Narrative Processes in Psychotherapy: Differences between Good and Poor Outcome Clients

Procesos Narrativos en Psicoterapia: Diferencias entre clientes con buen y mal resultado

 

 

Luis Botella and Joana Maestra Cutura

Universitat Ramon Llull (Barcelona), España.

Correspondence

 

 

---

ABSTRACT

This paper compares 30 patients with good therapeutic outcome to 30 with poor therapeutic outcome in terms of the differential distribution of (1) Intake Variables (2) Outcome and Process Variables, and (3) Narrative Variables. Results indicated that psychosocial functioning, motivation, pre-therapy symptoms, Working Alliance, total number of therapy sessions, total pre-post symptom reduction, and mean scoring for total working alliance in sessions 3, 4, and 8 discriminated between both groups. Results also showed that almost all narrative variables except some of them discriminated good outcome clients from poor outcome ones from the beginning, midpoint and final stage of their therapeutic process. These results are discussed according to their relevance for clinical practice.

Key words: narrative processes in psychotherapy; psychotherapeutic outcome; psychotherapeutic process.

---

RESUMEN

Este trabajo compara 30 pacientes con buen resultado terapéutico con 30 con un mal resultado terapéutico en términos de la distribución diferencial de (1) variables iniciales (2) variables de proceso y resultado y (3) variables narrativas. Los resultados indicaron que el funcionamiento psicosocial, la motivación, los síntomas pre-terapia, la alianza terapéutica, el número total de sesiones de terapia, la reducción total de los síntomas pre-post y la media de puntuación de la alianza de trabajo total en las sesiones de 3, 4 y 8 discriminaron entre ambos grupos. Los resultados también mostraron que casi todas las variables narrativas, excepto algunas de ellas discriminaron los clientes de buen resultado de los de mal resultado desde el principio, punto medio y fase final de su proceso terapéutico. Estos resultados se discuten en función de su relevancia para la práctica clínica.

Palabras clave: procesos narrativos en psicoterapia; resultado terapéutico; proceso terapéutico.

---

 

Introduction

Both in terms of proposing specific forms of narrative therapies (see, e.g., Angus & McLeod, 2004) and of studying the characteristics of client-generated narratives in psychotherapy (Neimeyer, 1995) story-telling and meaning construction have been added to the repertoire of constructs that help understand and foster psychotherapeutic outcomes and processes.

A detailed analysis of all the literature on client narrative processes in psychotherapy would be beyond the scope of this paper. However, we will summarize the main results from studies so far focusing on the main dimensions of narrative content, structure, and process. Research on narrative changes in psychotherapy until now demonstrates that all of the narrative dimensions summarized before are relevant to the clients' well being and, therefore, are affected by psychotherapeutic out-come and process.

Gonçalves and Machado (1999), focusing on the content of problematic patient narratives concluded that there was support for "the specificity nature of cognitive organization in these dysfunctions as well as to the pos-sibility of identifying this specificity in prototype narra-tives" (p. 1187).

Dimaggio and Semerari's (e.g., 2001; Dimaggio et al., 2008; Semerari et al., 2003) categorization of pathological narrative forms combines narrative structure (e.g., lack of integration), process (e.g., lack of hierarchical complexity), and content (e.g., lack of emotional states in the narrative). Their studies demonstrated that the client's narratives during sessions "made it possible to identify consistent clusters of constructs signaling the existence of different mental states" (Semerari et al., 2003, p. 349) and that these clusters changed as therapy advanced.

The study of narrative processes in Emotion Focused Therapy (EFT) by means of the Narrative Process Coding System (see, e.g., Angus, 2012) demonstrates that changes in the narrative process (in the direction of the development of new meanings through emotional awareness and symbolization of previously problematic ones) are the key to clients change.

Research by the group led by Miguel Gonçalves and Ines Mendes (see Gonçalves et al., 2011, 2012; Mendes et al., 2010, 2011; Matos et al., 2009) on new emerging therapeutic meanings that they refer to as Innovative Moments (IMs) demonstrate that IMs are transtheoretical since all therapies aim to create alternative meanings in clients' self-narratives. Again, therapeutic change is correlated with changes in narrative process and content.

In her detailed analysis Angus (2012) proposes that clients' disclosing and sharing personally relevant narra-tives in therapy has three main implications for psycho-therapeutic outcome and process:

(a) it is the basis of a strong therapeutic alliance;

(b) It helps therapists to understand their clients problems; and

(c) it allows the client to "access, explore and understand distressing life experiences with an empathic, non-judging other" (p. 375).

The main narrative dimensions that have been proposed to analyze client-generated narratives in psychotherapy, synthesized from a growing body of research including the one already summarized (see also Gergen, 1994; Gonçalves, 2000; Gonçalves & Henriques, 2000a, 2000b, 2000c, 2000d; McAdams, 2006; Neimeyer, 2006), are the following ones.

1. Narrative Structure and Coherence

1.1. General Orientation: (The Who, When, and Where of the narrative). General orientation of a narrative informs about the characters and the social, spatiotemporal, and personal contexts where the actions occur.

1.2. Structural Sequence: (The What of the narrative). In its minimal form it includes: (1) an initial event; (2) an internal response to the event (i.e., goals, plans, thoughts, or feelings); (3) an action and (4) its consequences.

1.3. Evaluative Commitment: (The Why of the narrative). Evaluative commitment informs about the significance that the narrator ascribes to the event being narrated, i.e., about its importance within his or her world outside the narrative itself.

1.4. Integration refers to the clarity of the narrative thread or plot, i.e., its overarching sense of connection to the variety of events being narrated.

2. Narrative Content and Multiplicity

2.1. Thematic Variety refers to the amount of different themes included in the narrative as well as to the description and detailed elaboration of their specific contents.

2.2. Variety of Events refers to the number of events included in the narrative.

2.3. Variety of Scenarios refers to the number of scenarios included in the narrative.

2.4. Variety of Characters refers to the number of characters included in the narrative.

3. Narrative Process and Complexity

3.1. Objectifying refers to the level of sensorial complexity of the narrative.

3.2. Emotional Subjectifying refers to the level of emotional complexity of the narrative.

3.3. Cognitive Subjectifying refers to the level of cognitive complexity of the narrative.

3.4. Metaphorizing refers to the level of metacognitive and meaning construction complexity of the narrative.

4. Narrative Intelligibility

According to the work of Gergen (1994), to be intelligible, a narrative should:

4.1. Establish a valued ultimate goal;

4.2. Make goals nonconflicting;

4.3. Make goals reasonable;

4.4. Select events relevant to the achievement of this goal;

4.5. Locate those events in a sequence;

4.6. Characterize its cast providing stability to its characters identity; and

4.7. Construe causal links among the events.

5. Other relevant dimensions (McAdams, 2001; Adler, Wagner, & McAdams, 2007)

5.1. Contamination sequences are episodes with a progressive beginning followed by a regressive ending.

5.2. Redemption sequences are episodes with a regressive beginning followed by a progressive ending.

6. Narrative position of the self

The client's position relative to his or her selfnarrative and its possibility of therapeutic change can be characterized as (i) victim, (ii) partially in control, (iii) moderately agentic, and (iv) highly agentic.

In a previous study (Botella & Gámiz, 2011) we compared 12 patients undergoing psychotherapy (five with good therapeutic outcome and seven with poor therapeutic outcome) in all of the relevant dimensions just mentioned. Moreira, Beutler, and Gonçalves (2008) conducted a similar study three years before and they found that good outcome cases presented a higher statistically significant total narrative change than poor outcome cases. These authors found no further statistically significant differences between groups even if nonstatistical analysis suggested some trends. A comparison of our previous results and the ones by Moreira, Beutler, and Gonçalves (2008) is relevant to the hypotheses guiding this paper.

(a) Nonstatistically significant trends in Moreira, Beutler, and Gonçalves' study (2008, pp.1190-1191) suggested that content multiplicity was the dimension for which the highest level of change was obtained, whereas the lower level of change obtained was for process complexity. This result is coherent with our own; we found that a higher level of variety of events and characters (two of the four subdimensions included as components of content multiplicity) distinguished good outcome patients' narratives from poor outcome ones.

(b) Nonstatistically significant trends in Moreira, Beutler, and Gonçalves' study (2008, pp.1190-1191) also suggested that changes in patients' narrative structural coherence were found throughout the therapeutic process and it seemed to be able to differentiate positive outcome cases from negative/poor outcome cases. This result is also coherent with our own; we found that a higher level of structural sequence and of integration (two of the four subdimensions included as components of narrative structural coherence) distinguished good outcome patients' narratives from poor outcome ones

(c) Nonstatistically significant trends in Moreira, Beutler, and Gonçalves' study (2008, pp.1190-1191) also suggested that integration appeared to be the most discriminative subdimension between positive and nega-tive/poor outcome cases. This result is only partially coherent with our own, because even if integration was indeed a discriminative dimension (as already discussed above) it was not the most discriminative one-which in our previous study turned out to be metaphorizing (a subdimension of Narrative Process and Complexity).

Our results (see Botella & Gámiz, 2011) also indicated that two narrative dimensions not included in Moreira, Beutler, & Gonçalves' (2008) study (i.e., redemption sequences and narrative position of the self as a victim) distinguished poor outcome patients' narratives from good outcome ones.

The present study's main goal was to retest the usefulness of a form of narrative analysis based on the multidimensional approach to narrative processes already discussed. In this case our goal was to test the commonalities and differences between a group of narratives from patients with good therapeutic outcome and another one from patients with poor therapeutic outcome, and with a bigger sample (60 patients instead of 12).

Regarding narrative processes in psychotherapy, our hypothesis was that for every narrative dimension included in our analysis there will be significant differences between good and pooroutcome patients.

Despite our main focus on differential narrative processes in good vs. poor therapeutic outcome, our study included also the assessment of Intake Variables (i.e. therapeutically relevant variables assessed during the intake session) and Outcome and Process Variables (symptoms and working alliance). The existing literature on the topic of these two clusters of variables and good vs. poor therapeutic outcome (see, e.g., Botella et al. 2008) led us to hypothesize that (a) therapeutic outcome would not be predictable exclusively on the basis of neither presenting complaint or problem's onset. Also, we hypothesized that (b) psychosocial functioning in good-outcome clients would be significantly lower than in poor-outcome one (due to the diminished quality of life entailed by psychological distress), (c) the mean scoring for motivation in good-outcome clients would be significantly higher than in poor-outcome ones, and (d) that the mean scoring for pretherapy symptoms in good-outcome clients would be significantly lower than in poor-outcome ones.

We also wanted to test the possible correlations between the Intake Variables assessed, their factorial structure, and the differences between good- and poor-outcome cases in Outcome and Process variables. Regarding this specific point (and again considering the existing literature on the topic of good vs. poor therapeutic outcome) we expected that the total number of therapy sessions and symptomatic level from sessions 1, 2, 3, 4 and 8 in good-outcome clients would be significantly lower than in poor-outcome ones. Conversely, we expected that averaged Therapeutic Alliance, total pre-post symptom reduction, and Working Alliance from sessions 1, 2, 3, 4 and 8 in good-outcome clients would be significantly higher than in poor-outcome ones.

 

Method

Participants

i. Clients: basic demographic variables

Participants in this study were 60 patients; comprising 43 women (69.4%) and 17 men (27.4%) who received outpatient psychotherapeutic treatment (one weekly session) in a University based psychotherapeutic unit in Barcelona, whose functioning is described below in the procedure section. Treatments analyzed corresponded to all those patients who began their therapeutic process at any point within the three years and four months included between the beginning of the unit and the analysis of the data included in this study. Participants' mean age was 33.27 years (SD = 10.48; range 19 to 57).

ii. Clients: clinical variables

The 60 patients in our sample were divided in two groups: 30 of them belonging to the good therapeutic outcome group (GTO) and the remaining 30 to the poor therapeutic outcome group (PTO). Patients belonging to the GTO group needed to fulfill all of the following criteria:

(1) Therapeutic treatment finished, i.e., therapist agrees that the client has reached the end of this therapy;

(2) Symptomatic improvement pre- post;

(3) No ruptures in the therapeutic alliance; and,

(4) No symptomatic relapse during the course of therapy.

Patients belonging to the PTO group must fulfill all of the following criteria:

(1) Therapeutic treatment finished, i.e., therapist agrees that the client has reached the end of this therapy or therapy dropout, i.e., clients that leave therapy without their therapist agreeing that he or she has reached the end of his or her therapy.

(2) No symptomatic improvement pre- post or symptomatic worsening.

Regarding the total number of therapy sessions per patient, the mean in our sample for the GTO group was 13.90 (SD = 5.49; range 4 to 25), and for the PTO was 21,65 (SD = 11,19; range 10 to 52). Differences between both groups regarding the number of sessions are logical taking into account that they were defined by their level of success: poor outcome clients do not succeed either by abandoning therapy or by staying for too long without success, that explains why the SD in the PTO group is higher and the range is wider than in the GTO one.

The distribution of the general sample, however, adjusts to criteria of normality (a Kolmogorov-Smirnov test was performed to assess this: Z = 1.76; Sig. (2-tailed) = .004). The extremes of this distribution (i.e., 10 patients with less than 10 sessions and 4 with more 39) can be explained by the sample being made up of good and poor outcome patients simultaneously. This accounts for the fact that some of the sample's patients were characterized by an abnormally low number of sessions (dropout) or an abnormally high number of them (stagnation).

Figure 1 depicts the distribution of presenting complains of our sample. As can be seen in the Figure, 78% of the sample requested therapy for one of the following reasons: relational difficulties, depressive mood or anxiety/stress.

 

 

Regarding problem onset, a 75% of clients (45) in our sample had experienced their present difficulty for more than one year previous to asking for help; a 13.3% (8 in total) for less than one year, and the remaining 11,7% (7 in total) reported having experienced it recurrently

Regarding patient's level of psychosocial functioning at the time of the consultation, they were asked to selfassess it by using a 0% to 100% scale (with 0% meaning not functioning at all, and 100% meaning functioning extremely well). Data from our sample indicated that the participants mean selfassessed level of psychosocial functioning was 53.13% (SD = 14.34; range 30 to 98).

ii. Therapists

Nine therapists participated in this study; seven women and two men. Their age range was 26-40 years (M = 33.22, SD = 6.48). All of the therapists were clinical psychologists and psychotherapists with a masters level training. Eight of them held a Ph.D. Clinical experience ranged from 3 to 17 years (M = 10.22, SD = 6.48).

Instruments

i. Outcome. To assess the clients' symptomatic level we used the CORE Outcome Measure (CORE-OM); a self-administered 34 item questionnaire that evaluates the following dimensions: (a) subjective wellbeing, (b) problems/symptoms, (c) functioning, and (d) risk-i.e., clinical indicators of suicidal attempts, selfharm or aggressive acts. Each one of the 34 CORE-OM items is scored along a Likert type scale ranging from 0 (never) to 4 (always or almost always).

The levels of symptomatic seriousness of the clients included in this study, as Corbella (2003) demonstrated, were equivalent to the ones in other CORE-OM validation studies (see Barkham et al., 2001; Evans et a., 2002).

ii. Therapeutic alliance. To assess the therapeutic alliance we used the short version of the Working Alliance Inventory (WAI) by Horvath & Greenberg (1986). The WAI, as a global measure of the working alliance, is made up of 3 subscales reflecting the three components of the therapeutic alliance described by Bordin (1979):

(A) Agreement on therapeutic goals.

(B) Agreement on therapeutic tasks.

(C) Emotional bond between therapist and client.

Each subscale of the WAI is made up of 12 items that are scored along a Likert type scale. The short version of the WAI used in this study (WAI Short Version; WAI-S) was proposed by Tracey and Kokotovic (1989) after factor analyzing the WAI and making it shorter so as to facilitate its use in clinical settings (see Corbella & Botella, 2004). The WAI-S demonstrated an adequate factor structure and retained its three subscales, each of which is composed of 4 items. The WAI-S is thus made up of 12 items that the client scores along a Likert-type scale ranging from 1 to 7. Thus, the total scoring of the WAI-S ranges from 12 to 84. Tracey and Kokotovic (1989) also demonstrated that the internal consistency of the WAI-S was strong-Cronbach's α = .98

The Spanish version of the WAI-S was created and validated by Corbella and Botella (2004) by means of a sample of 102 non-psychotic patients receiving outpatient psychotherapy. Internal consistency was also high-Cronbach's α = .93. Regarding the factor structure of the inventory, a Principal Component Analysis with Varimax rotation revealed three factors with eigenvalues of 5.66, 2.81, and 2.59 respectively. The three components accounted for a 65% of the total variance. The first factor was made up of items belonging to all the subscales of the WAI-S, the second of some items belonging to the subscale "agreement on goals", and the third of items belonging to the subscale "emotional bond". In summary, the first two factors were made up of a mixture of the subscales "agreement on tasks", "agreement on goals", and one item of the subscale "emotional bond".

iii. Narrative assessment. To assess the selected dimensions of clients' generated narratives in psychotherapy discussed in the introductory section of this paper, we used the Narrative Assessment Grid (NA-Grid; Botella & Gámiz, 2011). The NA-Grid is devised as a series of rating scales that combine the narrative dimensions already mentioned, i.e.:

1. Narrative Structure and Coherence

1.1. General Orientation of the narrative (GO)

1.2. General Structural Sequence of the narrative (SS)

1.3. General Evaluative Commitment of the narrative (EC)

1.4. General Integration of the narrative (I)

1.5. General Coherence of the narrative (GC) (1.1+1.2+1.3+1.4)/4

2. Narrative Content and Multiplicity

2.1. Thematic Variety (TV)

2.2. Variety of Events (VE)

2.3. Variety of Scenarios (VS)

2.4. Variety of Characters (VC)

3. Narrative Process and Complexity

3.1. Objectifying (O)

3.2. Emotional Subjectifying (ES)

3.3. Cognitive Subjectifying (CS)

3.4. Metaphorizing (M)

4. Narrative Intelligibility

4.1 Clarity of the narrative valued ultimate goal (C)

4.2. Conflict among different goals (Co)

4.3. Reasonability of the narrative valued ultimate goal (R)

4.4. Relevance of the narrated events (Re)

4.5. General Structural Sequence of the narrative (same as 1.2) (SS')

4.6. Stability of the Characters traits (SC)

4.7. Intelligibility of the causal links (In)

4.8. Narrative Form (specify the combination of rudimentary forms) (NF)

4.8.1. Contamination narrative

5. Narrative Position of the Self (NPS)

Clients' generated narratives are scored in each one of these rating scales along a Likert type scale that ranges from 1 to 3 (1 = Low; 2 = Medium; 3 = High). Higher scorings correspond to narratives that fulfill the definition of each dimension discussed above. Thus, a narrative that includes detailed and specific information about characters, social context, time and place, and personal circumstances, would be rated as "High General Orientation". If the narrative leaves all or many of these details unanswered, it would be rated as "Low General Orientation".

Each narrative is scored by a minimum of 4 independent raters that have been trained in using the NA-Grid for a minimum of 30 hours. The interjudge reliability amongst the panel of 4 raters was good (Cronbach's alpha = 0.81).

The NA-Grid has been previously used, tested, and refined in two unpublished research studies; a pilot one comparing narrative dimensions in good and poor-out-come therapies, and another one comparing narrative dimensions between depressive and anxious clients.

Procedure

The procedure followed in this study was the usual one according to the treatment protocol of the therapy service:

After the initial phone call of the client, an intake interview was conducted.

The intake interview followed the structure of the CORE Therapy Assessment Form. In addition, during the intake interview the client was asked to fill the CORE Outcome Measure already discussed.

The therapeutic team met so as to make the necessary decisions about therapy with the particular client.

Once therapy had begun, outcome and process were monitored by periodically asking the client to fill the CORE Outcome Measure (after each of the first four sessions and after every four subsequent sessions) and the Working Alliance Inventory (after each of the first four sessions and after every four subsequent sessions). The reason to assess outcome and process at these intervals is derived from the studies about the pace of therapeutic change (Howard, Kopta, Krause, & Orlinsky, 1986; see Corbella & Botella, 2004) indicating that 30% to 40% of patients experience significant changes between the first and the third session, and that between the fourth and seventh session this percentage increases to a 50-60%.

Once the therapeutic goals were attained, and provided therapist and client agreed, the end of therapy was scheduled. Again, as a means to monitor the therapeutic outcome and process the client was asked to complete the symptomatic and alliance measures.

As we already discussed, client generated narratives were defined as written accounts of the client's experiences and meaning-making processes that serve relevant therapeutic functions and that are generated in the course of therapy and as a part of it; examples include diaries, personal journals, letters, and other forms of reflective writing (Neimeyer, 1995).

Once the therapy was finished and included in the study, three client generated narratives were selected for each participant: one belonging to the beginning of his/her therapy (Narrative #1), one belonging to the midpoint of treatment (Narrative #2), and the third one belonging to the final stage of therapy (Narrative #3). Thus, if a particular therapy consisted on n sessions, Narrative 1 belonged to a sessions 1 to 3; Narrative 2 belonged to sessions (n/2)-1 to (n/2)+1; and Narrative 3 belonged to sessions n-3 to n.

These three client generated narratives were then scored by a minimum of 4 independent raters, as discussed in the previous section of this paper and according to the scoring criteria already specified.

An analysis of the therapeutic outcome and process for each participant allowed us to assign them either to the GTO group or to the PTO group (according to the criteria discussed in the "Participants" section).

 

Results

Clients Variables were divided in three groups:

1. Intake Variables (i.e. therapeutically relevant variables assessed during the intake session: i. main presenting complaint, ii. problem's onset, iii. psychosocial functioning, iv. motivation, and v. pretherapy symptoms).

2. Outcome and Process Variables (symptoms and working alliance).

3. Narrative Variables.

1. Differences between good and poor-outcome cases in Intake Variables

A chi square test of independence was performed to examine the relation between good- and poor-outcome and the first Intake Variable assessed (i.e., main presenting complaint). The relation between these variables was not significant, X² (6, N = 60) = 1.4, p = .97. Clients in our sample were equally likely to have a good or poor therapeutic outcome regardless of their main presenting complaint.

Table 1 presents the results of Student's t-test for the four continuous Intake Variables tested remaining (ii. problem's onset, iii. psychosocial functioning, iv. motivation, and v. pretherapy symptoms).

 

 

Our results indicated that the difference between the mean scoring for problem's onset in the GTO group and in the PTO one were not statistically significant.

Our results also indicated that psychosocial functioning in the GTO group (M = 49.17, SD = 13.71) was significantly lower than in the PTO one (M = 57.10, SD = 14.06)-notice that the mean score for the PTO group is actually higher, this is so because the CORE Outcome Questionnaire scoring scales assume that higher scorings indicate higher problematic levels and, thus, the "psychosocial functioning" scale is rather a scale of "problems in psychosocial functioning". Also, the mean scoring for motivation in the GTO group (M= 95.67, SD = 7.74) was significantly higher than in the PTO one (M = 74.97, SD = 17.45). Finally, the mean scoring for pretherapy symptoms in the GTO group (M = 1.43, SD = .54) was significantly lower than in the PTO one (M = 1.13, SD = .53).

2. Relations between Intake Variables

An analysis using Pearson's correlation coefficient (see Table 2) indicated that there were no significant correlations between any of the Intake Variables assessed.

 

 

3. Differences between good- and poor-outcome cases in Outcome and Process variables

Table 3 presents the result of Student's t-test for the following Outcome and Process Variables:

 

 

• Total number of therapy sessions,

• Averaged Therapeutic Alliance,

• Total Pre-post symptom reduction,

• Working Alliance from sessions 1, 2, 3, 4 and 8 (from then on the sample size was too small to compute),

• Symptomatic level from sessions 1, 2, 3, 4 and 8 (from then on the sample size was too small to compute).

Our results indicated that the difference between the mean scoring for working alliance in sessions 1 and 2, and symptomatic level in all sessions included in the study (except at intake) in the GTO group and in the PTO one were not statistically significant.

Our results also indicated that the mean scoring for total number of therapy sessions in the GTO group (M = 13.90, SD = 5.49) was significantly lower than in the PTO one (M = 21.65, SD = 11.19). Also, the mean scoring for averaged therapeutic alliance in the GTO group (M = 76.09, SD = 1.27) was significantly higher than in the PTO one (M = 71.47, SD = 3.63). The mean scoring for total pre-post symptom reduction in the GTO group (M = .63, SD = .27) was significantly higher than in the PTO one (M = -.07, SD = .17). The mean scoring for total working alliance (Session 3) in the GTO group (M = 74.10, SD = 1.81) was significantly higher than in the PTO one (M = 71.93, SD = 3.72). The mean scoring for total working alliance (Session 4) in the GTO group (M = 75.48, SD = 2.31) was significantly higher than in the PTO one (M = 70.90, SD = 3.03). Finally, the mean scoring for total working alliance (Session 8) in the GTO group (M = 76.88, SD = 1.83) was significantly higher than in the PTO one (M = 71.39, SD = 3.26).

4. Factor analysis of Intake, Outcome, and Process variables

A principal component analysis was performed on the correlation matrix of Intake, Outcome, and Process variables (varimax with Kaiser Normalization was the rotation method; rotation converged in three rotations).

The principal component analysis produced two factors with eigenvalues greater than 1. These two factors explain 52.18% of the variance in the correlation matrix and, as discussed before, varimax rotation was performed on them. The rotated factor patterns are presented in Table 4.

 

 

Component 1 (32.81% of variance) was mainly loaded by total number of therapy sessions (with a negative factor loading); motivation; total pre-post symptom reduction; and averaged therapeutic alliance.

Component 2 (19.37% of variance) was mainly loaded by total number of therapy sessions; psychosocial functioning (with a negative factor loading); problem's onset; pretherapy symptoms; total pre-post symptom reduction (with a lower factor loading than in Component 1); and averaged therapeutic alliance (also with a lower factor loading than in Component 1).

5. Differences between good- and poor-outcome cases in Narrative variables

Table 5 presents the result of Student's t-test for each narrative variable in each of the three assessment moments (Narrative #1, Narrative #2, and Narrative #3) comparing GTO and PTO.

 

 

Our results indicate that:

(a) At the beginning of their therapeutic process (Narrative #1) GTO clients were significantly different from PTO ones in all narrative variables (see Table 5) except Thematic Variety, Objectifying, Stability of the Characters traits, Narrative Form, and Narrative Position of the Self.

(b) At the midpoint of their therapeutic process (Narrative #2) GTO clients were significantly different from PTO ones in all narrative variables (see Table 5) except Objectifying and Narrative Form.

(c) At the final stage of their therapeutic process (Narrative #3) GTO clients were significantly different from PTO ones in all narrative variables (see Table 5) except General Evaluative Commitment of the narrative

Figure 2 summarizes and depicts graphically the averaged course of both groups therapy process in terms of their changes in all narrative variables.

 

 

Discussion

Neither presenting complaint, nor problem's onset predict therapy outcome. Psychosocial functioning, motivation, and pre-therapy symptoms discriminate between good and poor outcome. GTO cases are characterized by higher scorings in motivation and pre-therapy symptoms (as expected). Psychosocial functioning (assessed at intake), however, discriminates both groups in an unexpected direction: higher scorings in psychosocial functioning characterize PTO cases instead of GTO ones. This is not due to psychosocial functioning correlating negatively with motivation (as could be expected), because this is not the case.

Working alliance discriminates both groups after the second session (not before). Despite the fact that GTO cases are characterized by significantly higher scorings in pre-therapy symptoms at intake, this is not the case during the rest of the sessions included in this study. This may appear surprising, but a careful look at the data indicates the reason. The difference between both groups decrease after the first therapy session because the GTO group decrease their symptomatic level; however this is not the case with the PTO group, that does not decrease their symptomatic level-in fact, that was a condition for their being included in the PTO group.

As expected, the total number of therapy sessions was lower in the GTO group, while the averaged therapeutic alliance, total pre-post symptom reduction, and mean scoring for total working alliance in sessions 3, 4, and 8 were higher.

So far, our results indicate that the GTO group (vs. the PTO one) is made up of clients that at intake report any of the presenting complaints in our sample, and who have been experiencing it for no matter how long, but with a higher symptomatic level of distress. Their self-reported psychosocial functioning is lower; i.e., they report having more difficulties in coping with everyday life with their problem. Also, they report being more motivated towards therapeutic change.

During the course of therapy, they construe a stronger working alliance with their therapists already after the second session and maintain this stronger alliance on average and session-by-session during the whole of their therapeutic process. There symptomatic level of distress, despite being higher at intake, is no longer statistically different from the PTO group because they get better symptomatically already after the first session while the PTO group does not-i.e., the difference between both groups decrease asymmetrically. Accomplishing their goals takes them less sessions in total, and during these sessions they accomplish more.

This overall image is confirmed by the component analyses, which indicate the following. A first factor emerged that was composed by indicators of therapeutic success and GTO in general (i.e., few sessions, and high motivation, symptom reduction and therapeutic alliance throughout the whole process). This component can be called "Optimal Therapeutic Process" and reflects the results previously described as defining the GTO group. A second factor accounting for the data variance also emerged and was made up of indicators of poor therapeutic outcome, i.e., a high number of sessions, poor psychosocial functioning, early problem's onset, and pre-therapy symptoms. This component can be called "Complicated Therapeutic Process". Interestingly, total pre-post symptom reduction and averaged therapeutic alliance are also included in this component, although with a lower factor loading than in Component 1. This indicates the relevance of the therapeutic outcome and process irrespectively of optimal or complicated therapeutic process.

These results, coherent with a significant body of previous research but focused almost exclusively on a non-narrative view of therapy outcome and process, led us to a deeper exploration of what were the dimensions of client's self generated narratives that could potentially account for both outcome and process besides what we already know.

Good outcome clients enter therapy with a self-narrative significantly better than poor outcome clients in terms of structure, process, and content. However, some specific aspects of these narrative dimensions (Thematic Variety, Objectifying, Stability of the Characters traits, Narrative Form, and Narrative Position of the Self) do not discriminate both clients groups in the beginning. Other narrative variables (General Evaluative Commitment of the narrative, and Conflict among different goals) have lower scores in good outcome clients from the beginning and these scores remain lower as therapy advances to its midpoint. Conflict among different goals remains lower for good outcome clients during the whole therapeutic process, while General Evaluative Commitment of the narrative ceases being lower to become non-discriminative at the final stage of therapy.

As therapy advances, three of the five narrative variables that were not discriminative in the beginning become so (Thematic Variety, Stability of the Characters traits, and Narrative Position of the Self).

At the final stage of their therapeutic process, the narratives of good outcome clients become significantly different from the ones of poor outcome ones in all narrative variables except General Evaluative Commitment of the Narrative.

Thus, as discussed in the introductory section of this paper, narrative dimensions in the patients, discourse are relevant to their well being and, therefore, are affected by psychotherapeutic outcome and process. Again, as discussed by Angus (2012) our data demonstrates that changes in the narrative process are the key to clients change. Also coherently with the work of M. Gonçalves et al., 2011, 2012; Matos et al., 2009; Mendes et al., 2010, 2011; therapeutic change is correlated with changes not only in narrative process but also in its content.

More specifically, comparing the results of Moreira, Beutler, and Gonçalves' study (2008, pp. 1190-1191), our previous one (Botella & Gámiz, 2011) and the ones obtained in this study, it is confirmed again that changes in patients' narrative structural coherence are found throughout the therapeutic process and it seemed to be able they differentiate positive outcome cases from negative/poor outcome cases. We cannot affirm however, as Moreira, Beutler, and Gonçalves (2008) did that integration appeared to be the most discriminative subdimension between positive and negative/poor outcome cases, nor that content multiplicity was the dimension for which the highest level of change was obtained whereas the lower level of change obtained was for process complexity. Our study did not incorporate a mathematical method to calculate a single measure of discrimination. In fact, as already discussed, our data validate that there is not a single subdimension that surpasses all others, but a cluster of them showing sensitiveness to therapeutic process and outcome. Our results in this study confirm again that (as already found in Botella & Gámiz, 2011) two narrative dimensions not included in Moreira, Beutler, and Gonçalves' (2008) study (i.e., redemption sequences and narrative position of the self as a victim) distinguished poor outcome patients' narratives from good outcome ones.

These results can be used to suggest some implications for psychotherapy practice. Firstly, our study confirms once more that clients with good therapeutic outcome change the structure, process, and content of their narratives. An analysis of these changes can be useful to foster them from the beginning of the therapeutic process.

Even if clients' narratives change throughout the whole therapeutic process, they do not do so equally in all dimensions, and this suggests two points:

(1) Good outcome clients enter therapy with some "resilience" factors that make their narratives more prone to therapeutic change;

(2) Irrespectively of their initial state, all clients mobilize a series of processes that help them get better (be them learnt during the therapeutic process itself or activated by it);

(3) Since good outcome clients' narratives are already better from the beginning, the same level of narrative change leads to a higher point in them.

The vicissitudes of the diverging narrative dimensions along the therapeutic process, and specifically their differential patterns between good- and poor outcome clients suggests the following:

(1) Good outcome clients are capable of detaching themselves emotionally form their problem saturated narratives enough as to allow them a reflexive stance towards them but become emotionally involved in them again at the end of the therapeutic process once their narratives have been reconstrued through therapy.

(2) Good outcome clients' narratives are not as conflictive and dilemmatic as poor outcome ones during the whole of the therapeutic process.

(3) As therapy advances from the initial to the middle stages, the variety of topics that clients include in their narratives, the stability they attribute to their relevant others traits, and the level of reflexivity they contribute to their narratives becomes increasingly discriminative of good outcome. This suggests that good outcome clients become increasingly able to widen the focus of their narratives so as to encompass a broader array of relevant topics. Also that they become more reflexive and less informatively detached, and that they perceive more coherence and stability in their lives and significant others'.

The image that these results convey of successful therapy is one in which the good outcome client enters therapy with a self-narrative whose structure, process, and content make it more prone to therapeutic reconstruction. Because of this and of the change processes that therapy triggers, good outcome clients are increasingly capable of reconstructing their narratives through a reflexive stance towards them and becoming involved in them again once reconstrued. They are also increasingly capable of resolving their narratives inner dilemmas and conflicts, to include more topics in them, and to be more reflexive towards them. Such a combination of reconstructive processes leads to a more adaptive narrative as the therapeutic process advances.

As therapy reaches its final stages, all narrative dimensions (except the level of emotionally laden commitment) become significantly discriminant of good therapeutic outcome. The reason for emotional commitment to be higher again in good outcome clients has already been explained, and changes in the rest of dimensions suggest that clients' narratives are essential for assessing their change processes. Narratives are not a byproduct of one's life and actions; they are one's life and actions. In the words of Bertha Mook (1992): Life is lived while it is told, and it is told while it is lived.

 

References

1. Adler, J. M., Wagner, J. W., & McAdams, D. (2007). Personality and the coherence of psychotherapy narratives. Journal of Research in Personality, 41(6), 1179-1198. doi: http://dx.doi.org/10.1016/j.jrp.2007.02.006.         [ Links ]

2. Angus, L. (2012). An integrative understanding of narrative and emotion processes in Emotion-focused therapy of depression: Implications for theory, research and practice. Psychotherapy Research, 22(4), 367-380. doi: http://dx.doi.org/10.1080/10503307.2012.683988.         [ Links ]

3. Angus, L. E. & McLeod, J. (Eds.). (2004). The handbook of narrative and psychotherapy. Thousand Oaks, CA.: Sage.         [ Links ]

4. Barkham, M. & Mellor-Clark, J. (2000). Rigour and relevance: The role of practice-based evidence in the psychological therapies. In N. Rowland & S. Goss (Es.), Evidence-based Mental Health (pp. x-xx). London, UK: Routledge.         [ Links ]

5. Bordin, E. S. (1979). The generalization of the psychoanalytic concept of the working alliance. Psychotherapy: Theory, Research and Practice, 16, 252-260.         [ Links ]

6. Botella, L., Corbella, S., Belles, L., Pacheco, M., Gómez, A., Herrero, O., ... Pedro, N. (2008). Predictors of therapeutic outcome and process. Psychotherapy Research, 18(5), 535-542. doi: http://dx.doi.org/10.1080/10503300801982773.         [ Links ]

7. Botella, L. & Gámiz, M. (2011). Narrative assessment in psychotherapy: A constructivist approach. In P. Caputi, L. L. Viney, B. M. Walker, & N. Crittenden (Eds.), Personal Construct Methodology (pp. x-xx). New York: Wiley.         [ Links ]

8. Corbella, S. (2003). Compatibilidad entre el estilo personal del terapeuta y el perfil personal del paciente. Unpublished Doctoral Thesis, Universitat Ramon Llull, Barcelona, Spain.         [ Links ]

9. Corbella, S. & Botella, L. (2004). Investigación en psicoterapia: Proceso, resultado y factores comunes. Madrid, Spain: Vision Net.         [ Links ]

10. Dimaggio, G., Nicolo, G., Fiore, D., Centenero, E., Semerari, A., Carcione, A., & Pedone, R. (2008). States of minds in narcissistic personality disorder: three psychotherapies analyzed using the grid of problematic states. Psychotherapy Research, 18(4), 466-480. doi: http://dx.doi.org/10.1080/10503300701881877.         [ Links ]

11. Dimaggio, G. & Semerari, A. (2001). Psychopathological Narrative Forms. Journal of Constructivist Psychology, 14, 1-23. doi: http://dx.doi.org/10.1080/10720530125913.         [ Links ]

12. Evans, C., Connell, J., Barkham, M., Margison, F., Mellor-Clark, J., McGrath, G., & Audin, K. (2002). Towards a standardised brief outcome measure: Psychometric properties and utility of the CORE-OM. British Journal of Psychiatry, 180, 51-60. doi: http://dx.doi.org/10.1192/bjp.180.1.51.         [ Links ]

13. Gergen, K. J. (1994). Realities and relationships: Soundings in social construction.Cambridge, Mass.: Harvard University Press.         [ Links ]

14. Gonçalves, M., Mendes, I., Cruz, G., Ribeiro, A., Sousa, I., Angus' L., & Greenberg, L. (2012). Innovative moments and change in client-centered therapy. Psychotherapy Research, 22(4), 389-401.         [ Links ]

15. Gonçalves, M., Ribeiro, A., Mendes, I., Matos, M., & Santos, A. (2011). Tracking novelties in psychotherapy process research: The innovative moments coding system. Psychotherapy Research, 21,497-509.         [ Links ]

16. Gonçalves, O. (2000). Viver narrativamente: a psicoterapia como adjectivaçao da experiencia. Coimbra, Portugal: Quarteto.         [ Links ]

17. Gonçalves, O. & Henriques, M. (2000a). Manual de avaliaçao da estrutura e coerencia narrativa. Braga, Portugal: Universidade do Minho.         [ Links ]

18. Gonçalves, O. & Henriques, M. (2000b). Manual de avaliaçao do proceso e complexidade narrativa. Braga, Portgual: Universidade do Minho.         [ Links ]

19. Gonçalves, O. & Henriques, M. (2000c). Manual de avaliaçao do contéudo e multiplicidade narrativa. Braga, Portugal: Universidade do Minho.         [ Links ]

20. Gonçalves, O. & Henriques, M., (2000d). Método de avaliaçao de narrativas protótipo. Braga, Portugal: Universidade do Minho.         [ Links ]

21. Gonçalves, O. & Machado, P. (1999). Cognitive narrative psychotherapy: Research foundations. Journal of Clinical Psychology, 55, 1179-1191.         [ Links ]

22. Horvath, A. O. & Greenberg, L. S. (1986). The development of the working alliance inventory. In L. S. Greenberg & W. M. Pinsoff (Eds)., The psychotherapeutic process: A research handbook (pp. 529-556). New York: Guilford Press.         [ Links ]

23. Howard, K. I., Kopta, S. M., Krause, M. S., & Orlinsky, D. E. (1986). The dose-effect relationship in psychotherapy. American Psychologist, 41(2), 159-164. doi: http://dx.doi.org/10.1002/jclp.10167.         [ Links ]

24. Matos, M., Santos, A., Gonçalves, M., & Martins, C. (2009). Innovative moments and change in narrative therapy. Psychotherapy Research, 19, 68-80. doi: http://dx.doi.org/10.1080/10503300802430657.         [ Links ]

25. McAdams, D. (2001). The psychology of life stories. Review of General Psychology, 5(2), 100-122.         [ Links ]

26. McAdams, D. (2006). The problem of narrative coherence. Journal of Constructivist Psychology, 19(2), 109-125. doi: http://dx.doi.org/10.1080/10720530500508720.         [ Links ]

27. Mendes, I., Ribeiro, A., Angus, L., Greenberg, L., & Gonçalves, M. M. (2011). Narrative change in emotion-focused psychotherapy: A study on the evolution of reflection and protest innovative moments. Psychotherapy Research, 21, 304-315. doi: http://dx.doi.org/10.1080/10503307.2011.565489.         [ Links ]

28. Mendes, I., Ribeiro, A., Angus, L., Greenberg, L. S., Sousa, I., & Gonçalves, M. M. (2010). Narrative change in emotion-focused therapy: How is change constructed through the lens of the Innovative Moments Coding System? Psychotherapy Research, 20, 692-701.         [ Links ]

29. Mook, B. (1992). Intersubjetividad y estructura narrativa en la terapia familiar. Revista de Psicoterapia, 10-11, 13-22.         [ Links ]

30. Neimeyer, R. A. (1995). Constructivist psychotherapies: Features, foundations, and future directions. In R. A. Neimeyer & M. J. Mahoney (Eds.), Constructivism in psychotherapy (pp. 11-38). Washington: American Psychological Association.         [ Links ]

31. Neimeyer, R. A. (2006). Narrating the dialogical self: toward an expanded toolbox for the counselling psychologist. Counselling Psychology Quarterly, 19(1), 105-120. doi: http://dx.doi.org/10.1080/09515070600655205.         [ Links ]

32. Semerari, A., Carcione, A., Dimaggio, G., Falcone, M., Nicolo, G., Procacci, M., & Alleva, G. (2003). How to evaluate metacognitive functioning in psychotherapy? The metacognition assessment scale and its application. Clinical Psychology and Psychotherapy, 10, 238-261. doi: http://dx.doi.org/10.1002/cpp.362.         [ Links ]

33. Tracey, T. J. & Kokotovic, A. M. (1989). Factor structure of the Working Alliance Inventory. Psychological Assessment, 1, 207-210.         [ Links ]

 

 

Correspondence:
Luis Botella.
Facultat de Psicologia,
Ciencies de l'Educació i l'Esport (FPCEE) Blanquerna.
Universitat Ramon Llull (Barcelona), España.
Email: lluisbg@blanquerna.url.edu

Recibido: 02 de septiembre de 2015
Aceptado: 21 de octubre de 2015