assessing generalization in perceived self-efficacy

Upload: bldzsrnn

Post on 06-Apr-2018

230 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/2/2019 Assessing Generalization in Perceived Self-Efficacy

    1/9

    PERSONALITY ANDSOCIALPSYCHOLOGY BULLETINWeitlaufetal. /GENERALIZATION IN SELF-EFFICACY

    Assessing Generalization in Perceived Self-Efficacy:

    Multidomain and Global Assessments of the

    Effects of Self-Defense Training for Women

    Julie C. Weitlauf

    Daniel Cervone

    University of Illinois at Chicago

    Ronald E. Smith

    University of Washington

    Paul M. Wright

    University of Illinois at Chicago

    The authors assessed the impact of self-defense training forwomen on multifaceted aspects of perceived self-efficacy. As com-pared to a waiting list control condition, training increased self-efficacy perceptions not only for self-defense skills but also acrossa variety of domains, including self-defense abilities, sports com-petencies, and coping skills. Trained participants also experi-enced a significant increase inmore globalaspects of personality,including perceptions of physical self-efficacy and assertiveness.No changes were detected on a trait measure of global self-effi-cacy; however, there was a significant change on a compositescore of a multidomain self-efficacy questionnaire andon several

    domain-specific subscales, indicating that trained participantsexperienced a boost in multiple domains of self-efficacy notdirectly tapped by the intervention.Implications for constructingmore sensitivemeasures of coping skills generalization effectsarediscussed.

    Amajor challenge faced by practitioners who seek tofoster psychological change is to develop interventionswhose effects generalize (Goldfried & Robbins, 1982;Smith, 1989, 1999; Weitlauf, Smith, & Cervone, 2000).Ideally, interventionswillbenefit participants notonlyin

    the particular socialandbehavioral domains targeted bythe treatment but also throughout multiple aspects ofindividuals daily lives. This challenge has both practicaland theoretical components. Assessing generalizationrequires a theory that addresses two primary questions:(a) What are the psychological qualities that one shouldassess to investigate generalization? and (b) What is theproper way of assessing those qualities? The difficulties

    involved in answering these questions are evidenced bthe fact that the issue of cross-situational generality hbeen among the most vexing topics in all of personalipsychology (Caprara & Cervone, 2000; Mischel, 1968

    We address these theoretical and methodologicchallenges by drawing on the general framework osocial-cognitive theories of personality (reviewed iCervone & Shoda, 1999) and the more specific framwork of self-efficacy theory (Bandura, 1977, 1997Social-cognitive theory highlights the dynamic cognitiand affective processes underlying personality functio

    ing and coherence. In this perspective, assessment gobeyond the examination of overt psychological tendecies and focuses on dynamic psychological systems thcausally contribute to personality functioning. Seconassessment is contextualized; that is, psychologicalqua

    1683

    Authors Note: This article is based on an ongoing research prograthe Womens Self-Defense Project, directed by the first author undthe direction of the second author at the University of Illinois at Chcago (UIC).Portionsof this research were presented at the1999Amican Psychological Society Convention in Denver, Colorado. ThWomens Self-Defense Project was supported by a grant through thUIC Campus Research Board. The authors wish to thank UICs Redential Life Program andDepartmentof Kinesiology for their coopation in the facilitation of this project. In addition, we would like thankMary AnnPecfor herhelp inmanaging theprojectandto GlorBalague,DavidMcKirnan, andDonaldHellisonfor their contributioto the article. Correspondence concerning this article should be drected to Julie Weitlauf, VA Palo Alto Health Care System, PsycholoService (116B), 3801 Miranda Ave., Palo Alto, CA 94304; [email protected].

    PSPB, Vol. 27 No. 12, December 2001 1683-1691 2001 by the Society for Personality and Social Psychology, Inc.

  • 8/2/2019 Assessing Generalization in Perceived Self-Efficacy

    2/9

    ties are assessed with respect to the contexts that makeup peoples daily lives (Cervone, Shadel, & Jencius,2001). Contextualized assessment showcases patterns ofbehavioral variability across contexts; such idiosyncraticvariations in psychological functioning area basic fact ofsocial life (Bugenthal, 2000) and can critically distin-guish individuals from one another (Mischel, 1999;Mischel & Shoda, 1995).

    Perceived Self-Efficacy

    In the study of generalization of treatment effects,social cognitive theory places particular emphasis onpeoples subjective perceptions of their capabilities forperformance, or self-efficacy perceptions (Bandura,1997; Cervone & Scott, 1995). Self-efficacy theory positsthat self-efficacy perceptions govern a set of behavioral,cognitive, and affective processes that are critical to per-formance success (Bandura, 1999).Peoplewith a highersense of self-efficacy on tasks tend to persist in theirefforts when faced with challenges (Cervone & Peake,1986), display less anxiety in response to threats

    (Bandura, Taylor, Williams, Mefford, & Barchas, 1985),and report fewer anxiety-related cognitions (Sarason,Pierce, & Sarason, 1996).

    Research in many important activity domains docu-ments the critical role of self-efficacy beliefs in facilitat-ing adaptive behavior. For example, strong self-efficacybeliefs can facilitate the promotion of health behaviors(Schwarzer, 1995), including recovery from physical set-backs (Ewart, 1992), the control of eating (Glynn &Ruderman, 1986), and safer sex practices (Montoya,1998). Robust self-efficacy beliefs improve performancein the workplace (Stajkovic & Luthans, 1998) and pro-mote success in athletic pursuits (Escarti & Guzman,

    1999).

    Assessing Generalization in Perceived Self-Efficacy

    Throughout the social cognitive literature, efficacyperceptions generally have been assessed in a con-textualized manner. Measures tap peoples appraisals oftheir capabilities to attain specific levels or types of per-formances in designated contexts (Bandura, 1977;Cervone, 1985). The use of these contextualized mea-sures naturally raises the question of how one shouldassess cross-context generality. Here there are twooptions. The first is to employ measures that are decon-

    textualized. Instead of asking individuals to appraisetheir capabilities to handle designated challenges andcircumstances, questionnaire items might tap general-ized beliefs, such as whether one sees oneself asa compe-tent, efficacious person. Such procedures yield globalmeasures of the construct generalized self-efficacy(Coppel, 1980; Sherer et al., 1982; Schwarzer, Babler,Kwiatek, & Shroder, 1997). The question of whether

    intervention produces generalized effects can then boperationalizedby examiningthe influenceof the intevention on measures of this generalized construct (e.Smith, 1989; Weitlauf et al., 2000).

    Despite their popularity, the use of decontextualizemeasures of global self-efficacy clashes with the recommendations of social cognitive theory (Bandura, 1997Rather, Bandura (1997) recommends a contextualize

    approach to assessment. In this approach, self-efficaassessments tap beliefs about personal abilities to prduce specified levels of performance (Bandura, 199p. 45). In no case are these assessments dissociatefrom context (Bandura, 1997, p. 50). Instead, itemdescribe a behavioral setting and activity to be peformed and ask people to indicate their certainty ththey can perform the designated task in that settinItems linked to major activity domains are seen to ban improvement over omnibus measures that are dissciated from clearly defined activities and contextual fators(Bandura,1997, p. 48). Thus,domain-basedassements are favored because domain-linked measures ar

    stronger predictors of performance than areglobalindces (Bandura, 1997).

    A second possible empirical advantage of domaibased versus global assessments has received little attetion. This is the possibility that domain-based assesments will prove to be more sensitive indices of the geeralization of treatment effects. People commonly clinto abstract, global beliefs about the self that may be variance with their day-to-day experiences (Mische1968). As a result, assessments of global beliefs may sugest that a given intervention did not produce effecthat generalized. However, a domain-based approacmay reveal that the intervention did, in fact, generalibeyond the particular domain in which the interventiooccurred.This possibility is a central focus of thepreseresearch. To our knowledge, no prior studies havemployed both global and multidomain measures self-efficacy; thus, no studies directly speak to this potetial advantage of domain-based self-efficacy assessmen

    Global self-efficacy measures, such as the Sherer et a(1982) scale, assess global feelings of personal comptence by asking broad questions such as, When I makplans, I am certain that I can make them work. By cotrast, a domain-based measure of self-efficacy taps spcific, self-efficacy beliefs in a particular domainby askin

    highly contextualized questions (i.e., How confideare you that you can be on time for an 8:00 class everweek throughout a busy semester?). A number of stuies have employed contextualized domain-linked mesuresandfound that treatments cangeneralize from ondomain to another. These include studies of the effecof the clinical treatment of agoraphobia (WilliamKinney, & Falbo, 1989), self-defense training for wome

    1684 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

  • 8/2/2019 Assessing Generalization in Perceived Self-Efficacy

    3/9

    (Ozer & Bandura, 1990), and the mastery of a high-risksport (Brody, Hatfield,& Spalding, 1988). These studies,however, provide limited information about generaliza-tion in that the self-efficacy measures used tapped rela-tivelynarrow rangesof functioningin domains that werehighly related to the treatment domain.

    In the present study, we developed a contextualizedmultidomain measure that assesses self-efficacy in a vari-

    ety of activity domains (i.e., academic achievement,sports, interpersonal coping skills, assertiveness, andconscientiousness). The advantage of this MultidomainSelf-Efficacy Scale is that it allows us to assess generaliza-tion of self-efficacy beliefs to activity domains that areboth highly related to the training as well as to domainsthat are quite distinct.

    Self-Defense Training for Women

    Self-defense training for women provided an idealpsychosocial intervention to examine generalizationeffects. Self-defense training is well suited to the ques-tions asked here in that the intervention itself is highlycontextualized. Women are trained in skills that are spe-cificallydesignedto enablethemtocope with physicalorsexual assault. The question we ask, then, is whether thistraining produces changes in self-efficacy domainsbeyond physical self-protection. We hypothesized thatwomen who participated in self-defense training wouldexperience boosts in self-efficacy across a range of activi-ties that do not directly involve self-defense.

    Generalization was assessed using both decon-textualized global scales of generalized self-efficacy anda contextualized (multidomain) scale that assessed self-efficacy across a variety of activity domains. We also

    explored the relationship of participant schematic withseveral hypotheses. First, we expected to replicate theempowerment effect of self-defense training onwomen demonstrated by Ozer and Bandura (1990) andby Weitlauf et al. (2000). That is, we expected that train-ing would significantly enhance womens confidence intheir abilities to defend themselves if attacked but alsothat such training would have a positive impact on moreglobal beliefs of competence and physical ability. Sec-ond, wewantedtoexplore and compare thesensitivity ofcontextualizedmeasuresof generalizedself-efficacywithmore global, traitlike constructs. We suspected that the

    contextualizedmultidomainself-efficacymeasurewouldbe a more sensitive index of generalization than a globaltrait measure. That is, women who completed trainingwould show significant increases in self-efficacy across avariety of life domains, and this change would be moreevident than increases in global perceptions of efficacyas measured by a generalized self-efficacy questionnaire.Finally, we assessed generalization to several relevant

    personality domains, including self-esteem, assertivness, and aggression.

    In one experimental condition, we attempted tenhance generalization effects by embellishing traininwith structured discussions and writing assignmendesigned to help participants connect self-defense coing skills with other life activities. We reasoned that thgeneralization of self-efficacybeliefs fromone domainanother at least partly reflects constructive cognitivactivities. People may reflect on the skills they havgained in one domain, or on the new knowledge abouthemselves that they have gained while acquiring thoskills, and then apply their skills and self-knowledge tother life domains (Smith, 1999).

    METHOD

    Experimental Design

    We used a 3 (groups) 2 (measurement timerepeated measures mixed design in which participanwere randomly assigned to one of two training cond

    tions or a waiting list control condition. Assessments the outcome variables were completed at pretraining obaseline (Time 1) and immediately following the coclusion of the intervention at posttraining (Time 2).

    Participants

    Participants were 125 female undergraduate studenenrolledin introductory psychology at a large urban unversity in the midwestern United States. Participanrespondedto posted advertisements forfreeself-defentraining and were subsequently screened for eligibiliWomen were eligible to participate if they had no prevous or concurrent training in the martial arts, seldefense, or weapons training. Of the original 12women, 115 qualified to participate and 96 completethe training. The total attrition rate of qualified particpants was 17% and was not related to condition. Particpants ranged in age from 18 to 26 years (M= 19.02, SD1.50). Caucasians constituted 30.20% of the sampl17.70% were African American, 21.90% were Latin24.00% were Asian, 5.20% were multiracial, and 1.0were Native American.1

    Procedure

    Participants assigned to the self-defense condition

    were sent schedules of class times and location anbegan their training immediately. Pre- and posttrainindata collection occurred on-site during the first and laclass sessions. Participants assigned to the waiting licontrol condition were informed that they would begitheir training immediately following the completion the first session of classes (a waiting period of 4 weekand were instructed to attend two data collection se

    Weitlauf et al. / GENERALIZATION IN SELF-EFFICACY 168

  • 8/2/2019 Assessing Generalization in Perceived Self-Efficacy

    4/9

    sions that corresponded to the first and last weeks of theexperimental groups training. Waiting list participantsreceived no intervention during the waiting period andwere instructed not to pursue self-defense or relatedtraining during this time. Control participants also weresent information about thedates, times, and locationsoftheir self-defense course as well as dates and times fordata collection.

    Self-defense training condition. The basic experimentalintervention was a 16-hour, intensive, physical self-defense course specifically designed for college-agewomen. The training, consisting of four weekly, 4-hourtraining sessions,wasdesignedto teach verbaland physi-cal resistance to rape. The first and fourth authors, bothexperienced self-defense instructors, conducted all ses-sions. Make-up sessions were offered on several occa-sions so that participants who could not attend theirusual group could attend an alternate session. Thisresulted in full attendance by all experimental groupparticipants.

    Emotional resistance to rape was addressed throughcoping skills training that taught participants emotionmanagement and attentional control techniques thatwould help them focus their resistance efforts duringanattack. Participants were taught techniques to helpsecure their breathing, remain calm, and think theirway out of an assault. During the verbal resistancephases of training, participants were taught to use theirvoice as an active tool in resistance.

    Participants also were taught basic physical resistancetechniques. Techniques and moves were drawn primar-ily from the martial arts of Aikido, Shotokan Karate, andHapkido. Striking techniques included basic punches

    and blocks, groin and knee kicks, and several otherupper-body striking techniques such as the hammer fistand knife hand strikes. Participants learned to developtheir form and, using a blocking shield, practiced strikesat full force during each subsequent training session.Participants learned a variety of methods for freeingthemselves from an assailant when grabbed at the hand,wrist, arm, shoulder, throat, waist, or hair. Specialemphasis was placed on breaking out of strangulationholds and resisting choking.

    Participants wererepeatedlydrilledin allphysicalandverbal skills as a part of training. Continuous corrective

    feedback was provided to all students by both instructorsand, as the course progressed, by participants to oneanother. In addition, participants used classroom mir-rors as an added means of continuous feedback to helpwith the acquisition and correction of skills. To ensureskill mastery, participantscompleted2,000 repetitions ofeach striking technique and did not take on moreadvanced techniques until mastering more basic ones.2

    Enhanced self-defense training condition. Participants this condition were given an additional 30 minutes otraining per week during which they engaged group dcussions andwriting assignments designed to help thereflect on how the training might be applicable in thelives. Some of the discussion and journal exercises wefree form; however, much of this enhanced trainincomponentwasguidedby specific questions designedhelp them think broadly about how this training migaffect their lives.

    Measures

    Self-defense efficacy. Participants completed a 16-iteself-efficacy measure that assessed their perceived capbility to execute a variety of self-defense skills, recognithe warning signs of a potential attacker, and obtaaccess to the appropriate medical and legal resources help them cope with an assault. These items specificaltargeted the curriculum offered in the self-defencourse. Sampleitemsareas follows: Ihave theskill/abity toexecute (hurt ordisable anattacker via) a varietyclose-range techniques, including various forms ofbasic hand strike (straight punch, knife hand, hammfist, etc.)andI have theskill/ability to execute (hurt odisable an attacker via) a variety of kicking techniqueincluding a groin and knee kick. Participants ratetheir perceived ability to execute each of these skills o10-point scales, with 1 labeled not competent at alland 1labeled very competent. A score was computed by averaing across items. Reliability analyses indicated that thmeasure has an acceptable level of internal consistencCronbachs alpha at Time 1 was .94 and .98 at Time 2

    Global self-efficacy measures. The Physical Self-EfficaScale3 (Ryckman, Robbins, Thorton, & Cantrell, 1982was administered to assess perceived self-efficacy in thphysical or athletic domain. This 22-item scale assessperceptions of ones body, athletic ability, coordinatioand motor capabilities. Sample items are as follows: take little pride in my ability in sports and Because my agility, I have been able to do things which many oters could not do. Each item is scored on a 6-point scalInternal consistency was adequate; Cronbachs alphwas .82 at Time 1 and .78 at Time 2.

    To assess an even more global level of efficacy beliefSherer et al.s (1982) general self-efficacy subscale w

    used to assess feelings of general capability, productiviand self-regulation (see Note 3). Sample items includthe following: When I make plans, I am certain thatcan make them work and When I have somethinunpleasant to do, I stick to it until I finish it. Each of th17 items is scored on a 14-point scale, and a total scowas used for analysis. Reliability analyses indicated thCronbachs alpha was .86 at Time 1 and .81 at Time 2

    1686 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

  • 8/2/2019 Assessing Generalization in Perceived Self-Efficacy

    5/9

    Domain-specific self-efficacy. Our domain-linked measureof perceived self-efficacy was a 32-item multidomainscale designed to tap efficacy beliefs in a variety of com-monly occurring settings. Specifically, the scale tappedefficacy beliefs in five domains of functioning: sports(Ifyoure competing in a 5K/3-mile road race, how confi-dent are you that you would be able to finish the race inthe top 25% of the people of your age and sex?), aca-demic achievement (If youre taking a political scienceclass, how confident are you that you would be able toearn at least a B in the class?), conscientiousness (Ifyoure studying for an exam and your favorite showcomes on TV, how confident are you that you would beable to get yourself to turn off the TV and stick to yourstudying?), assertive communication (If you are feel-ing completely lost in a course, how confident are you inyour ability to go to the professor or a teaching assistantand gethelp?), andcoping4 (If you have to give an oralpresentationin class,how confidentareyou in your abil-ity to remain calm and collected when talking in front ofthe class?). For each item, participants responded on a

    100-point scale, with 1 being not at all confident, 50 beingsomewhat confident, and 100 being very confident. Itemswere summedtoprovide a measure ofperceivedself-effi-cacy within each of the five domains. A composite scorewas computed by aggregating across all items of thescale. Reliability analyses indicated that this compositehad an acceptable level of internal consistency, withCronbachs alpha being .92 at Time 1 and .94 at Time 2.On the individual domain-linked scales, Time 1 alphasranged from .60 (sports) to .81 (coping) and Time 2alphas ranged from .72 (sports) to .83 (coping).

    Personality Trait Measures

    The Rosenberg Self-Esteem Scale (Rosenberg, 1965)taps global feelings of personal value and self-worth.Sample items include the following: I feel that I am aperson of worth, at least on an equal basis with othersand I feel I am able to do things as well as most otherpeople.Eachof the 10 items is scoredona 4-point scale,with a higher score indicating a higher level of self-esteem. Reliability analyses indicated that Cronbachsalpha was .82 at Time 1 and .90 at Time 2.

    The Rathus Assertiveness Scale (Rathus, 1973) wasused to assess participants self-perceived levels of asser-tiveness.Sample items include thefollowing: I am open

    and frank about my feelings and I complain aboutpoor serviceina restaurant andelsewhere.This 30-itemmeasure features a possible range of 180 points (90 to+90).A total score wasused foranalysis.Reliability analy-sisindicated that Cronbachs alpha was .88at Time 1 and.90 at Time 2.

    The Aggression Questionnaire (TAQ) (Buss & Perry,1992)was used toassesslevels ofhostility andaggression.

    TheTAQcontains four subscales: Anger (Ihavetroubcontrolling my temper), Hostility (I wonder somtimes why I feel so bitterabout things), Physical Aggresion (Given enoughprovocation, I mayhitanother peson), andVerbal Aggression (When people annoymI may tell them what I think of them). Reliability analsesindicated that this measure hadan acceptable levelinternalconsistency. Forthetotal score(GeneralAggresion), Cronbachs alpha was .85 at Time 1 and .85 Time 2. Subscale reliabilities also were acceptable, witCronbachs alpha ranging from .62 to .74 at Time 1 an.75 to .78 at Time 2.

    Training Goals and Expectancies

    At the outset of the course, participants were askeddescribe their reasons for taking the course as well atheir specific behavioral goals. They also were asked tnote if they expected anypersonalskills orcapabilitieschange as a result of their participation. Participanthen were asked to circle their most important goal anto describe specific situations in which they felt th

    change in this particular characteristic or skill is highrelevant.

    Responses were transcribed verbatim. The firauthor and a research assistant developed a codinscheme allowing for independent themes to emergThe coding scheme included emergent themes of coingskills (i.e., improving communication witha spousehealth-related activities (i.e., improving fitness), anachievement (i.e., improving grades).Two independeraters used this scheme to code all responses. Kappa wat or greater than .87 for all responses.

    RESULTS

    Baseline Analyses

    Means and standard deviations for all variables in aconditions are provided in Tables 1 and 2. To assepotential baseline differences among conditionmultivariate analyses of variance (MANOVAs) were coducted for each conceptually related set of measureResults of MANOVAs on the self-efficacy variables anon the personality variables indicated no differencebetween participants in thethree conditions at baselinWilkss lambda = .90, F(8, 180) = 1.17, and Wilkslambda = .90, F(6, 182) = 1.63, respectively; ps > .0

    Follow-up one-way analyses of variance (ANOVAs) oindividual variables indicated no significant baselinedferences between conditions on self-defense selefficacy, F(2, 93) = .17, Ryckmans physical self-efficacF(2, 93) = .69, Sherer et al.s measure of general selefficacy, F(2, 93) = 2.31, or the composite score on thmultidomain self-efficacy questionnaire, F(2, 93) = 1.5allps > .05. Likewise,no baseline differences were foun

    Weitlauf et al. / GENERALIZATION IN SELF-EFFICACY 168

  • 8/2/2019 Assessing Generalization in Perceived Self-Efficacy

    6/9

    on Rosenbergs Self-Esteem Scale, F(2, 93) = 2.78, or for

    the total score on the Buss-Perry Aggression Scale,F(2, 93) = .16; ps > .05. However, a significant differencebetween groups was found on Rathuss trait measure ofassertiveness, F(2, 93) = 3.68, p< .05, 2 = .07. Follow-upcontrasts indicated that the basic self-defense cohortwas, at baseline, significantly lower on assertiveness thanboth the enhanced self-defense condition, F(1, 53) =5.93, p< .05, and the control condition, F(1, 65) = 6.87,

    p < .015. The groups also differed at baseline on th

    assertiveness subscale of the multidomain self-efficaquestionnaire, F(2, 93) = 3.28, p< .05, with participanin the enhanced self-defense condition scoring higheon this subscale than their counterparts in the basic sedefense condition, F(1, 53) = 9.10, p< .005. There weno baseline differences between the basic self-defenand control conditions or between the control anenhanced training conditions, ps > .05, ns. There we

    1688 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

    TABLE 1: Means and Standard Deviations for the Self-Efficacy Variables in All Conditions

    Condition

    Self-Defense Enhanced Self-Defense Control

    Variable M SD N M SD N M SD N

    Self-defense self-efficacyPre 67.00 18.87 26 71.10 30.75 29 69.00 25.72 41Post 124.35 25.08 26 133.90 16.68 29 73.66 28.51 41

    Physical self-efficacyPre 84.15 14.21 26 85.90 16.93 29 88.27 12.13 41Post 89.62 14.16 26 94.14 14.78 29 90.59 11.95 41

    Global self-efficacyPre 160.89 28.75 26 175.17 31.62 29 173.85 23.20 41Post 163.54 29.90 26 178.31 28.06 29 171.29 23.41 41

    Multidomain general self-efficacyPre 1906.50 354.72 26 2103.97 400.21 29 1968.93 497.11 41Post 2175.96 422.71 26 2260.86 393.15 29 1974.51 473.95 41

    Self-esteemPre 21.62 4.07 26 24.31 4.44 29 23.17 4.20 41Post 22.58 5.47 26 26.17 4.07 29 24.78 4.73 41

    AssertivenessPre .50 19.59 26 17.86 31.25 29 16.88 27.73 41Post 4.15 22.49 26 19.21 31.91 29 15.44 29.36 41

    Aggression

    Pre 44.73 12.78 26 46.07 13.64 29 44.20 14.98 41Post 43.35 14.03 26 47.93 15.22 29 40.46 14.30 41

    TABLE 2: Means and Standard Deviations of Average Item Responses for Self-Efficacy Domains

    Condition

    Self-Defense Enhanced Self-Defense Control

    Variable M SD N M SD N M SD N

    Sport self-efficacyPre 39.42 15.13 26 41.56 19.04 29 38.93 16.97 41Post 54.67 20.23 26 56.88 18.09 29 41.79 19.01 41

    Assertiveness self-efficacy

    Pre 63.64 12.08 26 75.00 15.41 29 69.70 19.21 41Post 71.32 13.76 26 78.52 15.34 29 69.31 16.89 41

    Coping self-efficacyPre 62.44 11.19 26 66.56 14.43 29 64.05 16.27 41Post 71.12 12.62 26 72.16 13.52 29 63.03 15.87 41

    ConscientiousnessPre 54.63 19.68 26 63.58 17.00 29 57.27 20.29 41Post 61.06 19.17 26 63.21 18.47 29 58.28 18.94 41

    Academic achievementPre 70.21 15.62 26 74.80 15.70 29 67.59 17.76 41Post 76.73 15.63 26 78.16 16.29 29 67.98 19.31 41

  • 8/2/2019 Assessing Generalization in Perceived Self-Efficacy

    7/9

    no additional baseline differences on any of the remain-ing self-efficacy subscales.

    Analyses of participants training goals and expectan-cies revealed highly consistent expectations for training.Nearly 100% of the women stated that they were takingself-defensetrainingto learnhowto betterdefendthem-selves should they be attacked. Some participants indi-cated a desire to become more physically fit through

    training, whereas others hoped for an improvement inself-esteem, confidence, and assertiveness.

    Treatment Ef fects

    Self-efficacy measures. A repeated-measures MANOVAincluding all self-efficacy variables revealed a significantGroups Time interaction on this variable cluster;Wilkss lambda = .36, F(8, 180) = 15.24, p< .001, 2 = .40.To assess the self-efficacy variables responsible for themultivariate effect, we conducted a series of two-wayrepeated-measures ANOVAs. These analyses revealedsignificant Groups Time interactions on all indices ofself-efficacy except for Sherer et al.s General Self-Effi-cacy Scale. Thus, we found a significant Groups Timeinteraction on the Self-Defense Self-Efficacy Scale, F(2,93) = 75.31, p < .001, 2 = .62. Planned comparisonsrevealed that both the basic and enhanced traininggroups exhibited significant increases in self-defenseself-efficacy relative to the waiting list control condition,F(1, 65) = 101.40,p< .001,2 = .61, andF(1, 68) = 155.47,p< .001, 2 = .70, respectively.

    A significant GroupsTimeinteraction alsooccurredonRyckmansmeasureof physical self-efficacy,F(2,93)=3.72,p< .05,2 = .07. Planned comparisons revealed thatparticipants in theenhanced conditionshoweda signifi-

    cant increase on this measure relative to untrained con-trols, F(1, 68) = 6.56, p< .015, 2 = .09. Although partici-pants in the basic self-defense condition also showedsignificant increases in physical self-efficacy after train-ing, t(25) = 3.75,p< .001, their improvement wasnotsig-nificantly different from that of the control participants,F(1, 65) = 3.53, p= .07.

    Turning to global self-efficacy, a two-way, between-group, repeated-measures ANOVA revealed no main orinteractioneffects ontheSherer et al.(1982) measure ofglobal self-efficacy, F(1, 93) = .30, p> .05, and F(2, 93) =.98,p> .05, respectively. Incontrast,a significantGroups

    Time interaction was found on the composite score ofthe multidimensional measure of general self-efficacy,F(2, 93) = 6.39, p< .005, 2 = .12. Planned comparisonsrevealed that relative to their control conditioncounter-parts, participants in the basic self-defense conditionshowed significant increases on the composite score ofthe Multidomain Self-Efficacy Scale,F(1, 65) = 15.72,p