Though originally developed to address substance use disorders (Miller, 1983
), MI has now been tested across a wide range of target behavior changes. It has been found to be effective both in reducing maladaptive behaviors (e.g., problem drinking, gambling, HIV risk behaviors) and in promoting adaptive health behavior change (e.g., exercise, diet, medication adherence). The clinical style and apparent mechanisms of change in MI thus seem to be related to generalizable processes of human behavior, and not limited to specific target problems. As discussed above, the effectiveness of MI also appears to be amplified when it is added to other active treatment methods. It therefore shows promise as one clinical tool, to be integrated with other evidence-based methods, for use when client ambivalence and motivation appear to be obstacles to change.
Therapist style and practice can substantially improve or degrade client outcomes. This has been reflected in the variability of outcomes of MI across therapists, sites, and studies. Research on MI sheds light on some of the underlying processes that may be operative well beyond the specific method of MI. Moyers and colleagues (2005
) have presented data indicating a complex relationship among therapist responses, client speech, and subsequent behavior change. Both the relational (MI spirit) and technical attributes of MI contribute to outcome as mediated by client change talk. Progress has been made toward constructing a causal chain that clarifies how MI affects behavior change. A large efficacy literature shows that MI can directly impact client outcomes (paths 6 and 7 in ). Linkage has also been established between specific MI practice behavior and client change talk (paths 1 and 2 in ), a hypothesized mediator of MI’s impact on behavior change. The strength of preparatory change talk predicts subsequent strength of commitment (path 3), both of which have been shown to predict client outcomes (paths 4 and 5). Furthermore, training in MI has been shown to improve clinician performance on MI skills (paths 8 and 9) that are themselves related to client outcome (paths 6 and 7), and to directly increase change talk among clients of trained clinicians (path 10). An independent review of MI process research found that MI implementation is discriminable by MI-consistent therapist behaviors, which in turn predict in-session client responses and post-session treatment outcomes in a manner consistent with the theory of MI stated here (Apodaca & Longabaugh, 2009
, p. 712). An obvious next step is the evaluation of full mediation models integrating the multiple links in this chain (Baron & Kenny, 1986
; Longabaugh & Wirtz, 2001
Even so, causal chain analyses are but a first step in understanding how and why MI effects behavior change. If therapist empathy does enhance client change talk or otherwise improve client outcomes, how does it do so? If the elicitation of client change talk is reliably linked to commitment and behavior change, why is that so? Is it literally the voicing of change talk that causes behavior change? Chanting aloud one hundred times, “I will change, I will change” seems unlikely to make it so. Instead, it is plausible that the processes of MI trigger covert events that are not directly observable, but which result in both increased commitment language and subsequent behavior change. In this case, the observed commitment language is not itself a cause of change, but represents a signal that the covert events are occurring and that change is likely to follow. If that is so, then the verbalization of commitment strength is not a necessary precondition for change. Often, we suspect, the tree falls in the internal forest and no one hears the sound of it.
What might such covert antecedent events be? Some possible descriptors are acceptance, readiness, or decision, with corresponding shifts in perception of self. A reasonable analogy is engagement. When a couple become engaged, they have reached a decision that they are ready (or at least preparing) to make a commitment to each other, which is accompanied by shifts in perception of themselves and their relationship. Engagement is often an emotionally charged, highly significant event, but is not in itself the act of commitment in the presence of witnesses. The public committing act of marriage follows from the private event of engagement. In American culture, at least, engagement does not typically involve binding legal documents. Most often it is a private event, the announcement of which is optional and may be formal or informal. There are some common outward and visible signs of engagement: a ring, statements made to others, focusing of intimacy on the betrothed. Yet none of these is in itself the act of engagement; they are simply reflections of the underlying event. So, too, readiness for change may emerge as a private, discrete shift that opens the door for public commitment.
There remain some interesting wrinkles to be ironed out in the fabric of MI, such as the role of disingenuous change talk. It was not the frequency or absolute level of commitment language (the intercept) that predicted behavior change, so much as a pattern of increasing strength of commitment (positive slope) during a counseling session (Amrhein et al., 2003
). Initial commitment level at the beginning of the MI session did not signal behavior change, and clients whose commitment strength did not increase during a session were less likely to be abstaining from drugs at follow-up. It follows that clients who enter a session already professing high commitment may not be the most likely to change. This in turn raises the issue of client honesty. People can offer dishonest change talk, signaling commitments that they have no intention of keeping. Amrhein’s psycholinguistic coding system included attention to nonverbal cues (such as a slight shrug of the shoulders) that when accompanying commitment language signal significantly decreased likelihood of behavioral follow-through. Such subtle cues probably contribute to clinicians’ impressions of client sincerity and motivation, which can in themselves be prognostic of behavior change outcomes (Dunn, Droesch, Johnston & Rivara, 2004
) . It is noteworthy that clients who were subsequently dishonest about abstinence showed the same pattern of in-session vacillating commitment as those who reported continued drug use (Amrhein et al., 2003
). The pattern of their in-session speech told the truth. Further study of the nature and patterning of client responses, as well as the manner in which intentionality is coded and decoded in everyday conversations (Malle, 2004
) may lead to more reliable markers of dissimulation and intentionality toward behavior change.
The relative contributions of the relational and technical components of MI also remain to be clarified. If therapists manifest a high relational level of accurate empathy and MI spirit, how much is efficacy further improved by adding the technical focus on eliciting change talk and commitment language? One randomized clinical trial (Sellman et al., 2001
) studied this question with moderately severe problem drinkers, comparing the effects of nondirective counseling, simple feedback, and MET. The MET intervention yielded significantly greater reduction in heavy drinking than did nondirective counseling or a single feedback session, indicating a large effect associated with the technical attributes of MI. Karno & Longabaugh (2005)
found an interaction between client anger, reactance and clinician interpersonal style; clients with high levels of anger and reactance did poorly with clinicians who demonstrated behaviors inconsistent with both the spirit and technique of MI. Thus, the answer to the relational versus technical contributions of MI may be a complicated one.
The opposite question – how much does MI spirit add to the technical components – would seem more difficult to evaluate, in part because MI without this underlying spirit is no longer MI. One trial of MI techniques delivered in what appears to be a more authoritarian overall style (Kuchipudi et al., 1990
) showed no effect on behavioral outcomes.
It is also likely that other factors will be discovered that play an important role in the processes and outcomes of MI. Clarification of these “active ingredients” could help to focus training on those components that are necessary and/or sufficient for the efficacy of MI, and thereby clarify what aspects can be modified (for example, in cross-cultural adaptations of MI) without compromising its efficacy (Miller, Villanueva, Tonigan, & Cuzmar, 2007
; Venner, Feldstein, & Tafoya, 2007