type and frequency of praise

21
Type and frequency of praise as determinants of favorability of self-image: An experiment in a field setting' Reuben M. Baron/ Alan R. Bass, and Peter M. Vietze/ Wayne Sfate University The overall purpose of the present research was to devise a set of procedures which would efFectively commumcate a "success expenence" to a sample of lower-class Black youth who were engaged in vanous job-traming programs The specific objective of the study was to ascertam how social evaluation might be most effectively presented m order to produce a more favorable self-image The major focus was on changes m self-image because it was assumed that satisfactory use of whatever skills an individual possesses will not be achieved unless an adequate level of self- esteem exists For example, it has been found that (a) indi- viduals of high self-esteem are more hkely to engage m success- ful goal-oriented behavior than those of low self-esteem (Stot- land, Thorley, Thomas, Cohen, & Zander, 1957, Denmark & Gut- tentag, 1967, Korman, 1970), (b) persons who are led to believe they are unqualified may react to a high level of pay by decreas- ing their productivity (Adams & Rosenbaum, 1962, Andrews, 1967), (c) mdividuals who have encountered repeated failure in an experimental situation lower their performance score in the face of success (Aronson & Carlsmith, 1963), (d) subjects who X This research was supported by Contract No 81-24-66-04, from the Manpower Adnimistration, U S Department of Labor We wish to express our appreciabon to Mr Donald Healas, Dr Wilham Welke, Mr Burt Bradley, Mr Jude Cotter, and Mr Haskell Stone of the Mayor's Youth Employment Project City of Detroit, for their cooperation in providmg subjects for this study and facilities for conducbng the research The wnbng of this report was supported by NSF grant CS-2373 to the first author 2 Requests for repnnts should be sent to Reuben M Baron, Department of Psychology, Wayne State University, Detroit, Michigan, 48202 3 Now at I>epartment of Psychology, Peabody College, Nashville, Tennessee

Upload: montclair

Post on 01-Mar-2023

1 views

Category:

Documents


0 download

TRANSCRIPT

Type and frequency of praise asdeterminants of favorability of self-image:An experiment in a field setting'Reuben M. Baron/ Alan R. Bass, and Peter M. Vietze/ WayneSfate University

The overall purpose of the present research was to devise aset of procedures which would efFectively commumcate a "successexpenence" to a sample of lower-class Black youth who wereengaged in vanous job-traming programs The specific objectiveof the study was to ascertam how social evaluation might bemost effectively presented m order to produce a more favorableself-image

The major focus was on changes m self-image because it wasassumed that satisfactory use of whatever skills an individualpossesses will not be achieved unless an adequate level of self-esteem exists For example, it has been found that (a) indi-viduals of high self-esteem are more hkely to engage m success-ful goal-oriented behavior than those of low self-esteem (Stot-land, Thorley, Thomas, Cohen, & Zander, 1957, Denmark & Gut-tentag, 1967, Korman, 1970), (b) persons who are led to believethey are unqualified may react to a high level of pay by decreas-ing their productivity (Adams & Rosenbaum, 1962, Andrews,1967), (c) mdividuals who have encountered repeated failure inan experimental situation lower their performance score in theface of success (Aronson & Carlsmith, 1963), (d) subjects who

X This research was supported by Contract No 81-24-66-04, from theManpower Adnimistration, U S Department of Labor We wish to express ourappreciabon to Mr Donald Healas, Dr Wilham Welke, Mr Burt Bradley, MrJude Cotter, and Mr Haskell Stone of the Mayor's Youth Employment ProjectCity of Detroit, for their cooperation in providmg subjects for this study andfacilities for conducbng the research The wnbng of this report was supportedby NSF grant CS-2373 to the first author

2 Requests for repnnts should be sent to Reuben M Baron, Department ofPsychology, Wayne State University, Detroit, Michigan, 48202

3 Now at I>epartment of Psychology, Peabody College, Nashville, Tennessee

494 Reuben M. Baron, Alan R. Bass, and Peter M Vietze

think poorly of themselves evaluate a person who gives thempositive feedback less favorably than someone who gives themnegative feedback (Deutsch & Solomon, 1959, Howard & Berko-witz, 1959, Wilson, 1965). The above lmes of evidence clearlysuggest the importance of devismg techniques to raise the level ofself-evaluation for persons havmg a negative self-image, a stateof afiFairs which is likely to obtain for a Black subject population(Dreger & MJler, 1968).

It was decided that our mitial studies would focus on thecomparative effectiveness of different kinds of social reinforce-ment parameters The general efficacy of short-term social rem-forcement procedures m modifying self-image has been suggestedby a number of recent investigators ( e g , Hass & Maher, 1965,Marlowe, 1962). In order to heighten the relevance of such anapproach for the present subject population, we further decidedto focus on reinforcement parameters that would be likely tobear on the unique remforcement histories of our subjects Itwas also hoped that this mvestigation would make a generalcontnbution to our understandmg of the conditions under whichself-evaluative processes may be modified

Unfortunately, most of the research which has investigatedthe differential efficacy of reinforcement parameters for lower-class populations has (a) utilized Caucasian children (e g., Zig-ler & Kanzer, 1962), and (b) focused on producmg mcrementsin specific task behaviors rather than self-image. Despite suchproblems it is possible to find variables of relevance to the presentconcerns by extrapolatmg from other contexts.

A number of recent mvestigations have demonstrated thatone difference m the sociahzation process between lower andmiddle-class children appears to be the differential reward valueplaced upon bemg personally "nice" as opposed to bemg corrector demonstratmg task mastery (Douvan, 1956, Terrall, Durkin,& Wiesley, 1959, Zigler & deLabry, 1962, Zigler & Kanzer, 1962)Although there has been some contradictory evidence (eg,Rosenhan & Greenwald, 1965) it has generally been suggestedthat lower-class children are likely to place a higher value onperson- as opposed to performance-oriented praise (e g., you'rea good guy vs. that's a good job), the reverse tends to be true for

Type and frequency of praise 495

middle-cIass children (see Zigler & Child, 1969, Havighurst, 1970,for detailed reviews of this position).

While there is a dearth of empirical data bearmg on the ante-cedents of such preferences, it has been reported, for example,that the Negro child in an academic settmg is reinforced largelyfor maximizmg neatness, good manners, and general docilityrather than for academic achievements (Dreger & Mdler, 1968)Further, the type of reinforcement Negro children receive is of" a more diffuse type than focusing on the adequacy of thechild's performance" (Dreger & Miller, 1968, p 9).

The specific problem we focused on m regard to this variablewas whether this hypothesized difference m the reward value ofpersonal- and achievement-onented praise, whether its origin becultural or developmental, would persist into late adolescencewith a sample of lower-class Black youth

Recently, the first author (Baron, 1966, Baron, Robmson &Lawrence, 1968, Baron, 1970) has suggested a model of the de-termmants of social remforcer effectiveness which relates theeffects of the current availabihty of a remforcer to past levels ofattamment. Baron assumes that based upon their past history ofreinforcement mdividuals develop a conception of what is anappropnate level of reward for them. Baron (1966) refers to thismtemal norm as a social reinforcement standard (SRS). Thisstandard defines a preferred region around which one seeks tosecure future reinforcement, so long as reinforcement levels fallwithin the region of the SRS, 1 e., are not substantially above orbelow, a person is able to mamtam a predictable mterpersonalenvironment. When a substantial dispanty exists a person isfaced with uncertamty as to how he should view himself aswell as uncertamty as to how to view the source of evaluation.Such evaluative uncertamty is hkely to prevent an "unequivocalbehavior orientation" and hence mterfere with a person's abilityto perform optimally (Stevenson & Weir, 1961, Baron, Robinson &Lawrence, 1968).

The present subject population may be assumed, m general,to have experienced a relative deficit of white authonty-figureapproval, this deficit being particularly pronounced m regardto achievement-oriented praise (Baron, 1970, Dreger & Miller,

496 Reuben M Baron, Alan R Bass, and Peter M. Vietze

1968). Given this assumption it should then follow that a rela-tively high level of praise from a white evaluator would be cat-egorized as less appropnate than a relatively low level of praise.This should be particularly true when achievement- as opposedto personally-onented praise is utilized

Fmally, it should be noted that although the evaluative feed-back m the present study is given m the context of actual taskperformance and hence can potentially aflFect performance, weview the task situation as essentially a vehicle to plausibly con-vey different types and frequencies of success experiences Thetasks were selected so as to be well within the already existingskill repertoire of our subjects, therefore, any effects one wouldobtam m the task realm are likely to reflect mcreases m generalmotivation to do well rather than the acquisition of new skills,1 e , performance rather than leammg.

In the selection of a particular method to dispense reinforce-ment we were primarily guided by a desire to achieve maximumcontrol over the number of success expenences the subjects en-countered rather than a desire to convey any specific mforma-tion to the subjects which might improve their performanceTherefore, we used a schedule where variations m the frequencyof praise were not tied to the quahty of the subjects' actualresponses but rather to randomly selected blocks of tnals Itshould be noted, however, that even a noncontmgent proceduresuch as this may be efFective m raising performance level if thetype of response called for is over-learned and hence is likely todepend more upon arousal or energy mobilization effects ratherthan the informational or cue function of a reinforcer Thepresent mode of dispensmg reinforcement, if it motivates thesubject to expend more effort and facihtates attentional processes,may, dependmg upon the task behavior required, improve per-formance (e.g, where speed of response is more important thanaccuracy of response)

In sum, while the effect of the present variables on task per-formance IS uncertain, if effects do occur they should be mostevident on simple, speeded tasks Furthermore, any effects wedo find should follow the same pattern as expected for the self-image measures, where achievement-onented praise is involved.

Type and frequency of praise 497

task performance should be best with low frequencies of praiseIt should also be noted that given the dearth of previous researchwhich: (a) has been conducted vidth reinforcement parameterssimilar to ours, (b) utihzed lower-class Blacks m their late teens,or (c) has investigated self-image effects with these vanables, thepresent study must of necessity be viewed as largely exploratory—an attempt to map a realm of investigation rather than an at-tempt to establish functional relationships of high reliability andgenerality

METHOD

Subjects

The subj'ects for this experiment were trainees enrolled m aNurses' Aide trammg program offered by the Mayor's Youth Em-ployment Project (MYEP) m the City of Detroit, Michigan Thetramees were referred to the trammg program by the Michigan StateEmployment Service, and were pnmanly girls who were consideredhy employment counselors to be imemployable at present but who,it was felt, would benefit from this type of trammg The final sampleconsisted of 28 Negro girls between 16 and 21 years of age with amedian age of 19* Most of these girls had had some high schooltrammg but did not have a high school diploma They participatedm the expenment durmg the onentation phase of the Nurses' Aidetrammg program before bemg exposed to any specific Nurses' Aidetrammg. Subjects were assigned to the various experimental conditionson a random basis.

Frocedure

The settmg for the study was the trammg center itself. Further-more, the experimenters, all of whom were white graduate students,were mtroduced as part of the training staff at tiie center These pro-cedures were followed so that our experimental mterventions wouldhe as "natural" and nondisrupbve to tiie trammg program as possible

Self-tmage measures A few days pnor to the expenmental pro-cedures and a few days after the expenment, each subject was ad-mmistered the Tennessee Self-Concept Scde (Fitts, 1965) and theInterpersonal Perception Questionnaire The pre- and post-measureswere admimstered by different experimenters, neither of whom was

4 Overall sample size was determined by the size erf the partaciilar trammggroup that we had access to at that tune Tlie sample size could not be readilyaugmented because it was impossible to predict when another comparable groiipof tramees would be processed at the center

498 Reuben M Baron, Alan R. Bass, and Peter M Vietze

involved m the administration of the experiment proper The TennesseeSelf-Concept Scale is a published, standardized scale which has beenpreviously used m research with subjects of similar age and classbackground as those m the present study (Fitts, 1965) iTus question-naire provides an overall self-concept measure as well as a number ofsubscale scores which focus on different sets of behefs about one'sself The test-retest rehabiLty for the total score is 92 (Fitts, 1965)The eight subscales used m the present research are (1) physicalself, (2) moral-ethical self, (3) personal self, (4) family self, (5)social self, (6) self-identity, (7) self-satisfaction, and (8) behaviorself Subjects responded to statements such as "I am an attractiveperson" on a five-point scale which ranges from Completely False toCompletely True The Interpersonal Perception Questionnaire wasconstructed for use in this research project to assess self-esteem, andconsists of twelve evaluative semantic differential scales such as good-bad, clean-dirty, and attractive-unattractive The subjects were askedto respond to this set of twelve scales six different times, as they seethemselves, and then as they feel their parents, teachers, employers,male fnends and female fnends perceive them This procedure yieldsSIX different measures of the subject's self-esteem

Since it was quite likely that the several measures of beliefs aboutself and self-esteem measures were mterrelated, a prebmmary studywas conducted to mvestigate the extent and nature of mtercorrelationswithm and between these self-image measures For this purpose asample of 111 high school students was obtamed from a local pubLchigh school where the students were of essentially similar backgroundand age as the subjects m the experimental sample Intercorrelationsamong the self-image measures obtamed for this high-school sample,m general, were low to moderate, with correlations withm the Ten-nessee Self-Concept measures and withm the Interpersoruil PerceptionQuestionnaire considerably higher than those between these sets ofmeasures Withm correlations for the Tennessee scale ranged from35 to 76, with the modal correlabons bemg m the 6o's For the mter-personal questionnaires, withm correlations ranged from 32 to 62, withthe bulk of the correlations bemg m the 40's and 50's Correlationsbetween these scales, on the other hand, ranged from .19 to 51, withthe majonty of the correlations m the 30's

Task MeasuresPerformance gams on three different tasks were also assessed,

thereby providing additional mformation about the motivational ef-fects of the reinforcer treatments The tasks were chosen so as to

Type and frequency of praise 499

simulate the range of operations that might be encountered dunng theactual training program. Such a procedure was assumed to mcreasethe likelihood that verbal reward treatments would be taken senously.In addition, the three tasks and the admmistrabon procedures weredesigned so that tiie subjects would have a great deal of difficulty mevaluating the quahty of their performance without external feed-back This was accomphshed by usmg ambiguous stimulus displays,by allowmg too little time for an adequate response, and/or by sconngthe task m terms of time to complete a trial These procedures wereintroduced m order to make an arbitrary, prearranged reinforcementschedule credible to the subject

TTie order of presentation of the three experimental tasks wasvaried randomly for the subjects The tasks were admmistered indi-vidually Each task was mtroduced with two practice tnals which werethen followed by two sets of twelve experimental tnals for a totalof 24 experimental tnals

Word recognition task For the word recognition task, fifteen four-and five-letter words were pnnted on a Ime with no divisions be-tween words and with several randomly selected letters capitalizedFor each tnal, which consisted of one such line of words, the subjectwas instructed to draw vertical hnes m front of the first letter andafter the last letter of each four- and five-letter word that she recog-nized m the Ime Subjects were informed that they would only have20 seconds to work on each list and that this would seldom allowtime to complete a trial After each tnal the experimenter reinforcedthe subject or remamed silent, accordmg to the schedule It wasassumed that the restncted time and the ambiguous nature of thetask would afford the subject little feedback as to the quality of herperformance in the absence of information from the experimenter

Visual percepttveness task For this task, on each tnal the ex-perimenter projected four slides containmg different geometnc de-signs, and then mstructed the subject to remember them so that shewould be able to recognize them as they appeared m a senes of nmeshdes which would follow All of the shdes were presented tachistis-copically. At one second intervals the subject indicated her choicesby checkmg the appropnate numbers on a slip of paper After thesubject fimshed, the experimenter reinforced or remained silent ac-cordmg to the appropriate schedule It seems reasonable to assumethat the tachistiscopic presentation of the rather similar geometncdesigns afforded little feedback to the subject concemmg her per-formance on this task

Manual dexterity task The Purdue Pegboard was utilized to

500 Reuben M. Baron, Alan R. Bass, and Peter M. Vietze

measure manual dextenty The subject was shown how to do themost difficult assembly used in the standardized procedtire with dem-onstrations by the experimenter After the subject had learned theprocedure for making this assembly, she was told that she would betimed on each trial The 24 experimental tnals were then administeredAfter the completion of a tnal the experimenter would record thetime and reinforce or remam silent according to the schedule It washoped that the subject's preoccupation with the task would tend tomask her perception of the amount of time it took

Independent Variables

The two independent vanables were frequency of praise (25%or 75%) and type of praise (person or task performance) The actualtnals on which the verbal reuiforcers were administered were selectedrandomly but were tiie same for all subjects, this procedure meant thatsubjects were receivmg or not receivmg praise mdependently of theactual quality of their performance In the 25% condition, six of the24 tnals were reinforced In the 75% condition, 18 of the 24 trials werereinforced There were two types of reinforcement at each level ofreinforcement person reinforcement was admmistered by the exam-iner saymg, "You're pretty good," or "You're fast," or "You're quick,"on arbitranly chosen tnals Performance reinforcement was accom-plished by saymg, "That's a good job," or "That's a fine job," or "That'sa nice job," on a fixed, arbitrary schedule of tnals Each of thevanous statements withm each condition was used equally often

Dependent VariablesThe dependent vanables used to measure changes m self-image

were test scores from pre- and post-admmistrations of the two paper-and-pencil measures of self-image (both total score and subscalescores were compared) The cntenon measures for the performancetasks mvolved the level of performance for the last six tnals of thereinforced phase of the task The specific nature of the score dependedupon the task for the Word Recognition Task mean number of wordscorrectly identified was utilized, for the Visual Perceptiveness Taskthe measure was the number of shdes per tnal correctly identified ashavmg previously been shown or not havmg been shown, on theManual Dextenty Task, the dependent measure was the mean tunein seconds taken to complete the four assembhes on each tnal Thesecntenon measures were adjusted for covanance analysis by the ap-propnate performance score for the two practice tnals

In order to assess the eflFects of the independent vanables in this

Type and frequency of praise 501

study, VIZ, type of reinforcement (person and performance) andfrequency of praise (25% or 75%) on changes in self-unage and ontask performance, a senes of 2 X 2 analyses of covariance with im-equal cell frequencies were used ^ For the self-image data, scores onthe premeasures served as the covanates while the delayed post-treatment scores were used as the dependent vanables For the per-formance data, average scores on the two nonreinforced practice trialsserved as the covanates while the average scores for the last block ofSIX tnals during the treatment phase served as the dependent vanables ®

RESULTS

Self-image Data'

With respect to the findings involving the self data, it be-comes quite evident that, overall, the experimental treatmentshad a strong impact on the tramee's self image. In general, it wasfound that lower frequencies of social reinforcement were moreeffective in enhancmg the mdividual's self-image than werebgher levels of social reinforcement (see Table 1) This efiFectreached sigmficance for all the "other" evaluation scales of theInterpersonal Perception Questionnaire (self as parents, teachers,employers, male and female fnends see me, "self as I see myself"was m the same direction but not significant) A significant mameffect mdicatmg greater positive self-change for lower frequen-cies of reinforcement was also found on the self-identity andfamily-self scales of the Tennessee Self-Concept Inventory (seeTable 1)

In addition, it appears that praise of the person is generallymore effective than praise of task performance m enhancmg self-concept This effect reached significance for two of the TeimesseeSelf-Concept scales (personal self and self-identity) and was of

5 Both the self-unage and performance data were originally analyzed usingdiange scores These analyses yielded essentially the same findings as thecovanance analyses but were not deemed appropriate because of difFerencesthat existed among some of the groups on the pre-measures Inspection of theindividual change scxjres as well as the group means revealed that we weredealmg with positive changes, 1 e , both self-image and performance wereenhanced over baseline levels m all condihons

6 The last block of six tnals was used because it was assumed that per-formance would stabilize by this pomt

7 The sample size for the self-image analyses was reduced to 23 because wedid not have pre-and-post scores available for 5 of the subjects, 1 e , they werenot present at both the pre- and post-admimstrabon of the self measures

502 Reuben M. Baron, Alan R. Bass, and Peter M Vietze

Table l Summary of analyses of vanance (F values) for self-imagemeasures.

Measure

Interpersonal Perception QuestionnaireSelf as 1 see myselfSelf as employers see meSelf as parents see meSelf as male friends see meSeif as teachers see meSelf as female friends see me

Tennessee Self-Concept ScaleTotal scoreIdentitySelf-satisfactionBehavior selfPhysical selfMorai-ethical selfPersonal selfFamily selfSocial self

Obiect ofreinforcement

(A)

2 7 72 674 68*3 383 761 49

2188 44 * *

< 1 00<1 00<1 00

3 527 05*4 1 5

<1 00

Frequency ofreinforcement

(B)

2 3411 70**5 34*6 50*5 70*9 72**

< 1 005 14*

< 1 00< 1 00< 1 00< 1 00

2 206 16*

< 1 00

A X B

< 1 003 41

<1 005 43*2 393 67

< 1 002 58

<1 00<1 00

1 152 23

<1 001 791 00

Note—N*p< 05

**p < 01

•• 2 3 for all analyses.

marginal significance on a third dimension of self, family-self(p = .06). Personal praise was also found to be significantly moreeffective than praise of performance for the "self as parents seeme" scale of the Interpersonal Perception Questionnaire

A significant interaction between type and frequency of re-inforcement was also obtamed on the "self as male friends seeme" scale (see Table 2)

A simple effects analysis elucidated this mteraction by dem-onstratmg that (a) at 75 percent remforcement, person-onentedpraise was significantly more effective than performance-orientedpraise, and (b) performance-focused praise delivered at a 25percent rate enhanced subjects' self-image significantly more thanperformance-onented praise given at a 75 percent level (p's <.05).

Performance Data

No significant results were obtamed for either the work rec-ognition or visual perception task For the manual dexterity task.

Type and frequency of praise 503

Table 2 Adjusted mean post scores for "myself as male fnends see me "

Rate of reinforcement25%

7 5 %

Totals

Ob|ect of reinforcement

Person

69 50(N = 6)

68 54(N = 6)

69 06(N = 12)

Performance

6972(N = 6)

65 99(N = 5)

68 23(N = 11)

Total

69 61(N = 12)

67 41( N = 11)

Note—Post-scores ore ad|usted by pre-scores. The higher the score the more favorablethe self-evaluation

Table 3 Adjusted mean number of seconds to complete manualdextenty task.

Rote of reinforcement25%

75%

Totals

Obiect of reinforcement

Person

24 59(N = 9)

23 88(N = 7)

24 28(N = 16)

Performance

21 74(N = 6)

24 64IN = 6)

23 19(N = 12)

Total

23 46IN = 15)

24 23(N = 13)

Note — T̂he higher the score, the poorer the performance Subjects' scores on the terminalblock of SIX trials during the treatment phase are ad|usted by their scores on the two nonreinforcedprachce trails.

however, several significant effects were obtamed (see Table 3)First, there was a significant effect for type of reinforcement,praise of the subject's task performance was found to be superiorto praise of the person herself m leadmg to improvement mperformance on this task (F = 672, p < 02) This findmg mustbe qualified, however, by the significant mteraction that occurredbetween type and frequency of reinforcement for this task (F =640, p < 02) An inspection of Table 3 reveals that this inter-action IS mamly accounted for by the superior performance ofsubjects m the praise of performance condition when there is alow frequency of praise

504 Reuben M Baron, Alan R. Bass, and Peter M Vietze

DISCUSSION

Changes in Self-image

Three major lands of effects emerged on the self-image data(a) praise of the person was generally more effective in raismgself-image than praise of task performance, (b) lower frequenciesof praise were generally more effective m producmg positivechanges m self-image than were higher frequencies of praise,(c) when a high frequency of praise was utdized and the referentwas "self as male fnends see me" praise of the person was sig-nificantly more effective than praise of performance, type ofpraise used did not produce differential effects under a low fre-quency of approval treatment

As noted previously, m evaluating these effects we are deahngwith degrees of positive change and are essentially trymg toestabhsh what vanable or combination of vanables is relativelymore effective m producmg such effects The superionty ofpersonal- to performance-oriented praise might be seen as re-flectmg the fact that the present measures focus a person onevaluatmg himself m terms of global personal attnbutes ratherthan m terms of specific task competence. It could also be arguedthat the present subject population is, m general, more hkely tosee vague evaluative praise as self-relevant since, as Hess andShipman (1965) have observed, lower-class mothers are moreprone than middle-class mothers to use vague personal impera-tives to shape their children's behavior

Both the mam effect for frequency of praise and the one sig-nificant interaction we obtained are consistent with the logic ofthe SRS model. Such findmgs suggest that low frequencies ofpraise from a white source, bemg more consistent with the sub-ject's past experiences than high frequencies of praise, are prob-ably viewed as more credible and hence have greater persuasiveimpact Such an explanation gains m cogency because of themteraction findmg smce it seems plausible to assume that m thepast It IS a high frequency of achievement-oriented praise thathas been most lackmg and it is this condition that stands out asthe least effective treatment

In terms of accountmg for the strength of the present self-

Type and frequency of praise 505

data the folloviong may be noted The experimenters were pre-sented as, and for the most part perceived as, part of a trainmgprogram which could affect the subjects' vocational future Sec-ondly, the present use of a noncontmgent reinforcement pro-cedure and a delayed measure of self-image may be viewed asequivalent to findmgs involvmg the delayed effects of ambiguouspersuasive communication m the general realm of attitudechange. McGuire (1969) pomts out that ambiguous persuasivecommunications frequently gam m eflScacy over tune becausethey motivate a person to make sense out of them, and hencerequire greater cogmtive work, a process which takes time (seealso Cohen, 1957). This hne of reasomng is further strengthenedhy the obtained supenority of personal praise to achievement-onented praise since it is this condition whtch is probably themost ambiguous cogmtively. It may also^be true, m general, thatfeehngs about self are more affected by stimuli which are rela-tively undifferentiated cognitively but which carry a clear af-fective connotation (see, for example, Staats, 1968, for a view ofatbtude formation and change which stresses the role of affectiveconditionmg).

The strength of the self-image data suggests that the use of anomntrusive laboratory-type experiment m a field setting mayprovide a methodology which combmes some of the potency ofnaturalistic mterventions with the control of a laboratoty study.

Performance Data

The lack of effects on the performance measures which m-volved visual perception and word recogmtion was not unex-pected. First, both Rosenhan and Greenwald (1965) and Mc-Grade (1966) also have failed to rephcate Zigler and Kanzer's(1962) finding that lower class subjects perform significantlyhetter when praise is person rather than achievement focused ®

8 It IS of course possible, given the relatively subtle nature of the differencebetween a verbal reinforcer bemg task and person onented, that failures toreplicate Zigler and Kanzer's (1962) findmgs are due to subjects' failure toperceive di£Ferences in type of praise For example, there might have been adifference m the way Zigler and Kanzer's experimenter dehvered the two rem-forcers which subsequent research has failed to rephcate This remains a pos-sibihty smce neither Rosenhan and Greenwald (1965) nor McGrade (1966),nor for that matter Zigler and Kanzer (1962), collected data concerning thesubjects' perception of type of reinforcer With regard to the preseat study

506 Reuben M. Baron, Alan R. Bass, and Peter AA. Vietze

Secondly, evaluation which is not contmgent upon the quahtyof a person's responses is not hkely to be very helpful in situationswhere task improvement depends pnmarily upon increasing one'saccuracy of information processmg. That is, the present rem-forcement procedure may result m subjects bemg praised whenthey were domg poorly and bemg ignored when they were doingwell The findmg, however, that the manual dexterity task wassensitive to our reward treatments does suggest that, where un-proved performance is largely a matter of motivational arousal,even noncontmgent reinforcement can serve as a context fordifferential reward effects (It may be recalled that the majorperformance measure here was speed of placement, a processhkely to be more reffective of energy mobilization than of theinformational value of a reinforcer).

The specific findmgs obtamed on this task are also of mterestThe greater effectiveness of achievement praise than personalpraise suggests that the subject is probably able on an extremelysimple task to evaluate his own performance despite our generalefforts to make self-evaluation of performance diflScult That is,the subjects m this condition are more able to obtam feedbackabout performance from the task itself—they can make judgmentsabout whether they are gomg relatively fast or slow If such aproblem-solving set exists, subjects are likely to be more mter-ested m achievement praise than personal praise smce such feed-back IS likely to appear more useful to them in tracking theirperformance.

The significant mteraction between frequency and type ofreward found on the manual dexterity task suggests that achieve-ment-oriented praise is most effective when given in small doses,statistically, the mteraction is carried by this one cell, although,

it should be noted that our findings mvolvmg significant mteractions betweentype of praxse and frequency of reinforcement suggest that at some level thetwo reaaforcers were being categorized as difFerent After the experimental tasbwere completed, a questiormaire was administered to ascertam whether thereward treatment was accurately perceived Unfortunately, immediately pnorto this administration a member of the skills center sta£E, who inadvertantlyhappened by, was asked by a number of the subjects why the mstructors weretelhng them they were doing well His response we later learned was "becauseyou did a good job " Given this confoundmg event it is difficult to evaluate thefindmg that subjects m bodi the task and person conditions tended to give atask mterpretabon of the eJ5)enmraiter's praise

Type and frequency of praise 507

as we might also expect from the SRS model, the 75 percent per-son-praise treatment is also relatively effective. The fact that ararely encountered type of reinforcer is most effective when dis-pensed at a low frequency supports the application of the SRSmodel to this settmg and subject population (see Baron, 1970).

Conclusions and Implications

The fact that we obtamed stronger findmgs on the self-imagethan performance data is really not surpnsmg given the presentuse of noncontmgent remforcement. While such a procedure mayhave prevented systematic improvement in task performance onthe "accuracy" type tasks, it was an effective procedure for dem-onstratmg that the present reinforcement parameters producedifferential changes in self-image Further, the fact that personalpraise was found to be more effective for changing self-imageand achievement-oriented praise more effective for simple taskperformance is consistent with the proposition that by late ado-lescence both kmds of remforcers have acquired adequate in-centive value even for a lower-class population, which type willbe more effective probably depends upon which is seen as morerelevant by the person m any given situation For example,achievement-oriented verbal approval may have high mcentivevalue m an academic settmg and low incentive value in a datingsituation (see Rosenhan & Greenwald, 1965, for supportive datafor this kmd of general position with a child population) Theinteraction findings mvolvmg both changes in self-image andperformance on the manual dexterity task suggest that msteadof thmking in terms of a single standard of social reinforcement,we might substitute the concept of multiple standards of socialremforcement (see Baron, 1970). These standards might be as-sumed to vary with such parameters of the remforcement situa-tion as type of reinforcer, source of reinforcement (eg , race orclass of reinforcing agent), and the type of task behavior ehcited

While the present data are useful from the pomt of view ofsuggestmg relationships among vanables that hitherto had notbeen related to one another, it should be noted that traditionalnotions of external validity call for a senes of control groupslnvolvmg subjects of varying sex, race, and social class in order

508 Reuben M. Baron, Alan R. Bass, and Peter M Vietze

to establish the hmits of our eflFects. Recent findings by Gruen,Ottmger, and Zigler (1970) and Baron and Ganz (in press), how-ever, suggest that race and class are less powerful predictors ofperformance than are more genotypic psychological vanablessuch as level of aspiration and locus of control, which cut acrossrace and/or social class. Such findmgs imply that the boundaryconditions for the present findmgs might better be sought m asearch for cross-cuttmg individual difference variables than inestabhshmg a plethora of control groups based upon phenotypiccategories. Further, it should be pomted out that other mdividualstudies m the current research program using similar reinforce-ment parameters (see Baron & Bass, 1969) have found (a) nodifferences between lower-class white and Black males, and (b)substantial uniformity between the results for lower-class malesand females

It IS felt that further extensions of the present design mvolv-mg (a) the use of a more contmgent reinforcement procedure,and (b) an exploration of the role of evaluator characteristics(eg., IS race of evaluator important per se or are more generalattributes such as status, attractiveness, etc, more important inmoderatmg remforcer effectiveness P) are in order. Furthermore,smce it IS always possible that present impressive findmgs canat least partially be accounted for by the paper and pencil natureof our measures, it is important that the present seliF-image datashould be cross-validated usmg more behavioral mdicants ofchanges in self-image For example, actual changes m job seek-ing behavior rather than pre-post comparisons of adjective checklists could be utilized (Bandura, 1969).®

SUMMARY

An experiment was conducted in a skills traming center witha sample of 28 Black female trainees The effects of variations intype (person- vs. achievement-oriented) and frequency (75%

9 It nnght be argued, for example, that the obtained effects merely reflectthe subjects' acquiescence to the "demand characteristics" of the expenmentalsettmg The use of (a) mdividuals other than the experimenters to administerthe pre-and-post measiu-es of self-image, and (b) a different setting and contextfor these measures (they were given as part of a general battery of tests), and(c) a delayed post-measure, all tend to weaken this land of altemabve explana-tion Furthennore, the existence of mteractions which are theoretically denvedlessens the cogency of simple response set mterpretations of these data

Type and frequency of praise 509

VS. 25%) of praise were ascertained on favorableness of self-image and task performance. It was found that, m general, lowerfrequencies of praise produced more positive self-evaluationsthan high frequencies of praise, and that person-onented praisewas more effective than achievement-onented praise m enhancmgself-evaluation A significant mteraction found for "myself asmale friends see me" demonstrated that (a) at a 75 percentfrequency of reinforcement, person-onented praise was signifi-cantly more effective than performance-onented praise, and (b)performance-focused praise dehvered at a 25 percent frequencyenhanced subjects' self-image significantly more than perform-ance-onented praise given at a 75 percent level. No significanteffects were found for the accuracy-type tasks (work recogni-tion, visual perceptiveness) On the task which involved asimple, speeded response (manual dexterity) there were signifi-cant effects (a) achievement-onented praise was more effectivethan person-onented praise, (b) a significant mteraction betweentype and frequency of praise occurred, indicating that perform-ance-oriented praise is most effective when given at a 25 percentlevel

REFERENCES

Adams, J S , & Rosenbaum, W B The relationsbp of worker productivity tocogmtive dissonance about wage lneqmties Journal of Applied Psychology,1962, 46, 161-164

Andrews, 1 R Wage meqmty and job performance An expenmental studyJournal of Applied Psychology, 1967, 51, 39-45

Aronson, E , & Carlsmith, J M Performance expectancy as a determinant ofactual performance Journal of Abnormal and Soctd. Psychology, 1962, 65,178-182

Bandura, A Principles of behavior modtficatton New York Holt, Rmehart,and Winston, 1969

Baron, R M Social reinforcement effects as a function of social remforcementhistory Psychological Review, 1966, 6, 527-539

Baron, R M, Robinson, E L , & Lawrence, S The effectiveness of social re-mforcement as a function of changes m rate of reinforcement Journal ofExpenmental Socuil Psychology, 1968, 4, 123-142

Baron, R M, & Bass, A R The role of social reinforcement parameters mimproving trainee task performance and self-image Fmal technical report,U S Department of Labor, 1969

Baron, R M The SRS model as a predictor of Negro responsiveness to reinforce-ment Journal of Soctal Issues, 1970, 26, 61-81

Baron, R M & Ganz, R L The effects of locus of control and type of feedbackon the task performance of lower dass Black children Journal of Personalityand Social Psychology (m press)

Cohen, A R Need for cogmtion and order of communication as a determinant

510 Reuben M. Baron, Alan R. Bass, and Peter M. Vietze

of opmion change In C I Hovland (Ed ), Order of preserttatton in per-suasion New Haven Yale Umversity Press, 1957

Denmark, F , & Guttentag, M Dissonance m the self-concepts and educationalconcepts of coUege and non-college onented women Journal of CounselingPsychology, 1967, 14, 113-115

Deutsch, M & Solomon, L Reactions to evaluations of others as influenced byself-evaluabon Soctometry, 1959, 22, 93-112

Douvan, E Social status and success stnvmg Journal of Abnormal and SoctalPsychology, 1956, 52, 219-235

Dreger, R M, & Miller, K S Comparabve psychological studies of Negroesand whites m the Umted States 1959-1965 Psychological BuUettn Mono-graph Supplement, 1968, 70 (whole No 3, Part 2) , 1-57

Fitts, W H The Tennessee self-concept scale manual Nashville, Tenn 1965Gruen, G, Ottmger, D , & Zigler, E Level of aspiration and the probability

learmng of middle- and lower-class children Developmental Psychology, 1970,3, 133-142

Hass, H 1, 6E Maher, M L Two expenments on the concept of self and thereaction of others Journal of Personality and Soctal Psychology, 1965, 1,100-105

Havighurst, R J Mmonty subcultures and the law of effect American Psy-chologist, 1970, 25, 313-322

Hess, R D & Shipman, V C Early expenence and the socialization of cog-mbve modes m children Chtld Development, 1965, 36, 869-886

Howard, I , & Berkowitz, L Reacbons to the evaluators of one's performanceJournal of Personality, 1959, 26, 494-507

Korman, A Toward an hypothesis of work behavior Jourrud of Allied Psy-chology, 1970, 54, 31-41

Marlowe, D Need for social approval and the operant condibomng of meanmgfulverbal behavior Journal of Consulting Psychology, 1962, 26, 79-83

McGuire, W J The nature of atbtudes and atbtude change In E Lmdzey &E Aronson (Eds ), The handbook of soctal psychology, Vol III Reading,Mass Addison-Wesley, 1969

McGrade, B J Effecbveness of reinforcers m relabon to age and social classJournal of Persorudtty and Social Psychology, 1966, 4, 555-560

Rosenhan, D & Greenwald, J The effects of age, sex, and socio-economic classon responsiveness to two classes of verbal remforcement Jourrud of Per-sonality, 1965, 33, 108-121

Staats, A W Social behaviorism and human mobvabon Pnnciples of theatbtude-reinforcer-discnminabve system In A G Greenwald, T C Brock& T M Ostrom (Eds ), Psychological foundation of attitudes New YorkAcademic Press, 1968

Stevenson, H W & Weir, M W Developmental changes m the effects ofreinforcement and non-reinforcement of a smgle response Chtld Development,1961, 3, 1-15

Stodand, E , Thorley, S, Thomas, E , Cohen, A R, & Zander, A The effectsof group expectabons and self-esteem upon self-evaluabon Journal of Ab-normal and Soctal Psychology, 1957, 54, 55-63

Terrall, G, Jr , Durkm, J , & Wiesley, M Social class and the nature of themcenbve in discrimmabon learmng Journal of Abnormal and Social Psy-chology, 1959 59, 270-272

Wilson, D Abihty evaluabon, post-decision dissonance and co-worker attracbve-ness Journal of Personahty and Soctal Psychology, 1965, 1, 486-489

Zigler, E , & Ghild, I L Sociahzabon In Lmdzey, G & Aronson, E (Eds ),The handbook of social psychology, Vol III Readmg, Mass Addison-Wesley, 1969

Type and frequency of praise 511

Zigler, E , & deLabry, J Concept-switchmg va middle-das^, lower-class andretarded children Journal of Abnormal and Socud Psychdogy, 1962, 86, 267-273

Zigler, E , & Kanzer, P The eflFectiveness of two classes of verbal reinforcerson the performance of middle- and lower-class children Journal of Personality,1962, 80, 157-163

Manuscript received October 19,