Original Research

Relationship Among Practice Change, Motivation, andSelf-Efficacy

BETSY W. WILLIAMS, PHD, MPH; HAROLD A. KESSLER, MD; MICHAEL V. WILLIAMS, PHD

Introduction: The relationship between an individual’s sense of self-efficacy, motivation to change, and the imple-mentation of improvement programs has been reported. This research reports the relationship among self-efficacy,motivation to change, and intent to implement continuing medical education (CME) activity learnings.

Methods: The measure of individual sense of self-efficacy was a 4-item scale. The measure of motivation wasa 4-item scale following on the work of Johnson, et al. The self-efficacy scale has been confirmed for structure,and together the 2 scales provide indicators of 3 underlying variables—2 self-efficacy constructs and a motivationvariable. In addition, a global intent to implement measure was collected.

Results: Preliminary analysis demonstrates a significant relationship between a self-efficacy construct, the mo-tivation to change construct, and global intent to change. Specifically, the sense of efficacy in effecting changein the practice environment is predictive of a high level of motivation to change, which, in turn, is predictive offormation of an intent to change practice patterns.

Discussion: Further inspection of the motivation to change construct suggests that it mediates the self-efficacyconstructs’ effect on intent. This is consistent with an earlier report on the relationship among self-efficacy, barriersto change, and stated intent. This new finding suggests that the proximal construct motivation completely masksan important underlying causal relationship that appears to contribute to practice change following CME: self-efficacy. A focus on the participants’ sense of self-agency may provide a path to practice change.

Key Words: self-efficacy, theory-planned behavior, theory-social learning, commitment to change

Introduction

Continuing medical education (CME) has served as the basisfor most postgraduate medical education among practition-ers. Reviews of the literature on the effectiveness of CME in-terventions have found differences linked to the type of learn-

Disclosures: The authors report none.

This paper was presented at the 39th Annual Meeting of the Alliance forContinuing Education in the Health Professions held in Orlando, Florida,January 15–18, 2014.

Dr. B. Williams: Assistant Professor, Department Behavioral Sci-ences/Clinical Program Director, Rush University Medical Center;Dr. Kessler: Professor of Medicine and Immunology/Microbiology, As-sociate Dean, Postgraduate Medical Education, Rush University MedicalCenter; Dr. M. Williams: Principal, Wales Behavioral Assessment.

Correspondence: Betsy W. Williams, Rush University Medical Cen-ter/Professional Renewal Center, 1421 Research Park Dr., Ste 3B, Lawrence,KS 66049; e-mail: bwilliams@prckansas.org.

© 2014 The Alliance for Continuing Education in the Health Professions,the Society for Academic Continuing Medical Education, and the Councilon Continuing Medical Education, Association for Hospital Medical Educa-tion. • Published online in Wiley Online Library (wileyonlinelibrary.com).DOI: 10.1002/chp.21235

ing activity (eg, didactic, audit, and feedback).1 Interestingly,they have also found substantial variation in effectiveness instudies of the same type of learning activity applied in dif-ferent contexts.2 This suggests that other variables are influ-encing effectiveness.

There has been a significant amount of work on the con-struct of motivation as an intervening variable between learn-ing and practice change. In addition, issues like commitmentto change3 and barriers to implementation4 have been identi-fied as potential intervening variables between learning andimplementation of practice. Other issues such as learningstyle5 as well as teaching style6 have also been reviewed.However, we need a better understanding of these and othervariables that contribute to changing behavior. Also impor-tant is an understanding of the relationships among thosevariables.

Many theoretical perspectives have informed medical ed-ucation, including learning theory, cognitive psychology, andhumanistic and social cognitive theory.7 Of the theoreticalmodels that have been employed, particularly relevant arethe social cognition models.8,9 These models underlie muchof health psychology and change theories. One such theoryis the Transtheoretical Stages of Change Theory. The model

JOURNAL OF CONTINUING EDUCATION IN THE HEALTH PROFESSIONS, 34(S1):S5–S10, 2014

Williams et al.

FIGURE 1. Working Model Used by the Team to Guide the Research.

has been applied in a variety of settings and has been the ba-sis for developing effective interventions to promote healthbehavior change, including addiction treatment.10–12

Social cognitive learning theories of behavioral changeposit that our actions, learning, and functioning are the resultsof a continuous, dynamic, reciprocal interaction among threesets of determinants: personal, environmental (situational),and behavioral.13 These models have been particularly ef-fective in predicting behavioral change where the choice isdifficult either because of the emotional content or becausethe behavior requires the compliance of others.14

One such model is the Health Action Process Approachmodel of Schwarzer.15 The authors have been using thismodel connected to the Theory of Reasoned Action modelby Fishbein and Ajzen16 as a general guide for investigat-ing these potential intervening variables (see FIGURE 1). Inprior work, the authors have found significant effects of mo-tivation, self-efficacy, and barriers to implementation, amongothers in the formulation of intent to change and/or practicechange.17 Theories such as the Theory of Reasoned Action,16

Social Cognitive Theory,18 and, as mentioned, the Health Ac-tion Process Approach (HAPA)19 all explore the interveningvariables between knowledge and action. They confirm theimportance of intent to change as well as self-efficacy in un-derstanding behavior change.

The current study focuses on the relationship among threevariables in the middle of such a model. The first variable, re-ported by many as an effective intervening variable betweenknowledge and practice change, is motivation to change (see,for example, Fox and Miner20 and Reed, Schifferdecker, andTurco21). The second variable, one we and others have re-ported as an effective mediator, is self-efficacy.17,22 The vari-able employed as the outcome measure is intent to changepractice, as this has been reported to be a good measure ofpractice change.23 Each of these variables has its own litera-ture as related to CME.

Methods

Motivation to change is one of the more studied interven-ing cognitive variables in the CME literature. There are

many models and theories that focus on this, including theTranstheoretical or Stages of Change Model.11 This modelconsists of 4 core constructs, including stages of changeas well as processes of change. In essence, this modelpostulates that motivation is not binary but rather a cre-ated internal state. It is viewed as ranging from precon-templative to a state of maintenance where the behaviorhas occurred. A number of authors have investigated theutility of this model in the CME literature. The recent arti-cle by Johnson and colleagues24 utilized secondary behav-ioral measures as indicators of the individual learner’s mo-tivational state within the transtheoretical construct. Theyfound that the further along the motivational progressionan individual was, the more likely change was made andthen maintained. A related area of research is commitmentto change with a number of studies demonstrating the use-fulness of commitment to change in promoting behavioralchanges.25–27

Self-efficacy as an intervening cognitive state betweenknowledge acquisition and behavior has a long historyamong learning theorists. While self-efficacy is part of theTranstheoretical Model of Change, Bandura and colleaguesare preeminent among those who have studied the impor-tance of this construct.8 It is less studied within the CMEliterature.22 The authors have employed this construct in sev-eral studies and have found that the variable appears to berobust.28

While any of the variables that have been demonstrated inother settings to measure processes that are active betweenlearning or knowledge acquisition and behavioral changemerit study, the authors were particularly interested in mo-tivation and self-efficacy. It should be noted that this is onestudy in a series of studies that will look at other relationshipsin this causal chain. The current study specifically focuses onthe relationship between motivation and self-efficacy, and thevariable intent to change, which is theoretically more proxi-mal to practice change.

Evaluation of many of these intervening variables hasbeen incorporated into the standard evaluation process usedas part of CME programming accredited through Rush Uni-versity Medical Center (RUMC). The data were collectedfrom a single didactic RUMC CME activity focused on hu-man immunodeficiency virus (HIV). Consistent with ourCME practices, participants in the activity were surveyedprior to and after completion of the activity. These resultsare 1 element of a larger evaluation of activity.

Scales

The measure of an individual’s sense of self-efficacy wasa 4-item scale explicitly built for CME/continuing profes-sional development (CPD). This scale was a heavily mod-ified version of the Perceived Competence Scale of Smith

S6 JOURNAL OF CONTINUING EDUCATION IN THE HEALTH PROFESSIONS—34(S1), 2014DOI: 10.1002/chp

Practice Change, Motivation, and Self-Efficacy

and colleagues.29 The scale was entered into a factor analy-sis, and each individual in the study was assigned a score oneach of the 2 factors identified. The scale was confirmed asto structure.17

The measure of motivation was also a 4-item scale andwas derived from measures used by Rush in its CME/CPDprograms. The approach used and items selected followedon the work of Johnson and colleagues.24 A set of behavioralcorrelates of implementation were taken as markers of gen-eral motivation. Intent to change was measured as a single,global, Likert-scaled item.

Approach

Two sequential sets of analyses were undertaken. The firstwas a series of regression studies that employed JMP(Version 9, SAS Institute, Cary, NC). The purpose of theseanalyses was to determine the existence of relationshipsamong the 3 constructs motivation to change, self-efficacy,and intent to change. A second element of these analyses wasa series of principal-components factor analyses to extractthe core measures of the constructs motivation to change andself-efficacy.

The direct analysis then served as the guide to the sec-ond set of simultaneous analyses to determine the relation-ships among the variables. As the independent variables werenot directly observed and likely intercorrelated, the analyseswere undertaken as a series of structural equation estima-tions. The focus of these estimations was not to confirm thescales but rather to observe the interrelationships betweenthe 2 constructs self-efficacy and motivation to change andthe directly assessed construct intent. We have previouslyconfirmed structure and reported this elsewhere.17 Also, theintent was not to produce a generalized measurement toolbut effectively employ extant measures. The second set ofanalyses was undertaken as a series of path analytic studiesemploying MPlus 7 (Muthen & Muthen, Los Angeles, CA).The purpose of these analyses was to determine the nature ofthe pattern of effects among the 3 constructs.

Results

These data reflect the responses of 51 participants in a CMEactivity accredited by Rush University Medical Center. Theparticipants were a subgroup of 123 participants in the activ-ity. These participants returned both a preparticipation and apostparticipation questionnaire. They were 34% female and66% male, with a mean age of 45.5 years.

Regression and Factor Analytic Results

Intent to change was regressed on the individual-level vari-ables of motivation to change and self-efficacy. The relation-

TABLE 1. Rotated Factor Loadings for Self-Efficacy

Variable Factor 1 Factor 2

I find that it is difficult to translate information

from scientific meetings to direct patient care.

−0.576868 0.128469

I succeed in changing patient regimens and my

clinical practice according to the latest

available data.

0.680898 0.161045

Typically, changing patient regimens has not

been as successful as I would like.

−0.050425 0.630329

I am as able to change my practice patterns in

response to new data as my colleagues.

0.746651 −0.329263

ship was found to not meet traditional levels of significance(p > .05, F = 1.55, df = 8/29). The relationship of interest in-volves the relationship of the constructs motivation to changeand self-efficacy to the formulated intent. To follow this con-struct further, estimates of the construct using the individualvariables as indicators were undertaken.

Four indicators of self-efficacy were employed. Theywere subjected to a principal components factor analysis.Two components were extracted with eigenvalues greaterthan 1. The first factor had primary loadings from 3 variables,the second from 1. TABLE 1 presents the factor loadings. Foridentification purposes, factor 1 was labeled P efficacy or Per-sonal Efficacy, and the second factor was labeled R efficacyor Results Efficacy. Cronbach’s alpha was 0.4, as would beexpected with a multiple factor solution; together, the 2 fac-tors accounted for 74% of the variance.

Respondents were scored based on the factor scoring co-efficients. Intent to change was regressed on their scores forboth Pefficacy and Refficacy. The regression was significant (p <

.05, F = 3.67, df 2/29). Further inspection of the regressionfound the relationship with Pefficacy to be significant (p < .05,t = 2.37), while the relationship with Refficacy did not meet thetraditional level of significance.

Four indicators of motivation to change were also em-ployed. They demonstrated a high intercorrelation (Cron-bach’s alpha = 91). They were subjected to a principal com-ponents factor analysis. One component was extracted withan eigenvalue greater than 1, consistent with 𝛼. The primaryfactor loadings are presented in TABLE 2. Respondent scoreswere generated using the factor scoring coefficients. Intent tochange was regressed on their scores for motivation. The re-gression was significant (p < .05, F = 5.78, df 1/33).

To further explore the relationships, intent to changewas then regressed on Pefficacy, Refficacy, and motivation tochange. The regression was significant (p < .05, F = 3.17,

JOURNAL OF CONTINUING EDUCATION IN THE HEALTH PROFESSIONS—34(S1), 2014 S7DOI: 10.1002/chp

Williams et al.

TABLE 2. Motivation Factor Loadings

Variable Factor 1

1. I frequently employ the clinical data from the

International AIDS Conference on the potential benefits

of antiretroviral (ARV) versus potential risks in selecting

ARV therapy for ARV-naıve patients when currently

treating my patients.

0.8756374

2. I frequently use clinical data presented from the

International AIDS Conference on efficacy, avoidance of

toxicity, and adverse events in selecting ARV therapy for

ARV-naıve patients when currently treating my patients.

0.9974074

3. I frequently use study results regarding the medical

management of HIV-positive patients from the

International AIDS Conference in my clinical practice.

0.7686701

4. I frequently use data presented from the International

AIDS Conference on viral hepatitis in my clinical

practice.

0.7376513

df 3/29). When tested individually, no single coefficient wassignificant.

The composite variable, motivation to change, was thenregressed on Pefficacy and Refficacy variables. The relationshipwas significant (p < .001, F = 5.75, df 2/35). Individual testsrevealed that the relationship between motivation to changeand the individual variable Pefficacy was significant (p < .01,t = 3.24), while the relationship between motivation tochange and the variable Refficacy failed to meet traditional lev-els of significance.

Path Analytic Results

Based on these results, 3 alternative models were tested aspotential causal paths. The 3 alternatives explored were pathsfrom motivation to change and Pefficacy to intent to change,paths from Pefficacy to motivation to change and motivationto change to intent to change, and paths from motivation tochange to Pefficacy and Pefficacy to intent to change.

The models did replicate the findings of the traditionalfactor analytic studies as to structure, with the 4 motivationto change indicators reflecting a latent variable and 3 of the4 self-efficacy variables indicating a latent variable consis-tent with Pefficacy. The issues of interest were the potentiallycausal relationships and their patterns.

The most saturated model with a path estimated directlyfrom each latent variable, motivation to change and self-efficacy, to the explicit variable, intent to change, yielded re-sults with a significant intercorrelation between self-efficacy

FIGURE 2. Final Path Analytic Solution Showing the Direction of the Re-lationships.

and motivation to change, but the path coefficients betweenthese latent variables and intent to change was nonsignificant(p < .05, chi-square 33.9, df 18).

The second model relaxed the requirement of a path be-tween self-efficacy and intent to change while maintainingthe path between motivation and intent to change. This modelalso estimated a causal path between self-efficacy and moti-vation to change (see FIGURE 2). The results in this casefound significant path coefficients between motivation tochange and intent to change as well as between self-efficacyand motivation to change (p < .01, chi-square 36.3, df 19).

The third model relaxed the requirement of a path betweenmotivation to change and intent to change while maintain-ing the path between self-efficacy and intent to change. Thismodel also estimated a causal path between motivation tochange and self-efficacy. The results in this case found animprovement in chi-square over the first model; however, theimprovement was less than in the second model. In addition,the path between self-efficacy and intent to change was notfound to be significant in this model (p< .05, chi-square 34.2,df 19).

Discussion

The results indicated that a measure of motivation to changeshowed a significant relationship to intent to change prac-tice. This finding is consistent with the Theory of PlannedBehavior, which states that attitude toward behavior, subjec-tive norms, and perceived behavioral control, together shapean individual’s behavioral intentions and behaviors.16 Therelationships between attitudes and subjective norms to be-havioral intention, and subsequently to behavior, has beenconfirmed in many studies.30

S8 JOURNAL OF CONTINUING EDUCATION IN THE HEALTH PROFESSIONS—34(S1), 2014DOI: 10.1002/chp

Practice Change, Motivation, and Self-Efficacy

This finding is also consistent with the broader health andCME literatures demonstrating the value of commitment tochange in promoting behavioral/practice changes. It shouldbe noted that the dependent variable in this study was intentto change, not commitment to change. It is possible, evenlikely, that if specifically modeled, commitment to changehas elements of both the construct motivation and intent tochange. The commitment to change literature predicated thatmotivation is clearly related to the formulation of intent.

The analyses of the raw data did not find a significantrelationship between the individual items and the criterionvariable. This indicates that the relationship is not a simpleone where the measures themselves are direct measures ofthe variables of interest. This finding, taken together with thelater findings from the abstracted measures of the unobservedvariables, is support for the existence of the unobserved vari-ables as active intervening processes that change the inten-tion of the learner.

This work extends these findings in that we used an indi-rect behavioral indicator of motivation to change. The scaleis an extension of the work by Johnson et al. 2012.24 Thisabstracted, unobserved variable evinced a strong correlationto stated intent to change. The approach used in the cur-rent study has the advantage that the behavioral indicator isclearly distinct from the criterion variable intent to change.The variables chosen by Johnson et al.,24 while perhaps moreface valid, might, in some instances, be viewed as alternativewording of the same construct rather than the separate vari-able motivation to change if intent to change is used as thecriterion.

Self-efficacy relates to one’s beliefs in the capability toorganize and execute the courses of action required to pro-duce given attainments. It has an extensive literature withinthe broader field of learning. Bandura’s work8,9 might be saidto provide one of the bases of the motivation-to-change lit-erature. In the current study, self-efficacy, using a measureheavily based on Smith and colleagues,29 was found to bean effective second-order variable. It appears to be active be-tween the acquisition of learning and the generation of in-tent to change practice.17 This relationship, however, is me-diated by motivation to change. This finding is also consis-tent with the prior literature. A number of theorists, amongthem Fishbein,31 Schwarzer,15 Wallston,32 and others haveargued for a learning model that has a sequence of cognitiveprocesses beginning in knowledge acquisition and ending inbehavioral change.

This finding adds to earlier findings of these authors onthe relationship between self-efficacy and perceived barriersto implementation.17 In that case, the authors found a sig-nificant relationship between self-efficacy and the degree towhich individuals reported barriers, and then between barri-ers and reported intent to change. Consistent with Mazma-nian et al.,33(p886) information about barriers alone does not

Lessons for Practice

● Evaluating the relationship among motiva-tion, barriers to change, and self-efficacycan provide valuable information about el-ements of CME activity design.

● Consistent with results of Mazmanian, ourfindings suggest that information aboutbarriers to change alone may not be suf-ficient to foster successful change.

● Methods of addressing activity partici-pants’ sense of self-efficacy have the po-tential for improving the efficacy of CME.

increase the likelihood of implementation of practice change:“Successful change may depend less on the information as-sociated with clinical effectiveness and barriers than on otherfactors that influence physicians’ performance.”

There are a number of limitations to the current study.First, results are based on a relatively small group of indi-viduals. It is possible that these findings are idiosyncraticto the nature of the activity and/or the individuals withinthe study. Perhaps more important, the finding introduces anew method of assessing the degree of motivation to change.As previously noted, this method has both strengths andweaknesses. It remains to be seen whether others are ableto replicate these findings. Finally, while intent to change isan indicator of practice change, it is not the same as a directmeasurement of practice change. The degree to which thesefindings are themselves efficiently related to practice changeremains to be tested.

These concerns, nonetheless, do not diminish the findingthat in two studies self-efficacy, a well-researched variablewithin the field of cognitive models of learning, has beenshown to be a powerful second-order variable as it relatesto the formulation of intent to change practice subsequentto a CME activity. These findings are noteworthy not onlybecause the study demonstrates that self-efficacy is an im-portant variable, but also because it demonstrates the limita-tion of a standard regression analysis where there is an or-dered path of cause. The proximal variable tends to absorball the explanatory variance, while secondary variables mayshow up as having insignificant coefficients, thus hiding a po-tentially critical variable in understanding the cognitive pro-cesses involved. Thus, for example, one may believe that thelimitation in the efficacy of the CME activity is the motiva-tion of the learner, whereas the true limitation may be thatlearners do not view themselves as being effective agents ofchange. These findings suggest that methods of addressing

JOURNAL OF CONTINUING EDUCATION IN THE HEALTH PROFESSIONS—34(S1), 2014 S9DOI: 10.1002/chp

Williams et al.

activity participants’ sense of self-efficacy may well be im-portant in improving the efficacy of CME.

References

1. Davis DA, O’Brien MA, Freemantle N, Wolf FM, Mazmanian P,Taylor-Vaisey A. Impact of formal continuing medical education: doconferences, workshops, rounds, and other traditional continuing edu-cation activities change physician behavior or health care outcomes?JAMA. 1999;282(9):867–874.

2. Marinopoulos SS, Dorman T, Ratanawongsa N, et al. Effectivenessof continuing medical education. Evid Rep Technol Assess (Full Rep).2007(149):1–69.

3. Wakefield J, Herbert CP, Maclure M, et al. Commitment to change state-ments can predict actual change in practice. J Contin Educ Health Prof.2003;23(2):81–93.

4. Price DW, Miller EK, Rahm AK, Brace NE, Larson RS. Assessment ofbarriers to changing practice as CME outcomes. J Contin Educ HealthProf. 2010;30(4):237–245.

5. Armstrong E, Parsa-Parsi R. How can physicians’ learning styles driveeducational planning? AcadMed. 2005;80(7):680–684.

6. Whitman N. Learning and teaching styles: implications for teachers offamily medicine. Fam Med. 1996;28(5):321.

7. Mann KV. Theoretical perspectives in medical education: past experi-ence and future possibilities. Med Educ. 2011;45(1):60–68.

8. Bandura A. Self-efficacy: toward a unifying theory of behavioralchange. Psychol Rev. 1977;84(2):191–215.

9. Bandura A. Human agency in social cognitive theory. Am Psychol.1989;44(9):1175.

10. Prochaska JO. Why do we behave the way we do? Can J Cardiol.1995;11:20A.

11. Prochaska JO, Velicer WF. The transtheoretical model of health behav-ior change. Am J Health Promot. 1997;12(1):38–48.

12. Russell B, Maher G, Prochaska JO, Johnson SS. Strategic approachesto continuing medical education: applying the transtheoretical model &diffusion of innovation theory. CE Measure. 2012;6(3):27–31.

13. Kaufman DM, Mann KV. Teaching and learning in medical educa-tion: how theory can inform practice. In: Swanwick T, ed. Understand-ing Medical Education: Evidence, Theory and Practice. Hoboken, NJ:Wiley; 2010:16–36.

14. Rosenstock IM, Strecher VJ, Becker MH. Social learning theory andthe health belief model. Health Educ Behav. 1988;15(2):175–183.

15. Schwarzer R. Self-regulatory processes in the adoption and mainte-nance of health behaviors. J Health Psychol. 1999;4(2):115–127.

16. Fishbein M, Ajzen I. Belief, Attitude, Intention and Behavior: An In-troduction to Theory and Research. New York, NY: Addison-Wesley;1975.

17. Williams B, Kessler HA. Self efficacy and structural barriersas mediators of practice change. Society for Academic Contin-

uing Medical Education, Spring Meeting; April 2013; Madison,WI.

18. Bandura A. Social foundations of thought and action. Englewood Cliffs,NJ: Prentice Hall; 1986.

19. Schwarzer R. Modeling health behavior change: how to predict andmodify the adoption and maintenance of health behaviors. Appl Psy-chol. 2008;57(1):1–29.

20. Fox RD, Miner C. Motivation and the facilitation of change, learning,and participation in educational programs for health professionals. JContin Educ Health Prof. 1999;19(3):132–141.

21. Reed VA, Schifferdecker KE, Turco MG. Motivating learning and as-sessing outcomes in continuing medical education using a personallearning plan. J Contin Educ Health Prof. 2012;32(4):287–294.

22. Peterson E. Measures of perceived self-efficacy as a method of evaluat-ing educational outcomes: an introduction. CE Measure. 2006;1(1):35–39.

23. Goulet F, Jacques A, Gagnon R, Boule R, Girard G, Frenette J. CMEworkshop in recognizing boundary limits during a medical visit. FamMed. 2006;38(8):570–576.

24. Johnson SS, Cummins C, Paiva A, Brown JJ. Measuring effectivenessof continuing medical education using the transtheoretical model of be-havior change. CE Measure. 2012;6(3):32–40.

25. Mazmanian PE, Daffron SR, Johnson RE, Davis DA, Kantrowitz MP.Information about barriers to planned change: a randomized controlledtrial involving continuing medical education lectures and commitmentto change. Acad Med. 1998;73(8):882–886.

26. Lockyer JM, Fidler H, Ward R, Basson RJ, Elliott S, Toews J. Commit-ment to change statements: a way of understanding how participants useinformation and skills taught in an educational session. J Contin EducHealth Prof. 2001;21:82–89.

27. White MI, Grzybowski S, Broudo M. Commitment to change instru-ment enhances program planning, implementation, and evaluation. JContin Educ Health Prof. 2004; 24:153–162.

28. Williams BW, Kessler HA, Saag M, Williams MV. Causal relationshipsin practice change subsequent to a CME activity. SACME Spring Meet-ing; 2008; Vancouver, BC.

29. Smith CA, Dobbins CJ, Wallston KA. The mediational role of perceivedcompetence in psychological adjustment to rheumatoid arthritis 1. JAppl Soc Psychol. 1991;21:1218–1244.

30. Sheppard BH, Hartwick J, Warshaw PR. The theory of reasoned action:a meta-analysis of past research with recommendations for modifica-tions and future research. J Consum Res. 1988;15:325–343.

31. Fishbein M. A reasoned action approach to health promotion. Med De-cis Mak. 2008;28(6):834–844.

32. Wallston KA. The validity of the multidimensional health locus of con-trol scales. J Health Psychol. 2005;10(5):623–631.

33. Mazmanian PE, Daffron SR, Johnson RE, Davis DA, Kantrowitz MP.Information about barriers to planned change: a randomized controlledtrial involving continuing medical education lectures and commitmentto change. Acad Med. 1998;73(8):882–886.

S10 JOURNAL OF CONTINUING EDUCATION IN THE HEALTH PROFESSIONS—34(S1), 2014DOI: 10.1002/chp