The Motivational Interviewing Treatment Integrity code, version 4.2 (MITI 4.2), is the 16-item tool that measures both the relational and technical skills of motivational interviewing. Four global scores capture the relational and technical spirit of a session, and 12 behavior counts tally specific clinician utterances. Trained coders apply it to recorded sessions, and it remains the most widely used fidelity instrument in MI research and training.

How the 16 Items Split Into Four Globals and Twelve Counts

The MITI 4.2 organizes clinician behavior along two axes. Relational items describe the interpersonal climate of the session. Technical items describe what the clinician actually says. The 16 items divide into four global ratings and 12 behavior counts, and each count is later rolled into a composite technical score.

The four globals are Empathy, Partnership, Cultivating Change Talk, and Softening Sustain Talk. Each is rated on a five-point scale, where 1 indicates poor practice and 5 indicates exemplary practice. Empathy and Partnership form the relational dimension. Cultivating Change Talk and Softening Sustain Talk form the technical dimension.

The 12 behavior counts divide into two technical families. The “cultivating change talk” family includes Persuade with Permission, Giving Information, and several reflective and question types that pull for client language favoring change. The “softening sustain talk” family includes the clinician moves that reduce or ignore client language against change. Every clinician utterance is assigned to one of these behaviors, which is why coder training takes considerable time.

The structure matters because it separates warmth from technique. A clinician can be deeply empathic and still fail to elicit change talk, and the MITI scores that gap directly rather than averaging it away.

Why Two Dimensions Beat a Single Fidelity Score

MI is defined by both its relational spirit and its technical method, and the MITI 4.2 was built to keep them distinct. The mechanism is straightforward: relational globals rate the clinician’s stance across the whole session, while behavior counts measure the frequency of discrete speech acts. Averaging the two would hide the most common failure pattern in practice, which is a warm clinician who argues for change.

The score logic follows from MI theory. The relational dimension predicts whether clients stay engaged and disclose. The technical dimension predicts whether clients talk themselves into change. When the technical scores are low, the session may feel pleasant but produce little client change talk, which is the proximal predictor of behavior change in the model.

The MITI 4.2 revision also refined the technical family. Earlier versions counted reflections and questions more generically. Version 4.2 narrowed the technical composites toward cultivating change talk and softening sustain talk, which sharpened the link between what coders tally and what MI theory says drives outcomes. Most published MI fidelity studies using the current code report on these composites rather than a single total.

This two-dimension design makes the tool useful for feedback, not just for research classification. A clinician can see whether the weakness sits in warmth or in technique, and coaching can target the weaker side.

Quantified Context: What the Scores Look Like in Practice

The MITI 4.2 uses explicit thresholds to classify an interview as proficient or competent. The commonly cited proficiency standard requires global scores of at least 4 on Cultivating Change Talk and 4 on Softening Sustain Talk, alongside minimum behavior-count ratios. These thresholds are the reference points most training programs use when they certify clinicians.

Behavior counts yield ratios that summarize technique. The reflection-to-question ratio compares the number of reflections to the number of questions, and higher values indicate a more listening-oriented style. The percent complex reflections measures the share of reflections that add meaning rather than simply repeat the client. The talk-time ratio estimates how much the clinician speaks relative to the client, and lower clinician talk time generally indicates better practice.

The proficiency benchmarks commonly cited in the MI literature are a reflection-to-question ratio of at least 1, ideally 2 or more, and roughly 50 percent complex reflections, with clinician talk time near or below half the session. Treat these as training targets rather than fixed laws. Different clinical populations and session lengths shift the numbers, and the MITI manual publishes the operative thresholds that should govern certification decisions.

Time and cost shape real-world use. Coding a single 20- to 30-minute session typically takes a trained coder roughly an hour, so a full training cohort generates substantial coder hours. Programs that code every trainee session quickly learn that sampling is necessary.

The Counterintuitive Detail: Empathy Alone Does Not Move the Needle

The most frequently misunderstood finding is that high Empathy and Partnership scores do not guarantee good technique. A clinician can earn a 5 on Empathy and still score poorly on Cultivating Change Talk, because the two dimensions measure different things. Warmth creates the conditions for change talk, but the clinician still has to elicit it.

This is why the MITI reports relational and technical composites separately rather than collapsing them. Programs that celebrate only the global scores can miss a clinician who listens beautifully and then lectures. The technical counts catch the lecture. Conversely, a clinician with strong counts but low Partnership may hit the numbers while damaging engagement.

A related caveat concerns the coding process itself. The MITI is a behavioral coding instrument, not a clinical judgment. High scores on the 16 items indicate that the observable behaviors of MI were present, not that the client improved. Fidelity and outcome are related but distinct, and studies sometimes find weak links between them depending on population and follow-up interval.

Comparison Table: MITI 4.2 Against Other Fidelity Tools

Tool Item structure Dimensions measured Best use
MITI 4.2 4 globals plus 12 behavior counts, 16 items Relational and technical Research fidelity, training certification
MISC Global ratings plus behavior counts across a broader session Client and clinician behavior Observing both parties in depth
VASE-R Written-response scoring MI-consistent responding Low-cost knowledge and skill screening
Behavior Change Counseling Index 12-item session rating Brief MI-consistent practice Routine clinical audit
Motivational Interviewing Skills Code Client language plus clinician behavior Change and sustain talk dynamics Mechanistic MI research

The MITI 4.2 is the only tool in this table purpose-built to score the relational and technical skill set with a fixed 16-item structure. Its closest research relative, the MISC, is richer but far heavier to administer. The lighter tools trade depth for practicality.

Edge Cases and Scoring Pitfalls

Several edge cases trip up even experienced coders. Very short sessions may not contain enough utterances to generate stable counts, and the manual addresses minimum coding requirements. Sessions dominated by assessment or crisis management may leave little room for MI-consistent behavior, so low scores can reflect context rather than clinician skill.

Cultural and linguistic factors complicate global ratings. Empathy and Partnership are inferred from tone, pacing, and phrasing, and coders may read these differently across cultural settings. Training programs increasingly include calibration exercises to reduce rater drift on the globals.

A specific red-flag checklist for anyone using MITI scores:

Treating a single coded session as a stable measure of clinician competence. Comparing scores across coder cohorts without checking inter-rater reliability. Reporting only global scores and omitting the behavior-count ratios. Using MITI thresholds as outcome measures rather than fidelity measures. Coding recordings with poor audio, which distorts counts of reflections and questions. Skipping recalibration, which lets coder drift silently inflate or deflate scores.

Inter-rater reliability is the load-bearing assumption of the whole system. When reliability is weak, the 16 items still produce numbers, but those numbers may not mean what the thresholds assume.

Practical Actions: Using the 16 Items Well

Start by deciding what question you need answered. If the goal is training certification, use the MITI 4.2 with the manual’s proficiency thresholds and sample at least two sessions per clinician. If the goal is routine quality improvement, consider a lighter tool and reserve the MITI for targeted review.

Before coding, confirm coder training. Coders should complete formal MITI training and demonstrate reliability against a gold-standard set before scoring independently. Recalibrate annually or whenever coder turnover occurs.

When interpreting scores, read the composites together, never in isolation. A strong relational composite with a weak technical composite signals a coaching target in eliciting and shaping change talk. A strong technical composite with weak relational scores signals a target in partnership and empathy. Use the behavior counts to specify which utterances to change.

Build feedback loops around the counts rather than the globals. Counts tell a clinician exactly what to do differently, such as raising the reflection-to-question ratio above 1 or increasing complex reflections toward half of all reflections. Globals tell a clinician how they came across, which is useful but harder to act on.

Limitations: What the 16 Items Cannot Tell You

The MITI 4.2 measures clinician behavior, not client outcomes, and it does not capture the full therapeutic relationship. It also depends on recording and coding infrastructure that many frontline settings cannot sustain. The cost in coder time is the main barrier to routine use, and sampling is the usual compromise.

The tool is also version-sensitive. Scores from MITI 3 and MITI 4.2 are not directly interchangeable, and comparing them without adjustment can mislead. Always report the version alongside the scores.

Finally, the thresholds are benchmarks, not diagnoses of competence. A clinician below threshold may still be effective with a particular client, and a clinician above threshold may still struggle with engagement. Use the MITI 4.2 as one input into supervision and training decisions, paired with direct observation and client feedback.

This article is general educational information about a research and training instrument. It is not a clinical protocol, a diagnosis, or a substitute for professional supervision. Clinicians should follow the current MITI manual and their own training standards, and readers with clinical concerns should consult a qualified local professional.