Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Jan 1.
Published in final edited form as: J Subst Abuse Treat. 2021 Oct 23;132:108642. doi: 10.1016/j.jsat.2021.108642

What can clients tell us about whether to use motivational interviewing? An analysis of early-session ambivalent language

David P Forman a,*, Theresa B Moyers a, Jon M Houck b
PMCID: PMC8671198  NIHMSID: NIHMS1751630  PMID: 34716039

Abstract

Background:

Although motivational interviewing (MI) is an effective method for promoting change in problematic alcohol and other drug use, it does not benefit all clients. Clinicians have little empirical guidance on who is likely to benefit from MI and who is not. We hypothesized that differences in clients’ spontaneously offered language early in the session would predict their responsiveness to MI during the remainder of the session.

Method:

The study obtained coding data from 125 counseling sessions from a large randomized controlled trial of clinician training. A cluster analysis created one group of clients whose language reflected ambivalence, and one group whose language reflected readiness to change. We conducted a univariate analysis of variance to compare the mean change in percent change talk across the session between groups.

Results:

Clients whose language reflected ambivalence early in the session had a greater change in their percent change talk during the remainder of the session, compared to those whose language reflected greater readiness to change (F(1,90) = 63.02, t = 7.94, p < .001). Surprisingly, the group whose language reflected readiness had a decrease in their percent change talk during the remainder of the session (M = −10.9%, SD = 16.3%). Adjusting the results for regression to the mean effects did not eliminate these differences.

Conclusion:

Clients’ language early in the session may offer clinicians some guidance on whether MI is likely to be useful or counterproductive in the treatment of substance use disorder.

1. Introduction

Although Motivational Interviewing (MI) is an effective method for promoting change in alcohol and substance misuse, it does not benefit all clients. Meta-analyses that support MI’s effectiveness recommend MI to front-line counselors as either a standalone approach or an adjunct to other therapies targeting substance use and high-risk behaviors (American Psychological Association, 2020; Substance Abuse and Mental Health Services Administration, SAMHSA, 2019). And indeed, 93% of surveyed substance use treatment centers in the U.S. report using MI as one of their treatment methods (SAMHSA, 2019). While this finding is encouraging for proponents of the empirically supported treatments movement, site differences within multi-site trials of MI are notable (Ball et al., 2007; Winhusen et al., 2008), and effect sizes have varied considerably across trials (Miller & Moyers, 2015). These findings pose a riddle for MI researchers and a clinical challenge for counselors aiming to deliver the treatment that is most likely to benefit the client in front of them. Some authors have suggested that these inconsistent findings are attributable to a lack of specification and monitoring of MI’s active ingredients during clinical trials, and emergent data support this view (DiClemente et al., 2017; Rowell, 2017). Yet even within well-controlled trials that use state-of-the-art fidelity monitoring and supervision, not all substance use clients improve with MI and some deteriorate (Miller, Yahne, & Tonigan, 2003). Research has yet to provide clinicians with consistent guidance on which clients are likely to be responsive to MI and which are not.

It seems sensible that client variables should moderate MI’s effectiveness, perhaps explaining the variability in addiction outcomes. Yet efforts to identify candidates have not been encouraging, with characteristics such as gender and ethnicity producing poor or inconsistent results across empirical reviews (Lundahl et al., 2010; Lundahl et al., 2013). Indeed, a comprehensive review of reviews of the efficacy and effectiveness of motivational interventions did not find any consistent client characteristics that moderate MI’s effects on outcomes for alcohol, tobacco, or drug use (DiClemente et al., 2017). At this point, clinicians providing addictions treatment are reliant on their clinical intuition to determine whether the client is a good “fit” for MI. This is a problem because research consistently shows intuition as inferior to statistical methods when making clinical predictions (Grove et al., 2000; Meehl, 1996). Yet reason for optimism still exists. Responsiveness to MI might be better predicted by variables such as ambivalence, which can vary across a session, as opposed to static variables assessed at baseline.

The psychological construct of ambivalence is central in MI theory and practice and research has described it as involving the coexistence of pros and cons regarding target behavior change (Miller & Rollnick, 2013). The primary insight of the MI method is that ambivalence is a particularly common barrier to meaningful change and that ambivalent language is highly responsive to the therapist’s efforts to influence it within a supportive and empathic relationship. It follows that the task of the MI therapist is to construct conversations about change so that clients’ language becomes more change-favoring, and less change-opposing as the session progresses. This is achieved primarily through a process of “evoking” in which the therapist both asks for and selectively responds to change-favoring language (change talk) while reframing and deflecting language that is change-opposing (sustain talk). The technical hypothesis of MI is that this therapist-guided change in client language contributes to positive behavioral outcomes (Miller & Rose, 2009), and some support exists for this hypothesis in studies investigating mechanisms of action in MI (Magill et al, 2018). The conflict resolution hypothesis of MI, which has received relatively little empirical investigation, proposes that the therapist’s thorough exploration of both sides of the client’s ambivalence naturally tends to resolve that ambivalence, clearing the way for future behavior change (Engle & Arkowitz, 2006; Greenberg, Rice, & Elliot, 1993). Therefore, investigations of clients’ ambivalent language contribute directly to the empirical foundations of each of these hypotheses for explicating MI’s mechanisms.

Studies have used paper-and-pencil readiness-to-change (RTC) measures such as the Stages of Change Readiness and Treatment Eagerness Scale (SOCRATES; Miller & Tonigan, 1996) and the University of Rhode Island Change Assessment (URICA; McConnaughy et al., 1989) to infer ambivalence as a static, stage-dependent characteristic of clients. Yet when research uses these measures to predict client response to MI, the findings are mixed, with some studies showing that higher readiness scores predict better outcomes with MI (Myers et al., 2016) and other studies showing no relationship (Murphy et al., 2012). Ambivalence may be a dynamic construct that fluctuates across a treatment session and that limited-item measures using closed-ended questions about readiness capture aspects that are related to, but distinct from, the spoken ambivalence that is the focus of MI practice. A measure of this interpersonal and fluid aspect of ambivalence, captured in client language, may offer promise in predicting responsiveness to MI. This possibility is bolstered by the consistent finding that client language in MI is predictive of outcomes (Moyers et al., 2009), particularly when measured proportionally (Magill et al., 2018).

The goal of the current study was to investigate whether substance use clients’ ambivalence, assessed via their language early in the session, would predict these clients’ differential responsiveness to MI. Our hypothesis was that differences in early-session language (before MI evoking began) would predict the malleability of substance use clients’ language during the remainder of the session (when MI evoking was intentional). Specifically, we hypothesized that clients who began an MI session with a relatively equal ratio of change to sustain talk (ambivalent) would show an increase in their percent change talk later in the session. Likewise, we hypothesized that those clients who began the session with a language ratio that favored either change talk or sustain talk (non-ambivalent) would not show a change in their language ratio.

2. Materials and methods

2.1. Participants

This secondary analysis reviewed 125 MI sessions from Project ELICIT, a randomized control trial that investigated whether substance use counselors could intentionally influence client language (Moyers et al., 2017). The study randomly assigned counselors to receive MI training-as-usual (MI-U) or an MI training that emphasized influencing client language (MI-Language Enhanced Attention and Focus, or MI-LEAF). Counselors submitted audiotaped MI sessions with real clients targeting substance use issues at 3, 6, and 12 months post-training. The clients in the audiotaped sessions provided consent through their respective agencies for their sessions to be audiotaped and reviewed by the study researchers. Because the study did not consider clients participants in Project ELICIT, it collected no client information. The main finding from Project ELICIT was that the frequency of sustain talk, but not change talk, was significantly lower for clients whose counselors were trained in the MI-LEAF condition, demonstrating that counselors can be trained to intentionally shape clients’ language. The Main-Campus Institutional Review board at University of New Mexico approved all procedures of Project ELICIT prior to the start of recruitment. A full description of Project ELICIT, including therapist-participant characteristics, is published elsewhere (Moyers et al., 2017).

The research team restricted the current analysis to 3-month follow-up sessions from both the MI-LEAF and MI-U conditions. We reasoned that if clients responded differently to the evoking strategies of MI depending on their initial level of ambivalence, as hypothesized, the effect was most likely to appear in these sessions where the therapist was using a full range of MI skills. The study did not formerly structure sessions nor did it instruct therapists to use any key questions or phrases. In addition, we selected the 3-month follow-up to ensure independence of data and because efforts to practice MI would be more accurate closer to, rather than further from, workshop training and enrichments, and prior to the onset of skill decay (Moyers et al., 2017).

2.2. Measures

2.2.1. MI skills code

As part of Project ELICIT, study staff coded client language from client sessions using the Motivational Interviewing Skills Code system (MISC 2.5, Houck et al., 2010). The MISC is a mutually exclusive and exhaustive coding system that evaluates both counselors’ and clients’ language relevant to the processes of MI. Interviewer language is evaluated for both global aspects of the sample (acceptance, empathy, direction, autonomy support, collaboration and evocation), and behavior counts for specific counselor statements that are consistent or inconsistent with MI (e.g., reflection, affirmation, confrontation). Every therapist reflection in the MISC is coded as either neutral or valenced. Valenced reflections are those that emphasize a particular direction about which the client is encouraged to speak. For example if a client said, “I wish I didn’t love drinking so much”, a neutral reflection would be “You wish you didn’t love drinking”. A positively valenced reflection would be “things would be easier if you could stop” and a negatively valenced reflection would be “You don’t want to give it up.” Positively valenced reflections, because they strategically move the client to respond in a way that encourages discussion of change, are a marker of the evoking process in MI. Although questions can also be valenced toward or away from change, the study did not code valenced questions using the MISC because we could not achieve acceptable interrater reliability.

Clients’ speech in the MISC is coded as either change talk, sustain talk, or neutral. Change talk is defined as “any language that moved in the direction of change” and sustain talk as “any language indicating a move away from change” (Houck et al., 2010). Interrater reliability for the coders in Project ELICIT ranged from good to excellent, with ICC’s ranging from .59 to .99 for MISC codes used in the analyses (Moyers et al., 2017).

2.3. Procedure and variables

To identify the point in the session where therapist-participant began use of the evoking strategies of MI, we divided the session into deciles and examined the sequence of the therapists’ coded language. We defined the evoking process as the decile in which the therapist began using positively valenced reflections. Our analyses suggested that, on average, therapist-participants began evoking at the end of the first decile (mean time = 4.967 min (SD 1.133). We named this part of the session “Pre-Evoking (PE)” and the remainder of the session “Evoking (E)”. Because therapists had not yet employed sustained efforts to shape client language in PE, we reasoned that speech in this segment was an indicator of clients’ baseline ambivalence about change. MI practice always begins with a process of engagement, during which the therapist and client explore the client’s ambivalence about the target change behavior with the goal of “understanding the dilemma” (Miller & Rollnick, 2011). Therefore, client ambivalence toward change, if present, would likely emerge during the initial segment of the MI conversation.

We then calculated the percentage change talk (PCT; i.e., Change Talk/[Change Talk+Sustain Talk]) for each client for both phases of their MI session. Mean PCT is typically high in MI clinical trials—for instance, prior work using sessions from Project MATCH showed a mean PCT of 75.6% (SD 12.8%) (Moyers et al., 2009). Mean PCT in the full ELICIT sample was similarly high: approximately 78.3% (SD 18.0%) at the 3-month follow-up. To identify differences in initial ambivalence in this sample, we asked SPSS to perform a two-step cluster analysis on client PCT in the PE phase, without specifying a number of clusters. The study team excluded sessions with no CT and no ST in the PE phase (n=32) or no CT and no ST in the evoking phase (n=1). The cluster analysis returned two groups: one with a mean 47.8 PCT (SD 20.3%), and one with a mean of 95.0% PCT (SD 8.2%). We labeled these groups Ambivalent (n=45) and Ready (n=47), respectively. We then subtracted the percent change talk from the evoking phase of the session from the percent change talk from the pre-evoking phase to give us a variable that represented the change in percent change talk (C-PCT), which was approximately normally distributed. Thus, our outcome variable reflected the change in the client’s positive language about change across the MI session.

2.4. Analyses

To determine whether members of the Ambivalent and Ready groups responded differently to the evoking strategies of MI, we conducted a univariate analysis of variance to compare the mean C-PCT between groups, with summary measures of Decile 1 MI skill (percent MI-consistent responses) as a covariate. The research team conducted post hoc analyses to determine if any changes detected were attributable to regression to the mean. To do this, we followed the procedures recommend by Barnett, van der Pols, and Dobson (2005) to adjust the between-groups effect using an estimate of the expected regression to the mean based on the population mean and standard deviation and within-subject standard deviation of PCT. We also used t-tests to compare measures of clinician skill in the Ambivalent and Ready groups. Finally, we conducted a univariate analysis of variance to compare the absolute value of mean C-PCT between groups. This is important because the mean PCT of the Ready group was near the ceiling of PCT and in practice could only remain the same or decrease, while the mean PCT of the Ambivalent group was distant from both the ceiling and the floor of PCT and could increase, remain the same, or decrease. By evaluating only the magnitude of change, ignoring the direction, this test permits equivalent change in both groups.

3. Results

The study team evaluated the quality of the cluster analysis using silhouette plots, which visually represent the squared Euclidean distance between clustered data (Rousseeuw, 1987). Most cases had silhouette values greater than 0.60, indicating good separation between the two clusters.

We conducted a sensitivity power analysis to determine the smallest effect that could reasonably be detected for this sample at a range of power levels. Because the study has a directional hypothesis, we estimated power assuming a one-tailed test. With n = 92 participants and power = 0.80, we could detect an effect of d = 0.522, a medium effect size (Cohen, 1988). At power = 0.70, we could detect an effect of d = 0.456. At power = 0.60, we could detect an effect of d = 0.399.

A univariate analysis of variance of the change in PCT from pre-evoking to evoking—that is, C-PCT—indicated a significant between-groups difference (F(1,89) = 64.20, t = 8.01, p < .001). The effect of Decile 1 clinician MI skill (percent MI-consistent responses) was not significant (F(1,89) = 1.13, t = 1.06, p = .29). Clients in the Ambivalent group showed an increase in PCT from pre-evoking to evoking (M = 24.2%, SD = 25.3%), while those in the Ready group showed a decrease in PCT (M = −10.9%, SD = 16.3%) (see Figure 1). The direction of these effects aligns with a decrease in change talk in the Ready group and a decrease in sustain talk in the Ambivalent group (see Figures 2–3). Paired t-test results showed that PCT changed significantly from pre-evoking to evoking for both groups with a greater change for Ambivalent (t(45) = 6.40, p < .001) than for Ready (t(46) = 4.60, p < .001). Within the first decile there was no evidence of differences on measures of clinician skill (Reflection: Question ratio, Percent complex reflections, Percent open questions, Percent MI-consistent) between Ambivalent and Ready groups (all p > .13). For the full session, a significant difference occurred between Ambivalent and Ready groups on Percentage MI-consistent (d = 0.44, t = 2.10, p = .04), such that Percentage MI-consistent was higher in sessions with Ready participants than in Ambivalent participants by about 2.5 percentage points (i.e., 98.0% vs 95.5%).

Figure 1.

Figure 1.

Open in a new tab

Between-group difference on Change in Percent Change Talk (C-PCT).

Figure 2.

Figure 2.

Open in a new tab

Change in clients’ speech from pre-evoking to evoking.

The research team calculated the expected regression to the mean using procedures recommended by Barnett et al. (2005). Adjusting the results for regression to the mean effects did not eliminate the observed between-groups difference. In addition, we tested the absolute value of C-PCT as a complementary means of eliminating regression to the mean as a potential explanation for the observed effects. Those results largely mirrored those of the main study analysis, with a significant between-groups difference (F(1,8990) = 16.36, t = 4.04, p < .001, d = 0.71). The effect of Decile 1 clinician MI skill (percent MI-consistent responses) was not significant (F(1,89) = 3.67, t = 1.92, p = .06). Clients in the Ambivalent group showed significantly greater absolute change in PCT from pre-evoking to evoking (M = 27.9%, SD = 21.1%) than did those in the Ready group (M = 14.3%, SD = 13.3%).

4. Discussion

These findings were consistent with our hypothesis that differences in clients’ motivational language at the start of the MI session would be associated with changes in their language by the end of the session. Specifically, those who began with a high degree of ambivalence showed an increase in their percent change talk when responding to the evoking strategies of MI. This effect persisted when we repeated the analysis ignoring the direction of change and focusing on the magnitude—clients who presented with a high degree of ambivalence showed a greater change in their percent change talk. Unexpectedly, a subgroup of clients whose language indicated that they were ready to change, showed a decrease in their percent change talk in the evoking phase of the session. We were unable to test our hypothesis that those who began the session with primarily change-opposed language—that is, with more sustain talk—would not show a change in their motivational language in response to the therapist’s evoking because the cluster analysis did not identify such a group in our sample. From a clinical perspective, we find it reasonable that data from a treatment study would mostly contain treatment-seeking participants who are either change-ready or ambivalent about change. Yet given the evidence that MI reduces sustain talk and the dissemination of MI into areas outside of the therapy room (e.g., emergency rooms, probation offices, etc.), this hypothesis remains highly relevant to the targeted practice of MI and should be tested in a sample containing a high proportion of participants who are likely to be change-opposed at the start of the session (e.g., mandated clients). Our findings support the value of ambivalence as a clinical indicator for the use of MI, and raise the possibility that knowledge of when not to use it may be critical. To our knowledge, this is the first study to evaluate differences in responsiveness to motivational interviewing by comparing ambivalent language before and during a therapist’s strategic evocation and reinforcement of change talk.

These results also provide some insight into the dynamics of within-session client speech in a relatively “pure” form of MI, delivered without personalized feedback. Much of the mechanistic work in MI has relied on highly structured interventions delivered in the context of clinical trials of substance use disorder treatment (cf. Magill et al., 2018) with all the support and supervision that that entails. In contrast, our data arise from a sample of sessions conducted by clinicians in frontline settings, with their usual clients. In these sessions, the speech offered spontaneously by the client during the opening minutes of the session predicts how that speech will develop over the remainder of the session. This occurred despite the absence of any prescribed structure across clinicians and sessions. This initial period appears to offer clinicians a snapshot of the client’s ambivalence, allowing them to decide whether to proceed with MI or to shift into a different type of treatment, based on each client’s speech. Previous work by Bertholet et al. (2010) has suggested that change language at the end of the MI session is also an important indicator of MI success. Bertholet et al.’s end-of-session effect was detected in the context of highly structured MI sessions, wherein the study instructed therapists to ask specific change-focused questions at the end of the interview, perhaps prompting the emergence of change language at this juncture. Our data support a somewhat different point: that indicators of ambivalence at the start of the MI session are associated with change language in the remainder of the session, including the end. Both indicators are potentially useful and may signal unique decision points for clinicians. For clients with high change talk at the end of the session, as found by Bertholet et al., change planning would be appropriate, as opposed to clients who are still voicing high levels of sustain talk. For clients with high change talk at the beginning of the session, bypassing MI in favor of a more active approach may yield better outcomes. The clinician could potentially use language at the beginning of the session, then, in real time to decide whether or not to use MI.

A larger conceptual question is how our data might inform a clinician’s understanding or use of the Stages of Change model (McConnaughy et al., 1989). Although our study did not directly measure clients’ stage of change, our results suggest that clinicians should tailor their use of MI to the client’s presenting level of motivation. In essence, our study directs clinical attention to an in-the-moment behavior that is characteristic of the Contemplation Stage of Change.

Our data add to a growing literature investigating the utility of the clients’ language construct in MI. They extend previous findings that clients’ language predicts outcomes (Magill et al., 2018) by demonstrating that even relatively small samples of clients’ expressions from theoretically relevant segments of the MI conversation may have predictive power. Although the team did code complete MI sessions, our analysis focused on the potential for therapists to use key language indicators early in the session to make a prediction about how useful MI might be. This approach is consistent with a robust body of literature from social psychology indicating that “thin slices” of verbal behavior can accurately predict human responses, including interpersonal deception, teacher bias, and voting behavior (Ambady & Rosenthal, 1993) as well as patient satisfaction and compliance using thin slices of a doctor’s tone of voice when speaking to alcohol misusing patients (Hall et al., 1981; Milmoe et al.,1967). Perhaps the most powerful example of how language from carefully constructed contexts can predict behavioral outcomes is the work on marriage stability (Gottman & Gottman, 2017). Using the Rapid Couples Interaction Scoring System, Krokoff, Gottman, & Hass (1989) were able to predict six-year marriage outcomes with 88% accuracy based on a couples’ 15-minute conversation about a high-conflict issue. Specifically in the area of MI, Caperton, Atkins & Imel (2018) have used thin slices of MI sessions to estimate therapists’ fidelity to the method. They found that sampling one third of an MI session (without interruption) was sufficient to approximate scoring of the entire session using the same rating instrument that our study used (MISC). Taken with our findings, this suggests that analysis of thin slices of therapist-client exchanges may provide additional insights into how therapists’ skills may interact with clients’ ambivalent language to influence outcomes.

Our study is the first that we know of to operationalize ambivalence by using a balance of the client’s own motivational language at the beginning of a clinical intervention. Similar work has operationalized ambivalence using only instances of change-opposing statements in the delivery of cognitive behavioral therapy (CBT; Westra & Norouzian, 2018). That promising work has found that change-opposing statements in early CBT sessions for generalized anxiety disorder predicted treatment response beyond symptom severity and self-report measures of motivation (Lombardi et al., 2014). Our study finds a similarly potent signal for change-opposing language for substance use issues, as the change in ambivalent clients’ language ratio was largely driven by reductions in sustain talk.

Although previous work has defined ambivalence using paper-and-pencil measures (Miller & Tonigan 1996; McConnaughy et al., 1989), a behavioral measure, such as language, would likely be more useful, particularly in therapeutic interactions. When grappling with a vexing behavior that has both costs and benefits, a difference may exist between talking to oneself about changing (intrapersonal) and talking to another person about it (interpersonal). Intrapersonal motivational talk is notoriously wearying when ambivalence is present, because it tends to be repetitive and unproductive. As the saying goes, a person who is grappling with a problem behavior first thinks about one side of it, then thinks about the other side of it, and then quits thinking about it altogether. Paper-and-pencil measures capture this intrapersonal expression of ambivalence in a snapshot; however, the interpersonal nature of speaking about change to another person may be an entirely different and more malleable experience of ambivalence. Our findings suggest that future work investigating clients’ early-session ambivalent language as a moderator of MI outcomes may prove fruitful. The role of ambivalence may also be important for understanding why MI appears to perform better for alcohol and substance use behaviors in adults (Lundahl et al., 2013) than with college drinkers (Foxcroft et al., 2016). Important differences may exist between older and younger adults’ readiness to recognize the costs of alcohol misuse and these differences may be related to variability in responsiveness to an intervention that requires the client to identify intrinsic motivation for changing.

Finally, the results of this study have implications for the testability of the conflict resolution hypothesis (Magill & Hallgren, 2019), which has received little research attention. By demonstrating that using clients’ motivational language to operationalize ambivalence across the counseling session can yield results, our study presents a potential methodological solution to the challenge of measuring the dynamic process of ambivalence.

A primary limitation of this study was the lack of a non-MI comparison group. Although previous research has shown that therapists’ use of MI-consistent behaviors elicits different motivational language than does a non-MI approach (Apodaca, et al., 2016; Magill et al., 2018), without a non-MI control group in this sample we cannot conclude that the changes we observed in clients’ motivational language were attributable to MI’s evoking process. The differences in language between the Ambivalent and Ready groups may have been evident in response to any treatment approach. Additional analyses that compare changes in clients’ ambivalent language during MI’s evoking with changes in response to a different approach would extend these findings and support the theory that MI exerts its effects by diminishing ambivalence about change. Our analysis also did not investigate the relationship between early session ambivalent language and subtypes of change talk later in the session (e.g. desire, reason, ability, need). Although doing so was not feasible with our sample due to low frequencies of the various categories, analyses of change language subtypes with larger samples may provide additional insight into the dynamics of motivational language.

Despite these limitations, our findings hint at an important bellwether for clinicians in busy substance use treatment settings. Careful assessment, in real-time, of the client’s language balance about their dilemma may inform the clinician about whether MI is likely to be a helpful next step.

Highlights.

  • Assessing clients’ language at the start of the session may tell clinicians whether or not to use motivational interviewing (MI)

  • Clients whose language is ambivalent early in the session may respond well to MI

  • For clients whose are ready to change, MI may be counterproductive

  • Measuring ambivalence as a balance of change and sustain talk may be a promising method for further investigations of MI process and theory

Footnotes

Credit Author Statement

David Forman developed the idea and hypotheses for the study and wrote the initial draft of the introduction, methods, and discussions sections. Theresa Moyers provided conceptual guidance throughout the manuscript and contributed significant writing to the discussion section. Jon Houck provided conceptual guidance and served as the data analyst for the project. He provided the analyses and figures and wrote the results section. All three authors reviewed, edited, and approved the final manuscript.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. American Psychological Association (2020, December 3). Psychological treatments. https://div12.org/psychological-treatments/
  2. Ambady N, & Rosenthal R (1992). Thin slices of expressive behavior as predictors of interpersonal consequences: A meta-analysis. Psychological Bulletin, 111 256–274. [Google Scholar]
  3. Apodaca TR, Jackson KM, Borsari B, Magill M, Longabaugh R, Mastroleo NR, & Barnett NP (2016). Which individual therapist behaviors elicit client change talk and sustain talk in motivational interviewing? Journal of Substance Abuse Treatment, 61, 60–65. 10.1016/j.jsat.2015.09.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Ball SA, Martino S, Nich C, Frankforter TL, Van Horn D, Crits-Christoph P, Woody GE, Obert JL, Farentinos C, & Carroll KM (2007). Site matters: Multisite randomized trial of motivational enhancement therapy in community drug abuse clinics. Journal of Consulting and Clinical Psychology, 75, 556–567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barnett AG, van der Pols JC, & Dobson AJ (2005). Regression to the mean: What it is and how to deal with it. International Journal of Epidemiology, 34(1), 215–220. 10.1093/ije/dyh299 [DOI] [PubMed] [Google Scholar]
  6. Caperton DD, Atkins DC, & Imel ZE (2018). Rating motivational interviewing fidelity from thin slices. Psychology of Addictive Behaviors, 32(4), 434–441. 10.1037/adb0000359 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cohen J (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Hillsdale, N.J.: Lawrence Erlbaum Associates [Google Scholar]
  8. DiClemente CC, Corno CM, Graydon MM, Wiprovnik AE, & Knoblach DJ (2017). Motivational interviewing, enhancement, and brief interventions over the last decade: A review of reviews of efficacy and effectiveness. Psychology of Addictive Behaviors, 31(8), 862–887. [DOI] [PubMed] [Google Scholar]
  9. Engle DE, & Arkowitz H (2006). Ambivalence in psychotherapy: Facilitating readiness to change. Guilford Press. [Google Scholar]
  10. Foxcroft DR, Coombes L, Wood S, Allen D, Almeida Santimano NML, & Moreira MT (2016). Motivational interviewing for the prevention of alcohol misuse in young adults. Cochrane Database of Systematic Reviews(7). [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Bertholet N, Faouzi M, Gmel G, Gaume J, & Daeppen J-B (2010). Change talk sequence during brief motivational intervention, towards or away from drinking. Addiction, 105(12), 2106–2112. 10.1111/j.1360-0443.2010.03081.x [DOI] [PubMed] [Google Scholar]
  12. Gottman J, & Gottman J (2017). The Natural Principles of Love. Journal of Family Theory and Review, 9, 7–26. [Google Scholar]
  13. Greenberg LS, Rice LN & Elliott R (1993). Facilitating emotional change: The moment-by moment process. New York: Guilford Press. [Google Scholar]
  14. Grove WM, Zald DH, Lebow BS, Snitz BE, & Nelson C (2000). Clinical versus mechanical prediction: A meta-analysis. Psychological Assessment,12, 19–30. [PubMed] [Google Scholar]
  15. Hall JA, Roter DL, & Rand CS (1981). Communication of affect between patient and physician. Journal of Health and Social Behavior, 22(1), 18–30. [PubMed] [Google Scholar]
  16. Houck JM, Moyers TB, Miller WR, Glynn LH, & Hallgren KA, (2010). Motivational Interviewing Skill Code (MISC) Version 2.5, Available from http://casaa.unm.edu/download/misc25.pdf. Retrieved from http://www.webcitation.org/6GjlS23Ac
  17. Krokoff LJ, Gottman JM, & Hass SD (1989). Validation of a global rapid couples interaction scoring system. Behavioral Assessment, 11(1), 65–79. [Google Scholar]
  18. Lombardi D, Button M, & Westra HA (2014). Measuring motivation: Change talk and counter-change talk in cognitive behavioral therapy for generalized anxiety. Cognitive Behavioural Therapy, 43, 12–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lundahl BW, Kunz C, Brownell C, Tollefson D, & Burke BL (2010). A meta-analysis of motivational interviewing: Twenty-five years of empirical studies. Research on Social Work Practice, 20(2), 137–160. [Google Scholar]
  20. Lundahl B, Moleni T, Burke BL, Butters R, Tollefson D, Butler C, & Rollnick S (2013). Motivational interviewing in medical care settings: a systematic review and meta-analysis of randomized controlled trials. Patient Education and Counseling, 93(2), 157–168. [DOI] [PubMed] [Google Scholar]
  21. Magill M, Apodaca TR, Borsari B, Gaume J, Hoadley A, Gordon REF, … Moyers TB (2018). A meta-analysis of motivational interviewing process: Technical, relational, and conditional process models of change. Journal of Consulting and Clinical Psychology, 86(2), 140–157. doi: 10.1037/ccp0000250 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Magill M & Hallgren K (2019) Mechanisms of behavior change in motivational interviewing: do we understand how MI works? Current Opinion in Psychology, 30, 1–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. McConnaughy EA, DiClemente CC, Prochaska JO, & Velicer WF (1989). Stages of change in psychotherapy: A follow-up report. Psychotherapy, 26(4), 494–503. [Google Scholar]
  24. Meehl PE (1996). Clinical versus statistical prediction: A theoretical analysis and a review of the evidence. Northvale, NJ: Jason Aronson. (Original work published 1954) [Google Scholar]
  25. Miller WR & Moyers TB (2015). The forest and the trees: Relational and specific factors in addiction treatment. Addiction, 110, 401–413. DOI: 10.1111/add.12693 [DOI] [PubMed] [Google Scholar]
  26. Miller WR, & Rollnick S (2013). Motivational interviewing: Helping people change (3rd ed.). New York, NY: Guilford Press. [Google Scholar]
  27. Miller WR, & Rose GS (2009). Toward a theory of motivational interviewing. American Psychologist, 64, 527–537. 10.1037/a0016830 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Miller WR, & Tonigan JS (1996). Assessing drinkers’ motivation for change: The Stages of Change Readiness and Treatment Eagerness Scale (SOCRATES). Psychology of Addictive Behaviors 10, 81–89. [Google Scholar]
  29. Miller WR, Yahne CE, & Tonigan JS (2003). Motivational interviewing in drug abuse services: A randomized trial. Journal of Consulting and Clinical Psychology, 71(4), 754–763. 10.1037/0022-006X.71.4.754 [DOI] [PubMed] [Google Scholar]
  30. Milmoe S, Rosenthal R, Blane HT, Chafetz ME, & Wolf I (1967). The doctor’s voice: Postdictor of successful referral of alcoholic patients. Journal of Abnormal Psychology, 72(1), 78–84. [DOI] [PubMed] [Google Scholar]
  31. Moyers TB, Martin T, Houck JM, Christopher PJ, & Tonigan JS (2009). From in-session behaviors to drinking outcomes: A causal chain for motivational interviewing. Journal of Consulting and Clinical Psychology, 77(6), 1113–1124. doi: 10.1037/a0017189 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Moyers TB, Houck J, Glynn LH, Hallgren KA, & Manuel JK (2017). A randomized controlled trial to influence client language in substance use disorder treatment. Drug and Alcohol Dependence,172 (43) 50. 10.1016/j.drugalcdep.2016.11.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Murphy CM, Linehan EL, Reyner JC, Musser PH, & Taft CT (2012). Moderators of response to motivational interviewing for partner-violent men. Journal of Family Violence, 27, 671–680. 10.1007/s10896-012-9460-2 [DOI] [Google Scholar]
  34. Myers B, Van Der Westhuizen C, Naledi T, Stein DJ, Sorsdahl K (2016). Readiness to change is a predictor of reduced substance use involvement: findings from a randomized controlled trial of patients attending South African emergency departments. BMC Psychiatry, 16(35). [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Project MATCH Research Group (1998). Matching alcoholism treatments to client heterogeneity: Project MATCH three-year drinking outcomes. Alcoholism: Clinical and Experimental Research, 22(6), 1300–1311. [DOI] [PubMed] [Google Scholar]
  36. Rowell LN (June 2017). Clearing the Muddy Waters: The Importance of Treatment Integrity in Motivational Interviewing Efficacy Trials. Oral presentation presented at the International Conference on Motivational Interviewing, Philadelphia, PA. [Google Scholar]
  37. Substance Abuse and Mental Health Services Administration. Enhancing Motivation for Change in Substance Use Disorder Treatment. Treatment Improvement Protocol (TIP) Series No. 35. SAMHSA Publication No. PEP19-02-01-003. Rockville, MD: Substance Abuse and Mental Health Services Administration, 2019. [Google Scholar]
  38. Westra HA, & Norouzian N (2018). Using motivational interviewing to manage process markers of ambivalence and resistance in cognitive behavioral therapy. Cognitive Therapy and Research, this issue. 10.1007/s10608-017-9857-6. [DOI] [Google Scholar]
  39. Winhusen T, Kropp F, Babcock D, Hague D, Erickson SJ, Renz C, et al. (2008). Motivational enhancement therapy to improve treatment utilization and outcome in pregnant substance users. Journal of Substance Abuse Treatment, 35, 161–173. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES