One Factor, Not Four: Validating the Modified Marijuana Craving Questionnaire in Spanish

Craving is one of the eleven DSM-5 criteria for cannabis use disorder and one of the few that maps directly onto a subjective state a clinician can track over a course of treatment. Yet the instruments for measuring it have a structural problem that this validation brings into focus. The original Marijuana Craving Questionnaire was built on a four-factor model – compulsivity, emotionality, expectancy, purposefulness – borrowed from tobacco and alcohol craving research. The short form inherited that architecture. The Chilean data show the architecture does not hold: the four factors, and even a reduced two-factor version, correlate so tightly that they describe one thing, not four.

For practice, the collapse is a feature rather than a defect. A unidimensional scale yields a single craving score that is simpler to administer, simpler to interpret, and simpler to chart across sessions. Clinicians monitoring a patient in cannabis-use treatment do not need four subscale profiles; they need a reliable number that moves when craving moves. The MMCQ supplies that, and the demonstrated invariance across sex and across community versus clinical settings means the same number means the same thing whether the patient is a help-seeking inpatient or a community participant, and whether they are a man or a woman.

The convergent pattern is clinically legible. Craving rising with cannabis withdrawal severity and with impaired control over use, and falling with later age of onset, is exactly the nomological network a craving measure should sit inside. None of these associations is large enough to make the MMCQ redundant with a withdrawal or control scale; it captures a related but distinct signal. That distinctness matters for treatment planning: a patient whose craving is high but whose impaired control is still modest is at a different stage than one in whom both have converged, and a scale that does not simply mirror its neighbours can mark that difference.

The Spanish-language scope is the quiet contribution. Validated cannabis-craving measures have been concentrated in English, and the authors note that no validation existed for Spanish-speaking populations. A scale that works in Chilean adults extends a usable craving metric to a large clinical population that previously had none – a reminder that measurement equity, not just treatment access, shapes who gets evidence-based care. It also makes cross-site comparison feasible within Spanish-speaking services that until now had to borrow English instruments and assume, rather than demonstrate, that they measured the same construct.

Two practical cautions apply. The validation rests on a single cross-sectional sample, so test-retest stability and sensitivity to change over treatment remain unestablished – and sensitivity to change is precisely what a craving monitor needs. The correlated residual in the final model signals item overlap that a future revision might prune. For now, the MMCQ is best read as a brief, structurally sound craving score for Spanish-speaking cannabis users, pending longitudinal data.

Where it fits in the assessment stack

The MMCQ belongs alongside withdrawal and impaired-control measures as a focused craving monitor, not a diagnostic instrument for cannabis use disorder. Its role is to quantify one DSM-5 criterion repeatedly and cheaply, giving the clinician a trend line rather than a single snapshot.

What still needs testing

Test-retest reliability, responsiveness to treatment, and predictive validity for relapse are all open. Cross-national invariance beyond Chile, and replication of the unidimensional structure in larger and more diverse Spanish-speaking samples, would strengthen the case before the scale is used to compare across countries.