Socialization and the Altruistic Behavior of Children

Psychological Bulletin1976, Vol. 83, No. S, 898-913

Socialization and the Altruistic Behavior of Children

J. Philippe RushtonUniversity of Toronto, Toronto, Canada

The recent research into altruistic behavior by children is reviewed under fourmain headings. The first concerns the generality of children's social altruism,and an average correlation of .3 appeared to be representative across behav-ioral situations. The second section concerns person variables with particularemphasis on age and the cognitive-developmental variables of moral judgmentand role taking. The third heading concerns environmental variables and in-volves more or less direct attempts to demonstrate socializing processes atwork. The items dealt with are reinforcement, the role of models, training inrole taking, and verbal socialization procedures such as preaching and induc-tion. Finally, theory is considered with particular emphasis on the theories ofsocial learning and cognitive development.

Since Bryan and London's (1970) review,research has proceeded apace into the deter-minants of altruistic behavior by children.One reason for this continuing interest is, nodoubt, that altruism, denned generally as be-havior carried out for the benefit of another,seems particularly important for understand-ing socialization processes. It will be a scien-tific theory of very real consequence thatwill account for how it comes about that abeing, brought into the world with appar-ently no other thought than its own gratifica-tion, eventually becomes capable of living itslife with concern for others. If it were pos-sible to specify some of the necessary andsufficient conditions that produce altruisticchildren (and adults), it would be an excel-lent indication that we were well on our wayto a general theory of the socialization proc-ess. The present review attempts to gauge howfar we have progressed in the last few yearstoward providing such a theoretical account.

Prior to attempting to review this progress,it might be beneficial to consider what hasbeen meant by "altruism" in the literature.Most researchers would probably acceptMacaulay and Berkowitz's (1970) conceptualdefinition of altruism as "behavior carried out

I would like to thank Kurt Danziger and JoanGrusec for their helpful comments on an earlierversion of this paper.

Requests for reprints should be sent to J. PhilippeRushton, Department of Psychology, University ofToronto, Toronto, Ontario, Canada M5S 1A1.

to benefit another without anticipation ofrewards from external sources" (p. 3). On theother hand, most researchers have generallyavoided the question of establishing the in-tentions behind the act by employing rela-tively narrow operational definitions withina context that allows for a reasonable infer-ence of behavioral altruism (e.g., donatingto a charity in an anonymous situation;sharing candies with an absent friend).Researchers do not directly investigate whatthe intention was behind the child's actions.As Krebs (1970) has discussed, this omissioncould be a serious problem because it isusually the intention behind an action, ratherthan the action itself, that determines itsmoral value. This would be particularly crit-ical if one's definition of altruism were to restupon a particular motivational basis, asAronfreed's (1970) and Hoffman's (1976)do on the concept of empathy. Other possiblecategories of altruism might include (a)normative, as when a child shares a toy be-cause he or she feels it is expected by an-other; (b) reciprocal, as when a child sharesa toy in the hope of later borrowing thefriend's toy; (c) principled, as when a childshares a toy in order to live up to a principlewithin himself or herself that prescribes ageneralized "ought"; and (d) fairness or jus-tice, as when a child shares a toy in order torestore a specific equitable situation withinhimself or herself. Very rarely, however, havethese various possible motivations been

898

SOCIALIZATION AND ALTRUISM 899

taken into account in choosing the dependentvariable.

For example, in one study (Emler &Rushton, 1974), 7- to 13-year-old childrenwere given an opportunity to donate tokensthey had won to a charity under conditionsof high or low sympathy eliciting instruc-tions. Such instructions might be thought topresume an empathy-sympathy motivationalbasis. On the other hand, the dependent mea-sure of donating to a child on a charityposter might be considered to be tappingmore of a normative or principled motivationthan a sympathetic one, a conclusion sup-ported by some of the data in the Emler andRushton (1974) study. If consideration ofthe motivational base does not occur, it couldmake the comparability of dependent mea-sures (based on different motives) somewhatproblematic. Thus, donation to a charity fora 7-year-old might not be motivated by thesame reasons as 11-year-olds. Similary, dona-tions to a friend, for all ages, might be quitedifferent from donating to a charity.

Perhaps such postulated motivational dif-ferences in producing altruism only becomeimportant, however, when antecedent condi-tions (independent variables) cease to havethe same effect across alternative measuresof altruism (dependent variables). There isthus a case to be made for the operationalequality of behavioral measures with the argu-ment that needless proliferation of "types"of altruism will only confuse the issue. Ac-cording to this latter point of view, types ofaltruism need be distinguished only whenirregularities in the empirical relationships,hopefully being discovered between indepen-dent and dependent variables, require it.There is some truth to both points of view.If truly systematic exploration of relationsbetween independent and dependent measureswas being undertaken across several mea-sures of altruism, one could sympathize withthe strict operationalists. However, to thedegree that research is being carried out inrelative innocence of the wider issues involved,a change is perhaps necessitated.

The measures of altruism that have beenused include the following: (a) donatingpossessions to a charity or to another child,(b) experimental measures of helping and

rescue behavior, (c) consideration for othersin competitive game situations, (d) teacherand peer sociometric techniques, and (e)naturalistic observation of helping and sharingbehavior. We turn now to an examination ofthe relationships that have been found amongthese measures.

THE GENERALITY OF CHILDREN'SALTRUISTIC BEHAVIOR

To test the hypothesis that generosity waspart of a pattern of prosocial moral charac-teristics including kindness and cooperation,Rutherford and Mussen (1968) initiallysampled 63 middle-class 4-year-old boys. Agenerosity score was found for each childbased on the number of candies given awayto a friend. On this basis the initial sampleof 63 was divided into 14 nongenerous chil-dren who gave no candies away at all and 17highly generous children who gave away alarge proportion of their candies. These ex-treme groups were then found to differ in avariety of ways. Specifically, teachers ratedthe generous children as more generous, moregregarious, less competitive, less quarrelsome,more kind, and less aggressive than the non-generous children. In addition, a behavioralmeasure of competitiveness based on a car-racing game showed the generous group to beless competitive than the nongenerous group.

In a subsequent study, these same authorsand their colleagues (Mussen, Rutherford,Harris, & Keasey, 1970) studied 95 childrenaged 11-12 years. Four measures of prosocialbehavior were used: (a) an honesty factorderived from a factor analysis of a socio-metric questionnaire, (b) an altruism factorderived by the same method, (c) a behavioralmeasure of honesty (not cheating in a ray-gunresistance-to-temptation game situation), and(d) a behavioral measure of altruism (gen-erosity in a prisoner's dilemma game). Forboys, of the six possible intercorrelationsamong the four measures, honesty in thesituational test correlated positively (r — .29,p < .05) with the honesty factor derived fromthe sociometric questionnaire; the altruismsociometric factor was reported as unrelatedto the altruism situational behavior; and theremaining four possible correlations were un-reported. For girls, the honesty sociometric

900 J. PHILIPPE RUSHTON

factor was positively associated (r — .27, p< .05) with the honesty situational behavior;the altruistic situational behavior was re-ported as unrelated to the altruistic factorderived from sociometric judgments, the hon-esty factor, or the behavioral test of honesty;and the remaining two possible correlationswere unreported. Any support for generalitythat might have appeared among the 4-year-old boys studied by Rutherford and Mussen(1968) had certainly disappeared by the ageof 12 (Mussen et al., 1970).

A study by Rush ton and Wiener (197S)also looked at the pattern of generality atdifferent ages. The following three behavioralmeasures of altruism were taken from 30seven-year-olds and 30 eleven-year-olds: do-nating tokens to a charity, sharing candywith a friend, and a competitiveness scorefrom a car-racing game. These last two mea-sures were the same as those used by Ruther-ford and Mussen (1968). For both ages com-bined, generosity to a friend related positively(r — .24, p < .05) with generosity to a char-ity, and negatively (r = —.55, p < .001) withcompetitiveness. Age differences were alsofound. The negative correlation between com-petitiveness and generosity to a friend de-clined sharply from the 7- to the 11-year-oldsamples (r = -.63, p < .001 to r = -.39,p < .05), whereas the positive relationshipbetween generosity to a friend and generosityto a charity rose from the 7- to the 11-year-old groups (r = .19, ns to r = .40, p < .01).For neither age group was competitivenessrelated to generosity to a charity.

Dlugokinski and Firestone (1973) also at-tempted to determine whether altruism, or asthey termed it, "other-centeredness," was ageneralizable phenomenon. From 164 chil-dren, aged 10 to 13, they took four measures:a pencil-and-paper measure of how one un-derstood the meanings of kindness; a pencil-and-paper measure of the relative importanceof altruistic as opposed to selfish values; so-ciometric peer ratings of considerateness andselfishness; and a behavioral measure con-cerned with donating money to a charity. Thesix possible correlations were all positive andranged from .19 to .38. Further, the authorsreported, multiple correlations of any threevariables as predictors of the fourth ranged

from .42 to .51. In a later paper, Dlugokinskiand Firestone (1974) replicated these rela-tionships.

Rubin and Schneider (1973) took two mea-sures of altruism from 55 five-year-olds. One,a measure of generosity, consisted of dona-tions of candy to a charity. The other, ameasure of helping behavior, consisted of theamount of work done for a peer. These twomeasures were positively intercorrelated (r= .40, p < .01). On the other hand, Greenand Schneider (1974) failed to find any sig-nificant relationships between three measuresof altruism taken from 100 boys aged 5 to 14.Their measures of altruism were sharingcandy with other children, picking up droppeditems for the experimenter, and volunteeringfree time to work with needy children.

Three recent studies have examined therelations among children's naturally occurringaltruism. Krebs and Sturrup (Note 1) reporta study of 23 seven- and eight-year-old chil-dren. Three altruistic coding categories wereused: offering help, offering support, and sug-gesting responsibly. Offering help was foundto correlate .21 with offering support and.09 with suggests responsibly, which in turncorrelated .24 with offers support. A some-what higher correlation was obtained when acomposite behavioral altruism score was cal-culated on the basis of the three precedingmeasures. This composite score correlated.47 (p < .01) with an independently derivedteacher rating of the child's overall altruism.Hansen, Goldman, and Baldwin (Note 2)carried out a naturalistic investigation of thealtruistic behavior of children having fourobservers code some 150 incidents over 40hours of observation on 23 four-year-olds ina university preschool. These authors, how-ever, failed to find any evidence for general-ity across such coded situations as respond-ing to distress, sharing material possessions inresponse to requests, and giving unsolicitedhelp on tasks. Finally, Yarrow and Waxier(1976) carried out a study with 108 childrenaged 3 to 7 that involved both experimentaland naturalistic measures of prosocial behav-ior. Six experimental measures were takenfrom two separate adult-child play periodsand three naturalistic measures were takenfrom observations of the child during "free-


play." The experimental measures involvedtwo measures of sharing, two measures ofhelping, and two measures of comforting allconcerned with the interaction between thechild and the experimenter. The naturalisticobservations concerned child-child interac-tions and were also coded into categories ofsharing, helping, and comforting. Unfortu-nately, the authors did not report any of thefirst-order correlations between their ninemeasures of altruism. Instead they point outthat in the experimental situation "scores onsharing" (presumably the two measures com-bined) correlate .32 with "scores of comfort-ing," although neither sharing nor comfortingwas related to helping. In addition, if sharingin the experimental situation were combinedwith comforting in the experimental situation,then this combined score related to a similarcombined sharing-comforting score from thenaturalistic data (r = .29).

It would appear from the studies reportingdata on the generality of altruistic behaviorthat Mischel's (1968) magic number of .3once again emerges as the overall represent-ative intercorrelation. A number of points,however, might be raised in connection withthe generality versus specificity of behaviorcontroversy.

First of all, the low correlations across be-havioral situations are usually found whenone specific task is correlated with one ormore other specific tasks. This might not bethe best way of approaching the problem. Itis possible that if a battery of behavioraltasks were given, individual differences in sub-sequent behavior might then become morepredictable. This predictability would be ex-pected if random error variance in each situ-ation averaged itself out—an expectationmade explicit in psychometrics where gener-ally the more items there are in the test, thehigher the reliability. Certainly combiningtasks led to the highest correlations in theDlugokinski and Firestone (1973, 1974),Yarrow and Waxier (1976), and Krebs andSturrup (Note 1) studies.

A second point is that there might be indi-vidual differences in both the amount and thepatterning of the generality versus specificityof the behavior under consideration. In anextreme case, for example, a priest might be

expected to show more generality in hisaltruistic behavior than a 7-year-old boy.

Finally, referring back to the opening re-marks on the possibility of different motiva-tional systems operating for altruism, it mightbe that higher correlations would be found ifmeasures were taken within motivational sys-tems rather than across them. Thus, it may bethat an individual behaving altruistically asa result of principled reasoning may showgenerality across situations that tapped thatmoral principle but show specificity of be-havior if measured in situations that tapped,say, sympathy. These are empirical ques-tions to be left to future research. From thepresent vantage point, however, the inter-relationships between children's altruisticbehaviors are of the magnitude of .3.

The question then emerges as to whethersuch a correlation is more indicative (a) of ageneral trait (either within or between dif-ferent categories of possible altruistic motiva-tions), of which there are underlying dispo-sitions, or (b) of situational specificity. Thereis no ready answer to such a question. A fig-ure of .3 can be used to support both a gen-erality and a specificity point of view. Itsuggests that there are some common determi-nants of the measures of altruism. Whetherthese common determinants are artifacts(response tendencies, observer biases, haloeffects, or uncontrolled variables such as IQ),internal dispositions (traits, such as empathy,or cognitive structures, such as role playingschemata), or common environmental control-ling conditions (model, cue, and reinforcingstimuli), remains to be decided.

On a methodological note, it is encouragingto see the increasing diversity of approachesto the problem of altruism. The experimentallaboratory is still the major source of datagathering (and is still being criticized, notalways justifiably, as artificial and subject todemand characteristics). The experimental ap-proach continues to be supplemented by pen-cil-and-paper "personality" measures andsociometric judgments. In addition a numberof recent studies have begun to describe natu-rally occurring altruism in children. Such stud-ies might prove exceedingly useful in thefuture, both in providing information abouthow children behave in a variety of settings,


and also in testing the generalizability of ex-perimental hypotheses. It is, of course, likelythat all approaches are valuable at this pointin the discipline. Problems inherent in onemethod (demand characteristics in laboratoryexperiments and observer bias in naturalisticstudies) can be counterbalanced and the find-ings validated by the use of alternative ap-proaches.

PERSON VARIABLESAge

A very recent study by Rheingold, Hay,and West (in press) demonstrated that aform of sharing is present in children as earlyas the first 2 years of life. Many other stud-ies have found that children's sharing in-creases over the age range of 6 to 12 (Elliott& Vasta, 1970; Emler & Rushton, 1974;Handlon & Gross, 19S9; Harris, 1971; Mid-larsky & Bryan, 1967; Rosenhan, 1969;Rushton, 1975; Rushton & Wiener, 1975;Ugurel-Semin, 1952; Wright, 1942). In addi-tion, Green and Schneider (1974) showedthat measures of helping also increased withage. Not all studies, however, found increasesover this age range. Staub (1970), for exam-ple, found a curvilinear relationship with agein rescuing behavior. While rescue behavior,in general, increased over the age range of 4to 9, it tended to fall off sharply at age 11.Furthermore, a number of studies have shownthat competitiveness rather than cooperative-ness increases with age, at least in Anglo-American cultures (Kagan & Madsen, 1971;Madsen, 1971; Madsen & Connor, 1973;Rushton & Wiener, 1975). Finally, Hart-shorne, May, and Mailer (1929) and Yar-row and Waxier (1976) found no relationshipbetween prosocial helpfulness and age.

Cognitive-Developmental Variables

That sharing behavior appears to increaseover the period of middle childhood has ledto widespread suggestions (e.g., Bryan &London, 1970; Krebs, 1970; Rosenhan, 1969,1972; Wright, 1971) that such age changesin behavior might be linked to cognitive-developmental changes, particularly changesin role-taking capacity and the basis of moraljudgment. It appears that middle childhooddoes see significant increments with age over

this time period in these areas of cognitivefunctioning. Flavell, Botkin, Fry, Wright,and Jarvis (1968) found, over a range ofrole-taking tasks, evidence of a developmentalshift from an "egocentric" to a "reciprocal"perspective during the period of 7-14 yearsof age. Piaget (1932) has documented move-ment over the same age period in children'smoral judgments which are viewed as show-ing a progression from an egocentric perspec-tive, based on authority and punishment, toone based on cooperation, concern for theother's intentions, mutual respect, and aware-ness of the other. Thus, the idea that pro-social behaviors, such as generosity, mightbe mediated by cognitive-developmentalprocesses, such as role taking and moral judg-ment, is an intriguing one. Some recent re-search has addressed itself to this question.

Moral judgment. Rubin and Schneider(1973), studying 55 seven-year-old children,found a relationship between moral judg-ment, assessed by the children's responses toa number of moral-conflict stories, and twodifferent measures of altruism. Moral judg-ment correlated r = .31 (p < .05) with theamount of candy donated to poor childrenand r = .40 (p < .01) with the amount ofhelp given to a peer on a task. Emler andRushton (1974), using moral judgment stor-ies concerned with distributive justice, foundthat predictions of 60 seven- to thirteen-year-old children's anonymous donations to acharity could be made with better than chanceresults from knowledge of the children's levelof moral judgment. Furthermore, this findingwas maintained when the effect of age wascovaried from the analysis.

The interesting question then arises as towhether the relationship found between moraljudgment and generosity (Emler & Rushton,1974; Rubin & Schneider, 1973) is one ofcausality or of covariance. Does the child'smoral reasoning cause his or her moral be-havior, as might generally be expected, or issome third factor accounting for the positivecorrelation between these responses (e.g., par-ents who socialize the child both to behave ina generous manner and to give particulartypes of moral judgment responses)? A re-cent study by Rushton (1975) with 140seven- to eleven-year old children attempted


to provide some initial data on this question.It was argued that if moral judgment was adeterminant of behavior rather than just acovariate, it might be expected to interactwith, and affect the reception of, such sociallearning inputs as modeling and preaching.The behaviors and preachings of a modelmight be expected to have had a differentialeffect depending on whether they were ob-served by a child with a high or a low levelof moral reasoning. The results of this lab-oratory study, which incorporated both animmediate and a 2-month delayed test, werehighly interesting. First, it confirmed the ex-pectation that there was an association be-tween the child's moral judgment and his orher generosity, although when age was co-varied from the analysis, the strength of therelationship was weakened. Second, moraljudgment did not interact with the social learn-ing inputs when the dependent variable was thechild's generosity score. High moral reason-ers, for example, were just as influenced by aselfish model preaching the virtues of beingselfish as were low moral reasoners. On theother hand, moral judgment did interact withthe social learning input when the child wasasked to evaluate the preacher model he orshe had observed. Children with a high moraljudgment score rejected the selfish preachermore than those with a low moral judgmentscore did. Thus, moral judgment was exert-ing an effect. It would seem that further re-search in this area might have importantimplications for socialization theory. One pos-sible strategy might be, • for example, to at-tempt to differentially alter moral reasoningon one hand and moral behavior on the other.If one changes a child's moral reasoning, willthis affect his or her behavior? Alternatively,will altering a child's behavior cause his orher moral reasoning to change? Or, indeed, isthere no necessary relationship between achild's judgments and behavior?

Role taking. As regards role-taking abilityand behaving altruistically, the results aremore equivocal. While Rubin and Schneider(1973) found a relationship in 55 seven-year-olds between the two measures of altruismcited above and "decentration" (a measureconceptually similar to "role-taking ability"),two other studies did not. Emler and Rush-

ton (1974) failed to find a relationship be-tween role-taking capacity and generosityusing two measures of role-taking, as opera-tionalized by Flavell et al. (1968) and shownto relate to moral judgment by Selman(1971). A study by Rush ton and Wiener(1975) with 60 seven- and eleven-year-oldsfailed to find a predictive utility for two dif-ferent measures of role-taking capacity, againtaken from Flavell et al. (1968), on threedifferent measures of altruism. On the otherhand again, Krebs and Sturrup (Note 1),using the same Flavell et al. (1968) role-taking tasks as Emler and Rushton (1974),found that role-taking ability correlated r =.46 (p < .02) with a composite altruism score.In addition, positive but nonsignificant corre-lations were found by Krebs and Sturrup be-tween role-taking and the component altru-ism scores. Additional significant correlationswere found between role-taking tasks andteacher ratings of the children's prosocial andcooperative behaviors (r = .41 and r = .42,ps < .05, respectively). Unfortunately, in theKrebs and Sturrup study, intelligence, asmeasured by both formal IQ test and teach-ers' ratings, correlated with all measures,thus confounding the relationship betweenrole taking and altruism. Finally, Hansen etal. (Note 2) found that a story-type taskdesigned to measure empathy failed to predictchildren's behavior across the several altruisticbehavior categories.

Further research using a wider range ofrole-taking tasks with particular emphasisperhaps on emotional role-taking (empathy)skills might prove useful. Most of the role-taking measures to date have stressed theperceptual-cognitive side of this skill. Worth-while research also might be spent providingadditional reliability and normative data onthe measures of role taking. It is difficult totest hypotheses using measuring instrumentsof uncertain reliability.

Generalized Cognitive Development

The possibility that some more general cog-nitive developmental variable would predictaltruism in children received no support froma study by Rushton and Wiener (1975). Abattery of cognitive developmental measureshad been taken from 7- and 11-year-old chil-


dren, including conservation judgments, intel-ligence test scores, tests of cognitive complex-ity, categorization responses, personal con-struct systems, and measures of egocentricity.Although the cognitive measures showed theexpected relationships to age and IQ, noneshowed any degree of relationship to threedifferent measures of altruism, including do-nating tokens to a charity, sharing candy witha peer, and scores on a competitive car-racinggame.

This was the general finding, too, of thestudy by Hansen et al, (Note 2) in which thePeabody Picture Vocabulary Test and suchstandard Piagetian tasks as seriation andclassification failed to predict children's natu-rally occurring altruism across such codedcategories as responding to distress, sharingpossessions, and giving help.

Sex

In his major review of the data, Krebs(1970) reported that out of a total of 17studies, including some unpublished material,no sex differences were found in 11 of them.However, when sex differences were found,they tended to favor girls. This remains thegeneral finding. An absence of sex differencesin altruistic behavior was reported in a num-ber of recent studies (Emler & Rushton,1974; Harris, 1970, 1971; Presbie & Coiteux,1971; Rubin & Schneider, 1973; Rushton,197S; Rushton & Wiener, 1975; Staub,1971a; Yarrow, Scott, & Waxier, 1973; Yar-row & Waxier, 1976; Krebs & Sturrup, Note1). Mussen et al. (1970) did find sex differ-ences, although not in any systematic man-ner. Dlugokinski and Firestone (1973, 1974)and Midlarsky and Bryan (1972) reportedsome tendency for females to be more altru-istic than males on some of their measures.

Few studies have reported the effects of IQon altruism. Krebs and Sturrup (Note 1)found positive relationships between both IQtests and teachers' ratings of intelligence anddifferent measures of altruism, including teach-ers' ratings of altruism. Three other studies,however, failed to find such a relationship(Rubin & Schneider, 1973; Rushton &Wiener, 1975; Hansen et al., Note 2).

ENVIRONMENTAL VARIABLES

Response Consequences

Very little research has been undertakeninto the effects of the consequences of behav-ing altruistically on subsequent altruistic be-havior. The little that has been undertaken isequivocal. While Fischer (1963) found thatmaterial reinforcement (candy) made con-tingent upon marble sharing in 4-year-oldchildren produced more sharing than socialreinforcement ("that's good, that's nice"),there were problems with the study. For ex-ample, data from extinction trials were notreported (showing how much sharing occurredwhen reinforcement was withdrawn). Further-more, the experimenter was present through-out the study. For generosity to occur theremust be an absence of such possible externalreward as experimenter approval. Aronfreedand Paskal (cited in Aronfreed, 1968) showedhow an adult's verbalizations of pleasure couldreinforce children's self-sacrificing behavior ifthose verbalizations had first been paired anumber of times with hugging the child. Mid-larsky and Bryan (1967) replicated theAronfreed and Paskal (cited in Aronfreed,1968) study and showed that children so re-inforced for one altruistic behavior (leverpressing carried out in the presence of theexperimenter) generalized this altruism to ananonymous candy-donating situation. Unfor-tunately, the authors felt that the originalconditioning of affect and the resultant per-formance curves failed to show certain ex-pected properties and, thus, open the possi-bility of demand characteristics and/or experi-menter effects having operated. Midlarsky,Bryan, and Brickman (1973) showed that anadult's social approval of 12-year-old chil-dren's donations to charity led to an increasein donating over no such approval. Interest-ingly, if the approval came from a previouslyselfish model, the approval appeared to be-come aversive and led to a depression in giv-ing. Unfortunately, here again extinction tri-als in the absence of the socializing agentwere not reported.

Finally, Gelfand, Hartmann, Cromer,Smith, and Page (1975) using a single sub-ject design showed that instructional promptsand social praise could increase children's do-


nations. Once again, however, our basic criti-cism remains. No attempt was made to assessthe effects of the durability of the independentvariables in the absence of the socializingagent. For "true" generosity to be inferred,there must be no possibility of external ap-proval from the presence of the experimenter,at least if we are to accept Macaulay andBerkowitz's (1970) conceptual definition ofaltruism cited earlier.

Observation of Models

There have now been many demonstrationsthat exposing a child to an altruistic modelcan enhance that child's subsequent altruisticbehavior (Bryan & Walbek, 1970a, 1970b;Grusec, 1971, 1972; Grusec & Skubiski,1970; Hartup & Coates, 1967; Presbie & Coi-teux, 1971; Staub, 1971a). Furthermore,studies have shown that a model's behaviorcan determine not only the amount but alsothe direction of altruistic behavior. Harris(1970) found that 10- and 11-year-old chil-dren would share with the model if the modelhad shared with them, would donate to acharity if the model had done so, or wouldretain their winnings if that was the examplethey had witnessed. In a subsequent study,Harris (1971) also found that children wereinfluenced by the model in the way in whichthey distributed their winnings across severalcharities.

Such laboratory modeling studies have oftenbeen considered by their authors to speakdirectly to important socializing processesoperating in the natural environment. Thus,such modeling studies are often interpreted asproducing new learning in observers. How-ever, Krebs (1970) has argued that if model-ing studies on altruism are to demonstrateinternalized new learning, then they mustdemonstrate both durability over time andgenerality across situations. Otherwise, thereis an alternative explanation of modeling stud-ies being more a function of demand char-acteristics and experimenter effects.

Several studies have attempted to provideevidence of the durability and generality ofbehavior change following observation of amodel. Rosenhan (1969) reported that theeffects on 6- to 10-year-olds of observing amodel and being able voluntarily to rehearse

generous behavior lasted for 7 days and,furthermore, generalized on a 3-week retestto produce more generosity in quite a differ-ent situation. Unfortunately, he failed to pro-vide complete details of the sample, signifi-cance tests used, or the generalizable effects ofthe other conditions. On the other hand,White (1972) reported a somewhat similarstudy with 9- to 10-year-olds in which re-tested generosity after a S-day delay stillexceeded that of controls. Elliott and Vasta(1970) showed generalization from the mod-eled sharing of candy to the very similar sit-uation in which the child had the opportunityto share pennies. They found a correlation ofr = .65 (p < .001) between the two mea-sures. Generalization did not occur at all,however, to a quite different kind of sharing(giving up a preferred toy to a stranger).Midlarsky and Bryan (1972) showed that anadult donating tokens to a charity affectedchildren's donations of candy to the samecharity 10 days later, even when the candydonations were solicited by a different experi-menter in a different setting. Rushton (1975)and Rice and Grusec (1975) showed that al-truistic modeling produced very strong dura-bility in 7- to 11-year-old children's generousbehavior over respectively 2-month and 4-month retest periods. Rushton (1975) alsoshowed that the modeled behavior, whethergenerous or selfish, generalized across suchchanges in the 2-month retest situation as adifferent experimenter in a different locale.

Despite the impressive number of studiesthat have demonstrated that a person's be-havior can change as a result of having ob-served a model, there is still uncertainty as towhy this happens. Many competing theoreti-cal accounts of the modeling process havebeen proposed (cf. Bandura, 1969; Gewirtz,1969; Kohlberg, 1969), and these will hope-fully generate productive research in thisfield in the future. Of more immediate concernhowever is the atheoretical criticism of mostof the sources of data which psychologistsmight use to test between these alternativemodels. Critics of laboratory modeling experi-ments suggest that "experimenter-bias" and"demand characteristics" may account forthe findings equally as well as formal theoriesof modeling.


One solution to this perplexing problem isto show that the processes that are discov-ered in the laboratory are also generalizableto the real world. This was the strategyadopted by Rushton and Campbell (in press),who studied modeling effects in the context ofblood donating among adults. In that natural-istic field experiment it was shown that ob-serving a model not only affected an immedi-ate test of volunteering to donate blood butalso affected whether blood was actually do-nated 6 weeks later in a naturalistic setting.Other studies with children have also recentlydemonstrated generalization effects to naturalsettings.

Thus, in a more extensive study than isusually carried out, Yarrow et al. (1973)provided preschool children with adult care-takers for several weeks. Then, in a series oftraining sessions, different types of sympa-thetic helping behavior were modeled (e.g.,sympathy statements to pictures of distress;sympathy statements and help for miniaturedoll dramas; real life sympathy and helpingbehavior). Modeling effects were assessed 2days and 2 weeks later. The degree of gen-eralization depended upon the type of train-ing. Generalization to "alternative forms" ofthe training session occurred in all trainingmethods and showed some durability. Notransfer occurred from the modeling of sym-pathy statements and helping with miniaturedoll dramas to the picture of distress situa-tion. Quite dramatic transfer effects werefound in children who had had nurturantcaretakers who had modeled helping andsympathy in both symbolic and live distresssituations. These children showed real-lifehelping on a 2-week retest in a situation andwith personnel quite different from those ex-perienced during training.

Friedrich and Stein (1975) also carried outa more extensive study than usual. In anearlier work (Friedrich & Stein, 1973; Stein& Friedrich, 1972) they had shown preschool-ers the prosocial television film "Mister Rog-ers' Neighborhood" three times a week duringa 4-week period and observed the resultantnaturally occurring behavior. Compared withneutral and aggressive control films, the pro-social film increased, over baseline, theamount of self-control, task persistence, and,

for children from lower-social-status families,prosocial interpersonal behavior (cooperation,nurturance, and verbalization of feeling). Inthis subsequent study, Friedrich and Stein(197S) showed four 20-minute prosocial"Mister Rogers' Neighborhood" films over a1-week period to kindergarten children aloneand in combination with special training.Television film modeling by itself led to gen-eralized content knowledge and some helpingbehavior increments on a fantasy puppet-playmeasure. However, the television film did notaffect real-life altruism by itself, although itdid do so when combined with other trainingmethods.

Coates, Pusser, and Goodman (1976) alsoassessed the effects of television film materialon children's naturally occurring social be-havior in a preschool setting. Compared withbaseline scores, both the programs "SesameStreet" and "Mister Rogers' neighborhood"significantly increased the giving of positivereinforcement to, and social contacts with,others in the preschool. The findings wereparticularly marked among those childrenwith low baselines.

From the studies cited above, it would ap-pear that relatively brief exposure to highlysalient models can produce durable and gen-eralizable behavior change in observers. Fur-ther such modeling effects are not limited tothe laboratory but also influence behavior inthe natural environment. Two further aspectsof models have been studied: consequencesto the model and characteristics of the model.

Consequences to the model. Doland andAdelberg (1967) had a child model whoshared receive "profuse" social reinforcementin one of their conditions. However, the effectof this manipulation was not independentlyassessed from a control condition and littlecould be concluded from it.

Elliott and Vasta (1970) showed children afilm of a 6-year-old boy giving f of the candyhe had won to a charity box. Following this,in one condition, the experimenter went up tothe film model and said, "That was very nice,Johnny, here's a toy to keep." Such "vicariousreinforcement" (Bandura, 1969, 1971), how-ever, failed to produce an increment in theobserver's subsequent altruism over themodel-no-reinforcement condition. Harris


(1970), too, failed to find that observation ofan experimenter praising a generous modelincreased the donation behavior of observingchildren.

Three other studies did, however, demon-strate vicarious reinforcement effects. Bryan(1971) showed that a generous model expres-sing positive affect ("This is fun" or "I feelwonderful") immediately after behaving gen-erously produced more subsequent imitativegenerosity than a model making the samestatements after a short delay. Presbie andCoiteux (1971) showed that praise to a modelfor his behavior, whether provided by themodel himself or by the experimenter, inducedmore subsequent imitation of the model bythe child than when the model was not sopraised. This effect was demonstrated forboth a generous model and a selfish model.Midlarsky and Bryan (1972) also showed vi-carious reinforcement effects for both gen-erous and selfish models. Their models smiledhappily and said, "It feels good to givemoney."

Characteristics oj the model. The effective-ness of modeling has been found to dependupon the characteristics of the model. Twoprincipal characteristics have been studied:nurturance and power. With regard to nur-turance, the findings are equivocal and seemto depend on whether the model is an adult ora peer and on whether or not the subject isaccustomed to being nurtured. For example,Hartup and Coates (1967) showed that nur-sery school children not used to being rein-forced by peers more readily imitated non-nurturant altruistic peers, whereas childrenwho were used to being reinforced by peersmore readily imitated a nurturant altruisticpeer.

On the other hand, studies using adults asmodels have often failed to find a positiveeffect due to the nurturance of the model onchildren's subsequent imitation of altruism.Rosenhan and White (1967) and Grusec andSkubiski (1970) found no main effect due totheir manipulations of nurturance, whereasGrusec (1971) actually found that nurturancetended to decrease imitation of generosity.Grusec suggested the provocative hypothesis(Grusec & Skubiski, 1970) that although nur-turance may facilitate the acquisition of neu-

tral, expressive behavior, it hinders acquisi-tion of behaviors that have some aversiveproperties for the individual engaging in them(e.g., those altruistic behaviors, such as shar-ing, which involve some cost to the individ-ual).

Yarrow et al. (1973), on the other hand,found in a much larger scale study, in whichnurturance was manipulated over severalweeks, that high nurturance was effective inincrementing the modeling of altruistic be-havior. Staub (1971a) found that a brief in-teraction with a nurturant rather than a non-nurturant adult significantly increased kinder-garten children's helping. However, this effectwas independent of the modeling. Modelingworked whether the model was nurturant ornot.

Yarrow et al. (1973) discuss the conceptand role of nurturance, and particularly themanipulation of nurturance in experimentalstudies, at some length and tentatively con-clude that nurturance will be most influential

when (a) it is a meaningful, warm relationship thathas been built up over time, (b) when it has in-cluded some withholding of nurturance, (c) whenit not only precedes the adult's modeling but iscontinuous throughout the entire modeling sequence,(p. 2S8)

Unfortunately, this analysis (and their study)confounds nurturance per se (noncontingentwarmth) with positive reinforcement (con-tingent warmth and approval) as in Items band c above. What the effects of nurturanceare separate from contingent warmth andapproval (i.e., positive social reinforcement)is still an open question, at least in relationto producing altruistic children.

Grusec (Grusec & Skubiski, 1970) sug-gested that instead of nurturant models,powerful ones would be more effective in pro-ducing imitation of behaviors involving coststo the individual. In her study she foundthat a powerful model induced more sharingthan a nonpowerful one (Grusec, 1971). Onthe other hand, Bryan and Walbek (1970b)failed to find a difference due to the power ofthe model. However, the Bryan and Walbekmanipulation of power (using the experi-menter as the model) did not appear asstrong as the Grusec (1971) manipulation(controller of important resources), and this


may have accounted for their failure. In theRush ton (1975) study the model was alsomade to appear powerful (destined as a poten-tial teacher in the child's school), and theeffectiveness of that model in inducing dura-ble imitation over an 8-week retest periodwas demonstrated. Unfortunately, the manip-ulation of power in that study was not inde-pendently assessed. One interesting finding inthe Rushton (197S) paper was that thepowerful model who made statements of posi-tive affect ("This is really fun" or "I likethis game"), prior to actually behaving, pro-duced the most subsequent imitative generos-ity or selfishness both on the immediate andfollow-up tests. Furthermore, it was thesemodels who were also evaluated most highlyby the children. These findings were repli-cated by Rushton and Owen (1975) using a4-minute television film instead of a livemodel.

Once again, most imitation occurred, onboth the immediate test and the delayed re-test, when the children viewed a powerfulmodel making statements of positive affect.Furthermore, these models were evaluatedmost highly by the children. Thus, possibly apowerful model demonstrating positive affecteither prior to (Rushton, 1975) or contingentupon (Midlarsky & Bryan, 1972) modeling isthe best inducer of imitative altruism,

Role Playing

Staub (1971b) trained children below theage of 6 to demonstrate a number of altru-istic helping behaviors (e.g., direct interven-tion, verbal consolation, calling for aid) bymeans of role playing situations in which thechild alternated playing the role of the helperand the helped. It was found that role play-ing led to more altruism on generalized tasks(helping the experimenter pick up somedropped objects and sharing of candy) andthat these effects lasted over 7 days. The dif-ferent patterning of the results for boys andgirls, however, tended to obscure the overalleffects. Furthermore, this procedure involvedmild reinforcement for playing helpfully, aswell as the experience of reversing roles, andthus the treatment was confounded.

Friedrich and Stein (1975) also carriedout a role-play training study. In addition to

the prosocial television films mentionedearlier, they provided children with trainingin role playing and assessed the effects on sub-sequent altruistic behavior. Role playing intheir study consisted of using "puppets in astructured rehearsal of key events and dia-logue from the program" (Friedrich & Stein,1975, p. 30). The results showed that chil-dren, particularly boys, who had watched theprosocial film and engaged in role-play train-ing engaged in more helping behavior in areal-life situation different from the trainingsituations. However, reinforcement for play-ing helpfully must have confounded this studyas well.

Rosenhan and White (1967) and White(1972) carried out studies in which they care-fully controlled for unwitting reinforcementeffects. They found that giving a model-ob-serving child the opportunity to rehearse(role play) the behavior in the presence ofthe model led to more subsequent imitativegenerosity than observation of the model byitself.

Verbal Socializing Events

Preaching. Bryan and Walbek (1970a,1970b) showed that a model's practices in-fluenced the child's behavior, but the samemodel's preaching did not. However, both themodel's preachings and behavior affected thechild's judgments of the model's attractive-ness. Grusec and Skubiski (1970) showedthat while a model's behavior was a clearsource of behavior change for both sexes,regardless of whether the model was nur-turant or not, the model's verbalizations wereonly effective for females who had been ex-posed to a nurturant model. Grusec (1972),however, showed that a model's verbalizationscould be as effective an influence on thechild's subsequent behavior as a model's be-havior, although not for 7-year-old boys.Rushton (1975) and Rushton and Owen(1975) failed to find direct behavioral effectsdue to preaching on an immediate test, de-spite attempts to strengthen the manipula-tion of the preaching variable substantially.Rushton (1975), however, found that preach-ing to a child had a substantial impact on thechild's behavior on an 8-week retest. Essen-tially, if the preaching were in the same di-


rection as the behavior, far less regression tothe mean occurred over the retest interval.Preaching was thus in some way either pre-venting regression to the mean of the mod-eled behavior in the congruent situation orfacilitating regression to the mean of themodeled behavior in the incongruent situa-tion. Rushton and Owen (197S) also found along-term effect due to preaching from amodel, although only in interaction with themodel's behavior. In this case, however, thepatterning of the results were weaker andsomewhat less clear. Thus, preachings from atelevision model might, in this context, haveless impact than a live one. With both thelive (Rushton, 197S) and the television model(Rushton & Owen, 1975), what the modelpreached had direct effects on the child'sevaluations of the model, thus replicating thefindings of Bryan and Walbek (1970a).

Two other studies found clear evidence fora preaching effect, both on immediate anddelayed tests. Midlarsky and Bryan (1972)found that a model's exhortations affected animmediate test of donating tokens to a char-ity and a 10-day generalization test of donat-ing candies to the same charity carried out ina different setting by a different experimenter.Further, on the immediate test there was aninteraction between the model's verbalizationsand the age of the listening child. Eleven-year-old children were more influenced bythe preachings than were 10-year-olds. Riceand Grusec (197S) found that a model'sverbalizations of what would be appropriatebehavior in a situation that allowed for do-nating to a charity significantly affected 9- to10-year-olds donations to the charity both onan immediate test and on a 4-month retest.It appears that what a model preaches canhave effects on the observer's subsequent be-havior. The conditions that maximize sucheffects are, however, still poorly understood.

Induction. Hoffman (1970, 1976) has sug-gested that the use of induction (i.e., theverbal elaboration of the good or bad natureand consequences of the act for other peopleand for the self) might play an importantrole in the socialization process. Hoffman andSaltzstein (1967) did indeed find a positivecorrelational relationship between the natu-rally occurring use of an inductive socializing

technique by parents and peer ratings of con-sideration for others in 12-year-olds. Elliottand Vasta (1970), in a modeling study men-tioned earlier, found that modeling plus vi-carious reinforcement plus the inductiveverbalization to the model of "If you do some-thing nice for someone else it means you area good boy" produced more subsequent do-nating behavior in observers than the model-ing conditions without the verbal elaboration.However, in a comparison of role playing andinduction techniques, Staub (1971b) foundthat induction procedures either had no effector actually decreased subsequent helping be-havior. Finally, Friedrich and Stein (1975)provided children with labels for behaviorsand feelings connected with altruistic acts andfound positive increments on measures of pro-social responding, particularly with girls incombination with watching prosocial tele-vision programs and taking part in the role-playing training procedure.

THEORETICAL CONSIDERATIONS

Currently, there appear to be two viabletheoretical approaches to the socialization ofaltruism in children: the cognitive-develop-mental stage theory approach (Piaget, 1932;Kohlberg, 1969; Hoffman, 1976) and thecognitive social learning approach (Bandura,1969; Mischel, 1973). Although these twotheories are often viewed as in conflict (Ban-dura & Walters, 1963, pp. 22-24, 206-210;Kohlberg, 1969, pp. 439-446; Rushton,1973), one of the basic differences appears tolie in the referents selected for the analysisof behavior. While the social learning ap-proach emphasizes the role of antecedent andconsequent environmental events (e.g., thepresence or absence of a model or reinforce-ment), the cognitive-developmental approachemphasizes the role of cognitive structures(stages) as measured, for example, by role-taking tasks and moral judgment stories. Letus consider the evidence in light of these twoapproaches.

From the social learning perspective, thetwo most powerful socializing techniquesavailable for producing altruistically behav-ing children are reinforcement and modeling(Bandura, 1969). The literature reviewed onaltruistic behavior by children suggests that


little evidence has been accumulated for theeffectiveness of direct reinforcement proce-dures although some of the existing researchis in the right direction. Modeling processeshave been studied at great length and recentstudies reviewed showed that modeling af-fected the amount, direction, durability, andgeneralizability of altruistic behavior—find-ings which strongly emphasize the power ofthis potential socializing force. Furthermore,increasing research is being directed at othermodeling parameters, demonstrating, we maytentatively conclude, that observation of re-sponse consequences to the model can be ef-fective in producing behavior change in ob-servers and that powerful models behavingwith some positive affect are the most effectivein producing behavior change. Thus, there ismuch of power and utility to the social learn-ing approach. It allows for a technology ofbehavior change, as well as an understandingof the kinds of processes responsible for na-turally occurring altruistic behavior. Thereare, however, limitations and incompletions tothis approach.

It is still a question of concern as to howmodels come to have their effects (cf. Ban-dura, 1969; Gewirtz, 1969; Kohlberg, 1969).What is it exactly that becomes internalized,and what are the laws governing these in-ternalizations? Bandura and Jeffrey (1973)have suggested an information-processingmodel. Mischel (1973) has suggested a num-ber of person variables to account for model-ing phenomena, such as expectancies, codingstrategies, and self-generated rules and plans.To date, however, only demonstrations of theeffects of some of these processes have oc-curred. Their parameters have not yet beenestablished.

Another problem with the social learningapproach is how to account for the othertraining procedures discussed (i.e., training inrole playing and verbal socialization proced-ures) . Inasmuch as training in role playing isreconceptualized as behavioral rehearsal, so-cial learning theory requires no great elabora-tion. The verbal socializing techniques thathave been referred to (preaching, induction,and labeling) have only recently begun to bestudied, and there appears to be little cer-tainty about how to categorize them, con-

ceptualize them, or account for them theoreti-cally. Since most formal and informal meansof exerting influence take part in a verbalmedium, it seems that a great deal more at-tention ought to be given to these socializa-tion processes. The recent moves of much ofsocial learning theory into a consideration ofmediational cognitive variables (Bandura &Jeffrey, 1973; Mischel, 1973) suggest thatsocial learning theory may potentially beelaborated to take some of these other factorsinto account.

From the cognitive-developmental point ofview, the most important socializing techniqueis that of role playing, or the provision ofperspectives separate and different from thoseof the child. Such different perspectives lead,through various cognitive processes, to a de-creasing amount of egocentricity and an in-creasing ability to see the world from differentperspectives and to organize the world in in-creasingly wider and more integrated sche-mata. Thus, the ability to decenter and seethe world (and presumably feel emotions)from another's point of view will be necessaryconditions for the occurrence of genuine con-cern for others. In support of the cognitive-developmental point of view are the findingsthat generosity increases with age and thatchildren within an age group, who have ahigher level of moral judgment ability or role-taking capacity, tend to be more generousthan children with lower levels of moral judg-ment or role-taking ability.

A cognitive-developmental account of thesocializing techniques reviewed would perhapsregard reinforcement, observation of models,and verbal socialization as either (a) provid-ing information to the child as to what wasexpected of him or her or (b) providing inputthat required the child to take new perspec-tives into account. Certainly role-play train-ing procedures such as those used by Staub(1971b) and Friedrich and Stein (197S)might be viewed as rather more than simplyproviding children with an opportunity torehearse behavior. Rather, such procedureswould be viewed as providing the child withperceptions of what it is like to be in an-other's situation. Regarding verbal socializa-tion, the role of induction would perhaps beof most importance because it would lead the


child to focus attention upon the perspectivesof others.

One possible difficulty with the cognitive-developmental approach resides in the degreeof generality found in the literature. If astrong version of the cognitive-developmentalstage position were adopted, one would expectchildren's responses to moral judgment storiesand role-taking tasks to serve as signs ofgeneralizable underlying cognitive schemataand that these would predict "generalizedaltruism." Or, in the words of Kohlberg(1969):

A given stage-response on a task . . . represents anunderlying thought organization . . . which deter-mines responses to tasks which are not manifestlysimilar, (pp. 3S2-3S3)

Such a viewpoint often finds expression inthe literature on cognitive development. Forexample, Flavell (1963) states that

a child of eight who possesses the grouping struc-ture will, by implication from the structure, showreversibility of thought, a relative lack of egocen-trism, a capacity for synthesizing rather than simplyjuxtaposing data, and a number of other charac-teristics, (p. 18)

Thus, the cognitive developmental position asformulated above requires generality of func-tioning at both the cognitive and the behav-ioral levels. The generality at the behaviorallevel appears to be of about 9% of the vari-ance for altruistic behavior. The generalityat the cognitive level does not appear to bemuch higher (see Shantz [197S] for a review,in addition to the research listed here). Itmight be that an increase in specificity incognitive functioning (e.g., moral judgmentsconcerned with distributive justice ratherthan, say, intentionality) might lead to evenbetter predictions of children's altruism.

So, it appears that if the social learningtheorists move further into cognitive proc-esses and the cognitive developmentalists be-come more specific about their cognitive con-structs, a useful integration might be possible.Certainly, neither approach can afford to ig-nore the empirical findings of the other. Ifsome degree of integration cannot occur,perhaps differential predictions from the twoapproaches can be formulated that will thentest the usefulness of one over the other.

For example, there is the interesting questionof whether there are empirical consequents tothe alternate conceptualizations of modelingand role playing espoused by the social learn-ing and cognitive developmental theories (cf.Bandura, 1969; Kohlberg, 1969). One theoryconceptualizes role play and modeling as pro-viding different perspectives for the observerthat increase his role-playing ability. Theother theory focuses upon the learning andrehearsal of specific new forms of behaving.Integration or opposition? These are, nodoubt, exciting prospects for the future.

REFERENCE NOTES1. Krebs, D. L., & Sturrup, B. Role-taking ability

and altruistic behavior in elementary school chil-dren. Paper presented at the annual meeting ofthe American Psychological Association, New Or-leans, August 1974.

2. Hansen, R., Goldman, B. D., & Baldwin, M.Towards a taxonomy of altruism: An observa-tional study of spontaneous prosocial behavioramong young children. Paper presented at theannual meeting of the Canadian PsychologicalAssociation, Quebec City, Canada, June 1975.

REFERENCES

Aronfreed, J. Conduct and conscience: The socializa-tion of internalized control over behavior. NewYork: Academic Press, 1968.

Aronfreed, J. The socialization of altruistic andsympathetic behavior: Some theoretical and ex-perimental analyses. In J. Macaulay & L. Berko-witz (Eds.), Altruism and helping behavior. NewYork: Academic Press, 1970.

Bandura, A. Principles oj behavior modification.New York: Holt, Rinehart & Winston, 1969.

Bandura, A. Vicarious and self-reinforcement proc-esses. In R. Glaser (Ed.), The nature of rein-forcement. New York: Academic Press, 1971.

Bandura, A., & Jeffrey, R. W. Role of symboliccoding and rehearsal processes in observationallearning. Journal of Personality and Social Psy-chology, 1973, 26, 122-130.

Bandura, A., & Walters, R. H. Social learning andpersonality development. New York: Holt, Rine-hart & Winston, 1963.

Bryan, J. H. Model affect and children's imitativealtruism. Child Development, 1971, 42, 2061-2065.

Bryan, J. H., & London, P. Altruistic behavior bychildren. Psychological Bulletin, 1970, 73, 200-211.

Bryan, J. H., & Walbek, N. H. Preaching and prac-ticing self-sacrifice: Children's actions and reac-tions. Child Development, 1970, 41, 329-353. (a)

Bryan, J. H., & Walbek, N. The impact of wordsand deeds concerning altruism upon children.Child Development, 1970, 41, 747-757. (b)

Coates, B., Pusser, H. E., & Goodman, I. The in-fluence of "Sesame Street" and "Mister Rogers'


Neighborhood" on children's social behavior inthe preschool. Child Development, 1976, 47, 138-144.

Dlugokinski, E. L., & Firestone, I. J. Congruenceamong four methods of measuring other-centered-ness. Child Development, 1973, 44, 304-308.

Dlugokinski, E. L., & Firestone, I. J. Other centered-ness and susceptibility to charitable appeals: Ef-fects of perceived discipline. Developmental Psy-chology, 1974, 10, 21-28.

Doland, D. J., & Adelberg, K. The learning ofsharing behavior. Child Development, 1967, 38,695-700.

Elliott, R., & Vasta, R. The modeling of sharing:Effects associated with vicarious reinforcement,symbolization, age, and generalization. Journal ofExperimental Child Psychology, 1970, 10, 8-15.

Emler, N. P., & Rushton, J. P. Cognitive-develop-mental factors in children's generosity. BritishJournal of Social and Clinical Psychology, 1974,13, 277-281.

Fischer, W. F. Sharing in pre-school children as afunction of amount and type of reinforcement.Genetic Psychology Monographs, 1963, 68, 215-245.

Flavell, J. H. The developmental psychology ofJean Piaget. Princeton, N. J.: Van Nostrand, 1963.

Flavell, J. H., Botkin, P. T., Fry, C. L., Wright,J. W., & Jarvis, P. E. The development of role-taking and communication skills in children. NewYork: Wiley, 1968.

Friedrich, L. K., & Stein, A. H. Aggressive and pro-social television programs and the natural behav-ior of preschool children. Monographs of the So-ciety for Research in Child Development, 1973,38(4, Serial No. 151).

Friedrich, L. K., & Stein, A. H. Prosocial televisionand young children: The effects of verbal labelingand role playing on learning and behavior. ChildDevelopment, 1975, 46, 27-38.

Gelfand, D. M., Hartmann, D. P., Cromer, C. C.,Smith, C. L., & Page, B. C. The effects of instruc-tional prompts and praise on children's donationrates. Child Development, 1975, 46, 980-983.

Gewirtz, J. L. Mechanisms of social learning: Someroles of stimulation and behavior in early humandevelopment. In D. A. Goslin (Ed.), Handbookof socialization theory and research. Chicago:Rand-McNally, 1969.

Green, F. P., & Schneider, F. W. Age differences inthe behavior of boys on three measures of altru-ism. Child Development, 1974, 45, 248-251.

Grusec, J. E. Power and the internalization of self-denial. Child Development, 1971, 42, 93-105.

Grusec, J. E. Demand characteristics of the modelingexperiment: Altruism as a function of age andaggression. Journal of Personality and Social Psy-chology, 1972, 22, 139-148.

Grusec, J. E., & Skubiski, S. L. Model nurturancedemand characteristics of the modeling experiment,and altruism. Journal of Personality and SocialPsychology, 1970, 14, 352-359.

Handlon, B. J., & Gross, P. The development of

sharing behavior. Journal of Abnormal and SocialPsychology, 1959, 59, 425-428.

Harris, M. Reciprocity and generosity: Some deter-minants of sharing in children. Child Development,1970, 41, 313-328.

Harris, M. Models, norms and sharing. PsychologicalReports, 1971, 29, 147-153.

Hartshorne, H., May, M. A., & Mailer, J. B. Studiesin the nature of character. Vol. 2. Studies in ser-vice and self-control. New York: Macmillan, 1929.

Hartup, W. W., & Coates, B. Imitation of a peeras a function of reinforcement from the peer groupand rewardingness of the model. Child Develop-ment, 1967, 38, 1003-1016.

Hoffman, M. L. Moral development. In P. H.Mussen (Ed.), Carmichael's manual of child psy-chology (Vol. 2, 3rd ed.). New York: Wiley,1970.

Hoffman, M. L. Empathy, role taking, guilt, anddevelopment of altruistic motives. In T. Lickona(Ed.), Moral development and behavior: Theory,research, and social issues. New York: Holt, Rine-hart & Winston, 1976.

Hoffman, M. L., & Saltzstein, H. Parent disciplineand the child's moral development. Journal of Per-sonality and Social Psychology, 1967, S, 45-47.

Kagan, S., & Madsen, M. C. Cooperation andcompetition of Mexican, Mexican-American, andAnglo-American children of two ages under fourinstructional sets. Developmental Psychology, 1971,5, 32-39.

Kohlberg, L. Stage and sequence: The cognitive-developmental approach to socialization. In D. A.Goslin (Ed.), Handbook of socialization theoryand research. Chicago: Rand-McNally, 1969.

Krebs, D. L. Altruism—An examination of theconcept and a review of the literature. Psycho-logical Bulletin, 1970, 73, 258-302.

Macaulay, J., & Berkowitz, L. (Eds.). Altruism andhelping behavior. New York: Academic Press,1970.

Madsen, M. C. Developmental and cross-culturaldifferences in the cooperatvie and competitive be-havior of young children. Journal of Cross-Cultural Psychology, 1971, 4, 365-371.

Madsen, M. C., & Connor, C. Cooperative and com-petitive behavior of retarded and nonretardedchildren at two ages. Child Development, 1973,44, 175-178.

Midlarsky, E., & Bryan, J. H. Training charity inchildren. Journal of Personality and Social Psy-chology, 1967, 5, 408-415.

Midlarsky, E., & Bryan, J. H. Affect expressions andchildren's imitative altruism. Journal of Experi-mental Research in Personality, 1972, 6, 195-203.

Midlarsky, E., Bryan, J. H., & Brickman, P. Aver-sive approval: Interactive effects of modeling andreinforcement on altruistic behavior. Child Devel-opment, 1973, 44, 321-328.

Mischel, W. Personality and assessment. New York:Wiley, 1968.

Mischel, W. Toward a cognitive social learning re-


conceptualization of personality. Psychological Re-view, 1973, 50, 2S2-283.

Mussen, P., Rutherford, E., Harris, S., & Keasey, C.Honesty and altruism among preadolescents. De-velopmental Psychology, 1970, 3, 169-194.

Piage't, J. The moral judgment of the child. London:Routledge & Kegan Paul, 1932.

Presbie, R. J., & Coiteux, P. F. Learning to begenerous or stingy: Imitation of sharing behavioras a function of model generosity and vicariousreinforcement. Child Development, 1971, 42, 1033-1038.

Rheingold, H. L., Hay, D. F., & West, M. J. Sharingin the second year of life. Child Development, inpress.

Rice, M. E., & Grusec, J. E. Saying and doing:Effects on observer performance. Journal of Per-sonality and Social Psychology, 197S, 32, 584-593.

Rosenhan, D. L. Some origins of concern for others.In P. Mussen, J. Langer, & M. Covington (Eds.),Trends and issues in developmental psychology.New York: Holt, Rinehart, & Winston, 1969.

Rosenhan, D. L. Learning theory and prosocial be-havior. Journal of Social Issues, 1972, 28, 151-163.

Rosenhan, D. L., & White, G. M. Observation andrehearsal as determinants of prosocial behavior.Journal of Personality and Social Psychology,1967, 5, 424-431.

Rubin, K. H., & Schneider, F. W. The relationshipbetween moral judgment, egocentrism, and altru-istic behavior. Child Development, 1973, 44, 661-665.

Rushton, J. P. Social learning and cognitive devel-opment: Alternative approaches to an understand-ing of generosity in 7 to 11 year olds. Unpublisheddoctoral dissertation, University of London, Eng-land, 1973.

Rushton, J. P. Generosity in children: Immediateand long term effects of modeling, preaching, andmoral judgment. Journal of Personality and SocialPsychology, 1975, 31, 459-466.

Rushton, J. P., & Campbell, A. C. Modeling, vicari-ous reinforcement and extraversion on blood do-nating in adults. Immediate and long-term effects.European Journal of Social Psychology, In press.

Rushton, J. P., & Owen, D. Immediate and delayedeffects of TV modeling and preaching on children'sgenerosity. British Journal of Social and ClinicalPsychology, 1975, 14, 309-310.

Rushton, J. P., & Wiener, J. Altruism and cognitive

development in children. British Journal of Socialand Clinical Psychology, 1975, 14, 341-349.

Rutherford, E., & Mussen, P. Generosity in nurseryschool boys. Child Development, 1968, 39, 755-765.

Selman, R. L. The relation of role taking to thedevelopment of moral judgment in children. ChildDevelopment, 1971, 42, 79-91.

Shantz, C. U. The development of social cognition.In E. M. Hetherington (Ed.), Review of childdevelopment research (Vol. 5). Chicago: Univer-sity of Chicago Press, 1975.

Staub, E. A child in distress: The influence of ageand number of witnesses on children's attempts tohelp. Journal of Personality and Social Psychol-ogy, 1970, 14, 130-140.

Staub, E. A child in distress: The influence of nur-turance and modeling on children's attempts tohelp. Developmental Psychology, 1971, 5, 124-133. (a)

Staub, E. Use of role playing and induction inchildren's learning of helping or sharing behavior.Child Development, 1971, 42, 80S-816. (b)

Stein, A. H., & Friedrich, L. K. Television contentand young children's behavior. In J. P. Murray,E. A. Rubinstein, & G. A. Comstock (Eds.),Television and social behavior (Vol. 2, Televisionand social learning). Washington, D.C.: Govern-ment Printing Office, 1972.

Ugurel-Semin, R. Moral behavior and moral judg-ment of children. Journal of Abnormal and SocialPsychology, 1952, 47, 463-474.

White, G. M. Immediate and deferred effects ofmodel observation and guided and unguided re-hearsal on donating and stealing. Journal of Per-sonality and Social Psychology, 1972, 21, 139-148.

Wright, B. Altruism in children and the perceivedconduct of others. Journal of Abnormal and So-cial Psychology, 1942, 37, 218-233.

Wright, D. The psychology of moral behavior.Harmonds worth, Middlesex, England: PenguinBooks, 1971.

Yarrow, M. R., Scott, P. M., & Waxier, C. Z.Learning concern for others. Developmental Psy-chology, 1973, 8, 240-260.

Yarrow, M. R., & Waxier, C. Z. Dimensions andcorrelates of prosocial behavior in young children.Child Development,, 1976, 47, 118-125.

(Received May 13, 1975)

Documents

Socialization and the Altruistic Behavior of Children