Empirical Testing

Study: Meta-analyses of cross-sectional literature of TTM application across health problems to examine relationship between stages and processes of change (Rosen, 2000).

Research Questions

  • How much does use of change processes vary across stages and is this a function of study design, publication status or type of health problem?
  • Is the trajectory of change processes consistent across different health problems?
  • Which of the 10 change processes vary most and least strongly across stages and does this differ by type of health problems?
  • Searches from PsychLit, Medline, Dissertation Abstracts international, National Institute of Alcohol Abuse and Alcoholism Science Data base (ETOH) were done between 1980 to 1999 to retain 47 studies.
    • 10 smoking cessation, 7 substance abuse, 13 exercise, 5 diet change, 6 psychiatric disorders or counseling and 6 different health problems were assessed.


  • Use of behavioral processes increased linearly for most health problems.
  • Between action and maintenance, use of behavioral processes increased slightly or remained constant except for smoking, where it decreased.
  • Behavioral processes peaked during action and maintenance stages for all health problems.
  • Cognitive-affective processes peaked during contemplation or preparation for less than 50% of all studies and was not associated with any stage in other studies.
  • Self-liberation and counter-conditioning, both used most often in action or maintenance stage in >84% of studies, varied the most by stage with large effect sizes (d= 1.0)
  • Self-reevaluation, consciousness-raising, reinforcement management and stimulus control varied moderately to strongly by stage, while the effect sizes varied by health problems.
  • Helping relationships, social liberation, environmental reevaluation and dramatic relief varied moderately across stages.

Results Summary

  • On average, 11% of the variance in cognitive-affective processes and 14% of the variance in behavioral processes were explained by stage assignment.
  • No single trajectory of processes of change was established for all health behaviors – process of change is not generalizable across health behaviors.
  • Contrast between precomtemplation and all other stages (combined) accounted for 70% of between-stage variance in use of experential processes and 50% of between-stage variance for behavioral processes.

Study: Meta-analysis of findings from empirical applications of TTM on physical activity and exercise-related behavior change (Marshall, 2001).


  • MEDLINE, PsycLIT, Sports Discus, and UnCover were searched from 1983 to 2000.
  • 1998-2000 issues of several relevant journals were searched.
  • Ninety-One (91) independent samples from 71 published reports (61 articles, 10 conference abstracts) were retained.
  • The study designs varied: 54 cross-sectional designs, 6 longitudinal, 10 quasi-experimental and 1 randomized controlled trial.
    • Across all studies (N=68,580), 14% of individuals were in precontemplation,16% in contemplation, 23% in preparation, 11% in action and 36% in maintenance.
    • Stage assignment varied by country of study (US, UK, Canada, Australia), age, sampling method (random vs. nonrandom), recruiting method (active vs. passive), definition of regular physical activity (frequency vs. duration vs. intensity), response format of surveys (Likert measures vs. fixed format measures).
      • As individuals moved across stages of change, their levels of physical activity increased.
      • The largest effect was seen in preparation stage, where individuals meet established criterion for physical activity.
      • Small to moderate effect size increases were seen between precontemplation and contemplation stages - moving between "inactive" stages also produce increase in physical activity.
      • There was effect moderation between three of the four stage transitions.
      • Positive and significant effect estimates were noted for self-efficacy across all stage transitions - confidence increases with each stage of change.
      • Self-efficacy increased across all stage transitions in a non-linear fashion with small to moderate (Contemplation to Preparation), moderate (Precontemplation to Contemplation) and moderate to large (Action to Maintenance).
      • Perceived benefits of change increase while perceived disadvantages decrease with each stage transition.
      • Behavioral pros - Precontemplation to Contemplation had the largest and most robust effect size while Contemplation to Preparation had the smallest and least robust effect sizes.
      • Behavioral cons - Precontemplation to Contemplation had the most pronounced decline while the smallest decline was during the Action to Maintenance stage.
      • Precontemplation to Contemplation had the largest effect size, followed by Preparation to Action
      • Across all processes and stage transitions, self- liberation from Precontemplation to Contemplation had the largest single effect while the smallest single effect was seen in self-evaluation from Action to Maintenance stage.
      • From Contemplation to Action and from Action to Maintenance, Counterconditioning had the largest effect size.
      • From Preparation to Action, self-liberation has the largest effect.
      • Similar to epidemiologic findings of physical inactivity, 30% of the study sample were inactive (Precontemplation or Contemplation).
      • While current projections for those meeting physical activity requirements are at 22%, this study reported 50% of its sample in Action or Maintenance stage.
      • Stage distribution is associated with sampling and recruitment methods.
      • Unstandardized measures of physical activity due to differences in classification based on intensity, duration and frequency may introduce bias into stage classification. Appropriate use of TTM necessitates standardized stage measures with a consistent response format and validity and reliability data specific to the target population.
      • There isn't a clear distinction between behavioral and experiential processes in physical activity as reported in smoking cessation. The correlation between constructs from both processes was 0.91, implying a large shared variance.
      • Cross-sectional studies have reported associations between stages and processes. Future research efforts should be on implementing longitudinal studies that examine the mediators and moderators of change transition.
      • Given the small effect size seen between Precontemplation and Contemplation, care must be taken to differentiate between individuals in both stages and not to misclassify individuals as similarly "inactive."
      • The differential magnitude of effect size between stage transitions suggests discontinuity of self-efficacy across stages, opposing results presented in narrative reports.
      • Importance of timing of decisional balance is unclear, i.e. focusing on the point at which behavioral pros outweighs behavioral cons isn't as relevant as the magnitude of pros and cons associated with behavior change.