The past 100 years of psychotherapy research has sought not only to examine the efficacy and effectiveness of psychotherapy, but also to identify the causal mechanisms and processes underlying therapeutic change (Lambert, 2013; Wampold & Imel, 2015). The existing research on psychotherapy processes has provided us with a rich understanding of several variables that are linked to treatment outcomes, including attendance to the therapeutic alliance, eliciting and responding to client feedback, addressing client expectations and motivation, and collaboration (Duncan, Miller, Wampold, & Hubble, 2010; Norcross, 2011; Swift & Greenberg, 2015). Yet, although significant gains have been made through this research, as a field we still do not know exactly why or how change occurs for clients.

Our limitations in knowing the exact why and how of change in psychotherapy may be partially due to the limitations in the research methods we frequently use. A large portion of the current research being conducted focuses on testing the efficacy of specific treatments or treatment techniques. Although this research plays a valuable role, therapist, client, and process variables are often suppressed or ignored in these types of studies. Much of the research focusing specifically on psychotherapy processes utilizes a global rating of the process variable for the entire session. These studies may miss details of the psychotherapy process that could help to explain the how and why of client change.

Micro-process research is one method for gaining a deeper understanding of these dynamics. This research typically involves a moment-to-moment analysis of treatment sessions, focusing on a segment of a single session, an entire session, a specified number of sessions, or the entire course of therapy for some clients. One example of micro-process research was provided in the last issue of the Psychotherapy Bulletin. In that issue, Pace and colleagues (2016) discussed research they are doing, examining natural language processing in the therapeutic dyad. In these studies, they are examining transcripts from psychotherapy sessions in order to identify relational/emotional processes in word usage that are associated with therapeutic change. Another example comes from a study of treatment outcomes for generalized anxiety disorder (Fluckiger, Zinbarg, Znoj, & Ackert, 2014). In this study, trained coders conducted minute-by-minute ratings of the fifth session for 20 different clients. While observing these sessions, coders were identifying clients’ experiences of negative moods, expressions of strength, working toward goals, and positive reinterpretations of problems. In another interesting study, Altenstein, Krieger, and Holtforth, (2013) had trained coders watch a middle therapy session for a client receiving exposure-based cognitive therapy for depression. While watching, these coders used a joystick to rate the therapists’ and clients’ circumplex of dominant and affiliative behaviors throughout every second of the session.

Most of the micro-process research that has been conducted has used outside/neutral parties to evaluate the session content (e.g., trained researchers to code behaviors and emotions, a computer program to code text); however, it may also be valuable to have clients and their therapists provide moment-to-moment ratings of their sessions. Given their level of involvement, clients’ and therapists’ moment-to-moment ratings may be able to provide unique insight into the why and how questions of therapeutic change. With this goal in mind, we recently completed a study in which we asked 18 clients and their therapists to provide moment-to-moment ratings of the effectiveness of a single psychotherapy session. Specifically, we were seeking to (1) examine whether variability in the moment-to-moment ratings within sessions was present; (2) identify therapist and client behaviors associated with higher and lower rated segments of the sessions; and (3) test whether clients and therapists were congruent in their moment-to-moment ratings. Here we provide the results from one dyad as an example of this type of micro-process research.

The Client-Therapist Dyad

The client was 30 years old, identified as an American Indian female, and was married in a heterosexual relationship. She was seeking treatment primarily for concerns with depression, but also experienced some anxiety symptoms and alcohol use concerns. At the time of her participation in this study (after the fifth session), the client rated the therapeutic relationship as a 47 (of 60) on the Working Alliance Inventory-Short Revised (Hatcher & Gillaspy, 2006), and the total session effectiveness as a 32.5 (of 40) on the Session Rating Scale (Johnson, Miller, & Duncan, 2000).

The therapist was a similar-aged female (specific demographic information is not reported in order to protect the individual’s identity) who was receiving training in a clinical psychology Master’s program. The therapist reported working with this client from an integrated approach. The therapist rated the therapeutic relationship as a 40 (of 60) on the Working Alliance Inventory-Short Revised.

Moment-to-Moment Coding of Sessions

The client and therapist had been working together for five sessions with no knowledge of this study. Per routine clinic procedures, all sessions were recorded. The client volunteered to participate in this study after seeing a flier posted in the clinic waiting room. After informed consent for the study was provided, the client watched a video recording of the most recent treatment session. While watching the recording, she held a dial rating device with a pivoting central knob and a digital numeric display screen. She was instructed to rate every moment of the session on a scale ranging from helpful (100, dial pointing to the far right) to hindering (0, dial pointing to the far left), with 50 being neutral. The anchors (helpful and hindering) were labeled on the device and the digital screen reported the numeric rating the knob was recording. After watching and rating the entire session, the client was asked to watch their three highest and the three lowest rated segments and describe why those segments were rated as helpful or hindering.

After the client completed her ratings of the session, her therapist was invited to come in and watch the same session and complete the moment-to-moment ratings in a similar way. At the end of the session, the therapist was shown her three most helpful and three most hindering rated segments, and was asked to describe what was occurring in each. Last, the therapist was asked to watch the client’s highest and lowest rated segments, but was not told which was which. Instead, the therapist was asked to guess which three segments were rated more helpful and which three segments were rated as more hindering by her client.

Evidence for Variability Within the Session

Figure 1 displays the client’s and therapist’s moment-by-moment ratings for the session. One immediately notices the variability within the session. For example, the client tended to rate the session fairly neutral, but then had several segments that were rated very positively and some rated negatively. The therapist, on the other hand, tended to fluctuate more in the positive and negative ratings, and rarely identified the segments of the session as just being neutral. This finding in particular demonstrates the need for micro-process research. Based on the variability that was observed, it is clear an end-of-session global rating would not represent the many highs and occasional lows that occurred during this session. By pinpointing the highs and lows, researchers can more specifically identify the processes leading to change.

Explanations for the Variability

Helpful segments. When asked to describe what was occurring in the segments rated most helpful by the client, the client indicated the therapist was helping her to become more accepting of her positive and negative thoughts; the therapist allowed her to express her own motivation for change, rather than judge her for doing something wrong; and the therapist validated her progress. When asked to describe her most helpful rated segments of the session, the therapist indicated they were characterized by when she was able to just listen and provide support while the client engaged in change talk or made her own connections leading to insight. In general, the therapist indicated all three were characterized by her feeling like she and the client were on the “same page.”

Hindering segments. The client described her most hindering rated segments as characterized twice by the therapist asking a question, and once by the therapist offering advise that did not seem to her to be related to what they were talking about or the direction the session was moving. The therapist indicated her three most hindering rated segments included two times where she tried to provide validation to the client and the client did not accept the validation, and one time where there was an awkward silence where she, as the therapist, forgot the point of what she was sharing with the client.

A review of the qualitative descriptions of helpful and hindering segments provides us with some insight into clients’ and therapists’ opinions about how and why psychotherapy works. For this pair, the session seemed to be more helpful when the client and therapist felt like they were on the “same page.” This seemed to occur when the therapist was providing support and validation to the client, allowing the client to lead, and helping the client become more accepting of herself. These segments were not characterized by the use of any specific treatment techniques or the therapist providing any specific education or insight to the client. For this pair, the hindering segments seemed to be characterized by the therapist trying to lead the client in a direction the client did not understand or was not ready to go.

Congruence between the Client and Therapist

Client and therapist ratings on the 0 (hindering) to 100 (helpful) scale were matched for time and a Pearson’s correlation coefficient was calculated. A significant, but modest, correlation between the client and therapist ratings was found, r = .13, p < .01. The client’s average rating of the session, M = 52.20 (SD = 8.44), and the therapist’s average rating of the session, M = 53.07 (SD = 13.96), were similar, t(457) = 1.22, p = .22. By visually examining Figure 1, it appears that the client and therapist similarly recognized one hindering segment toward the end of the session. It also appears the therapist and client matched for about half of the positively-rated segments. However, the therapist rated several segments as hindering that the client felt were neutral or positive. In addition, the therapist missed several of the segments rated positively by the client. The client’s three most helpful and three most hindering-rated segments were shown to the therapist and the therapist was asked to make a categorical guess as to which ones the client rated as helpful or hindering. The therapist guessed four of the six segments correctly.

The level of congruence, or lack thereof, between this client and therapist is interesting. One might expect, given these two individuals were providing their opinions about every moment of the session, a relatively small correlation would be found. In addition, given they had only worked together for four sessions prior to this point and that the therapist was a trainee, the significant correlation found is actually fairly impressive. However, the divergent segments (at about one-quarter and five-eighths of the way through the session, respectively) are important areas of concern. Some of these were areas where perhaps the therapist felt like she was being helpful in what she was doing, but the client did not feel like she was getting much out of what was happening. Some of these were also times when the therapist felt like things were going poorly, but the client did not feel the same way—this was seen for most of the hindering-rated segments by the therapist. These discrepancy segments would be important areas to target, because they represent clinical errors in therapist judgment, which in turn could lead to small treatment decisions by the therapist that could then lead therapy in the wrong direction.

Conclusions

The micro-process data that were gathered provided important insight into what was working and what was not working during this session for this individual therapist and client. Using similar micro-process data from several dyads, it will be interesting to see if the level of congruence within a pair is significantly related to the number of sessions the client and therapist have had together as well as their ratings of the therapeutic alliance and treatment outcomes. Future studies are needed that focus on gathering moment-to-moment evaluations of treatment sessions by clients, therapists, and outside observers. Data from these types of micro-process studies can provide a rich and detailed picture of how and why psychotherapy may be effective—an understanding that can complement findings from the other frequently used process and outcome research methods.

+6

0

Joshua K. Swift, Ph.D.

Joshua K. Swift, Ph.D. is an Assistant Professor in the Department of Psychology at Idaho State University. He conducts psychotherapy process and outcome research, with a special focus on client factors (preferences and expectations) and premature termination. He is also a licensed psychologist.