Surprisingly, considering the popularity of motivation rulers in clinical practice and their cornerstone in MI-based interventions [1, 8], there are very few published studies on their psychometric properties. Our results provide general support for the validity of the three rulers, with the confidence ruler showing slightly better overall performance in predicting behavior change after adjusting for other potential confounding variables.
Construct validity of the rulers was demonstrated through the expected positive associations with stage of change at baseline, which reinforce similar patterns observed in adolescent smokers undergoing addiction treatment . Subjects in the preparation stage of change reported the highest importance, readiness, and confidence. In addition, predictive validity was supported by significant prediction of changes in smoking behavior in the two weeks after the index ED visit. The ability of the motivation rulers, independently and as a group, to predict smoking persisted in the multivariable analysis even after controlling for demographic variables and nicotine dependence. Moreover, the magnitude of this predictive ability was on par with stage of change in the fully adjusted models. This is a robust test of predictive validity, considering that both nicotine dependence and stage of change have historically been strong replicable predictors of smoking behavior [18–20]. In the final combined-ruler model, the confidence ruler appeared to have the strongest and most consistent relation with smoking behavior change. This supports the extant literature showing that self-efficacy, or confidence in one’s ability to change, is a preeminent predictor of change [21, 22].
Notably, those who attempted to quit but were unsuccessful and relapsed back to smoking looked remarkably similar in terms of motivation to those who had achieved seven-day abstinence. Both of these groups of changers perceived the importance of smoking cessation and their readiness to quit as significantly higher than those who continued to smoke, but relapsers were not markedly different from successful quitters across these two rulers (e.g., importance and readiness). A similar but slightly more complex trend was noted with confidence: although successful quitters endorsed stronger confidence in quitting than relapsers in the bivariable analysis, thus differentiating the two groups of changers, this effect was attenuated in the multivariable analyses. Only nicotine dependence remained an independent predictor, able to differentiate between those quitters who relapsed back to smoking and those that achieved successful change. This same pattern held true for stage of change. Stage was able to distinguish continuous smokers from both groups of changers but did not differentiate relapsers from successful changers.
As a whole, these data suggest measures of motivation are much better at predicting who will initiate change than they are at predicting transition to successful change. This pattern is consistent with the health-behavior change literature in general, which has prompted recent calls for reformulating traditional health-behavior theories to more proactively distinguish between predictors of behavioral initiation from predictors of behavioral maintenance [23–25]. This same recommendation has been echoed for conceptual model building to study health-behavior change in acute medical settings . Our results suggest that motivational readiness may predict who tries to change, but nicotine dependence predicts who will relapse back to smoking.
In our exploratory analyses, we found that greater nicotine dependence was associated with lower importance, readiness, and confidence ratings—a pattern already observed in the literature [13, 14]. It is difficult to know exactly why this association exists. However, some theorists have appealed to cognitive dissonance theory to explain it . Individuals who smoke heavily and are, therefore, strongly dependent on nicotine, and who rate their motivation to change as high, are in a dissonant state. Their behavior, heavy smoking, conflicts with their perceptions, that cessation is important. It is difficult to maintain such a dissonant state for long; one or the other must change. Either motivation wins out, and the individual reduces his or her smoking and, therefore, becomes less dependent, or the individual continues to smoke at a high rate and devalues the importance of cessation. The collective effect is to produce a negative correlation between dependence and motivation. The only way to truly test the dissonance reduction hypotheses is through longitudinal or experimental study designs.
The data were collected in the ED setting, and, consequently, the results should be generalized to other settings with caution. Additional work replicating our results across other settings is needed. The sample sizes for the relapsers and successful quitters were small, making it difficult to detect differences between the predictors, like the motivation rulers. This may have obscured actual differences (i.e., Type II error). Further, the rulers we used were anchored by “0” and “10.” This differs from rulers sometimes employed, which can be anchored by “1.” This difference is subtle and seems unlikely to exert a powerful influence on results or interpretations. Nevertheless, further inquiry into how different rulers perform may be warranted. We used the transtheoretical model’s stages of change to establish construct validity. Although widely used and studied more than any other motivation measure, it is nevertheless controversial, with some scholars suggesting the construct is invalid . Interestingly, one of the main arguments against the stages of change is that motivation is likely to be on a continuum rather than threshold- or stage-based. Rulers and scales that are measured in a more continuous manner are often appealed to as a means of addressing this very limitation. Finally, the effect sizes of some of the association, like the correlations between the readiness rulers and nicotine dependence, were small, prompting caution when interpreting the strength of the results. This limitation is partially mitigated by the fact that, although small, they were generally on par with the effects sizes found in the extant literature on predictors of cessation [19, 20, 29].