This article was downloaded by: [University of Newcastle (Australia)]On: 03 September 2014, At: 11:45Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: MortimerHouse, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Sports SciencesPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/rjsp20

The self‐efficacy‐performance link in maximumstrength performanceChristie M. Wells a , David Collins b & Bruce D. Hale aa Pennsylvania State University , University Park, Pennsylvania, 16802, USAb St Mary's College , Waldegrave Road, Twickenham, Middlesex, TW1 4SX, UKPublished online: 14 Nov 2007.

To cite this article: Christie M. Wells , David Collins & Bruce D. Hale (1993) The self‐efficacy‐performance link inmaximum strength performance, Journal of Sports Sciences, 11:2, 167-175

To link to this article: http://dx.doi.org/10.1080/02640419308729980

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensorsmake no representations or warranties whatsoever as to the accuracy, completeness, or suitability for anypurpose of the Content. Any opinions and views expressed in this publication are the opinions and viewsof the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sources of information. Taylorand Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses,damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connectionwith, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Journal of Sports Sciences, 1993, 11, 167-175

The self-efficacy-performance link in maximumstrength performance

CHRISTIE M. WELLS,1 DAVID COLLINS2* and BRUCE D. HALE1

'Pennsylvania State University, University Park, Pennsylvania 16802, USA and 2St Mary's College,Waldegrave Road, Twickenham, Middlesex TW1 4SX, UK

Accepted 4 September 1992

To examine whether manipulating self-efficacy affects strength performance on a bench press, and to see if thesesituation-specific changes would affect levels of physical self-efficacy, 24 undergraduates untrained inweightlifting were randomly assigned to three groups: 'light', who lifted less weight than they believed; 'heavy',who lifted more weight than they believed; and control, for whom there was no manipulation. Self-efficacymeasures were taken before and after the manipulation. Physical self-efficacy was measured using the PhysicalSelf-Efficacy Scale (PSE). 'Light' subjects lifted significantly greater increases in weight than the other subjects.'Heavy' subjects significantly decreased self-efficacy following the manipulation. Initial self-efficacy was foundto be a significant predictor of baseline maximum, while manipulated self-efficacy was significant forperformance change. The PSE scores did not change pre- to post-study. The results suggest that self-efficacy is asituation-specific construct which can be manipulated, and which relates to both past performance experienceand future performance.

Keywords: Self-efficacy, strength, physical performance, self-confidence.

IntroductionSelf-confidence is an important contributory factor tothe eventual level of athletic performance (Feltz, 1982,1988a) and, as such, has justifiably received a great dealof research attention in sport psychology. Confidence inone's own ability plays a role in enabling athletes toovercome specific performance 'barriers' which ofteninvoke a psychological challenge quite out of proportionto their relative difficulty. Such barriers - the 4-minmile was a classic example - act to impede progress bothwithin and between performers. But once one person isable to surpass the barrier, suddenly more and moreathletes follow suit. As Bandura (1990, p. 129) stated,'once extraordinary performances are shown to bedoable, they become commonplace'. This implies adefinite psychological impact on athletes' perceptions oftheir own abilities. This perceived ability, more gener-ally called self-confidence, can be operationalized in avariety of ways (Feltz, 1988b), but is most commonlyexpressed by the construct of self-efficacy. As defined byBandura (1977, 1982, 1986), self-efficacy is the convic-

* To whom all correspondence should be addressed.

0264-0414/93 © 1993 E. & F.N. Spon

tion that one can successfully execute the behaviourrequired to produce a certain outcome. The strongestand most dependable source of self-efficacy is perform-ance accomplishment. With repeated successes, efficacyexpectations rise; if experiences are perceived as fail-ures, expectations fall. Therefore, regardless of theparticular task involved, the important factor is how thetask is perceived by the individual. It is the individual'sown cognition about the task and the situation whichdetermine the expectations of performance outcome,and hence, have a significant influence on progress.

Self-efficacy and its relationship with performancehas been widely studied. Recent investigations haveconsidered youth gymnastics (Weiss et al., 1989) andtennis (Ransom and Weinberg, 1985; Weinberg andJackson, 1990), and have found self-efficacy to be asignificant factor in the prediction of performance orsuccessful performance itself. Most studies, however,have involved endurance-type strength activities and/orperformance prediction (Feltz, 1982; Feltz and Mugno,1983; Gayton et al., 1986; McAuley and Gill, 1983;Weinberg et al., 1979, 1981). Feltz (1982) found self-efficacy, along with past performance accomplishments,to be significant predictors of performance for divers.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ewca

stle

(A

ustr

alia

)] a

t 11:

45 0

3 Se

ptem

ber

2014

168 Wells et al.

These findings were consistent with those of Weinberget al. (1981), who found pre-existing self-efficacymeasures to be predictive on only the initial trial of a leg-extension task, with subsequent trials dependent uponmanipulated self-efficacy via the 'source' of previousperformance. This was a follow-up study to that ofWeinberg et al. (1979) who, in a similar leg-extensionstudy, demonstrated support for the self-efficacy-per-formance relationship, as subjects with high self-efficacy held their legs extended significantly longer thansubjects with low self-efficacy. These studies suggestthat self-efficacy can indeed be modified to achievedifferences in performance.

In terms of strength performance, there has not beenmuch emphasis on self-efficacy per se. Instead, studieshave involved concepts similar to self-efficacy, butwhich have gone by different names, such as externalcues (Ness and Patton, 1979), expectations (Nelson andFurst, 1972), motivational cues (Carnahan et al., 1990)and 'psyching-up strategies' (Shelton and Mahoney,1978). Ness and Patton (1979), for example, usedmanipulation of external cues to facilitate performanceon a bench press apparatus. The resultant performanceenhancement reflected possible changes in the cogni-tions of the subjects, but the specific constructs were notassessed. If self-efficacy (a specific construct) was foundto have a significant impact on performance, then therange of mental strategies which contribute to enhancedstrength could be narrowed. As a consequence, inter-ventions in this specific area could be designed moreeffectively.

Self-efficacy can be said to be a more specificpsychological construct than self-confidence in that self-confidence relates more closely to feelings about self-esteem, whereas self-efficacy relates more to confidenceabout specific performance potential. It is also said tofluctuate across situations and tasks. A more generalmeasure of self-efficacy, called physical self-efficacy andmeasured by the Physical Self-Efficacy Scale (PSE:Ryckman et al., 1982), has also been developed andapplied to such sports as women's gymnastics (McAuleyand Gill, 1983), marathon running (Gayton et al., 1986)and various motor tasks and sport activities (Ryckman etal., 1982). This can be conceptualized as a measure ofseveral related concepts such as confidence in one'sphysical appearance and self-esteem, as well as theability to perform in general physical/motor tasks. ThePSE was originally described as a trait measure, but onestudy found evidence suggesting the PSE could be usedto predict performance (Gayton et al., 1986). Althoughat first sight this appears contradictory (e.g. self-efficacyis defined as a stare-specific measure), the potentialapplications for a global predictor of performance arehighly desirable. It would therefore seem logical toexamine the potential of the PSE in such a role.

This study was therefore designed to investigatewhether specific manipulation of self-efficacy levelswould affect strength performance. It was expected thatthese levels of self-efficacy would be greatly enhanced bypast experience in that successful performance would beassociated with increases in both self-efficacy and weightlifted. Conversely, perceived performance decrementswould result in decreased self-efficacy and poorerlifting. Additionally, data were collected to determine ifthe manipulation of self-efficacy in a specific situationwould affect PSE (assumed to be a trait factor) and ifPSE would itself have any impact on performance of thespecific task.

Method

Subjects

The 24 subjects (aged 19.5-22,1 years) were volunteercollege students enrolled in summer school courses.These subjects had little or no weightlifting experience,which made them more amenable to change via thechosen manipulation.

Measures

A pre-study inventory was developed which determinedtrie subjects' present activity level (aerobic activity andstrength training), height, weight and an estimate of theamount of weight they believed they could bench pressat that time (to be used as an initial target maximum). Italso served to get an initial measure of efficacy in theform of how the subject felt he or she would compare tosimilar individuals in a weightlifting task.

The Physical Self-Efficacy Scale (PSE) developed byRyckman et al. (1982) was used as a global trait self-efficacy measure. The scale is made up of two sub-scales,the Physical Self-Presentation Confidence Scale(PSPC) and the Perceived Physical Ability Scale (PPA).The scale was found to have an overall internalconsistency (Cronbach alpha) of 0.81.

Situation-specific self-efficacy was measured usingthe Situational Self-Efficacy Scale (SSE). The SSE is aLikert 7-point scale developed specifically for thisstudy, which was administered prior to any lifting tonote each subject's belief in his or her ability to lift themaximum weight expected for that session. The ques-tion was, 'How confident are you in your ability to lifttoday's max?' The scale ranged from 1 ('I don't think Ican lift as much as last time') to 7 ('I definitely can lifttoday's maximum'). It should be noted that strictextrapolation of Bandura's methodology would requirethat the subject be presented with a series of weights at,say, 10-lb (4.5-kg) increments, to each of which a

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ewca

stle

(A

ustr

alia

)] a

t 11:

45 0

3 Se

ptem

ber

2014

Self-efficacy and strength 169

confidence rating typically from 1 to 10 (no chance tocompletely confident) would be required. With thenovice population used in this study, the use of such asystem was deemed unsuitable because their lack ofexperience in weightlifting would have mitigatedagainst meaningful or reliable estimates. The systemused in the present study was designed to cope withnovices by relating every challenge (and associatedestimate) closely to their personal experiences.

A post-study inventory included general questionsdesigned to check whether the subjects had been awareof manipulations, any changes in difficulty of their lifts,and to ensure they had indeed felt they have been giventhe opportunity to lift to their maximum during eachsession.

Procedure

The subjects were randomly assigned to one of threegroups - a control group, a light (L) group and a heavy(H) group. Males and females were evenly assigned (twofemales and six males per group). The study consisted offour individual sessions over a 5-day period, with a dayof rest between the first and second sessions. In eachsession, a procedure was followed which resulted in eachsubject recording a one-repetition maximum (1RM) lifton the bench press (judged by modified power-liftingrules). The subjects performed a warm-up set of fiverepetitions at approximately 50% of their perceivedmaximum, three repetitions at approximately 75% oftheir perceived maximum and, finally, attempted a set ofone at 100% maximum. Prior to this attempt, as in eachsubsequent session, the subjects were reminded that 'itis very important that you try your best to improve yourprevious maximum'. This was important, as one of theconditions of self-efficacy theory in modifying behav-iour is that proper incentives must be offered for change(Bandura, 1977,1986). In all the sessions, if an attemptcould not be successfully completed, weight wasremoved from the bar in 5-lb (2.25-kg) increments untilthe subject could perform the lift. If the subject was ableto lift this weight easily (a determination made by thesubject and the experimenter together), then an addi-tional 5 lb were added until a final maximum value wasreached. Between each of the sets, the subjects had a 30-srest period during which the investigator placed theweights on the bar in such a way so as not to arouse thesuspicion of the subject that any manipulation wastaking place. The subject prepared for the lift by lyingon the bench with his or her eyes closed (suggested to thesubjects as a preparation strategy), during which timethe collars were placed - actually or purportedly - onthe bar for the appropriate group, 'for safety purposes,to make certain the weights are on the bar securely',without arousing the suspicion of the subject. This was

the procedure underlying the experimental manipula-tion.

From the first two sessions, a baseline measure of theactual maximum weight lifted (BM) on the bench presswas obtained. The light group and the control groupused two 5-lb collars, whereas the heavy group did notuse collars. However, all of the subjects were told thatcollars were being used as an essential safety precaution.This illusion, together with subject safety, was facili-tated by the use of'spotters' who loaded the bar, lifted iton and off the subject and made sure that the weightdiscs stayed in place. The subject was unable to see theends of the bar (their view was blocked by the discs) andinspection of the post-test questionnaires showed thatthey were unaware of the deception. This procedure,under which subjects still lifted exactly the weight theyrequested, facilitated the later manipulation. Prior toany lifting at the first session, the subjects filled out thepre-study inventory. The PSE was also administered(PSE 1) to measure the level of trait self-efficacy in thesubjects prior to any experimental manipulations.

At the second session, the first measure of situationalself-efficacy (SSE1) was taken. The day's maximum forsession two was the maximum from session one. At theend of each session, the subjects were asked to circle apercentage which corresponded to the amount of effortthey put into their lifting that day. They were also askedto indicate anything that may have hindered theirperformance that day. These questions were included asa measure of any additional factors which the subjectsbelieved may have influenced their performance duringthat particular session. Once again, post-hoc inspectionsdid not display any differences or appreciable problemswith the subjects' performances.

The third session followed the same procedures as thefirst and second sessions; however, the warm-up repeti-tions were with dumbbells rather than the bar itself, toensure that the initial lift of the bar would be 'spared'until the subject attempted his or her maximum. Thiswas in accordance with Feltz's (1982) idea that self-efficacy manipulation would only affect the trial im-mediately following the change. The subjects were toldthat 'by the end of today's session, we would like to seeyou lift 10 lb more than yesterday', to create a situationwhich would be perceived as a 'barrier'. It was also anattempt to provide an incentive and motivation assomething to 'shoot for'. The subjects responded to thisfirst target maximum via SSE2. For this session,though, the light group now did not use collars, theheavy group used the collars and the control groupcontinued as before. As a result, actual weight lifted waseither the same (control), 10 lb lighter (light) or 10 lbheavier (heavy) than what the subjects requested andhence believed it to be. Those subjects who queried theweight they had just attempted were given false

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ewca

stle

(A

ustr

alia

)] a

t 11:

45 0

3 Se

ptem

ber

2014

170 Wells et al.

feedback to maintain the illusion. After this attempt (atwhat the subjects believed to be the previous day'smaximum), the next self-efficacy measure (SSE3) wastaken. This asked: 'Now, having attempted your pre-vious maximum, how confident are you in your ability tomeet today's maximum?' This was a key check of themanipulation, to determine if any cognitive changesoccurred. The subjects then had the opportunity toattempt 10 lb more than what they believed they hadattempted earlier.

For the fourth session, the subjects first estimated theamount of weight they felt they could successfully liftthat day (perceived maximum). Next, they responded toSSE4, with their stated perceived maximum being theday's target. The use of collars for this session was as forthe first two sessions (i.e. the light and control groupsused collars, the heavy group did not) and, as a result,the subjects attempted exactly the weight theyrequested. Following the dumbbell warm-up, the sub-jects attempted the weight they had indicated. Thisweight was recorded as the final maximum (FM). Afterthe final lift, the subjects were given the PSE again(PSE2) and the post-study inventory, which included acheck to ensure the subjects had indeed felt they hadbeen given the opportunity to lift to their maximumduring each session. Table 1 summarizes the procedure.

Results

Weight lifted

The means and standard deviations of all groups' lifts,both baseline (BM) and final maxima (FM), aresummarized in Table 2 and presented graphically in Fig.1. Inspection of the mean levels suggested cause forconcern since, despite the random allocation of subjects,the heavy group appeared to be very different from theother two. A simple contrast on either final or differencescores may be confounded by these initial differences.Therefore, to control for this potential confound, thefirst stage of analysis was completed by means of a one-way between-groups ANCOVA on the final maximawith the baseline values as covariates. The results arepresented in Table 3. Further analysis showed that

the homogeneity of regression assumption was satisfied,so the results presented in this table represent theanalysis of choice (Ferguson and Takane, 1989).

Not surprisingly, and indeed as a requirement for theeffective use of the ANCOVA, the regression betweenBM and FM is highly significant. The results also showthat, even allowing for initial differences, the groups aresignificantly different on FM. Follow-up to this resultwas performed by multiple t-tests on changes inperformance (FM—BM difference score means) withinthe three groups, using Holm's sequentially rejectedBonferroni adjustment to control for inflation of type 1error (Seaman et al., 1991). The use of change (i.e.difference) scores in this follow-up analysis was furtherprotection against the potential confound of initial, pre-manipulation differences between groups. The results,summarized in Table 4, show that the significantdifferences are due to changes in the light group.

The change in weight lifted from pre- to post-test wassignificantly greater in the light group, whereas theincreases achieved by both the heavy and control groupsdid not differ from each other. Despite the apparentmagnitude of this difference, however, effect sizecomparisons (Thomas et al., 1991) yielded only a smalldifference, with an effect size of 0.25 between the heavyand light groups on BM to FM. This apparentcontradiction in the results can be attributed to the highvariability in the subjects, an almost inevitable conse-quence of using a small group of novices rather thanexperienced and committed lifters.

Situational self-efficacy

A 3x4 (groups x trials) ANOVA with repeatedmeasures on trials (SSE1-SSE4) was performed inorder to determine any differences between groups andacross scores for self-efficacy. The results are presentedgraphically in Fig. 2 and the means and standarddeviations by group across SSE1—SSE4 are shown inTable 5.

A significant F value (F221 = 6.68, P<0.001) for themain effect of group indicated a difference betweengroups across the four measures. Further comparisonusing a Tukey test showed the heavy group (tn = 4.66) to

Table 1 Summary of procedures

Session 1 Session 2 Session 3 Session 4

• Pre-study inventory• PSE1• Lift to maximum

SSE1Lift to maximum

• SSE2 • SSE4• Prior maximum attempt (with the • Maximum lift: FM (without the

manipulation) manipulation)• SSE3 • PSE2• Second maximum attempt • Post-study inventory

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ewca

stle

(A

ustr

alia

)] a

t 11:

45 0

3 Se

ptem

ber

2014

Self-efficacy and strength 111

Table 2 Mean ( + S.D.) baseline maximum andfinal maximum weight lifted (lb) for each group

Group Baseline (BM) Final (FM)

Light 146.25 + 67.12 164.38±68.74Heavy 116.25 + 42.40 118.75 + 45.18Control 160.63 + 59.97 165.00+60.89

110BASELINE MAX. FINAL MAX.

Figure 1 Means for each group for baseline maximum (BM)and final maximum (FM) weight lifted (lb).

Table 3 Between-groups ANCOVA on final maximum(FM) with baseline maximum (BM) as covariate

d.f. M.S. Significance

Within-cellsRegressionGroup

2012

31.4272 680.96

558.292313.16

17.77P<0.001P< 0.001

Table 4 Inter-group contrasts on FM—BM scoresbetween groups

Group

LightControlLightHeavyHeavyControl

X + S.D.

18.1±6.54.37 ±4.218.1+6.52.5 + 5.92.5 + 5.9

4.37±4.2

t

r14 = 5.03

rH = 5.00

r1 4=-0.73

Significance

P<0.001

P<0.001

P<0.5

Note: All significance levels are two-tailed and adjusted by Holm'ssequential rejected Bonferroni method.

oza.

gu.

SSE 1 SSE 2 . S S E 3 SSE 4

Figure 2 Mean SSE ratings by group. * Measure takendirectly after manipulation.

be significantly different from both the light (m = 6.03)and control (jn = 5A\) groups. A significant main effectof trials was also found (F3j63 = 4.12, P<0.01). Specificcomparisons of each SSE rating showed the differencewas manifest between SSE2 (m = 5.13) and SSE3(m = 4.92), the measures prior to and following themanipulation, respectively. Finally, a significant inter-action effect was also apparent (.F663 = 6.38, P<0.001).

Follow-up tests indicated that while all groups' SSEratings appeared to be affected by the manipulation, theheavy group was negatively affected with a mean of 4.88at SSE2 and of 2.75 at SSE3, in contrast to the light andcontrol groups, for whom the effect was positive butnon-significant. Additional comparisons involvingeffect size revealed very large differences between thelight and heavy groups on SSE3 especially, with aneffect size of 3.8.

Self-efficacy - performance relationship

To further examine the extent to which self-efficacy wasrelated to eventual performance, two stepwise regres-sion analyses were completed. The first used FM as thedependent variable with SSE ratings 1-4. The secondused an index of improvement (FM—BM) as thedependent variable.

With FM as the dependent variable, only SSE1 wasaccepted into the equation as a significant variable(i7

122 = 5.03, P<0.05), although it only accounted for18% of the variance. Thus SSE1 probably represents ameasure of pre-existing self-efficacy. Since all SSEratings would have been highly correlated, the pin value

Table S Mean (±S.D.) scores on the four SSE measurements for each group

Group

LightHeavyControl

"P<0.001.

SSE1

6.25 ±1.495.63+1.065.00±1.31

SSE2

5.38+1.064.88 ±1.36"5.13±1.25

SSE3

6.38 ±0.742.75 + 1.17"5.63 ±1.06

SSE4

6.13 + 0.845.38+1.195.88 ±0.99

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ewca

stle

(A

ustr

alia

)] a

t 11:

45 0

3 Se

ptem

ber

2014

172 Wells et al.

(probability of F to include) of 0.05 would haveprecluded the inclusion of the other independentvariables. This is shown to be the case by the correlationmatrix presented in Table 6. Therefore, in relation toFA1, which was heavily dependent on strength level,SSE1 appears to have been the best indication of anindividual's general opinion on his or her bench pressability. In fact, all the other SSE ratings showed littlerelationship with FM.

When the change in lifting performance (FM —BM)was used as the dependent variable, however, the samestepwise regression yielded the third SSE rating (SSE3,i.e. that taken immediately after the manipulation) asbeing the only independent variable to be included.Furthermore, in this case, there was a significantlygreater percentage of variance (almost 40%) accountedfor by the equation. This measure (SSE3) probablyrepresents a better indication of the specific self-efficacy-performance relationship and further demon-strates the power of the manipulation in affectingperformance. Not surprisingly, Table 6 shows that therewas an extremely significant correlation between SSE3and lift difference.

PSE scores

Finally, in order to determine if a manipulation ofsituational self-efficacy would have an effect on PSE,scores for pre- and post-test PSE measures (PSE1 andPSE2) were analysed in a 3 x 2 (groups x trials) ANOVAwith repeated measures on trials. A non-significant F(Fl25 = 0.25, P< 0.05) indicated there was no significantchange over time for any group. Thus, manipulatedsituational self-efficacy did not have a significant effecton physical self-efficacy. In addition, as Table 6 shows,the correlation between PSE1 and FM —BM was notsignificant. This suggests PSE, given that it is a validtrait measure of self-efficacy, should not be used as apredictor of situationally specific performance. Step-wise regressions with both BM and FM as dependent

variables also failed to support initial PSE as a signifi-cant factor in predicting performance.

By contrast, inspection of the relationships betweenPSE2 (the post-manipulation measure) and the SSE/performance change measures is suggestive of a closerrelationship. Although PSE did not change significantlyfrom pre- to post-test, it appears that the subjects'experiences with the bench press and experimentalmanipulations may have influenced the way in whichthey view their general physical prowess.

Discussion

The results of the present study are in partial accordancewith recent research on self-efficacy. It appears that self-efficacy can be manipulated and does indeed have arelationship with strength performance.

The present study showed significant changes in self-efficacy immediately following a manipulation, i.e. theamount of weight a subject actually lifted in contrast tothe amount of weight he or she believed was lifted. Nessand Patton (1979) achieved similar results, in that thosewho originally lifted less than they believed, actuallylifted greater weight in the final session than did thosewho originally lifted more than they believed. For thepresent study, it was hypothesized that what was takingplace cognitively was a re-evaluation of the subject'sability to perform the strength task that would manifestitself in the very specific measure of self-efficacy.Indeed, the subjects in the present study who lifted agreater weight than they believed (the 'heavy' group)immediately made a connection that their ability to lift aspecific weight previously attained was now changed.Thus the SSE rating immediately following the mani-pulation plummeted for this heavy group. This findingwas taken as evidence that the manipulation had beensuccessful on a cognitive basis - the subjects' beliefsabout performance capabilities were changing. While

Table 6 Correlation matrix of self-efficacy measures (PSE and SSE) with final maximum(FM), performance change (FM-BM) and PSE

FMP

F M - B MP

PSE1P

PSE2P

PSE1

0.310.07

0.2690.102

0.7460.000

PSE2

0.2080.164

0.3310.057

0.7460.000

SSE1

0.4230.02

0.4090.024

0.1070.310

0.4050.025

SSE2

0.1160.295

0.3280.059

0.1460.248

0.3890.03

SSE3

0.2480.121

0.6980.000

0.0720.369

0.3910.03

SSE4

0.010.48

0.3390.053

0.1120.292

0.4200.02

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ewca

stle

(A

ustr

alia

)] a

t 11:

45 0

3 Se

ptem

ber

2014

Self-efficacy and strength 173

not significant, the 'light' group's ratings for SSE3increased slightly following the manipulation.

The SSE3 rating was also highly correlated with liftdifference. This may have been because, while the lightgroup has higher ratings on SSE3, they also had agreater lift difference, with some subjects increasing theweight lifted from baseline by as much as 25 lb. On theother hand, the heavy group had lower ratings on SSE3and subsequently either stayed the same with a liftdifference of zero, or increased only slightly. Onesubject in the heavy group even had a negativedifference score. Although initial group differencescould be advanced as a counter-explanation, the use ofANCOVA followed up by the difference scores miti-gates against this factor confounding the results. Inaddition, it should be noted that the heavy group did notdiffer significantly from the other two groups on eitherthe general (PSE) or initial specific (SSE) measures ofself-confidence. This strong effect was a most interest-ing finding, which has many implications for overcom-ing 'psychological barriers'. Simply by changing theirbeliefs - in this case using false performance feedback(Mahoney, 1979) - individuals can lift beyond theirpresent maximum performance. It would appear thatnegative accomplishments which serve to lower aperson's self-efficacy also keep that person from makingany significant gains in performance. But of equalinterest, the control group had neutral messages, nonegatives and made no attempt to enhance their self-efficacy. As a result, there was no significant perform-ance gain, even though the control subjects were alsogiven the opportunity to improve in the final session.This particular finding could have implications forcoaches seeking to improve performance via feedbackwhich affects self-efficacy.

Another interesting finding was that the heavy groupsomehow 'recovered' from their doubt (noted in SSE3ratings) as reflected in SSE4, which was more consistentwith SSE1 and SSE2. A possible explanation for thiscould be the wording of the SSE4 question, as itfollowed the subjects' predictions of their day's maxi-mum lift. In this matter, the subjects would likely havebeen extremely confident in their ability to complete thelift successfully when it was a weight they had chosenthemselves. They may not have been particularlypleased with the weight, but they had little doubt theycould lift it!

Another possible explanation lies with pre-existingself-efficacy as Weinberg et al. (1981) noted. Pre-existing self-efficacy had an effect on the first trial only,and beyond the first trial a better measure was previousperformance. Feltz (1982) further supported this asser-tion. If this is so, then it would be expected that the moregeneral PSE measure, representative of trait self-efficacy (here in terms of 'physical' self-efficacy), would

affect ratings of self-efficacy on this task. Thus, whileSSE may fluctuate across trials, self-efficacy ratings atthe final session may have been influenced more by thelevel of stable trait PSE than by the more unstablesituational self-efficacy. Unfortunately, as the earliercorrelation matrix and regression analysis showed, thisexplanation was not supported by the results.

As expected in the present study, the PSE scores didnot change significantly from PSE1 and PSE2 in anygroup. What makes PSE different from self-efficacy isthat it encompasses self-confidence and self-esteemmore closely and their relation to general physical/motortasks and appearance, whereas self-efficacy focuses onbeliefs about specific performance ability. The PSE wasincluded in this study to test its possible contribution asa factor affecting strength performance. Unfortunately,the pre-manipulation measure did not correlate signifi-cantly with lift differences, contrary to the results ofGayton et al. (1986), who claimed that PSE was a majorpredictor of marathon performance. Of course, it ispossible that the questions used in the PSE are morerelated to aerobic fitness or endurance than to strength.Inspection of the questions yields only two which relateto strength (items 4 and 21), while other items eitherpertain directly to aerobic fitness or to constructs whichresearch suggests are more closely associated with it (e.g.stress response, sweating, etc.).

The different picture provided by PSE2 (the post-manipulation measure) must also be considered. Theresults are suggestive of an influence of the specificphysical experience (bench press) on the purportedlystable trait measure of PSE. Of course, PSE changeswere non-significant and it may be that the degree ofinfluence was greater than usual since the task washighly novel to these novice subjects. Whatever thereason, further investigation is indicated and, based onthe present study, application of the PSE to specificsituations must be questioned.

A desirable extension of the present study may proveto be a valuable tool in validating the assertions madehere. By using a double-dissociation model and simplyextending the study by an additional four sessions, anumber of factors could be examined by use of pathanalysis statistics. In the second four sessions, the groupwould be reversed, the 'heavy' group repeating foursessions but under the 'light' conditions, and vice versa.In so doing, it would be interesting to see if PSEremained constant throughout the second condition aswell. It may also be possible to note directionalinfluences or an order effect upon performance at theend of the second four sessions in relation to the first foursessions. Changes in SSE that parallel original changesfor the new 'light' or 'heavy' group would be furtherevidence of the manipulation being successful in chang-ing one's self-efficacy via past performance.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ewca

stle

(A

ustr

alia

)] a

t 11:

45 0

3 Se

ptem

ber

2014

174 Wells et al.

The overall importance of this study was that whileother researchers were manipulating general conceptssuch as motivational cues (spotter behaviour: Carnahanet al., 1990), expectations (opponent's strength: Nelsonand Furst, 1972) and external cues (apparent weightlifted: Ness and Patton, 1979), this study linked each ofthese general constructs to a very specific piece of thecognitive puzzle - self-efficacy. It appears that each ofthe general constructs is either in part determined bylevels of self-efficacy for the particular task (motiva-tional cues and expectations) or has a cyclic effect oncognitions and self-efficacy (external cues). This meansthat external cues play a part in engendering cognitivechanges that are reflected by ratings of self-efficacy.Therefore, the suggestion supported by this study thatself-efficacy and performance are indeed related is apromising finding and has many implications forperformance enhancement.

In conclusion, consideration should be given to asimilar study on manipulation of self-efficacy amongexperienced weightlifters (Fitzsimmons et al., 1991),which was reported subsequent to the completion andsubmission of the present paper. The results of theFitzsimmons and co-workers' investigation matchedthose reported here and were, once again, in accordancewith Ness and Patton (1979), although the improve-ments obtained by their subjects were smaller. This isnot surprising, however, given that the focus of theinvestigation was experienced lifters, who would inevit-ably have less room for improvement. Two notabledifferences in methodology between Fitzsimmons andco-workers' study and the present experiment meritattention.

First, in assessing self-efficacy, Fitzsimmons et al.asked their subjects to provide a figure (weight) whichthey were 100%, 75% or 50% confident they could lift.This approach, once again a departure from the typicalBandura approach of set increments which was des-cribed earlier, appears arbitrary and somewhat ambi-guous. What is the personal meaning of a 50%confidence for example? Does this mean 'I can make thelift 50% of the time' or 'This is a risky attempt - I mightmake it with luck?' Unfortunately, lack of clarificationor post-hoc enquiry means that this question is leftunanswered. Secondly, given that the experiencedpopulation of lifters used in this study would appear tohave a better internalization of the skill demands andhence be harder to fool, the lack of manipulation checkswas a surprising omission. On the positive side, thestudy did attempt a double-dissociation paradigm, butunfortunately failed to consider order effects, probablythe most interesting factor in such a manipulation.However, this study by Fitzsimmons et al. does serve tofurther support the application of self-efficacy based

manipulations in work with athletes, albeit such mani-pulations would be most effective with novices.

References

Bandura, A. (1977). Self-efficacy: Toward a unifying theory ofbehavioral change. Psychological Review, 84, 191-215.

Bandura, A. (1982). Self-efficacy in human agency. AmericanPsychologist, 37, 122-147.

Bandura, A.V. (1986). Social Foundations of Thought andAction: A Social Cognitive Theory. Englewood Cliffs, N.J.:Prentice-Hall.

Bandura, A. (1990). Perceived self-efficacy in the exercise ofpersonal agency. Journal of Applied Sport Psychology, 2,128-163.

Carnahan, B.J., Shea, J.B. and Davis, G.S. (1990). Motiva-tional cue effects on bench press performance and self-efficacy. Journal of Sport Behavior, 13, 240-254.

Feltz, D.L. (1982). Path analysis of the causal elements inBandura's theory of self-efficacy and an anxiety-basedmodel of avoidance behavior. Journal of Personality andSocial Psychology, 42, 764-781.

Feltz, D.L. (1988a). Gender difference in the causal elementsof self efficacy on a high avoidance motor task. Journal ofSport and Exercise Psychology, 10, 151-166.

Feltz, D.L. (1988b). Self confidence and sport performance.In Exercise and Sport Science Reviews (edited by K.B.Pandolf), pp. 423-457. New York: Macmillan.

Feltz, D.L. and Mugno, D.A. (1983). A replication of the pathanalysis of the causal elements in Bandura's theory of selfefficacy and the influence of autonomic perception. Journalof Sport Psychology, 5, 263-277.

Ferguson, G.A. and Takane, Y. (1989). Statistical Analysis inPsychology and Education. New York: McGraw-Hill.

Fitzsimmons, P.A., Landers, D.M., Thomas, J.R. and van derMars, H. (1991). Does self-efficacy predict performance inexperienced weightlifters? Research Quarterly for Exerciseand Sport, 62, 424-431.

Gayton, W., Matthews, G. and Burchstead, A. (1986). Aninvestigation of the validity of the Physical Self-EfficacyScale in predicting marathon performance. Perceptual andMotor Skills, 63, 752-754.

Mahoney, M.J. (1979). Cognitive skills and athletic perform-ance. In Cognitive-behavioral Interventions: Theory,Research, and Procedures (edited by P.C. Kendall and S.D.Hollon), pp. 205-241. New York: Academic Press.

McAuley, E. and Gill, D. (1983). Reliability and validity of thePhysical Self-efficacy Scale in a competitive sport setting.Journal of Sport Psychology, 5, 410-418.

Nelson, L.R. and Furst, M.L. (1972). An objective study ofthe effects of expectation on competitive performance.Journal of Psychology, 81, 69-72.

Ness, R.G. and Patton, R.W. (1979). The effect of beliefs onmaximum weight-lifting performance. Cognitive Therapyand Research, 3, 205-211.

Ransom, K. and Weinberg, R. (1985). Effect of situationcriticality on performance of elite male and female tennisplayers. Journal of Sport Behaviour, 8, 144-148.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ewca

stle

(A

ustr

alia

)] a

t 11:

45 0

3 Se

ptem

ber

2014

Self-efficacy and strength 175

Ryckman, R.M., Robbins, M.A., Thornton, B. and Cantrell,P. (1982). Development and validation of a Physical Self-Efficacy Scale. Journal of Personality and Social Psychology,42, 891-900.

Seaman, M.A., Levin, J.R. and Serlin, R.C. (1991). Newdevelopments in pairwise multiple comparisons: Somepowerful and practicable procedures. Psychological Bulletin,110, 577-586.

Shelton, T.O. and Mahoney, M.J. (1978). The content andeffect of 'psyching up' strategies in weight lifters. CognitiveTherapy and Research, 2, 275-284.

Thomas, J.R., Salazar, W. and Landers, D.M. (1991). What ismissing in P<.05? Effect size. Research Quarterly forExercise and Sport, 62, 344-348.

Weinberg, R.S. and Jackson, A. (1990). Building self-efficacyin tennis players: A coach's perspective. Applied SportPsychology, 2, 164-174.

Weinberg, R.S., Gould, D. and Jackson, A. (1979). Expec-tations and performance: An empirical test of Bandura'sself-efficacy theory. Journal of Sport Psychology, 1, 320-331.

Weinberg, R.S., Gould, D., Yukelson, D. and Jackson, A.(1981). The effect of pre-existing and manipulating self-efficacy on a competitive muscular endurance task. Journalof Sport Psychology, 4, 345-354.

Weiss, M.R., Wiese, D.M. and Klint, K.A. (1989). Head overheels with success: The relationship between self-efficacyand performance in competitive youth gymnastics. Journalof Sport and Exercise Psychology, 11, 444-451.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ewca

stle

(A

ustr

alia

)] a

t 11:

45 0

3 Se

ptem

ber

2014