 2004): it effects how we perceive personal success (Liu, Soleck, et al., 2004; Storck, 1997); it interacts with other identity characteristics (e.g., gender, race); it affect one’s overall quality of life (D’Andrea & Daniels, 2001). The economic context that people find themselves in can, therefore, affect how they view themselves and how others view them (Liu, Soleck, et al., 2004). This understanding of SES is an important consideration because it can affect the quality of a therapeutic relationship and how a psychotherapist responds to diverse clients. In fact, the differing values associated with SES can introduce both conscious and unconscious biases into a psychotherapist’s clinical judgments (Sue & Sue, 2003).

Early research by Abramowitz & Dokecki (1977) found evidence of negative bias against clients of lower SES in the form of less favorable mental health diagnoses. A later review of related research (Garb, 1997) suggested that this form of bias may not occur in all clinical samples. However, in his review, Garb (1997) reported that studies that showed less evidence of clinician bias included larger sample sizes (e.g., over 200 participants). Inconsistency among studies has been attributed to factors such as differences in sample sizes and the operational definitions of SES. However, one similarity among the few studies in this area is the general method in which psychotherapist bias was measured. In most SES studies, clinical bias has been measured through a psychotherapist’s clinical judgments of a client or simulated client. These clinical ratings include client diagnosis, prognosis, degree of pathology, and recommended treatment options (e.g., Bamgbose & Edwards, 1980; Di Nardo, 1975; Lee & Temerlin, 1970; Routh & King, 1972; Trachtman, 1971; Umbenhauer & De Witte, 1978). Although clinical judgments are important considerations for examining psychotherapist bias, they do not capture all aspects of a psychotherapist’s reactions to a client. Psychotherapist cognitive and countertransference reactions represent two important considerations not adequately studied in SES-related research.

It has been recommended that future SES-related research on psychotherapist bias be extended to include perceptions of a client’s motives, attitudes, and causal explanations for client behaviors (Garb, 1997). These are examples of a psychotherapist’s cognitive reactions, or attributional judgments, of a client. Psychotherapist attributional judgments are causal explanations made about the source of a client’s problems or behaviors (Chen, Froehle, & Morran, 1997; Rabinowitz, Zevon, & Karuza, 1988; Strohmer, Biggs, Keller, & Thibodeau, 1984; Weiner, 1979). For example, some may explain another person’s problems or behaviors as a result of his or her own internal personality characteristics, that is internal cognitive attributions, or from environmental influences or stressors, that is external cognitive attributions (Rotter, 1966; Sue & Sue, 2003; Weiner, 1979). Because cognitive attributions shape what is perceived as the genesis of someone else’s circumstances, these attributions can guide the direction of therapy. This is particularly significant because cognitive attributions about others can originate from one’s own beliefs or values toward SES in general. That is, psychotherapists who hold certain preconceived values related to higher versus lower SES may conceptualize the origins of clients’ concerns differently based on this aspect of cultural diversity, rather than on objectively based client-reported symptoms and life circumstances. In turn, this can lead to stereotyping, negative treatment implications (Fiske, 1993), and the potential for psychotherapists unconsciously socializing clients into accepting certain views of themselves and their problems (Karuza, Zevon, Rabinowitz, & Brickman, 1982).

How client SES impacts a psychotherapist’s countertransference reactions to a client is another important consideration for clinical judgments. Just as psychotherapists’ cognitive reactions can impact their clinical judgments, their countertransference reactions to clients can also significantly impact their decision-making process. Classically, the psychoanalytical definition of countertransference is a psychotherapist’s reaction to a client originating from a psychotherapist’s own unresolved intrapsychic conflicts (Gelso & Hayes, 1998). However, more recent definitions have added to our understanding of countertransference and these have included personal reactions evoked by the interpersonal style or characteristics of the client (Schwartz, Smith, & Chopko, 2007). This understanding of countertransference considers the influence of certain aspects of the client on a psychotherapist.

More classical understandings of countertransference have viewed psychotherapist countertransference as more of a threat to the well-being of the therapeutic process (Gelso & Hayes, 1998). However, contemporary views of countertransference acknowledge the benefits that recognized countertransference can have on furthering the therapeutic process. If psychotherapists learn to acknowledge, understand, and manage their feelings of countertransference, they can actually advance the therapy process in more appropriate and healthy ways (Gelso & Hayes, 1998; Kiesler, 2001). The idea that certain client characteristics can influence a psychotherapist’s countertransference reactions has been researched to a small degree. According to one research review, the few studies that have been conducted in this area focused mainly on certain client diagnoses, client sexuality, clients coping with past traumas, and child and adolescents (Schwartz & Wendling, 2003). The results from most of these studies reported evidence for the presence of psychotherapist countertransference reactions in relation to client characteristics. Therefore, the examination of client SES will help to expand further our understanding of the many aspects of client diversity and its influence on psychotherapist countertransference.

The purpose of the present study was to expand the scope of previous SES bias studies and to empirically investigate differences in psychotherapist response, through attributional bias and countertransference reaction, to persons of various SES backgrounds.

Methodology

Participants

Participants were 141 professional counselors and counselor-trainees and ranged in age from 22 to 74 years (M = 46.65, SD = 12.67). Eighty-four (59%) were female and 57 (40%) were male. Out of the total sample, 122 (86%) were self-identified as European American, 7 (5%) as African American, 7 (5%) as Hispanic American, 2 (1%) as Asian American, and 3 (2%) as “Other”. Participants were members of the American Counseling Association, state counseling organizations across the United States, or students associated with a university accredited by Council for the Accreditation of Counseling and Related Educational Programs.

With the exception of those earning less than $25,000 annually (7%), reported income levels of participants appeared to be evenly distributed among four categories: $25,000 to $50,000 (22%), $50,000 to $75,000 (28%), $75,000 to $100,000 (23%), and $100,000 and above (20%). The vast majority of participants rated themselves as being at a “middle” socioeconomic status level (41%). The highest percentage of participants (27%) identified themselves as having 1 to 5 years of professional experience; however, two other categories (6 to 10 years and 26 or more years of experience) had the next highest percentages of participants (11% and 18%, respectively). The majority of participants identified themselves as being Licensed Professional Counselors (50%) or Licensed Professional Clinical Counselors (29%), and having a master’s-level education (64%) or higher.

Instruments

Demographic Questionnaire

The demographic questionnaire, developed by the researchers, included the following information: age, sex, self-identified race, household income, self-perceived socioeconomic status/class, relationship status, licensure status, number of years of professional counseling experience, and number of years of counselor education completed.

Marlowe-Crowne Social Desirability Scale

The Marlowe-Crowne Social Desirability Scale (SDS) is a well-known measure for test bias related to an individual’s need for approval (Leite & Beretvas, 2003). The SDS consists of 33 items, 18 are keyed “true” and 15 are keyed “false” (Crowne & Marlowe, 1960). The internal consistency coefficient for the SDS, using the Kuder-Richardson Formula 20, is .88 (Crowne & Marlowe, 1960). The alpha coefficient for the SDS in the present study was .86, indicating good internal reliability for this measure.

Clinical Attribution Scale

The Clinical Attribution Scale (CAS) is an 18-item, five-point Likert-type measure of dispositional bias (i.e., attributions of personality characteristics) and situational bias (i.e., attributions of external influences) published by Chen et al. (1997). According to Chen et al., this 18-item scale was adapted from a similar 33-item attributional scale used by Storms (1973), Russell (1982), and Batson, Jones, and Cochran (1979). Participants rated each response using a five-point Likert-type scale ranging from A = Strongly Agree, B = Agree, C = Undecided, D = Disagree, to E = Strongly Disagree (Chen et al.). Participants responded to statements such as, “Aspects of the person’s personality caused him to behave the way he did” (i.e. dispositional bias) or “The person’s behavior might be different if the situation were different” (i.e. situational bias). In the present study, lettered scores were translated to point values for statistical purposes: A = 1, B = 2, C = 3, D = 4, and E = 5. Higher scores on the CAS indicated a more internal (dispositional) inclination and lower scores indicated a more external (situational) inclination to explain a person’s problems. An item analysis conducted on the revised CAS revealed a Cronbach alpha reliability coefficient of .87 (Chen et al.). Chen et al. assessed validity by using expert rater judgments to determine where each item fell on the dispositional-situational continuum. In terms of reliability, their assessment revealed an average intraclass correlation coefficient of .96 and an average interrater agreement of .90. In the present study, statistical analysis revealed a Cronbach alpha of .81, indicating good internal reliability for the CAS.

Impact Message Inventory

The Impact Message Inventory (IMI) is a self-report transactional inventory (Kiesler & Schmidt, 2006) that was designed to measure how one interprets or characterizes the personal style of another person (Kiesler, 1987). The IMI has shown to be a useful measurement in countertransference research (Schwartz et al., 2007; Schwartz & Wendling, 2003) because it identifies the kinds of feelings, or distinctive covert reactions, that a person (i.e. client) can elicit in another person ([i.e., a psychotherapist] Kiesler & Schmidt, 2006).

The Impact Message Inventory Circumplex (IMI-C) is a briefer 28-item version of the full 56-item version of the IMI. The IMI-C, utilized in the present study, employs the four main anchor characteristics of the original octant subscales used in the 56-item measure, these are dominant, submissive, friendly, and hostile (Kiesler & Schmidt, 2006). Each item is rated on a 4-point Likert-type scale. Each subscale reflects the emotions produced within an individual (participant) by indicating how he or she has been impacted by another person’s behavior (simulated client). Each item is rated on a 4-point Likert-type scale according a person’s strength of agreement with each statement (1 = Not at all; 2 = Somewhat; 3 = Moderately so; 4 = Very much so).

Examples of participant responses to a statement using the different subscales, may be

Higher scores on the subscales reflect stronger emotional reactions. The internal consistency reliabilities from 14 studies (from 2002 to 2005) using the IMI-C reported Cronbach alphas ranging from .61 to .87 for the four primary subscales (Kiesler & Schmidt, 2006).

In the present study, the reported internal consistency reliabilities (Cronbach Alpha) for each of the subscales was .72 for the Dominant subscale, .78 for the Hostile subscale, .72 for the Submissive subscale, and .74 for the Friendly subscale. These results indicate adequate internal reliability for the IMI used in the current study.

Clinical Judgments

Clinical judgments (i.e., diagnosis, symptom severity, prognosis, treatment approach, length of therapy, and therapy outcome) have been a traditional measurement of bias in psychotherapy research (Garb, 1997; Lopez, 1989). For the purposes of comparison to past research, we incorporated four clinical judgment ratings into the present study. Participants were asked to rate the client along the following clinical judgments using a Likert scale. The four items included:

These questions received face validity after feedback from a committee of psychotherapy professors, and the questions were subsequently adapted from the research of Murdock & Fremont (1989).

Procedures

This study was developed as an analogue research design. Participants were contacted via e-mail invitation to participate in an online survey. This was the first known empirical study using an online survey with video simulation to assess attributional biases and countertransference reactions. Groups of participants were systematically chosen to receive a survey invitation that directed them to the case with a client from either a higher or lower SES. Neither participant group was aware of the other survey option. Each participant was asked to read a written case vignette describing a simulated client and his presenting problem. The client would be from either higher or lower SES. To supplement the written vignette, participants were also asked to view a 4-minute video segment of the simulated client presenting for a first session. Data collection materials included: (a) an informed consent statement, (b) written simulated client case vignette, (c) a demographic questionnaire, (d) the Clinical Attribution Scale, (e) the Impact Message Inventory-Circumplex, (f) the Marlowe-Crowne Social Desirability Scale, and (g) the Clinical Judgment items. Total completion time for reading and viewing the case study, and completing all surveys, was approximately 20 minutes.

Video Simulation

The videotape simulated a first-time client case assessment. The socioeconomic status was differentiated in the written client summary and video script by client-reported education (i.e., high school versus graduate degree), occupation (i.e., auto body service worker versus district manager), income (i.e., $20,000 versus $150,000), and lifestyle characteristics (i.e., bowling versus golfing as personal hobbies). All other personal and clinical information, such as family structure (i.e., wife and two children) and presenting problems (i.e., trouble establishing a social network), remained the same for both simulated clients. The actor (i.e., client) in the video presentation was the same person for both charcters, but SES was differentiated by appearance (i.e., unshaven versus shaven, t-shirt and jeans versus suit and tie). Face validity was determined by reviews of each video by a committee of psychotherapy professors, examining similarities of presenting problems and differences only related to higher or lower SES. Videos can be viewed online at: http://learn.uakron.edu/video/files/clinic/clinic_BWcheck.htm (lower SES client video) and http://learn.uakron.edu/video/files/clinic/famclinic_BWcheck.htm (higher SES client video).

Data Analyses and Results

In this study, the categorical independent variable was higher and lower client SES and the dependent variables were CAS scores (one continuous variable measuring attributional bias), IMI-C scores (four continuous variables measuring emotional countertransference reactions), and Clinical Judgment scores (four quantitative variables measuring therapeutic difficulty). Null hypothesis one was: “There is no statistically significant group difference in scores on the CAS between clinicians reacting to a lower SES client versus a higher SES client.” Null hypothesis two was: “There is no statistically significant group difference in scores on the IMI-C between clinicians reacting to a lower SES client versus a higher SES client.” Null hypothesis three was: ”There is no statistically significant group difference in Clinical Judgment scores between clinicians reacting to a lower SES client versus a higher SES client.”

To rule out potential effect from participants’ personal or professional characteristics, or social desirability bias, preliminary analyses were conducted to determine whether significant relations existed among demographic variables or social desirability (i.e., SDS scores), and the instruments used in the study. Pertinent demographic variables used in preliminary analyses included number of years working as a counseling professional, self-perceived SES level, self-reported income level, highest educational degree received, and licensure status. Because all variables were treated as continuous, Pearson correlations were used to determine whether these variables should be included in the main analyses as covariates. Because 20 separate Pearson correlations were conducted, a Bonferroni correction was used. The alpha level employed to determine statistical significance was, therefore, .003 (i.e., .05/20). No statistically significant correlations were found among any of the demographic variables in the study’s instruments. In addition, no statistically significant correlations were found between the SDS and the four IMI-C subscales. Therefore, the main inferential analyses were conducted without the use of covariates.

Main Inferential Analyses

Hypothesis One

A one-way ANOVA was used to test the significance of group differences between participant responses to the CAS for each level of an independent variable of the client videos for higher and lower SES, while also analyzing variation between and within each of these participant groups (Mertler & Vannatta, 2002; Weinfurt, 1995). Results showed no significant main effect between the two groups of participants, F (1,139) = .06, p = .81, partial η² = 0. Thus, the type of client participants responded to (i.e., higher SES or lower SES client) did not differentiate participants’ CAS scores. Therefore, null hypothesis one was not rejected.

Hypothesis Two

A one-way MANOVA was used to examine differences in the four IMI-C subscale scores among the participants viewing the higher or lower SES client videos, while controlling for the correlations among these dependent variables, in addition to studying the variation between the independent variables (i.e., higher or lower SES client videos) (Mertler & Vannatta, 2002). Results revealed a statistically significant main effect for IMI-C ratings, F (4,136) = 4.21, p = .003, observed power = .92. Partial η² = .11. This was a medium-effect size for the model (Cohen, 1988). Because there was a statistically significant main effect, post hoc analyses were conducted. Univariate ANOVA results indicated that only one of the four dependent variables, the IMI-C Dominant subscale, had significantly different scores among participants reacting to a client of higher versus lower SES, F (1, 139) = 9.58, p = .002. Partial η² = .06. This was a small effect size (Cohen, 1988). Participants rated the higher SES client (Mean = 9.90, Standard Deviation = 2.71) as being more dominant than the lower SES client (Mean = 8.61, Standard Deviation = 2.23). Therefore, null hypothesis two was rejected.

Hypothesis Three

Independent sample t-tests were used to evaluate the four-participant clinical judgments and whether they differed according to higher versus lower SES client. Ratings related to the severity of the client’s problems were significantly different between the participant responses to the two different client SES videos, t = –2.65 [df = 139], p <.05. The pattern of participant responses indicated that more mild problems were rated for the client of higher SES (Mean = 4.87, Standard Deviation = 1.5) and more severe problems were rated for the client of lower SES (Mean = 5.49, Standard Deviation = 1.29). The other three clinical judgments did not differ according to client type. Therefore, null hypothesis three was rejected.

Discussion

Results showed that there were no significant differences on scores of attributional bias between participant responses to higher or lower SES client videos. The average CAS total scores for each set of responses for participants viewing the higher or lower client videos were very close to one another: the client with the higher SES was 51.57, and client with the lower SES was 51.91. According to the CAS, this means on average the two groups of participants answered the questions in a somewhat neutral manner, without much differentiation toward internal or external attributional bias. Thus, no attributional bias was found among the responses of participants viewing the higher or lower SES client videos.

Overall, there are two important considerations related to the nonsignificant findings of the present study as they relate to previous studies. One consideration is how attribution is measured. In some previous studies, attributions were judged by rating a client’s traits or skills (Baron, Albright, & Malloy, 1995; Darley & Gross, 1983; Stevens, 1980, 1981). In these studies, participants ascribed traits to a person without necessarily being aware of the types of causal attributions to which they belonged. Considering the absence of a validated attributional trait list from previous research, the present study used the CAS, which asked participants to make causal explanations for the client’s problems. Perhaps the results of the present study would have been different if a different measurement of attributional characteristics was utilized.

One other important consideration when comparing the present study with previous attributional bias studies relates to how information about the client is presented (i.e., ambiguously or unambiguously). Our study deliberately attempted to present the client in a somewhat neutral and ambiguous way. This approach was taken in a purposefully, to be conservative given the potential training and practice implications of finding SES-related bias among psychotherapists. However, some previous studies were very direct in identifying the client as having positive or negative characteristics or ability levels (i.e. poor grades), and then participants were asked to rate the causal attributions of the individual’s success (Baron et al., 1995; Calhoun, 1975; Charles & Littig, 1982; Mann & Taylor, 1974). The manner (i.e. ambiguously or unambiguously) in which client information is provided to the participants has been shown to make a significant difference in how participants respond (Baron et al., 1995). Therefore, future research should consider not only the form of attributional measurement used but also how information about the observed individual is presented to the participants. Perhaps, the combination of these factors might have made a difference in the nonsignificant results of the present study.

Countertransference Reactions

Participants rated the interpersonal behavior of the client with higher SES as evoking feelings of dominance more so than the lower SES client. The “dominant” scale used in the IMI-C describes the interpersonal behavior of someone who leads, directs, influences and controls others (Kiesler & Schmidt, 2006).Our results show that client SES did appear to evoke differential countertransference reactions from the participants. Finding the presence of psychotherapist countertransference in relation to client SES in the present study is consistent with the results of previous research, in relation to psychotherapist countertransference and other forms of client diversity. Previous studies have also found that counselors had some type of countertransference reaction (i.e, anxiety, dislike, etc.) to clients with specific diversity characteristics (Gelso, Fassinger, Gomez, & Latts, 1995; Hayes & Gelso, 1993; Milton, Coyle, & Legg, 2005). Some of these studies used client simulations similar to the methodology of the present study (Gelso et al., 1995; Hayes & Gelso, 1993). However, the biggest difference between the present study and previous studies relates to how countertransference was measured. Previous client diversity studies primarily assessed countertransference through the measurement of anxiety or discomfort (Gelso et al., 1995; Hayes & Gelso, 1993; Milton et al, 2005), but the more specific characteristics of the interpersonal reactions of the counselor toward client were not measured. The present study not only found evidence of countertransference reactions, but also was able to identify what type emotional impact (i.e, dominance) the client had on the counselor. Overall, these results support the theory of countertransference, which suggests that a counselor can experience an evoked or elicited response to certain client characteristics or behaviors, which would parallel the responses of other typical counselors in similar circumstances (Kiesler, 2001; Schwartz et al., 2007).

Finally, significant results were found in the present study with regard to clinical judgments. It was found that participants tended to rate the client of lower SES as having more severe problems while client of the higher SES problems received more mild ratings. This was a general rating of how participants perceived the client’s severity of problems, and it was not necessarily based on actual problems presented in the video. Therefore, it is interesting to note that despite the presentation of ambiguous information about the client’s presenting problems, participants had opposite ratings of problem severity for the lower (i.e. rated with more severe problems) and higher SES client (i.e. rated with more mild problems) than was actually presented verbally during the case vignettes. Differences in clinical judgments between lower and higher SES clients have been found in past research. Previous studies conducted in the 1960s and 1970s concluded that negative clinical judgment bias was found in relationship to lower SES clients (Abramowitz & Dokecki, 1977). Past research also found that lower SES clients tended to receive ratings of increased pathology (DiNardo,1975; Lee and Temerlin, 1970; Trachtman, 1971). These past findings highlight the presence of negative clinical judgments toward clients of lower SES. Similarly, in the present study ratings suggested that client SES influences how psychotherapists perceive a client’s problems. In this case, the client’s problems were either over-pathologized or minimized, in relation to client SES, despite the same information presented to the participants.

Clinical and Research Implications

The overall implication from the results of this study is that SES as a client diversity characteristic can have an impact on countertransference reactions and clinical judgments of psychotherapists. Countertransference reactions and clinical judgments can influence how psychotherapists make treatment decisions for the client, and they also influence the process of therapy. With the consideration of multicultural principles in clinical practice, psychotherapists are encouraged to be aware of the differences between themselves and their clients (Association for Multicultural Counseling and Development, 2009).

Beyond the traditional findings of clinical judgments related to client SES, the results from this study indicate a need for psychotherapists to consider that diverse client characteristics can have an impact on a psychotherapist’s countertransference reaction. In this study, psychotherapist countertransference was elicited from the higher SES client indicating that they were impacted by the client in a dominant way. These results give support to the suggestion that countertransference is a common factor influencing psychotherapists’ work with clients (Pillay, 2009), and that unrecognized countertransference can have a negative impact on the therapy process (Kiesler, 2001).

Countertransference has the ability to positively influence the psychotherapy process (Gelso & Hayes, 1998) because it can provide information to psychotherapists about a generalizable client population or characteristic. More specifically, if psychotherapists are able to make themselves aware of their countertransference reactions, they may prevent themselves from acting out in a biased manner with their clients. It may also inform psychotherapists about how to approach or not to approach their clients, or even allow psychotherapists to disclose some aspects of a countertransference reaction to their client in order to deepen the therapeutic relationship (Gelso & Hayes, 1998). In turn, this form of therapeutic disclosure may expose clients to one perspective of how others may emotionally react to them in outside world. It is recommended that psychotherapists learn to understand countertransference by identifying its origins, triggers, and manifestations (Hayes & Gelso, 2001). In addition, psychotherapists should learn to identify their feelings of countertransference in the therapeutic process using the skills and abilities, such as: self-integration, anxiety management, conceptualizing skills, empathy, and self-insight (Gelso & Hayes, 1998).

Results found here also show that it is important to consider the impact of SES-related differences between a psychotherapist and a client. Differences in worldviews and value systems can have a significant impact on the therapeutic process (Sue & Sue, 2003). The results of the present study found that participants reacted to clients of higher and lower SES in different ways, and they highlight the importance of psychotherapist consideration of SES as a part of one’s cultural identity and how thereapist SES might compare with client SES value systems. Otherwise, inappropriate or incompatible treatment approaches may be used if SES value differences go unrecognized.

Limitations

It is important to point out that although the present study showed significant findings, the effect size was small. Therefore, psychotherapists should use these results as starting point for understanding clients of differing SES. In addition, the small effect size (Partial η² = .06) indicated that the variables tested in the present study accounted for a minority of the variance in the results. There are several other limitations to this study.

First, a person’s perception of another’s SES can include many subjective elements, even those we define as “objective” (Liu et al., 2004). Therefore, the client presentations may not have achieved the clear-cut dichotomy of a higher or lower SES client because of the subjective nature of the observer’s interpretation.

Second, the CAS, a measurement of causal attributions, may not have captured the entire range of attributional biases shown by psychotherapists. Future research may consider developing a new, valid, and reliable measure of attributions, which would consider the different forms of attributional bias and how they relate to psychotherapist decision-making relative to client diversity.

Third, the use of the IMI-C in a simulated therapy scenario versus a real counseling situation should be taken into consideration. Although the IMI-C’s psychometric properties have been shown in various studies over the course of approximately 20 years, it is difficult to know with certainty how psychotherapists might have responded to clients in their own practice. In addition, the psychometric properties indicated for the IMI-C in this study, were in the range of .72 to .78 for internal validity, indicating fair results. Therefore, caution should be exercised for future research when utilizing this instrument.

The fourth limitation of this study is related to the use of video simulations instead of clients. The use of client simulations avoids the ethical implications of using actual psychotherapists and clients (Heppner, Kivlighan, & Wampold, 1999). However, the use of clients in actual therapy situations in future research can provide a more natural view into the actual experiences of psychotherapists and their clients (Pope-Davis et al., 2001).

Finally, there may have been some limitations related to using an internet survey. Although an online survey was purposely used in the present study because it was user friendly, cost effective (Heppner et al., 1999), and had the potential to increase sample size and increase the breadth of participants (i.e., a national sample), future research should take into consideration the level of computer literacy of participants and any technical issues that may arise related computer compatibility.