Computer-Based Motivational Interviews for Cannabis Showed Promise but Were Less Effective Than Face-to-Face | Research

The takeaway

Both computer-based and face-to-face motivational interviews generated "change talk" about marijuana use, but only the face-to-face format showed a trend toward actually reducing use at two-month follow-up.

Researchers developing digital interventions for cannabis use and clinicians using motivational interviewing.

Both formats generated similar change talk, but only face-to-face showed a trend toward reduced marijuana use

What the researchers found

Both interview formats generated similar amounts of change talk and sustain talk after adjusting for verbosity. However, the relationship between change talk and actual behavior change differed by format: in face-to-face interviews, stronger change talk trended toward reduced marijuana use at 2 months (though not significant, p=0.08), while in the computer format, change talk did not predict behavior change.

Why it matters

Scaling up motivational interviewing through computer platforms could reach more cannabis users. This study suggests the format shows promise for generating the right kind of conversation, but translating that into behavior change may require the human connection of face-to-face interaction.

The numbers in context

150 participants (frequent users, occasional users, non-users). Face-to-face produced significantly more words (p<0.001). After controlling for verbosity, change talk was similar (p=0.47). Face-to-face showed trend toward use reduction with stronger change talk (p=0.08). Computer format: no relationship between change talk and use (p=0.16).

How the study worked

Randomized controlled trial with 150 marijuana users and ambivalent non-users assigned to face-to-face or computer-mediated motivational-type interviews. Change talk was coded using Amrhein's manual. Marijuana use was assessed at 2-month follow-up.

What this study cannot tell us

Small sample size (150) spread across user types. Two-month follow-up is short. Self-reported marijuana use. The motivational interview format was adapted, not standard MI. Computer-mediated format may improve with technological advances.

How to read the evidence

Preliminary: small RCT with short follow-up and non-significant primary outcome trends, though rigorous language coding methodology.

When this study was published

2025 study.

The bigger picture

Digital health interventions for substance use are expanding rapidly. This study provides important nuance: computers can facilitate meaningful therapeutic conversations, but the mechanism by which those conversations change behavior may require elements unique to human interaction.

Questions still open

Would video-based computer interviews bridge the gap between text-based and face-to-face? Could AI-powered conversational agents improve on static computer interviews? What elements of face-to-face interaction drive behavior change?

Common questions

Could a computer program help someone quit marijuana?

Computer-based interviews generated meaningful conversations about change, but in this study, only face-to-face interviews showed a trend toward actually reducing use. More research is needed to improve digital formats.

What is change talk?

Language that expresses desire, ability, reasons, or commitment to change behavior. In motivational interviewing, stronger change talk typically predicts actual behavior change.

Read the original research

Impact of Computer-Mediated Versus Face-to-Face Motivational-Type Interviews on Participants' Language and Subsequent Cannabis Use: Randomized Controlled Trial.

Journal of medical Internet research, 27, e59085

View on PubMed ↗Publisher’s paper ↗

Citation

Llanes, Karla D; Amastae, Jon; Amrhein, Paul C; Lisha, Nadra; Arteaga, Katherina; Lopez, Eugene; Moran, Roberto A; Cohn, Lawrence D. (2025). Impact of Computer-Mediated Versus Face-to-Face Motivational-Type Interviews on Participants' Language and Subsequent Cannabis Use: Randomized Controlled Trial.. Journal of medical Internet research, 27, e59085. https://doi.org/10.2196/59085