Client Perceptions of the Most and Least Helpful Aspects of Couple Therapy

Abstract

Couples have a unique perspective to share about the therapy they receive. The current study uses a mixed-methods design to examine what couples report about most and least helpful elements of two behaviorally-based treatments tested in a large clinical trial of couple therapy. Results indicate that responses are highly variable and fall into five main themes, which are then compared between treatment conditions, genders, and outcome groups. One interesting finding is that all groups reported wanting more discussion of sexual issues. Findings are discussed in the context of common factors research, recent developments toward unified principles of change in couple therapy, and model-specific differences.

Couples often seek therapy as a last resort, reluctantly, or not at all (Doss et al., 2003), despite the availability of effective forms of treatment (Benson & Christensen, 2016; Carr, 2014; Lebow et al., 2012). There has been significant progress toward determining levels of empirical support for couple and family therapies (Sexton & LaFollette, 2016) and toward understanding the specific mechanisms contributing to treatment effectiveness (Knobloch-Fedders et al., 2015). Similarly, evidence-informed practice remains a worthy goal for the field, with research programs moving toward systematic dissemination of evidence-based approaches into widespread clinical settings (Monson et al., 2014). Despite these advances, many practitioners find the research literature too removed from clinical practice to be sufficiently informative in their daily work with clients (Sexton & Datchi, 2014). The current study bridges this gap by utilizing a mixed methods design to examine couples’ experiences in therapy to inform researchers and practitioners about client perspectives on the most and least helpful aspects of couple therapy. Because two distinct treatments are examined, the study also advances efforts toward identifying common factors, universal principles/processes, and model-specific mechanisms of change in couple therapy from the perspective of couples themselves.

Behavior-based Approaches to Couple Therapy

The treatments experienced by couples in this study are both behavior-based approaches with empirical support, delivered by trained and supervised therapists as part of a randomized clinical trial (Christensen et al., 2004). Traditional Behavioral Couple Therapy (TBCT; Jacobson & Margolin, 1979) is a change-based and skill-based approach with methods such as behavioral exchange, in which partners identify reinforcing behaviors they can each increase, communication training focused on listening and expressing skills, and problem-solving training focused on skills for collaborative definition and resolution of problems. Partners work with the therapist to learn, practice, and use new behaviors both in and out of session with one another to improve their relationship satisfaction. Integrative Behavioral Couple Therapy (IBCT; Christensen et al., 2014, 2020) integrates elements of TBCT with methods designed to foster emotional acceptance of previously unacceptable characteristics of one’s partner. It includes methods such as formulation and unified detachment to foster non-blaming dyadic understandings of problems, empathic joining to increase vulnerable expressions of emotion, and tolerance-building to reduce the intensity and impact of problems on the relationship. A core aspect of the IBCT approach is the non-blaming DEEP formulation of problems, in which the therapist understands and fosters a view of problems as stemming from normal Differences between partners instead of underlying deficiencies, understandable Emotional sensitivities to those differences based on historical and cultural influences, External stressors impacting one or both partners, and the Pattern of interaction they engage in when painful differences arise. This formulation may be particularly effective with intercultural couples (Kalai & Eldridge, 2021).

Despite similarities in long-term outcome between IBCT and TBCT (Christensen et al., 2010) and overlap in some aspects of treatment (e.g., dyadic focus, use of behavioral functional analysis, support for changing behaviors), the two treatments also differ substantially. For example, TBCT represents a traditional first-wave behavioral treatment that emphasizes behavior change and skill-building while IBCT represents a third-wave approach that integrates traditional behavior change methods with emotional and cognitive shifts toward acceptance of problems and aversive partner behaviors. In fact, observational assessments of treatment sessions showed that IBCT therapists engaged in three times more acceptance strategies than TBCT therapists, who engaged in three times more change strategies than IBCT therapists (Christensen et al., 2004). In addition, the two treatments created different kinds of changes in reports of couples’ out of session behavior. Although both treatments produced both behavioral changes and acceptance, couples in TBCT reported more changes in behavior than couples in IBCT, while couples in IBCT reported more increases in acceptance than couples in TBCT (Doss et al., 2005).

Common Factors and Principles of Change

Because these two treatments share some similarities and yet have distinct emphases, studying them together is a natural opportunity to examine common factors or principles of change in couples therapy as well as distinct mechanisms and processes from the viewpoints of couples themselves. A variety of perspectives have been proposed for understanding the common elements across approaches to couple therapy, such as a common factors approach (Blow & Sprenkle, 2001; Sprenkle et al., 1999), a moderate common factors approach (Davis et al., 2012; Sprenkle & Blow, 2004), and common principles of change (Benson et al., 2012). The common factors approaches have tended to emphasize the importance of therapist qualities, therapeutic alliance, and client hope/expectancy over specific treatment methods, citing the lack of significant differences in efficacy between treatments (Wampold & Imel, 2015). The common principles approach tends to emphasize the promise of identifying overlapping principles of change across treatment approaches (Christensen, 2010), so the focus is more on the changes therapists help couples make in treatment than the qualities of the therapist, couple, or alliance.

Couple Perceptions of Helpful Factors

Prior studies have asked couples for their perceptions of the helpful aspects of therapy. Some of these were conducted on specific forms of treatment, such as social constructionist, narrative, feminist, and solution-focused therapies (Bowman & Fine, 2000), or integrative systemic and emotionally focused therapies (Goldman & Greenberg, 1992; Greenberg et al., 1988). Other studies examined general, eclectic, or systemic approaches to couple therapy (Alexander, 1997; Helmeke & Sprenkle, 2000; Olson, 2002; Wark, 1994). These studies vary in methodology, although all used interviews conducted during or after treatment with a relatively small number of couples (3–42 couples), and all couples (with the exception of Alexander, 1997) were seen in a training clinic by pre-licensed therapists. Only one study was conducted in the context of a clinical trial (O’Leary & Rathus, 1993), and this study was focused on perceptions of 31 individual women who received either individual or couple therapy for depression. Some researchers conceived of the change process as gradual (Christensen et al. 1998), while others assessed perceptions of critical incidents or pivotal moments (Greenberg et al., 1988; Helmeke & Sprenkle, 2000; Wark, 1994). Among the latter, there tended to be low concurrence in reports of critical incidents or pivotal moments when both therapists and couples were queried, indicating that couples’ perceptions offer unique and valuable information.

Themes in couples’ responses include reports about the helpfulness of trust in the therapist, safety in session structure like ground rules, client agency and choice, therapist refocusing on core issues, having a neutral and dedicated time devoted to focusing on the relationship, and the equal treatment of partners (Bowman & Fine, 2000). Other couples report increased trust, safety, support, ability to discuss thoughts and feelings about the relationship, greater awareness of their own and their partner’s thoughts and feelings (Goldman & Greenberg, 1992), conflict management, improved communication, a coherent understanding of the underlying conflicts and causes of their problems, and the therapist’s ability to refocus the tasks or goals of therapy sessions (Alexander, 1997). In addition, couples report the following themes in the change processes during critical incidents: expression of underlying feelings that leads to interpersonal change, expressing feelings and needs, acquiring understanding, taking responsibility for experience, receiving validation from the therapist (Greenberg et al., 1988), positive results, the routine provided by structure, alternative perspectives offered by the therapist, non-directive style of the therapist, directive therapist, and focus on positives by the therapist (Wark, 1994).

One study made a distinction between facilitators of change that occur in and out of session (Olson, 2002), noting that couples report in-session facilitators as directives given by the therapist for between-session activities such as homework, the therapist acting as a mediator and facilitator of sessions, and the therapist creating space for the clients to see things that could make a change in the couple’s life. Facilitators of change that occur both in and out of session were reported and categorized as shifts in affect (such as less anger and defensiveness), behavior (such as development of communication skills and learning new ways of approaching one another), and cognition (such as recognizing relationship patterns and one’s own role in maintaining the pattern). These three clusters of change in affect, cognition, and communication were also reported in Christensen et al. (1998), along with five preconditions for change: safety, fairness, normalization, hope, and pacing. When treatment conditions were compared, some differences between treatments emerged, such as emphasis on emotional expression, awareness of partner’s sensitivities and vulnerabilities, and therapist empathy and caring in EFT, while couples in IST emphasized the treatment team’s expertise, presence, neutrality, and positive connotations (Goldman & Greenberg, 1992).

Couple Perceptions of Unhelpful Factors

Recent reviews of couple therapy (Carr, 2014; Lebow et al., 2012; Snyder & Halford, 2012) note that a sizable percentage (20–30%) of couples do not show significant posttreatment improvement, or decline at follow up. These findings remind us that it remains just as important to discover what doesn’t work well. Very few studies have explored what is unhelpful by directly asking couples. Bowman and Fine (2000) asked couples what was unhelpful, and participant responses highlighted the unequal treatment of partners, the therapist talking too much, the use of the word “therapy,” and the constraints of the 1-hour session. Similarly, Alexander (1997) summarized perceptions of couples who described their therapy as unsuccessful. These couples reported that the following elements were missed or lacking: conflict management and improved communication, a coherent understanding of the underlying conflicts and causes of their problems, and the therapist’s ability to refocus the tasks or goals of sessions. Finally, Wark (1994) asked couples about hindering events in therapy, and their responses fell into three categories: lack of therapist follow-through on assignments, therapist imposing misplaced clinical judgment, and lack of problem resolution.

Current Study

Couples’ written responses to an inquiry about the most and least helpful aspects of therapy were analyzed qualitatively and quantitatively. To the authors’ knowledge, this is the first study on this topic to examine couples’ perceptions using mixed methods. It is the largest sample size among studies of this topic, allowing both rich descriptions and statistical comparisons between groups in the study. Each method is used to make better sense of the other, and the combination provides a clearer picture of couples’ perceptions. This study examined whether variables such as treatment condition (IBCT, TBCT), gender (wife, husband), and treatment outcome (recovered, deteriorated) were associated with client reports of the most and least helpful aspects of therapy. The following research questions were proposed:

Qualitative Research Questions:

1. What themes emerge from clients’ responses to a question about the most helpful things about therapy?

2. What themes emerge from clients’ responses to a question about the least helpful things about therapy?

Quantitative Research Questions:

• 3. Do partners in IBCT and TBCT treatments differ significantly in their reports of the most and least helpful aspects of therapy?

• 4. Do husbands and wives differ significantly in their reports of the most and least helpful aspects of therapy?

• 5. Do husbands and wives within IBCT and within TBCT differ significantly in their reports of the most and least helpful aspects of therapy?

• 6. Do partners who show clinically significant deterioration at 2-year follow up differ significantly from partners who show clinically significant recovery at 2-year follow up in their reports of the most and least helpful aspects of therapy?

Materials and Methods

Participants

Participants included 134 heterosexual married couples seeking therapy in the context of a clinical trial of two forms of behavioral couple therapy (Christensen et al. 2004, 2006, 2010). Wives reported a mean age of 41.62 years (SD = 8.59) and husbands reported a mean age of 43.49 years (SD = 8.74). Couples had been married for an average of 10.00 (SD = 7.60) years, and about half had children (68 of 134, 50.7%). Mean years of education was 16.97 (SD = 3.23) for wives and 17.03 (SD = 3.17) for husbands, including kindergarten. Participants were Caucasian (husbands: 79.1%, wives: 76.1%), African American (husbands: 6.7%, wives: 8.2%), Asian or Pacific Islander (husbands: 6.0%, wives: 4.5%), Latinx (husbands: 5.2%, wives: 5.2%), and Native American or Alaskan Native (husbands: 0.7%). Almost half of all couples disclosed that they had attended marital therapy together in the past.

Participants were required to be over 18 years of age, fluent in English, legally married, cohabiting, and have a high school diploma or equivalent. They reported clinically significant levels of marital distress at three time-points prior to starting therapy, suggesting the distress was chronic. Participants could not have current diagnoses of alcohol or drug dependence, bipolar disorder, schizophrenia, or selected personality disorders (antisocial, borderline, and schizotypal), and could not be participating in additional therapy. Participants could be taking psychotropic medication if they were on a stabilized dose and did not anticipate changes in dose for the duration of the study. Information about relationship violence from the wives was used to exclude couples in which husbands had reportedly engaged in dangerous levels of violence.

Therapists were licensed clinical psychologists practicing in the local communities (Los Angeles or Seattle). They had between 7 and 15 years of experience post-licensure. They received training in both forms of therapy via treatment manuals and attendance at workshops. Therapists were provided with supervision from experts in each approach, consisting of weekly audio- and/or videotape reviews of sessions, with feedback provided prior to their next session.

Researchers conducting the study were trained in both treatment approaches, and the third author developed IBCT. Despite knowing that IBCT was designed to improve upon the long-term efficacy of TBCT, and the context of this study being a clinical trial comparing the two treatments, researchers assumed that couple perceptions may be entirely different from those of the researchers. Therefore, researchers maintained an open curiosity about couples’ perceptions instead of having a priori expectations or hypotheses.

Procedures and Measures

Each couple was screened for inclusion and exclusion criteria in a three-stage process including a telephone interview, questionnaires, and one in-person pre-treatment assessment. Eligible couples were randomly assigned to one of two treatment conditions. Sixty-eight couples received TBCT, 66 couples received IBCT, and therapy was free. While a maximum of 26 sessions was offered to each couple, an average of 22.9 (SD = 5.35) sessions occurred over an average of 36 weeks. One hundred twenty-six of the 134 participants were considered “treatment completers,” having attended over 10 sessions.

Couples were paid to complete routine assessments. Specific to this study, a Client Evaluation of Services (CES) questionnaire was administered to the couples immediately following the last session. The CES was developed based on a Client Satisfaction Questionnaire (Nguyen et al., 1983) to efficiently measure satisfaction of service across eight items on 4-point Likert scales that each partner completed. The data for this study come from an additional open-ended question added to the end of the CES, “What were the most helpful and least helpful things about the therapy?” Couples were instructed to complete the questionnaire independently at home and were provided materials and postage to mail it back to the research team. To maximize honesty couples were also informed that their therapists would not have access to their responses. All study methods were reviewed and approved by the Institutional Review Boards (IRBs) at the participating universities.

Mixed Methods Design

This study follows Creswell’s (2003) description of mixed methods data transformation utilizing a sequential exploratory strategy, in which the researcher quantifies the qualitative data first, followed by quantitative data analysis. Codes and themes are qualitatively extracted from the data, then counted for the number of times they occur in the data, which then allows the researcher to produce quantitative results with the qualitative data.

Qualitative Methods.

The qualitative component involved a content analysis of written responses, by extracting information in a systematic and replicable manner (Smith, 2000). The content analysis procedures included determining and specifying units of analysis, and determining coding categories based on patterns and evidence in the data (Flick, 2006). All qualitative coding and data analyses were done on Atlas.ti.

Prior to coding, a “start list” (Miles & Huberman, p. 58, 1994) was developed by the second author in consultation with the first author, which included descriptive categories that considered the conceptual framework, research questions, hypotheses, knowledge of the two forms of therapy, clinical experience, previous research findings, and understanding of the therapeutic alliance, therapist factors, and client factors in couple therapy. As recommended by Miles and Huberman (1994), categories within the start list remained flexible to allow revision of categories as the responses were evaluated.

To prepare the data for coding, information regarding treatment condition, gender, and treatment outcome was removed from responses to minimize unintended researcher bias. Distinct units of information in each response were then identified to be coded (i.e., the phrases, sentences, etc. that represent distinct thought units, each to be coded separately within the response; Smith, 2000).

Using content analysis methods, the second author extracted and sorted key words, statements, and phrases into broad categories (Smith, 2000). Responses were coded in as many ways as possible through a line by line analysis (Miles & Huberman, 1994). In addition, thematic categories that emerged from the responses that were descriptive, as opposed to interpretative, were identified in order to most closely follow each individual’s words and meanings (Miles & Huberman, 1994). The initial coding process also involved the constant comparative method, constantly taking new information from data collection and comparing it to the emerging categories in order to establish and refine the categories (Orcher, 2005). This entailed a code-revise-code-revise process, in which several new codes were added to the list as themes emerged, and others were eliminated from the list when they were not represented in the data. In addition, codes began with very detailed descriptions then became more global categories over time after the entire response set had been reviewed at least three separate times. Thematic categories of meaning that were distinct ideas remained separate, whereas multiple categories that represented one meaning set were clarified and combined, thereby minimizing overlap between categories. Ultimately, after five reviews and regular consultations with the first author, a detailed coding system of the most helpful and least helpful aspects of couple therapy emerged.

Using this coding system, the second author coded responses one more time. Each response was coded as a whole, and could have multiple codes. However, any given code was only assigned to a response once, even if several parts or sentences in the response referred to that code. Also, one sentence could receive more than one code if multiple categories were included in it. The first and second authors, joined by an advanced graduate student in clinical psychology, examined the unclear statements (n=22, 10% of responses), interpreted their meaning, and assigned the appropriate code after it was agreed upon by the group. Following this last review, a final list of codes and a frequency count for each code was generated.

Credibility, trustworthiness, and transferability were addressed using methods similar to those used by Davis and Piercy (2007a). Steps to ensure credibility and trustworthiness included the use of rich, thick description, the presentation of negative or discrepant information, the discussion of researcher bias, and the use of the constant comparative method of data analysis (Creswell, 2003). Triangulation was also addressed by analyzing and cross-checking a variety of data from multiple perspectives in order to assign codes to difficult or unclear responses. Finally, steps to ensure transferability included reporting unique client characteristics and the possible resulting effects on the data.

Quantitative Methods

Reliability of the coding system used in the qualitative portion of the study was then examined by recruiting and training four master’s level psychology students to use the coding system. About 1/6 of the 210 responses were randomly selected for training purposes in weekly coding meetings. Reliability was regularly calculated throughout the coding process using the formula suggested by Miles and Huberman (1994) for content analysis research. Each coder independently coded the remaining responses for the occurrence of 28 potential codes (14 most helpful and 14 least helpful), which fall under the five larger categories of therapist, client, therapy interventions and process, outcome, and logistical factors. As Table 1 shows, the inter-rater reliability was above .80 for all most and least helpful factors across all coders, except for most helpful client factors, which was slightly below at .79, giving evidence of the dependability of the coding system. Of note, there were no changes to the coding system and no new codes emerged during coding, suggesting that the coding system closely represented the data.

Table 1.

Inter-rater Reliability among the Five Domains

Response domains	Most helpful	Least helpful
Therapy factors	0.92	0.92
Therapist factors	0.93	0.87
Logistical factors	0.90	0.90
Outcome factors	0.95	0.92
Client factors	0.79	0.88

After the content analysis was complete, frequencies of codes were examined for any statistical differences that existed between groups. The independent variables were treatment group (IBCT and TBCT), gender (husbands and wives), and clinical significance of couples’ outcome using Dyadic Adjustment Scale (Spanier, 1976) scores at 2-year follow-up, including 29 deteriorated (14 IBCT and 15 TBCT) and 52 recovered (30 IBCT and 22 TBCT) couples. Chi-square tests examined differences between the treatment (TBCT and IBCT) and outcome (recovered and deteriorated) groups. Due to the categorical and dependent nature of couples’ data, McNemar’s test (McNemar, 1947) examined differences between genders.

Results

A total of 210 individual responses emerged, ranging in length from one to 336 words, varying from list format to highly descriptive sentences. Some participants responded that there were only most helpful aspects of therapy, while others focused on the least helpful aspects.

Themes in Responses

Five themes (presented in order from highest to lowest frequency) represented all responses about what was most and least helpful: Therapy factors, therapist factors, therapy outcome factors, client factors, and logistic factors. Overall, the number of themes assigned to a single response ranged from one to six, out of ten possible most and least helpful codes. Notably, each theme was found to be both most and least helpful by different individuals, so that one person may have described the therapist as most helpful whereas another person described the therapist as least helpful. Also, some individuals described aspects within one theme that were both most and least helpful, such as describing therapy factors as most and least helpful.

What Clients Report as Most Helpful in Couple Therapy

Below is a summary description of each theme in order of highest to lowest frequency of responses, with verbatim examples. Table 2 contains frequencies of responses. Statistical group comparisons follow the description of themes.

Table 2.

Frequency of Responses: What Was Most Helpful About the Therapy

Theme	Total Frequency (n= 210 responses)	IBCT (n = 136)	TBCT (n = 132)	Husbands (n = 134)	Wives (n = 134)	Recovered (n = 104)	Deteriorated (n = 58)
Therapist	160	52	106	75	85	78	22
Qualities	82	26	57	34	49	46	8
Behavior	78	26	49	41	36	32	14
Therapy	152	102	52	73	78	72	28
Interventions	133	95	40	62	71	63	23
Process	19	7	12	11	7	9	5
Client	39	10	29	18	21	21	7
Self	31	9	22	14	17	15	7
Partner	2	0	2	0	2	1	0
Couple	6	1	5	4	2	5	0
Outcomes	37	6	21	15	21	15	4
Logistics	21	6	15	11	10	12	2
Amount of time	10	3	7	8	2	6	2
Getting to therapy	2	0	2	0	2	0	0
Research project details	9	3	6	3	6	6	0

Therapist Factors

Therapist Qualities.

This includes qualities such as therapist caring (“[The therapist] is an excellent therapist, as well as a caring human being.”), understanding (“[The therapist] was most helpful. He was patient, understanding and neutral.”), sense of humor (“Also, the kindness, empathy and humor of the therapist made it easier to be open and honest.”), and cultural sensitivity (“The therapist was culturally sensitive which was an extremely important component for the success of the program.”). Eight respondents simply stated that “Our therapist was the most helpful” or some variation of this, directly implicating the therapist as the most helpful aspect of the therapy.

Therapist Behaviors.

This includes behaviors such as the therapist giving feedback (“Most helpful was counselor feedback.”), listening to the couple (“She listened to both of us, allowing us to speak both through her and directly to each other.”), identifying themes or patterns in the couple’s behavior (“Therapist’s insights about themes in our relationship were helpful.”), understanding and presenting viewpoints (“I felt that she worked to thoroughly understand us and worked to present other viewpoints.”), and reducing criticism (“He didn’t let us get away with spending sessions just criticizing each other.”).

Therapy Factors

Interventions.

This includes references to specific methods (“The actual communication skills and suggestions are the most helpful.” “Learning to discuss a problem and staying level headed; learning how not to fight at every disagreement.” “I like the “positives” list doing something nice for your spouse.” “The book is quite helpful; I realize others have similar problems and there are ways to cope, and deal with them.”).

Processes.

This includes processes that occurred in the therapy session or one’s overall experience of the therapy such as safety or neutrality that was not a technique or assignment (“Providing a place and time where we felt safe and could take risks.” “The most helpful was being restricted to a process, and not really being allowed to just complain for an hour.”).

Client Factors

Self Factors.

This includes responses about oneself, such as thoughts, feelings, behaviors, homework compliance, commitment, and self-disclosure: “I felt very comfortable during our sessions.” “Therapy helped me most by being able to express how I truly feel inside to my spouse.”

Partner Factors.

This includes responses about partner behavior: “My spouse did not really want to come to counseling and really came to enjoy [the therapist] and open up.”

Couple Factors.

This includes responses such as joint motivation, disclosures, or uncovering misconceptions: “Getting to know each other’s thoughts and expressions about certain things.” “Both of us do a huge amount of reading, but actually doing the work and not shortcutting them to say “oh yeah, that’s how it works” was very helpful.”

Outcome Factors

This includes responses regarding something that the couple achieved in therapy, such as an improvement or increase deemed positive, or the ability to use techniques learned in session, outside of session, or after therapy had ended. Clients frequently reported outcomes of gaining tools to solve their own problems at home, improved problem solving, and improved communication: “This therapy helped us to communicate better.” Outcomes of increased acceptance and understanding were also reported.

Logistics

This includes responses such as amount of time and scheduling flexibility: “Regular sessions over a reasonable duration.” “Getting together to talk one time weekly.” A few responses mentioned research project details as most helpful: “The questionnaires helped me to be clear about my feelings about the marriage.” Time factors often included the regularity and duration of sessions. Some clients reported office location and parking as most helpful.

Statistical Group Comparisons on Most Helpful Responses

The first aim of the quantitative analyses was to examine TBCT and IBCT group differences in their reports of the most helpful aspects of therapy. Chi-squares showed significant differences between treatment groups on the following most helpful response frequencies: Therapy (χ² = 8.35; df = 1; p = .004), Therapist (χ² = 8.37; df = 1; p = .004), and Client (χ² = 6.73; df = 1; p = .010). Specifically, partners in TBCT noted therapist and client factors more than partners in IBCT, while IBCT partners noted therapy factors more than partners in TBCT.

McNemar’s tests using binomial distribution (McNemar, 1947) were performed to determine differences between genders and between genders within each treatment group. Results did not show significant differences between husbands and wives. Within treatment groups, there were no differences between husbands and wives in TBCT, and one significant difference between husbands and wives within IBCT. Among IBCT wives there were nine responses identifying Client factors as the most helpful aspect of therapy, while there was only one Client response from IBCT husbands (p = .008).

Finally, chi-squares revealed no significant differences between the two outcome groups (recovered and deteriorated at 2-year follow up), although it is important to note that frequencies among deteriorated partners were below 5 for the Outcomes and Logistics responses.

What Clients Report as Least Helpful in Couple Therapy

Below is a summary description of each theme that emerged in responses about what was least helpful, in order of highest to lowest frequency of responses. Table 3 contains frequency counts of these responses.

Table 3.

Frequency of Responses: What Was Least Helpful About the Therapy

Theme	Total Frequency (n = 210 responses)	IBCT (n = 136)	TBCT (n = 132)	Husbands (n = 134)	Wives (n = 134)	Recovered (n = 104)	Deteriorated (n = 58)
Therapist	10	3	7	4	6	3	4
Qualities	1	0	1	0	1	1	0
Behavior	9	3	6	4	5	2	4
Therapy	45	20	21	20	28	18	8
Interventions	32	14	15	14	21	16	4
Process	13	6	6	6	7	2	4
Client	20	10	9	9	11	6	5
Self	6	2	3	1	4	3	1
Partner	6	3	3	4	3	0	2
Couple	8	5	3	4	4	3	2
Outcome	28	12	15	11	15	8	9
Logistics	64	31	34	31	33	29	9
Amount of  time	31	13	18	15	15	10	3
Getting to therapy	8	3	6	5	4	7	0
Research project details	25	15	10	11	14	12	6

Logistics

There were a greater number of logistics responses than any other least helpful domain. All groups often reported wanting more frequent sessions, longer sessions, and more time in general: “Truthfully, the least helpful aspect was the logistics. Due to no fault of [the therapist], it was difficult to arrive on time and have the benefit of full sessions. His office is a long distance from our home, and the meetings were during rush hour. This was frustrating to me. I would have liked more time.” Although logistics related to getting to the therapy session were reported by all comparison groups except for deteriorated partners, this was reported infrequently. More often logistics related to the research project were reported, especially by IBCT partners and recovered and deteriorated partners. Wives more frequently reported research project details such as the videorecording and questionnaires, whereas husbands more frequently reported the lack of individual sessions: “Some of the questionnaires are extremely repetitive in the types of questions asked and the quantity of questions is somewhat cumbersome.” “Being very emotional for more sessions might have brought more things to light but having the video camera there kept the lid on for me.”

Therapy Factors

Processes.

This includes responses such as: “Initially, first few sessions, lacked any structure. I was not sure where we were headed until after a few visits.” “Occasionally I felt we didn’t really get to the point and were discussing extraneous issues that weren’t really helpful.”

Interventions.

All groups except for deteriorated partners frequently reported reading assignments as least helpful (“The readings were the least helpful” “Written literature too wordy.”). All groups except for wives had reports about the problem-solving training as least helpful: “Problem solving strategies (been there, done that).” Deteriorated partners and wives were the only groups not to find some aspect of communication skills training to be least helpful.

Outcome Factors

This includes responses about something that the couple did not achieve in therapy, such as an issue that was not addressed, inability to find solutions to long-standing problems, or lack of understanding of the couples’ underlying conflicts and causes of their problems: “Did not get as much insight as I might have liked regarding understanding the causes of our problems.” All groups reported wanting more discussion of sexual issues: “Little or no focus on sex issues.” All groups reported lacking enough tools or exercises to use at home after the therapy ended. TBCT partners, husbands, and deteriorated partners more frequently reported a lack of individual focus in the therapy than did IBCT partners, wives, and recovered partners.

Client Factors

Self Factors.

This includes responses about one’s thoughts, feelings, behaviors, difficulty incorporating skills learned: “Some of the dialogue was not helpful, but mostly because I didn’t listen at times. I was not ready to.” “Attempting to incorporate skills at home environment.”

Partner Factors.

This includes responses such as lack of openness and motivation (“When my spouse cancelled or didn’t show up.”) or perceived traits or personal problems of the partner (“The fact that I’m married to a totally sexless, affectionless, loveless spouse.”).

Couple Factors.

This includes responses such as not completing homework assignments (“Some suggested exercises weren’t helpful because we didn’t do them!”) or arguing (“Sometimes we got into arguments, because of the issues that were raised during the sessions.”).

Partner factors that were least helpful were often a lack of motivation or openness in therapy, and couple factors were often a lack of completing homework assignments. Wives were more likely than husbands to report something about their own behavior that was least helpful. Husbands were more likely than wives to report something about their spouse’s behavior that was least helpful. All groups reported client factors as being least helpful, but most groups focused more on partner or couple factors than self factors.

Therapist Factors

The therapist was reported to be the least helpful aspect of therapy fewer times than any other domain across all comparison groups.

Therapist Behaviors.

This referred to things that the therapist failed to do sufficiently or at all, such as a lack of assistance, inability to refocus goals, and not treating partners equally: “Wish [the therapist] could not so much take sides, but be more assertive in recognizing mistakes made by myself and spouse.”

Therapist Qualities.

Only one response indicated therapist’s qualities were least helpful: “I feel [that the] therapist’s style was not as effective as other therapists I have worked with.”

Group Comparisons on Least Helpful Responses

The analyses for least helpful responses parallel those used for most helpful responses. Results did not show any significant differences between treatment groups, genders, or genders within each treatment group. However, chi-squares indicated significant differences between the two outcome groups on the following least helpful response frequencies: Therapist (χ² = 4.08; df = 1; p = .043) and Outcomes (χ² = 5.63; df = 1; p = .018), although the Therapist domain had frequencies under 5 for both outcome groups. Those who had unfavorable treatment outcomes at 2-year follow-up reported a higher percentage of least helpful outcome factors at post-treatment than those who had more favorable long-term outcomes. The mixed-method design of the study provides a richer understanding of this expected statistical difference. Qualitatively, the most frequently reported least helpful outcome factor was sexual issues not being addressed.

Discussion

The current study utilized mixed-methods process research to examine the common and model-specific mechanisms of change in couple therapy from clients’ perspectives on the most and least helpful aspects of two models of therapy. There are patterns in the data that support prior research and other patterns that are unique to this study.

Integration of Common Factors and Unified Principles of Change

Couples’ perceptions about the most and least helpful aspects of couple therapy fell into five domains: therapy, therapist, logistical, outcome, and client factors. These domains, found in the responses of both treatment groups and genders, can be compared to those in the common factors literature and principles of change approach. Specifically, Davis and Piercy’s (2007b) model-independent variables, including client variables, therapist variables, and therapeutic process factors, closely compare to the client, therapist, and therapy factors that emerged in the current study. Davis and Piercy (2007b) also include expectancy and motivational factors with the subcategories of faith in the referral source and fit of the model, the latter of which emerged in the current study as a subcategory. In contrast, the therapeutic alliance variable in Davis and Piercy’s (2007b) study did not emerge as a separate theme in the current study, and was instead incorporated into therapist factors. Blow et al.’s (2007) examination of the role of the therapist in common factors resulted in therapist variables including observable traits and states, and inferred traits and states that also resemble the two therapist subcategories of behaviors and qualities that emerged in the current study.

In comparing the five domains in the current study to the five proposed principles of change (Benson, et al. 2012), we notice that couples report some of the same principles, which were primarily categorized in the outcomes, therapy, and client domains in the current study. Clients’ reports confirm that the following three principles of change are viewed by them as particularly helpful: (a) altering the view of the problems to be more objective, dyadic, and contextualized, (b) eliciting emotion-based, avoided, private behavior, and (c) increasing constructive communication. A fourth principle, decreasing dysfunctional behavior, refers primarily to more severe dysfunction such as abuse, substance use, and impulse-control problems, while couples in this study did not tend to have these characteristics due to the inclusion and exclusion criteria. However, couples clearly appreciated and reported as helpful therapist attempts to maintain a safe, productive, and constructive atmosphere in the sessions instead of allowing dysfunctional behavior such as criticism. Regarding the fifth principle, emphasizing strengths and gains, clients did mention therapist focusing only on problems as a least helpful aspect of therapy, and affirmation of partners, evaluating strengths, and reviewing pleasant incidents over the past week as most helpful.

Model-Specific Findings and Differences

A notable finding emerging from this study was significant treatment group differences on three most helpful domains: therapy, therapist, and client factors. Specifically, partners in TBCT noted therapist and client factors more than partners in IBCT, while IBCT partners noted therapy factors more than partners in TBCT. This difference may be driven by the structured and directive nature of TBCT, which makes the therapist and client activity more prominent.

Of the three primary behavior-change treatment strategies, couples in TBCT found communication training the most helpful, followed by problem-solving training, and lastly behavioral exchange. This is not surprising, since when partners have limited or different levels of motivation to engage in the positive behavioral exchanges, particularly at the outset of treatment, this aspect of therapy is less effective. Also, the behavior exchange methods are designed to provide temporary improvement, while the skills training that follows is designed to provide more lasting benefits. Indeed, the Shadish and Baldwin (2005) meta-analysis on TBCT tested components of treatment and found that communication and problem-solving strategies (combined) were reliable predictors of effect size, while behavioral exchange was not, so couples may be noticing this impact as well. Out of these two methods, the preference for communication training over problem solving training is interesting. Often couples present with problems that are unresolvable, despite use of effective problem-solving skills. In these situations, use of good communication and listening skills about the feelings each person experiences around the unresolved problem can provide closeness and connection despite the chronic nature of the problem.

Couples in both treatment conditions frequently reported outcomes of improved communication, indicating that a change of communication very likely occurred in both forms of therapy. Prior research has identified the development of communication skills as an important in-session facilitator of change (Olson, 2002), and has found that increases in positive communication and increases in acceptance of partner problem behaviors are important mechanisms of change (Doss et al., 2005). It is notable that both mechanisms of change were reported by couples in this study, and that outcomes of improved communication that are behavioral and visible to others are reported more often by couples than outcomes of improved acceptance, an individual inner-experience.

While communication and problem-solving training are components of both treatments, in IBCT these methods are typically done in a more flexible and less structured manner, and are integrated into acceptance methods. Therefore, empathic joining and unified detachment methods do focus on communication and increase couples’ ability to openly communicate, and IBCT couples noted this focus on communication and problem-solving as helpful. Also notable is that IBCT partners reported a few instances of desiring “more behavioral modification” between them. Unfortunately, there is no further information on what those participants meant by behavioral modification. The safe assumption is that these partners were hoping to achieve individual, partner, and/or couple behavioral changes, and that this need was not met. Although behavioral change is a component of IBCT, it is often secondary to acceptance-promoting methods. Each form of change (acceptance and behavior change) is designed to facilitate the other, but this may not come about without intentional focus on both. This highlights the challenge in delivering the desired balance of behavior change and acceptance, unique to each individual and couple, and the importance of open communication and seeking feedback from couples throughout treatment about this balance. These findings also remind us of the importance of change-based methods within acceptance-based treatments for couples.

A few other qualitative differences between treatment approaches emerged. For example, some TBCT partners reported aspects of the therapy that were least helpful that are often found in IBCT treatment. These included difficulties learning to accept the partner’s character and faking arguments. Acceptance and faking arguments are methods used in IBCT, which raises curiosity about their use in TBCT, despite the high level of treatment fidelity and adherence in this clinical trial (Christensen et al., 2004). Also, the least helpful responses about acceptance in TBCT may reflect the difficulty effectively utilizing acceptance interventions in the context of a treatment focused on changing (TBCT) rather than accepting (IBCT).

Methodological Limitations and Contributions

Despite the large sample size, particularly in comparison to prior studies on this topic, some codes had low frequency counts, such as the least helpful therapist code and least helpful responses from deteriorated couples. While these low numbers are informative qualitatively, they prevented clear quantitative results in those analyses. The ethnic and educational diversity of the sample is limited, with about 20% of the sample identifying as ethnic minorities and most participants having college degrees. The therapy context included recording every session, free treatment offered by highly reputable universities, payment for completion of assessments, highly trained and closely supervised therapists, and a mean number of 23 sessions. While these characteristics of the study are beneficial as they represent a therapy experience that is unique compared to prior studies on this topic, it can be expected that participant and treatment characteristics influenced perceptions of therapy and willingness to share those perceptions. Future research should examine the most and least helpful aspects of couple therapy according to couples other than those represented in this sample, and with samples of more educational, racial, ethnic, socioeconomic, and relational diversity, including gay and lesbian couples, unmarried couples, and cohabitating couples.

It is important to note the limitations of a mixed-methods study that employs a content analysis followed by statistical tests. Beyond the challenging nature of the tasks for the researchers, it is particularly important to address issues of researcher bias (Creswell, 2003). As Davis and Piercy (2007b) note, a challenge the qualitative researcher faces is reducing the effects of his or her preferences on reporting data. In other words, the data should reflect what the couples say as closely as possible. It is hoped that the nature of responding to an open-ended question in writing minimized this effect. Additional steps were taken to minimize bias, such as de-identifying the gender, treatment condition, and therapist name of each response. The second author who initially developed the coding system met with a team of other individuals to collaboratively interpret unclear responses. Although bias may have existed in the interpretations and conclusions, authors highlighted examples of contrary findings to minimize the influence of this bias. The combination of authors, one who is a common factors proponent (author 2) and others who are model-specific experts (authors 1 and 3) allowed for an integration of these perspectives in the current study.

A final limitation of the study is the retrospective nature of couples’ reports, since they reported on a whole course of therapy as they completed treatment. Although there are also benefits to getting their perceptions of the treatment as a whole, future examinations could integrate an assessment strategy focused on key sessions as they occur, rating sessions immediately afterward on helpfulness then following up with further queries when sessions are rated particularly helpful or unhelpful.

Considering the shortage of in-depth literature on clients’ perceptions of helpful and unhelpful aspects in marital therapy, this data set is unique and contributes new information to therapists and researchers. It illustrates the importance of understanding the therapeutic process from each couple’s point of view and demonstrates that when asked, couples are forthcoming about their experiences and often have much to say. This study is also one of just a few to examine what is unhelpful about couple therapy from the couple’s point of view.

The coding system that evolved from the current study contributes to process research by offering a systematic conceptualization of the way clients view therapy. It is not meant to be the only categorization of most and least helpful elements of couple therapy; however, it is clearly distinct from other categorizations found in literature. First, the current study did not divide responses into model-dependent or model-independent factors. Rather, elements that were both model-dependent and model-independent were incorporated into the coding system. Second, logistics were a separate, distinct set of responses found in the current study not found in prior research. Third, whereas therapist factors were divided into categorizations of observable or inferred traits and states in other research (Blow et al., 2007), therapist factors were simplified into behaviors and qualities in the current study. The simplification allows therapists to examine themselves within two domains, and may aid in educating future mental health practitioners on the need for certain actions as well as certain attributes when conducting therapy. Fourth, the current study also paid greater attention to the depth and variety of outcome factors, beyond the “softening” and “making space for the other” subcategories found in Davis and Piercy’s (2007a) research. In fact, the current researchers did not further subcategorize the outcome factors, finding that this theme was better represented within its own domain.

Clinical, Training, and Research Implications

One of the patterns in the data with clear implications for therapists and trainees is the high frequency of responses referring to the therapist as most helpful, which supports the long history and recent advances on the importance of the therapeutic relationship (i.e., Friedlander et al, 2018; Garfield, 2004; Norcross & Lambert, 2018; Sprenkle & Blow, 2004; Sprenkle et al., 2007). Within this study, the frequency of responses referring to the therapist’s helpful qualities highlights the misconception of behavioral approaches as more mechanistic than humanistic. Couples appreciated the effective attention to the therapeutic relationship in these approaches.

Some gender nuances emerged regarding the therapeutic relationship, in which husbands were more likely to report what behaviors they saw from the therapist, and wives were more likely to report what they saw in therapist qualities. These qualitative differences alert us that couple therapists will benefit from understanding the unique needs of each partner when forming and maintaining the therapeutic alliance. Therapists will want to consider how they balance their use of empathetic qualities with more directive behaviors.

It should also be noted here that both genders reported the importance of cultural considerations in therapy, and therapist cultural competence was listed as a most helpful therapist quality, confirming that this is an integral component for therapists to address (American Psychological Association, 2017). The therapists in this study included men and women of color along with white men and women, and 28% of the couples involved one or both partners who identified as a person of color. When therapy outcomes among ethnic minority couples were compared with those of white, non-Hispanic couples over the course of 5 years of follow-ups, results from the two groups were found to be comparable (Yi, 2007). However, of concern is that African American women may have had less satisfying long-term results from treatment. Aligned with the increase in racial consciousness in this country, these results and increases in overt white supremacy ideologies and violence implore our field to move swiftly and thoughtfully toward elevated cultural competence and humility, antiracist therapist practices, therapist activism, and recruitment and retention of BIPOC students, trainees, and therapists. Training programs need to pay particularly close attention to preparation of trainees for effectively meeting the needs of clients who have been racially marginalized.

It is encouraging that therapist factors were the lowest reported category among the least helpful aspects of the therapy, implying that the therapist is rarely considered a least helpful factor by couples in therapy. However, examining those responses is particularly helpful for clinicians and trainees to discern what to avoid. A clear pattern in these responses emerged, indicating that all but one of the responses focused on therapist behaviors, not qualities. Behaviors noted by couples included ineffective instruction, not assisting with problem resolution, not assisting with materials given in session, not self-disclosing, not treating partners equally, and not refocusing session goals. These responses paint a picture of couples wanting an open, active, equally-aligned therapist who maintains constructive and goal-focused sessions.

Another pattern for therapists and trainees to note in the responses about least helpful aspects is that partners wanted more discussion about the sexual relationship, and this was reported by all groups in the study. This may be one of the most interesting findings, since the therapists were licensed, skilled and supervised weekly, and therefore would have pursued discussion of sex had the couple indicated it as a problem or expressed an interest in discussing it. This is an indicator to therapists that it is important to assess and address this aspect of relationships more thoroughly, even if the partners seem embarrassed to discuss it or don’t mention it as a presenting problem (Rothman et al., 2020). This is particularly true since many couples who enter therapy are unsatisfied with their sexual relationship (Doss et al., 2004).

Also of note for therapists and trainees is that reading assignments were reported by both treatment groups as least helpful. The reason for this finding is unclear, but warrants further attention, as it could stem from the materials used, low compliance, unequal compliance between partners, lack of integration of the readings into treatment, or other reasons.

Finally, the logistics factors are noteworthy in suggesting that aspects of the therapy such as commuting to, parking at, and time of sessions are noticed by and affect clients. It is especially helpful for couple therapists to remain aware that logistics influence the couple prior to entering the therapy room and become part of their entire therapy experience. The move to video sessions, out of necessity with stay-at-home orders in the pandemic or simply out of preference, may alleviate these added stressors for couples and improve accessibility. Therapists may also consider the benefit of longer sessions, as couples often suggested that the standard one-hour therapy session was not enough time. Researchers may be interested in examining differences between typical and lengthier sessions.

One overarching message in the data is that there is significant variability in couples’ responses, meaning that therapists, not surprisingly, need to respond in highly individualized ways to maximize helpful methods and minimize unhelpful ones for each couple. Based on how the findings compare to the common factors literature and emerging unified principles approach, it seems warranted for therapists and training programs to integrate common factors and unified principles to capture all of the important therapy elements as perceived by couples themselves.

A number of research implications have emerged from this study as well. In this study, the need to focus more on the sexual relationship was an important finding revealed through the mixed-methods design, drawing on both quantitative and qualitative understanding of the data, and asking about what is least helpful, not just most helpful. Future studies may want to integrate these methods to glean similarly specific and clinically-relevant implications.

Although there were several implications emerging from couples’ responses, future research would benefit from a more in-depth investigation of couples’ perceptions, such as in the form of an interview and use of member checks (Creswell, 2003), as well as assessments of these perceptions throughout the therapy, collected at different times over the course of treatment. A more systemic lens would compare perceptions between partners within each couple, and between therapists and couples. The current study involved two types of treatment that were behavioral in nature, so it would be interesting to study more distinct treatments as a way to continue examining common factors and principles along with model-specific factors of therapy.

Finally, the coding system developed in this study is appropriate for use with approaches other than TBCT and IBCT, even individual, group, and other systemic approaches, making its use easily transferable to future investigations. Although the coding system would benefit from refinement through its use in future research, it is clearly a distinctive and effective way of examining client’s experiences in therapy, common factors, and unified principles of change.