Countertransference Feelings when Treating Teenagers. A Psychometric Evaluation of the Feeling Word Checklist–24

Introduction

In child and adolescent outpatient clinics, individual therapy sessions serve as a central part of the assessment and treatment of patients. The relationship between the patient and the therapist is an important tool to improve the patient’s mental health, and the emotional exchange between patient and therapist is a major part of all treatment relationships (Hill & Knox, 2009). Binder and colleagues (2011) found in a group of adolescent patients treated by therapists with different theoretical approaches, the need for a certain type of therapist behavior and attitude to establish a working relationship imbued with trust and autonomy.

Countertransference was first described by Sigmund Freud as therapists’ emotions, feelings, attitudes and behaviors that were not neutral and were result of the therapists’ neurotic conflicts and, therefore, a disturbing factor to be kept out of therapy (Freud, 1910; Freud, 1912). Kernberg (1965, pp. 38-56) labeled Freud’s description as a “narrow,” excluding the relational perspective in individual psychotherapy. Since then the concept of countertransference has been vividly described and theoretically discussed, and the understanding of it has developed over the years. Different authors underlined a relational aspect in countertransference. For example, Winnicott (1949), when reporting psychotherapy process with a child, described the therapist’s negative feelings as a normal reaction. Sullivan (1953) felt the therapists’ reaction could be understood as a response to the patients’ invitation to an interpersonal interaction. The totality of what the therapist experiences and feels, together with the patient might be seen as countertransference. Thus, the psychotherapist’s emotional reaction is the therapist’s most important instrument to understand the patient and the relationship between them (Heimann, 1950). This “totalistic” position serves to normalize the therapist’s feelings, and brings the inner life of the therapist, as well as the therapist’s contributions to the relationship, into the theoretical and clinical discussions (Kernberg, 1965; Dahl, 2011).

Even though countertransference originally was described in psychodynamic theory, empirical studies indicate that countertransference is a universal phenomenon, and it is not associated with only one theoretical orientation (Betan et al., 2005; Dahl et al., 2011). Also, authors of empirical studies (Holmqvist et al., 1996a; Holmqvist et al., 2004; Røssberg et al., 2003b) hold that countertransference develops in the interaction between the therapist and the patient, and is a result of their unique relationship. Betan and colleagues (2005) explored self–reported countertransference reactions. Clinicians from a variety of theoretical orientations filled in a countertransference questionnaire reporting their conscious, cognitive, affective, and behavioral responses to their patients. The authors posited that the questionnaire provided clinicians across theoretical orientations a method of capturing information about the patient and of offering treatment that could be therapeutically significant (Betan et al., 2005). To explore countertransference feelings empirically, different versions of countertransference questionnaires called the Feeling Word Checklists (FWC) have been developed. Despite the fact that the countertransference construct was developed from individual therapy, most studies using FWCs are performed in inpatient departments. As far as is known, FWC has not been used in individual therapy sessions with adolescents.

However, in another line of empirical work with countertransference, the rater–coded Core Conflictual Relationship Theme Method (CCRT), which measures countertransference in individual sessions with adolescents, showed high repetitiveness in the therapists’ narratives about their parents and their patients (Luborsky et al., 1990; Tishby et al., 2011).

Aims of the Present Study

The primary aim of this study was to explore the factor structure and psychometric properties of the Feeling Word Checklist–24 (FWC–24) used in individual sessions with adolescents. How many clinically meaningful factors do the items in FWC–24 represent and what are their psychometric properties?

A secondary aim was to examine the relationship between the factors and the therapist variables: age, gender, number of years of college or university education, number of years of clinical education after college/university, years of clinical practice, years with clinical supervision, and therapist–rated alliance.

Method

The study was performed in outpatient child and adolescent clinics in Vestfold County, Norway. Data was collected in May and June 2011. Therapists in four outpatient clinics filled in the FWC–24 after each session with a teenager, aged 13–18 years. Some therapist data were collected; age, years of college/university education, years of clinical education after college/university, years of clinical practice, years of clinical supervision.

The study (Ulberg et al., 2012) was approved by the Norwegian Social Science Data Services (Project number: 24880 FEEL–IT) as a pilot in the First Experimental Study of Transference work in Teenagers (FESTIT).

Participants

All therapists in the outpatient clinics (N= 50) were invited to participate. The therapists included specialists and trainees in child and adolescent psychiatry, child and adolescent psychology, and counselors. All therapist data were anonymous. The therapists were instructed to fill in FWC–24 (maximum 20 forms per therapist) after each session with adolescents, aged 13–18 years.

Adolescents referred to outpatient clinics in the Norwegian mental health system have a broad range of clinical problems, such as depression, anxiety disorders, personality disorders, behavioral disturbances, ADHD, and psychosis. Since the primary aim was to explore the factor structure and psychometric properties of the Feeling Word Checklist–24 (FWC–24) used in individual sessions with adolescents, and because of limited economic resources and time constraints, no patient data was included.

Feeling Word Checklist

Whyte and colleagues contributed to the empirical research on countertransference by developing the first version of the therapist rated Feeling Word Checklist (FWC) (Whyte et al., 1982). The FWCs developed differ, with various numbers of feeling words, and therapists are asked to fill in questionnaires with either yes/no or a Likert-scale answers. Only two studies have used the FWC in individual therapy to examine the factor structure. The FWC in both those studies used a Likert scale (Holmqvist et al., 2002; Ulberg et al., 2009; Dahl et al., 2011). No study has explored the factor structure of FWC in individual therapy sessions with adolescents.

No agreement has been reached on the number of factors underlying the FWC. These factors vary in number from three to seven. Each study uses its own FWC, thus explaining the variation in number of factors and whether the FWC uses a Likert scale or yes/no questions (Røssberg et al., 2003; Holmqvist et al., 2001). In addition, the studies involve different adult patients, e.g., those with borderline disorders and psychosis, anxiety, or large heterogeneous patient groups (Holmqvist & Armelius, 1996b; Hoffart & Friis, 2000; Røssberg et al., 2003b).

Associations between the FWCs and patient variables, such as gender, diagnosis, and use of medicine have been reported by a few authors, and have especially studied staff feelings when working with inpatients. The patient variables suicidal ideation and aggression were connected to negative feelings among the staff members (Røssberg et al., 2003a). Holmqvist and colleagues reported that 60% of the variance was explained by three components: the personal style of the staff, personal style of the patient, and the unique interaction effect between patient and therapist. These components were larger than therapist response to patient diagnosis (Holmqvist et al., 1996a; Holmqvist et al., 1996b).

In the present paper the therapist’s emotional reactions to adolescent patients were explored. The Feeling Word Checklist–24 (FWC–24) was used (Holmqvist et al., 1996a; Holmqvist et al., 1996b; Holmqvist et al., 2007). The FWC–24 is a questionnaire designed to capture the therapist’s self–reported feelings; that is, the feelings of which therapists become aware, acknowledge, remember, and are willing to report after each session (Najavits, 1995; Dahl, 2011). The 24-item version, also used in the Swedish Erica Process and Outcome Study (EPOS) (Odhammar et al., 2011), was chosen. The form was translated from Swedish into Norwegian and back into Swedish.

Alliance

The working alliance refers to the collaboration between the therapist and patient, and captures the quality of the working relationship in psychotherapy (Horvath et al.,1986; Horvath et al.,1989,). Different instruments are used to assess the working alliance; however, Bordin’s (1976) definition of alliance is applicable across different therapeutic models and techniques. According to Bordin, the working alliance incorporates a mutual understanding of the following factors:

However, in the present study a therapist-rated visual–analogue scale (VAS) (Borkovec & Costello, 1993; Borkovec & Nau, 1972; Borkovec & Sibrava, 2005) was used as the measure of alliance. The therapists rated the alliance on a scale from 0 (The patient showed resistance. Would not engage in therapy) to 10 (The patient showed no resistance. Was highly engaged in therapy).

As far as is known, only two other studies have explored the correlation between therapists’ feelings and the alliance (Najavites, et al., 1995; Dahl, et al., 2011). Najavites and colleagues (1995) reported that therapists’ positive feelings showed a positive correlation with therapist evaluation of alliance, and negative feelings correlated significantly negative with therapist evaluation of alliance. Dahl and colleagues (2011) reported a strong negative correlation between “disengaged” countertransference and therapist–rated alliance.

Statistics

The analyses were performed in SPSS–18. To verify that our data were suited for analyses, the Kaiser–Meyer–Olkin Measure was calculated. We also used Bartlett’s test of sphericity x² to decide if the correlation between the feeling words was strong enough for factor analyses. To investigate the factor structure in the FWC–24, a principal component analysis (PCA) was conducted on the 24 items, with promax rotation. We used the Kaiser eigenvalue “greater-than-one” criteria, a screen plot, variance accounted for by the solution, and interpretability to select the number of factors to rotate. For discriminating purposes we wanted to keep the items that correlated highly with only one subscale. Hence, we excluded feeling words that loaded with less than 0.4 on single factor and feeling words loaded with more than 0.3 on multiple factors. We repeated the procedure to yield a final factor solution that included only feeling words that loaded >0.4 on only one factor. The internal consistency of the factors was measured by Chronbach’s alpha. The factors were validated with analysis of the correlation with therapist variables and alliance by using Pearson Correlation or t–test.

Results

Data was collected from 41 therapists. The therapists mean age was 40. 2 years (SD =9.5); mean years of college or university education was 6.7 (SD = 1.7), mean years of clinical education after college or university was 4.2 (SD = 2.6), mean years of clinical practice was 6.4 (SD = 6.4), and mean years of clinical supervision was 4.3 (SD = 2.6). A total of 410 forms were collected and included in the study. Each therapist filled in at least one, but no more than 20, form[s].

Factor Analyses

The Kaiser–Meyer–Olkin Measure verified the sampling adequacy (0.87), and the Bartlett’s test of sphericity, x² (276) = 4532,87, p<0.000, indicated that correlations between items were sufficiently large for PCA.

The analysis revealed four factors with an eigenvalue >1.0. These four factors explained 63% of the total variance. The first factor (Table 1) was clearly the strongest, and this factor alone explained 31% of the variance.

When looking into the feeling words included in each factor, they all seemed clinical meaningful. The four factors were named Confident (included: playful, vigorous, enthusiastic, happy, open, warm, content), Inadequate (included: overwhelmed, tense, nervous, thunderstruck, helpless), Disengaged (included: indifferent, tired of, cold, irritated) and Pragmatic (included: prudent, neutral)

Validation

The correlation between the factor scores in the reduced set of items and the subscales scores were > 0.93 on all factors. This implies that the subscale scores may substitute the factor scores without loss of information.

The internal consistencies of the subscales were measured with Chronbach’s alpha: For the Confident subscale alpha was 0.86, for Inadequate it was 0.80, for Disengaged it was 0.71, and for Neutral it was 0.92.

The mean values over all sessions was highest on Confident, where the mean = 1.45 (SD=0.64). The Neutral subscale was slightly lower, with a mean = 1.20 (SD = 0.77). Inadequate has a mean = 0.49 (SD = 0.53), while the Disengaged subscale had a mean = 0.36 (SD = 0.46).

To further validate the countertransference (CT) factors, the relationship between the factors and the following therapist variables were explored; age, gender, number of years of college or university education, number of years of clinical education after college/university education, years of clinical practice, years with clinical supervision. Also the relationship between the CT factors and alliance were explored (Table 2).

Correlational analyses with therapist variables, revealed positive correlation between Confident CT and the therapist’s age, length of education, supervision and more years working in child– and adolescent psychiatry. Inadequate CT showed a negative correlation with years of supervision. Disengaged CT had a negative correlation with the therapist’s age and years working in child– and adolescent psychiatry. Neutral CT correlated positively with age, years of supervision and postgraduate clinical education.

Alliance showed a positive correlation with Confident CT and a negative correlation with Inadequate CT and Disengaged CT.

Discussion

The primary aim of the present study was to explore the factor structure and psychometric properties of the FWC–24. The number of clinically meaningful factors the items in FWC–24 constituted and their psychometric properties were explored. In the present study 18 items from FWC–24 constituted four subscales that were conceptually coherent, psychometrically acceptable, and clinically recognizable. The four subscales were named Confident, Inadequate, Disengaged and Pragmatic. These seem to represent different feeling aspects.

Comparing this study with other factor analyses of FWCs, common factors were observed. However, because of the use of different FWCs, with different numbers of feeling words, the studies and factors are not directly comparable. For example, two of the seven subscales found by Røssberg and Friis (2003), Inadequate and Confident, conceptually overlap with the factors Inadequate and Confident in the present study. Comparing this study with the two studies of individual psychotherapy (Dahl et al., 2011; Holmqvist et al., 2002), three of the factors conceptually overlap with the study from Dahl and colleagues (Dahl et al., 2011); Confident, Inadequate, and Disengaged. Two factors in the study from Holmqvist and colleagues (2002) showed similarities with some of the factors in the present study. Their factor, Positive, seemed to conceptually overlap with the first factor, Confident, and their factors, Negative and Distant, seemed to be comparable with the feeling aspects in the third factor, Disengaged. The other studies report factors with more negative and aggressive feeling words. However in FWC–24, only the words “cold” and “frustrated” describe negative feelings.

Correlational analyses with therapist variables and alliance seems to support the conceptual meaningfulness of the four factors; Confident CT was positively associated with more educational and practical experience, as well as with an increased amount of supervision. Whether the positive correlation between Confident CT and age is a result of age, or is explained by the probable high correlation between experience and age, we do not know. In line with other studies (Dahl et al, 2011; Najavits et al., 1995), Confident CT also showed a strong correlation with the therapist-rated alliance.

Inadequate, on the other hand, showed a strong negative correlation with therapist rated alliance, which is in line with Najavits and colleagues (Najavits, et al., 1995) study. However, in Dahl et al. (2011) there was no correlation between therapist rated alliance and Inadequate CT. In their study, the therapists were all very experienced, which may explain why the findings are divergent on this point. In addition, Inadequate CT correlated with amount of supervision, that is, more supervision, lower instances of Inadequate CT. However, the amount of education or of clinical practice do not seem to affect the therapists’ Inadequate CT reaction.

Therapists who were older and had longer clinical experience had lower score on the words describing the Disengaged factor. Therapist-rated alliance showed a negative correlation with Disengaged, as was the case with the results reported by Najavits and colleagues (1995). The Neutral factor showed a positive correlation with therapist age, years of clinical experience and education, and years of supervision. The factor Neutral might mirror a pragmatic stance in more experienced, highly educated and older therapists. The correlational analyses imply that those therapists feel more confident, more neutral and pragmatic, and less inadequate and less disengaged.

One limitation of the present study is the inconsistency in the number of forms filled in by the therapists, for example a therapist may have filled in more than one form after different individual sessions with the same patient. Thus, there is some interdependence in the data. If more specific patient information were available, correlational analyses could have been performed on the patients’ contribution to countertransference. However, very few studies have reported an effect of patient variables or patient–reported alliance on countertransference (Najavits et al., 1995; Dahl et al., 2011; Røssberg et al., 2003a).

Binder and colleagues (Binder et al., 2011) reported the need for the therapist to appear comfortable with being a therapist and to help the adolescents to make the therapeutic experience understandable and meaningful. The FWC–24 might be a valuable and useful instrument in individual therapy with adolescents by increasing the therapist’s awareness of countertransference feelings in the sessions with the patient, and by helping the therapist to focus more precisely on relationship themes. This, in turn, may facilitate the therapeutic process and enhance the improvement of the patient’s interpersonal relationship.

Conclusion

The four factors revealed were Confident, Inadequate, Disengaged and Neutral. The subscales had satisfactory internal consistency and described meaningful emotional profiles, which partly overlapped with factors from other studies. The factors revealed might be useful in further research on patient variables concerning the therapist–patient relationship in individual therapy with adolescents.