Cognitive Development in Retarded and Nonretarded Persons ... · Cognitive Development in Retarded and Nonretarded Persons: ... 1977; Weisz & Achenbach, 1975). The emphasis on pro-cess

Psychological Bul le t in1981, Vol.90, No. I, 153-118

Copyright 1981 by the American Psychological Association, Inc.0033-2909/81 /9001-0153S00.75

Cognitive Development in Retarded and Nonretarded Persons:Piagetian Tests of the Similar Structure Hypothesis

John R. Weisz and Keith Owen YeatesUniversity of North Carolina at Chapel Hill

Proponents of "developmental" and "difference" theories of mental retardationhave long debated the similar structure hypothesis. It holds that when nonor-ganically impaired retarded and nonretarded persons are similar in developmentallevel (operationally defined as mental age), they are also similar in the processesand concepts by which they reason. The developmental position endorses thishypothesis; the difference position opposes it. The hypothesis and the debate overits validity have been strongly influenced by Piaget's cognitive developmentaltheory. In this article we survey 30 studies involving 104 separate tests of thehypothesis with Piagetian conceptual measures. We distinguish between studiesthat did and did not meet an important procedural requirement: screening or-ganically impaired subjects from their mentally retarded samples. Studies thatdid not meet this requirement yielded findings inconsistent with the similar struc-ture hypothesis; retarded groups were significantly inferior to nonretarded groupsmatched on mental age. Studies that did meet the requirement, however, yieldedfindings supporting the hypothesis. This pattern of findings carries significantimplications for the developmental versus difference controversy and for diagnosisand training of the mentally retarded.

When a mentally retarded child and ayounger nonretarded child happen to be atthe same level of intellectual development,say at the same mental age, how similar arethe two children in the processes by whichthey reason? For a decade and a half thishas been a central issue in an often heateddebate. The debate has far-reaching impli-cations for the study of both mental retar-dation specifically and cognitive develpmentgenerally; it is called the developmental ver-sus difference controversy.

On one side of the controversy is the de-velopmental position, advanced by Zigler

The review was facilitated by support from NationalInstitute of Mental Health Research Grant 1 RO3MH34652 01 to John R. Weisz and National Instituteof Health and Human Development Training Grant 1T32 HDO7178 to Keith Owen Yeates.

The material discussed in this article is also reviewedin the forthcoming book, Mental Retardation: The De-velopmental-Difference Controversy, edited by EdwardZigler and David Balla.

We are grateful to Thomas Achenbach and DavidMacPhee for their helpful comments on an earlier ver-sion of this article.

Requests for reprints should be sent to John R. Weisz,Department of Psychology, University of North Caro-lina, Chapel Hill, North Carolina 27514.

(1969) and elaborated by Weisz, Yeates,and Zigler (in press). This position appliesonly to people not suffering from organicimpairment. It holds that retarded and non-retarded people pass through cognitive de-velopmental stages in an identical order.These stages are usually defined as thosedescribed by Piaget (e.g., 1964, 1970). Re-tarded people are said to traverse thesestages more slowly and attain a lower upperlimit than nonretarded people. But when thetwo groups are equated for level of cognitivedevelopment, which is most often operation-ally defined as psychometric mental age(MA), they are not expected to differ in thecognitive processes by which they reason.

On the other side of the controversy aretheorists who hold what Zigler (1969) haslabeled the difference position. One tenet ofthis position is that the cognitive develop-ment of the retarded person differs from thatof the nonretarded person in ways other thanmere differences in developmental rate andceiling. Even when retarded and nonretardedpeople are equated for level of development,they are expected to differ in the cognitiveprocesses they use in reasoning. Milgram(1973), for example, has argued that the

153

154 JOHN R. WEISZ AND KEITH OWEN YEATES

cognitive stages of retarded individuals arestructurally different from those of the non-retarded. The retarded person's stages, heargued, are likely to contain traces of moreprimitive developmental stages and are aptto give way to a regression to those earlierstages. Consequently, he maintained thatretarded persons employ less advanced rea-soning than MA-matched nonretarded per-sons.

By its emphasis on processes of reasoning,the developmental versus difference contro-versy has helped to stimulate a significantshift in the focus of comparative research.There has been a reduced emphasis on suchend products of learning and problem solvingas trials to criterion on experimental tasks.Investigators have been increasingly atten-tive to the dynamics underlying such learn-ing and problem solving (see e.g., studies ofhypothesis behavior by Weisz, 1977; Weisz& Achenbach, 1975). The emphasis on pro-cess has also stimulated a burgeoning inter-est in comparative research on the conceptspeople use in reasoning. These concepts, asdescribed by Piaget (e.g., 1970), are the veryimplements of thought, the tools by whichreasoning occurs.

Along with heightened interest in Piage-tian theory has come an awareness that thedevelopmental versus difference debate reallyinvolves two hypotheses. Either may be trueor false independent of the other. One is thehypothesis that retarded and nonretardedindividuals pass through cognitive develop-mental stages in the same order. This hasbeen labeled the similar sequence hypothe-sis. Evidence germane to it was examined byWeisz & Zigler (1979), and most of the ev-idence supported the hypothesis. The secondhypothesis addressed the question with whichwe began this article, and it is the centralfocus of this review. We describe the hy-pothesis and discuss its importance in detail.

Similar Structure Hypothesis andPiagetian Research

As noted earlier, the developmental po-sition holds that retarded and nonretardedpeople who are matched for level of devel-opment are also alike with respect to their

formal processes of reasoning. We label thistenet the similar structure hypothesis. Weuse this label because the hypothesis is basedon the view that such carefully matched re-tarded and nonretarded people are similarin cognitive structure as described by Piaget(see Zigler, 1969, pp. 537-541). In thePiagetian literature the term cognitive struc-ture is used to represent the way intellect isorganized (e.g., the logico-mathematicalforms that underlie particular patterns ofreasoning). Cognitive structure cannot bedirectly observed. Instead it is inferred fromvarious overt manifestations of reasoning.These manifestations constitute operationaldefinitions of cognitive structure. Typicallythey take the form of Piagetian measures ofthe concepts children use in reasoning andsolving problems. Thus, studies using suchmeasures form the pool of evidence used inthe present review.

The similar structure hypothesis holdsthat when these Piagetian measures are usedwith retarded and nonretarded groups ofsimilar MA, and when appropriate experi-mental control has been exercised (as de-scribed below), the groups will not differreliably. In sharp contrast, Milgram (1973)argues that the retarded will prove inferiorto their nonretarded MA peers in the con-ceptual processes required to perform Piage-tian tasks. He argued that IQ is an index ofproblem-solving ability, and he cites In-helder's (1968) view that the retarded child'sreasoning shows "viscosity, oscillation," andregression to immature levels. Building onthis rationale, Milgram (1973) predictedthat "given tasks without a ceiling or flooreffect for the general MA of the subjectsbeing used, the retarded will dependablydemonstrate an equal-MA deficit" (p. 209).

In arguing that the retarded will be in-ferior to their nonretarded peers, Milgramadopted a view that is typical of most dif-ference positions. A rather unconventionaldifference position has been taken by Kohl-berg (1968). He believes that the kind ofconceptual development described by Piagetrequires massive doses of "general experi-ence." Consequently, Kohlberg reasons thatthe retarded should be conceptually moreadvanced than their nonretarded MA peersbecause they have lived longer and thus have

SIMILAR STRUCTURE HYPOTHESIS 155

acquired a more extensive base of experi-ence.

For two reasons, then, evidence from thePiagetian tradition seems ideally suited toa test of the similar structure hypothesis.First, the hypothesis itself was stimulated byattention to the kinds of underlying processesPiagetian conceptual measures are designedto tap. These measures are quite diverse;they cover a broad spectrum of the child'sexperience with the world. Second, it is spe-cifically within the Piagetian domain that anunusual triangular theoretical conflict hastaken shape. There are three possible out-comes of experimentation comparing MA-matched retarded and nonretarded groups.Each outcome can add to our knowledge bysupporting one of three incompatible views—the developmental position and its similarstructure hypothesis (Zigler, 1969), the con-ventional difference position (Milgram,1973), or an unconventional difference po-sition (Kohlberg, 1968). This three-way the-oretical conflict presents an opportunity forthe kind of strong inference rarely seen inpsychological research. In the present articlewe attempt to take advantage of this unusualopportunity. First, certain caveats are inorder.

Methodological Issues: Ground Rules for aFair Test

Matching for Developmental Level

The similar structure hypothesis predictssimilar reasoning in children who are equatedfor developmental level. The most widelyemployed operational definition of develop-mental level in research on the developmen-tal-difference controversy is the psychomet-ric MA. Certainly, MA is at best a roughindex; it could possibly be improved upon bya collection of measures derived from de-velopmental theory. Nonetheless, the MA,as obtained on such standardized instru-ments as the Stanford-Binet IntelligenceTest, is a broad-based measure that taps awide range of cognitive activities. Moreover,nonorganically impaired mentally retardedand nonretarded children who are matchedfor total Stanford-Binet MA (both shortform and standard) are evidently similar in

the patterns of items they pass (Achenbach,1970, 1971). Although the MA may be aless than ideal measure of developmentallevel, it appears to be an adequate matchingdevice for tests of the similar structure hy-pothesis.

The Question of Organic Impairment

The developmental position generally, andthe similar structure hypothesis in particu-lar, both apply only to people not afflictedwith specific genetic or other physiologicaldefects. Thus the theory is not intended toapply to people whose retardation is linkedto such causes as Down's syndrome or braininjury. Some critics of the developmentalposition (Ellis, 1969; Milgram, 1969, 1973;Spitz, in press) have maintained that thesekinds of organic impairments do not influ-ence behavior significantly. Consequently,they have argued that it is not necessary toexclude organically impaired persons whenstructuring tests of the developmental posi-tion. Ellis, (1969) for example, insisted thatonly "rarely have behavioral differencescharacterized different etiological groups"(1969, p, 561). Reviews touching on thisquestion reveal rather mixed evidence (e.g.,Blount, 1968). It should be noted, however,that a number of investigators (e.g., Cruik-shank, 1967; Reitan, 1973) have devotedmuch of their lifework to the study of idio-syncratic behavior distinguishing organi-cally impaired individuals from those withintact nervous systems. Furthermore, thereis some evidence that organic impairmentmay affect performance on the specific kindsof reasoning and problem-solving tasks mostoften used in research on the developmentalversus difference controversy; effects havebeen found in mentally retarded groups (seeBalla, Styfco, & Zigler, 1971; Balla & Zig-ler, 1964; Harter, Brown, & Zigler, 1971)and in nonretarded groups (Elkind, Koegler,Go, & Van Doornick, 1965).

In summary, some but not all studies haverevealed performance anomalies associatedwith organic impairment. The developmen-tal position specifically excludes organicallyimpaired individuals from its predictions.Furthermore, difference theories (with thepossible exception of Luria's, 1963, position)


can be adequately tested with samples thatexclude organically impaired subjects. Thusthe clearest tests of the similar structurehypothesis are those that exclude organicallyabnormal cases. This does not mean thatstudies failing to satisfy this condition havebeen excluded from the present review. Itdoes mean, however, that such studies mustbe regarded as yielding less than ideal evi-dence. At the end of our review, we try todetermine whether studies in which organi-cally impaired individuals were excludedshowed a different pattern of findings thanstudies in which organically impaired indi-viduals were not excluded.

Role of Noncognitive Factors

Another consideration is that the behaviorof retarded people reflects more than theirformal cognitive processes. There is generalagreement that a relative deficiency in cog-nitive functioning is the essential definingfeature of mental retardation. However,noncognitive factors distinguishing retardedfrom nonretarded persons may have ob-scured the precise nature of this cognitivedeficiency. Over the past 2 decades the workof the Peabody group (see Cromwell, 1963),Gordon and O'Connor (reviewed by Heber,1964), Stevenson and his colleagues (Ste-venson, 1965), and the Yale group (see Zig-ler, 1971), has revealed motivational char-acteristics of the retarded that underminetheir experimental task performance. Forexample, numerous studies by the Yalegroup (see Zigler, 1971) have exposed theimpact of such factors as institutionaliza-tion, heightened desire for attention andpraise, wariness of adults, idiosyncratic re-ward preferences, tendency to anticipatefailure, and willingness to settle for relativelylow levels of success. There is also a growingbody of evidence on learned helplessnessamong the retarded; it indicates that re-tarded children, particularly older ones, areless likely than nonretarded children of sim-ilar MA to ascribe their failures to control-lable factors and are less likely to persist inthe face of failure and negative feedback(see Gibson, 1980; Weisz, 1979, 1981, inpress-b).

Given such findings on motivation and

helplessness, it is clear that for tests of thesimilar structure hypothesis to be fair, re-tarded and nonretarded subjects should be(a) equally motivated to perform at theirbest and (b) equally confident that successis possible. This standard reflects a simpleprinciple: One cannot conclusively attributea group difference on a dependent variableto one known difference in subject charac-teristics (e.g., IQ) when the subjects alsodiffer on other factors that affect perfor-mance on that variable. The importance ofthis principle is underscored in the writingsof Cole and Bruner (1971), Labov (1970),and Tulkin and Konner (1973) on the dan-gers of attempting to infer competence di-rectly from performance.

Having stressed the importance of con-trolling noncognitive factors, we must notethat few of the studies we review show ex-plicit efforts to exercise such control. At thispoint in the evolution of the developmental-difference controversy, however, we believeit is important to take the proverbial bull bythe horns and use such data as are availablein the literature. We have therefore includedall the studies we could find that we consid-ered relevant, whether or not they includedefforts to control motivation and expectancy;however, we have attempted to provide suf-ficient details of each study to enable thereader to judge whether noncognitive factorsmay have influenced the findings.

Variability and the Reliability of "No-Difference" Findings

The reader should be sensitive to one otheraspect of the studies reviewed. The similarstructure hypothesis is fundamentally a nullhypothesis. That is, a conclusion that thehypothesis is supported can result only fromfindings of no significant difference betweenMA-matched groups differing in IQ. Yet no-difference findings have traditionally beenregarded by social scientists as inconclusive.We discuss the implications of this paradoxafter reviewing the data. During the review,however, we attempt to shed light on thequestion of whether the no-difference find-ings might have resulted from excessive vari-ability in the data. Such an interpretationof a no-difference finding loses considerable


credibility when other analyses of the samesubjects yield statistically significant differ-ences. We therefore indicate, with each no-difference finding reported, whether otherdifferences among the same subjects at-tained significance.

Having registered these caveats, we nowproceed to the evidence on the similar struc-ture hypothesis, drawn from several differ-ent conceptual domains.'

Tests of the Similar Structure Hypothesis

Moral Judgment

Retarded people have often been labeledimmoral by the nonretarded in our society(see accounts in Kamin, 1975; Sarason &Doris, 1969; Zigler & Harter, 1969). Con-sequently, research comparing the moraljudgment of retarded and nonretarded per-sons should be of particular interest. Ourliterature search yielded two relevant exper-iments, both carefully designed. In the first,Taylor and Achenbach (1975) compared thejudgments of MA-matched retarded andnonretarded subjects at three MA levels: 6'/2,8, and 9!/2 years. Retarded and nonretardedgroups had mean IQs of 75 and 111, andmean chronological ages (CAs) of 11.4 and6.9 years. Subjects judged what various pro-tagonists should do in stories involving moraldilemmas. The dilemmas concerned prom-ises, stealing, punishment, and the value oflife. The youngsters' rationales for theirjudgments were scored according to Kohl-berg's (1969) stages of moral reasoning. Allsubjects were from public schools: recordswere used to select only children showing noindication of organic impairment. The re-tarded and nonretarded groups did not differsignificantly in their level of moral judgment.There was a significant effect of MA levelon moral judgment, however. This suggeststhat the no-difference finding did not resultfrom excessive variability in the data. TheTaylor-Achenbach study included Piagetianmeasures in other conceptual domains. Theseare discussed later in this review.

The second study was designed partly asa sequel to the Taylor-Achenbach experi-ment. Kahn (1976) compared moral judg-ments of three MA-matched groups in afive-story interview based on Kohlberg's pro-

cedures. His subjects were all public schoolchildren: (a) nonretarded subjects averaging101 in IQ, 7.3 years in MA, and 6.9 yearsin CA; (b) mildly retarded subjects aver-aging 66 in IQ, 7.5 years in MA, and 10.7years in CA, all screened to eliminate or-ganic cases; and (c) moderately retardedsubjects averaging 45 in IQ, 7.3 years inMA, and 18.2 years in CAs, with etiologiesincluding Down's syndrome, rubella, con-genital syphilis, and other organic abnor-malities. Consistent with the similar struc-ture hypothesis, the first two groups did notdiffer in level of moral judgment; in fact, thesecond group was not significantly differentfrom either of the other groups. However,the group suffering from organic impair-ment was significantly inferior to the non-retarded group in moral judgment. Kahn'sconclusion about this pattern of findings di-rectly addresses the etiological issue we dis-cussed earlier: "Similarity of performanceby [MA-matched] retarded and nonretardedchildren does not extend to retarded childrenwith etiologies other than cultural-familialretardation" (1976, p. 213). Other measuresused by Kahn are discussed below.

Role Taking

One conceptual skill that many considerrelated to moral judgment is role taking. TheTaylor and Achenbach (1975) investigationincluded two role-taking tasks. In the first,the subject's level of awareness of the ex-perimenter's thought processes was inferredfrom behavior on successive games of tic-tac-

1 The evidence that is reviewed with regard to thesimilar structure hypothesis unfortunately does not in-clude any of the data from the Temple longitudinalstudy (Stephens, et al., 1974), described in the reviewof the similar sequence hypothesis (Weisz & Zigler,1979). Although the Temple data are potentially ap-plicable to the similar structure hypothesis, the pub-lished analyses of the data (Mahaney & Stephens, 1974;Stephens, 1972; Stephens et al., 1974) are not. Retardedand nonretarded groups in the study are not matchedfor MA, and when analysis of covariance is employed(see Stephens & McLaughlin, 1974) in contrasting theperformance of retarded and nonretarded subjects, it isused to control for both MA and CA. This analysis,which in principle also controls for group differences inIQ, is not appropriate for a test of the similar structurehypothesis (see Weisz, 1976) and poses other statisticalproblems as well (see Kappauf, 1976).


toe. The second was a guessing game de-signed to tap higher levels of role-taking skill(adapted from Flavell, Botkin, Fry, Wright,& Jarvis, 1968). In this game the child triedto fool an observer who had to guess whichof two boxes the child had taken penniesfrom. When responses to both tasks werescored according to Kuhn's (Note 1) role-taking criteria, the retarded and nonretardedgroups did not differ. There was a significantMA effect on the simpler role-taking taskand a significant sex effect (boys superior)on the more advanced task.

A large-scale study by DeVries (1970,1973a, 1973b) included a role-taking taskin which the child tried to fool an experi-menter who guessed which hand the childhad hidden a penny in; at other times thechild did the guessing. Intellectually averagechildren with a mean IQ of 105, mean MAof 6.8 years, and mean CA of 6.5 years wereincluded. These youngsters were comparedwith a retarded group with a mean IQ of 72,mean MA of 6.5 years, and mean CA of 8.8years. Retarded children were evidently notscreened for organic impairment. In contrastto the findings of Taylor and Achenbach(1975), DeVries found retarded children tobe significantly inferior to the nonretardedin role-taking performance. DeVries usedother measures that we discuss later.

Perspective Taking and Perceptual-Spatial Concepts

A number of problems outside the socialrealm per se involve reasoning that is con-ceptually similar to role taking.

Perspective-taking {spatial egocentrism),One problem is that of identifying how aperceptual array would look from a per-spective other than one's own. Rubin andOrr (1974) compared MA-matched re-tarded and nonretarded children on two suchtasks; one task involved identifying and theother, constructing the perspective of an-other person on a stimulus array. Bothgroups of children were from public elemen-tary schools. Their IQs averaged 73 and 105,their MAs averaged 7.7 years and 8.5 years,and their CA means were 11.9 and 7.9 years,respectively. There was apparently noscreening of organic cases. Nonetheless, the

two groups did not differ reliably on eithertask; however, the retarded group was sig-nificantly inferior to a CA-matched groupon both tasks.

Perceptual decentering. Smith (1977b)assessed a related skill using a modificationof Elkind, Koegler, and Go's (1964) PictureIntegration Test. The test consisted of sev-eral complex pictures (e.g., a face composedof a bat, balls, acorns, and a drum). To assessthe child's ability to decenter, the experi-menter first asked what the picture lookedlike and then whether the child saw anythingelse. Responses were scored for the numberof responses involving parts, the number in-volving whole "gestalts," and the number ofpart-whole integrations (the most advancedresponse developmentally). Again the sub-jects were MA-matched retarded and non-retarded public school children. The twogroups had mean IQs of 69 and 115, the MAmeans were 8.6 years and 9.1 years, and CAmeans were 12.5 years and 8.5 years, re-spectively. The retarded sample was evi-dently not screened for organically impairedchildren. The two groups did not differ sig-nificantly on the decentering task; however,both groups had significantly more part-whole integrations than did a retarded groupmatched to the nonretarded group for CA.

Relative thinking. Another related prob-lem is that of relative thinking as measuredby Piaget's (1928) "brother-sister" and"right-life" tests. The first test taps chil-dren's grasp of the relation between beinga sibling and having a sibling. A sampleitem: George has three brothers, Paul, Henry,and Charles. How many brothers does Paulhave? How many brothers are there in thefamily? On the right-left test, children pointto their own right and left hands, then to theexperimenter's. McManis (1969f) gave bothtests to 15 retarded and 15 nonretardedyoungsters at each yearly MA level from 5to 10 years. The IQs averaged 58 for re-tarded subjects and 100 for the nonretarded;CAs ranged from 7.7 to 21.2 years in theretarded group and from 5.3 to 10.8 yearsin the nonretarded (CA means were not re-ported). The source of subjects (publicschools vs. residential institutions) was notspecified. McManis (1969f) reported no sig-nificance tests. He indicated, however, that


on both tasks the pattern of results was sim-ilar: as level of abstractness in the test ques-tions increased, retarded subjects showed in-creasing MA deficits (i.e., correctly answeredquestions at later MA levels than nonre-tarded subjects). Given the well-establishedlinkage between brain damage and difficultywith abstraction, it is important to note thatMcManis's (1969f) retarded sample in-cluded substantial percentages suffering fromorganic impairment: 62% at MA 5, 43% atMA 6, 43% at MA 7, 23% at MA 8, 8% atMA 9, and 46% at MA 10.

Lane and Kinder (1939) used the sametwo Piagetian tasks to assess relative think-ing among five groups of institutionalizedretarded subjects of unspecified etiology.Two of the groups differed in IQ but werematched for MA. A group with a mean IQof 51 averaged 8.3 years in MA and 16.3years in CA. A group with a mean IQ of 64averaged 8.1 years in MA and 12.1 years inCA. No significance tests were reportedanywhere in the study. However, the twoMA-matched groups were quite similar,both in the gross percentage of questionsanswered correctly and in the percentage ofsubjects answering each individual questioncorrectly. Other groups in the sample dif-fered markedly in both respects. Lane andKinder (1939) note that "the records of thetwo groups with the same mental age arethus found to be strikingly similar, whereascomparisons on the basis of chronologicalage or IQ show marked differences in per-formance" (p. 113).

In a third study of relativism, Prothro(1943) administered the right-left test toMA-matched institutionalized retarded andpublic nursery and elementary school non-retarded subjects. The nonretarded samplehad a mean IQ of 113, a mean MA of 6.2years, and a mean CA of 5.6 years. The re-tarded sample had a mean IQ of 39, a meanMA of 6.1 years, and a mean CA of 27.3years; only retarded individuals diagnosed"familial" or "undifferentiated" were in-cluded; cases involving "motor defects" orepilespy were excluded. It is not clear whatetiologies may have characterized membersof the undifferentiated group. It is possible,however, that some of them were organicallyimpaired, since an IQ of 50 is the approxi-

mate lower limit for the familial retarded(see Zigler, 1967), and Prothro's groupmean was 39. Performance of the nonre-tarded children on the right-left test, thoughat a very low level, was significantly superiorto that of the retarded. We return to theProthro study when we discuss animism,causality, and quantitative concepts.

The Devries (1970, 1973a, 1973b) andTaylor and Achenbach (1975) studies alsoassessed relative thinking. Recall that De-vries's study compared retarded children av-eraging 71 in IQ and 6.5 years in MA withnonretarded children averaging 105 in IQand 6.8 years in MA; the retarded group wasapparently not screened for organic impair-ment. Taylor and Achenbach did screen outorganically impaired cases. Their sample in-cluded retarded and nonretarded childrenwith mean IQs of 75 and 111, respectively;the groups were MA-matched at three MAlevels: 6'/2, 8, and 9'/a years. Devries foundno significant difference between retardedand nonretarded subjects on either thebrother-sister or right-left test, but thebrother-sister test did differentiate signifi-cantly between high-IQ and average-IQchildren of the same CA. Similary, Taylorand Achenbach (1975) found that whereasperformance on the two tests improved sig-nificantly with MA, there were no significantdifferences between retarded and nonre-tarded children matched for MA.

Logical contradiction. As a child movesinto the concrete operational stage of cog-nitive development, an important transitiontakes place, Perceptually dominated thoughtgives way to logically guided thought. Thatis, cognition begins to be less "stimulusbound." Smith (1977a) investigated the rel-ative progress of retarded and nonretardedsubjects into logically guided thought. Hecompared the performance of 30 nonre-tarded and 30 retarded school childrenmatched for MA on a task involving logicalcontradiction. The IQ, MA, and CA meanswere not reported, and there was no indi-cation that the children were screened fororganic impairment. The task was a modi-fication of the Jastrow ("ring segment") Il-lusion. In this illusion, the child is first askedto pick the "big" ring segment. The positionof the two ring segments is then reversed,


and the subject is asked to pick the "little"ring segment; when position is changed, thering first identified as big by the child ap-pears to be little. Smith (1977a) and an in-dependent observer rated reactions to theillusion, looking for evidence of surprise atthe contradiction between logic and percep-tion. They found that the MA-matched re-tarded and nonretarded children did not dif-fer significantly in frequency of surprisereactions. However, retarded children showedsurprise significantly less often than a CA-matched nonretarded group.

Conceptions of the World and PhysicalCausality

Animism. One of Piaget's earliest inter-ests concerned children's interpretations ofnaturalistic phenomena. Of the many thatPiaget investigated, one of the first subjectedto American comparative research was thechild's concept of life. Russell, Dennis, andAsh (1940), administered a standardized in-terview in which subjects were asked to clas-sify a series of objects as living or nonlivingand to justify each classification. The re-tarded subjects were inmates in two insti-tutions for the "feeble-minded"; they werenot described with respect to etiology or IQ.A nonretarded contrast group (no IQ or in-stitutional status reported) was matched tothe retarded group for MA range at two MAlevels (i.e., 6-7 and 8-9 years). Neither MAnor CA means were reported for retardedor nonretarded subjects. At the lower MAlevel the two groups did not differ signifi-cantly in animism stage level. At the higherMA level, however, the retarded subjectshad significantly more advanced life con-cepts than their nonretarded MA peers. Fol-lowing up on this finding, Russell et al. com-pared that portion of the retarded samplewho were under 21 years of age with theirolder retarded MA peers; the older groupshowed significantly more advanced life con-cepts. Russell et al. concluded that althoughMA imposes restrictive limits on the effectsof CA, experience does seem to enhance thedevelopment of the life concept over andabove the effects of MA alone. This conclu-sion and the finding on which it is basedrepresent the first support we have found for

Kohlberg's (1968) unconventional differ-ence position on retardation.

However, when Prothro (1943) used theRussell et al. procedure he failed to find sup-port for Kohlberg's (1968) position. InProthro's study, described earlier, a retardedgroup screened for motor defects and epi-lepsy averaged 39 in IQ and 6.1 years inMA. The nonretarded comparison groupaveraged 113 in IQ and 5.6 years in CA.The two groups did not differ significantlyin life concepts.

Granich (1940) used an interview similarto but more diversified than that used byRussell et al. and by Prothro. Granich's sam-ple included public school boys of averageand retarded intelligence matched for MArange (7.2-8.4 years). The nonretarded boysaveraged 102 in IQ, 7.8 years in MA, and7.6 years in CA. The retarded boys averaged56 in IQ, 7.7 years in MA, and 13.9 yearsin CA. The etiologies for the retarded sam-ple were not reported. The two subjectgroups were very similar in frequency of at-tribution of life (or traits of living things)to animate and inanimate objects; bothgroups were markedly inferior to a group ofnormal children averaging 13.6 years in MA.No significance tests were reported, how-ever. The Granich interview included ques-tions on several other conceptual domainsthat are discussed later in this review.

More recently, Smeets (1973, 1974) stud-ied children's attributions of life and specificlife traits (e.g., dying, growing, feeling) toanimate and inanimate objects. In his firststudy, children's attributions were followedby the question, What makes you think so?The sample included retarded and nonre-tarded subjects matched for MA range. Thesource of subjects was not specified. Groupmeans were not reported, but the nonre-tarded group ranged from 90 to 110 IQ, from5.8 to 7.2 years in MA, and from 5.8 to 7.2years in CA. The retarded group rangedfrom 50 to 65 in IQ, from 5.8 to 7.2 yearsin MA, and from 11.0 to 12.0 years in CA.In addition, an older nonretarded group (IQs90-110) were matched to the retarded groupfor CA range (11.0-12.0 years). Errors weremost frequent among the young normal sub-jects and least frequent among the older nor-mal subjects, but no tests of significance


were reported. However, a probability anal-ysis of the distribution of errors indicatedthat the errors made by the retarded andyounger normal groups were largely due toguessing; the errors made by the older nor-mal group were less frequent and less ca-pricious. In a second report, apparentlybased on the same experiment, Smeets (1974)presented significance tests for differencesamong the groups. In the attribution of lifeto animate objects, the three groups differedsignificantly overall, but the retarded andMA-matched nonretarded did not. In theattribution of life to inanimate objects andin the attribution of life traits to animateobjects, the three groups did not differ sig-nificantly. Finally, attribution of life traitsto inanimate objects was significantly lessfrequent among the older normal than theyounger normal group, but neither groupdiffered significantly from the retarded.

Artificialism. In Piaget's (1929) book,The Child's Conception of the World, heincluded among the child's "prelogical" be-liefs not only animism but artificialism—thetendency to explain natural phenomena asresulting from deliberate activity of man oranthropomorphic beings. The study byGranich (1940), described previously, in-cluded a number of questions about the or-igin of natural phenoma (e.g., flower seeds).Granich found that responses indicativeof artificialism were virtually nonexistentamong the 13- and 14-year-old nonretardedboys in his sample. Such responses wereslightly less frequent among the sample ofretarded boys than among their normal MApeers. The difference between thes latter twogroups was not tested for significance, butit is very small and would almost certainlyhave been nonsignificant.

Realism and the dream concept. A thirdprelogical belief attributed by Piaget (1929)to the young child is realism. One reflectionof this belief is the inability to distinguishbetween external objects and one's thoughtsand dreams. Granich's (1940) interviews in-cluded questions designed to tap realism; forexample, one asked whether dreams actuallyappear during the night on the ceiling of thebedroom. Granich found that 13- and 14-year-old boys showed virtually no realism.The retarded group showed slightly fewer

such responses than their nonretarded MApeers; this difference would almost cer-tainly have been nonsignificant had Granichtested it.

DeVries's (1970, 1973a, 1973b) investi-gation also included an assessment of chil-dren's dream concepts. In this portion of herstudy, average-IQ children differed signifi-cantly from their CA-peers of high IQ, butthe MA-matched low- and average-IQ chil-dren did not differ significantly.

Magic. In a related feature of DeVries'sstudy, children's notions with regard tomagic (see Piaget, 1930) were assessed.Here again, the CA-matched average- andhigh-IQ children did differ significantly,but the MA-matched average- and low-IQsubjects did not.

Causality. In Judgment and Reasoningin the Child, Piaget (1928) discussed howthe child's developing ability to understandcausal relations is reflected in a capacity touse the word because correctly. In Prothro's(1943) comparative study, the subjects wereall asked to complete five sentence begin-nings that ended in the word because, (e.g.,"The boy hurt his leg because . . ."). Re-sponses were scored for whether they ac-tually described a causal relation. Prothro'sretarded sample was significantly less adeptat this task than his MA-matched group ofnonretarded children.

Quantitative Concepts (Other ThanConservation)

Classification and class inclusion. Group-ing objects into classes is one of the simplestquantitative behaviors. Apparently the ear-liest comparative study of classification thatbears on the similar structure hypothesis wasthe Prothro (1943) investigation. Prothroused three tasks that called for the child tosort objects into groups of things "that arejust alike" or "that go together." The tasksinvolved sortings beads, pencils, and paste-board shapes. Classification on the basis ofcolor and form were both possible in eachtask. With scores summed across the threetasks, Prothro's institutionalized retardedsubjects were significantly inferior to non-retarded children of similar MA. Since this ,task involves abstraction, there is some ques-


tion as to whether Prothro's findings mayhave been influenced by the possible inclu-sion of organic cases among his retarded sub-jects.

At first glance, the Taylor and Achenbach(1975) study appears to provide a useful con-trast to the Prothro experiment, since Taylorand Achenbach excluded all cases with signsof organicity, and their procedure includeda classification task quite similar to thoseused by Prothro. Their finding of no-differ-ence between MA-matehed retarded andnonretarded subjects must be regarded asnot very meaningful, however; the fact thatall of their subjects succeeded at the taskindicates that it was too simple to revealgroup differences.

Finally, the DeVries( 1970, 1973a, 1973b)investigation included a classification ("sort-ing") and a class inclusion problem, thoughneither problem is described in detail. Onneither problem did DeVries's retarded andaverage-IQ children of similar MA differsignificantly. This may reflect a floor effecton the sorting task, which did not show aCA effect either. But the no-difference find-ing with the class inclusion task appears tobe more meaningful: this task did show ahighly significant effect of CA (p < .0001).

Additive composition. A concept closelylinked to classification is additive composi-tion. Piaget (1952) described a three-stageprocess in the development of the ability toadditively compose classes. Children in thefirst stage cannot compare subclasses withwhole classes but can compare separate sub-classes; during the second stage, childrenidentify subclasses with the whole; finally,at the third stage, children have an abstractconception of subclasses within wholes. Toillustrate, Piaget showed children an arrayof brown beads and white beads, all of whichwere wooden. When asked whether therewere more brown beads or wooden beads,Stage 1 children answered by comparing thebrown subclass to the white subclass. Stage2 children identified the brown subclass withthe whole class "wooden," and answered"the same." Only in Stage 3 did childrenrecognize that there were more beads in thewhole class "wooden" than in the subclass"brown." McManis (1968) presented a sim-ilar task to 90 nonretarded elementary school

children and 90 institutionalized retardedchildren, matched at six MA levels (5, 6, 7,8, 9, and 10-11 years). The retarded subjectswere evidently not screened for organic im-pairment. Group means were not reported,but the nonretarded group ranged from 85to 115 in IQ, from 5.3 to 11.9 years in MA,and from 5.3 to 10.8 years in CA. Retardedsubjects ranged from 47 to 73 in IQ, from5.0 to 11.8 years in MA, and from 7.7 to21.1 years in CA. McManis (1968) founda significant increase in the proportion ofStage 3 responses and a decrease in propor-tion of Stage 1 responses across MA levelsfor both retarded and nonretarded children.(No Stage 2 responses were obtained.) Ateach of the five MA levels, the retarded sub-jects showed more Stage 3 responses thandid the MA-matched nonretarded group.This difference favoring the nonretarded wasnot statistically significant, however.

Concept of probability. Children's con-cepts of probability were studied by Steven-son, Hale, Klein, and Miller (1968). The in-vestigators employed filmed instructions anda test booklet with intact school classes, eachseen as a group. To introduce the probabilityquestions, the film noted that in a group ofboys pictured in the test booklet, some haddark hair and some light, whereas some woreglasses and some did not. Subjects wereasked to infer which of two characteristicswas more probable, given certain informa-tion about base rates in the target population(e.g., "I am thinking of a certain boy in thegroup who wears glasses. What color do youthink his hair is, light or dark?" [Stevensonet al., 1968, p. 19]). In a second task, theyoungsters were told how may pegs of var-ious colors were inside a certain box and thenwere asked to predict the outcome of blinddrawings from the box. The original sampleincluded high-IQ children from a universitylaboratory school, average-IQ children froma public junior high school, and low-IQ chil-dren from public school classes for the ed-ucable retarded. Mean IQs of the threegroups were approximately 120, 100, and 70,respectively. Retarded subjects were re-ported to be "primarily" of the familial type,with cases of "gross motor or sensory im-pairment" excluded. Stevenson et al. (1968)did not attempt to achieve MA matching but


they did compare the performance of re-tarded subjects with a fourth-grade, nonre-tarded group whose MA was said to be"comparable with that of the retarded 5V'(p. 29). Mean MAs of the two groups werenot reported, nor was any test reported toindicate whether the two groups differed sig-nificantly in MA. On both tests of the prob-ability concept, the retarded subjects weresignificantly inferior to the nonretardedgroup. The study included other measures,most of which are not relevant to this review.But one, a conservation task, is discussedbelow.

Comparison of gross, intensive, and ex-tensive quantities. In his analyses of thedevelopment of quantitive thinking, Piaget(1952) described three kinds of perceivedquantity by which things are compared with-out actual measurement. The first to developis gross quantity; it is in evidence when chil-dren consider only uncoordinated perceptualrelations of gross quantitative equality ordifference. In the second, intensive quantity,children are said to compare quantities byseriating them along more than one dimen-sion (e.g., width and height) concurrently.The third and most advanced was labeledextensive quantity; using this type of per-ceived quantity, children can overrule ap-parent differences between two equal quan-tities by imposing equal units of measurementon them. McManis (1969a) used this typol-ogy to score children on their use of com-parison processes with sticks, colored water,and beads. His retarded subjects were insti-tutionalized; more than one third were or-ganically impaired. Their mean IQ was 58;they averaged 7.9 years in MA and 14.4years in CA. The nonretarded subjects werefrom public schools. They averaged 100 inIQ, 8.5 years in MA, and 8.0 in CA. Acrossall three types of comparison combined, non-retarded subjects significantly outperformedtheir retarded MA peers. Separate analysesfor each task revealed that the two groupsperformed similarly in comparing grossquantities but that the retarded were signif-icantly inferior in comparing intensive andextensive quantities. To understand this pat-tern of findings, McManis (1969a) exam-ined the cognitive processes required forcomparison of intensive and extensive quan-

tities. This led him to hypothesize that re-tarded persons may show deficits in seriationrelative to their nonretarded MA peers.

Seriation. In a subsequent study Mc-Manis (1970) tested this hypothesis. His re-tarded sample again consisted of institution-alized individuals who apparently had notbeen screened for organic impairment. TheirIQs ranged from 46 to 72, their MAs from5.4 to 11.9 years, and their CAs from 8.8to 22.3 years. (Means were not reported.)The nonretarded comparison group weredrawn from public elementary schools. Theyranged from 85 to 116 in IQ, from 5.3 to11.9 years in MA, and from 6.0 to 11.8 inCA. To equate the two groups in develop-mental level, MA-matched retarded-non-retarded subject pairs were formed. Subjectswere asked to arrange five sticks in order ofincreasing length, then to insert four addi-tional sticks of differing lengths into the ar-rangement in their appropriate position. Aspredicted earlier by McManis (1969a;1969b), significantly more nonretarded thanretarded subjects succeeded at the task.When the analysis was broken into five dif-ferent MA levels, McManis found that theretarded-nonretarded difference was onlysignificant for subjects in the 5-7 year MArange. This study also involved measures oftransitivity and conservation, which are dis-cussed later.

Unlike the McManis (1970) study, theinvestigation by Taylor and Achenbach(1975) systematically ruled out retardedchildren suffering from organic impairment;the Taylor and Achenbach (1975) study alsoincluded a measure of logical seriation. Itinvolved the following problem: "There arethree brothers—Bob, Joe, and Dick. Bob istaller than Joe, and Joe is shorter than Dick.Who is the shortest brother?" (p. 47). Thisproblem requires the ability to coordinatequantitative relations into a triadic series.Taylor and Achenbach found no significantdifference between their MA-matched re-tarded and nonretarded subjects, but theyalso found no significant effect of MA(p — .13). This raises a question as towhether the task employed may have simplybeen insensitive to group differences withinthe 5- to 10-year-old MA range employed.

Transitivity. A transitive inference in-


volves reasoning like the following: If A >B, and B > C, then A > C. Most Piagetiansbelieve such reasoning is closely tied to se-riation. We have found five studies of tran-sitivity that have some bearing on the similarstructure hypothesis. One is the DeVries(1970, 1973a, 1973b) investigation. Her bat-tery of tasks included a "length transitivity"task. (No other task details were reported.)The task did differentiate significantly be-tween CA-matched high- and average-IQsubjects, but low- and average-IQ subjectsof similar MA did not differ significantly.

Gruen (1973) assessed transitivity oflength among MA-matched familial re-tarded and nonretarded children. Both groupswere from public schools. The retardedgroup averaged 64 in IQ, 6.5 years in MA,and 9.6 years in CA. Nonretarded childrenaveraged 102 in IQ, 6.6 years in MA, and6.5 years in CA. Three sticks of differinglength were used, with Stick A longer thanB, and B longer than C. Subjects made com-parisons between A and B and between Band C; then they were asked to infer therelative sizes of A and C. To prevent solu-tions based on perceptual acuity alone, sticksA and C were placed inside the arms of aMueller-Lyer illusion. This made the shorterstick appear to be the longer one. Half ofthe subjects in each group were given a mem-ory aid to remind them of the results of theA versus B and B versus C comparisons. Inthe memory-aid condition, retarded subjectsmade significantly fewer correct transitivejudgments than the nonretarded subjects; inthe no-memory-aid condition, the two groupsdid not differ significantly. For nonretardedsubjects performance was significantly bet-ter in the memory-aid than in the no-mem-ory-aid condition; among retarded subjectsthe memory aid made no significant differ-ence. Gruen's interpretation was that thenonretarded children showed superior logi-cal inference. That is, when all the infor-mation ordinarily stored in memory was con-cretely available, retarded children did notuse this information in the same logical, se-quential manner as their nonretarded MA-peers.

The Mueller-Lyer illusion procedure wasalso used by McManis (1970) to assess tran-sitivity of length among the nonretarded and

unscreened retarded samples described ear-lier. Recall that MAs in both groups rangedfrom about 5 years to about 12 years, andthe IQ ranges were 85-116 and 46-72, re-spectively. Although no concrete memoryaid was employed, subjects were required torecall the results of the original A versus Band B versus C comparisons before beingasked to make the A versus C judgment.Nonretarded subjects made significantlymore correct transitive inferences than didthe MA-matched retarded subjects.

In another comparative investigation,McManis (1969c) assessed transitivity oflength and weight. In the length assessmentthe Mueller—Lyer procedure was used. Tran-sitivity of weight was assessed via a concep-tually similar procedure in which the weightof clay balls of different sizes was varied byembedding differing amounts of metal withinthem. The sample appears to be the sameas in the McManis (1969a) study. There theretarded subjects were all institutionalized,and more than one third were organicallyimpaired. They averaged 58 in IQ and 7.9years in MA. The public school nonretardedcontrast group had a mean IQ of 100 anda mean MA of 8.5 years. McManis (1969c)did not report significance tests for the re-tarded versus MA-matched nonretarded con-trast in his sample; however, our own chi-square analyses of his data indicated thatthe retarded group made significantly fewercorrect transitive inferences than the non-retarded, on both the length and the weight(both ps < .01).

Although the magnitude and consistencyof Gruen's (1973) and McManis's (1969c)findings on transitivity are striking, theirmethodology has come under fire. Lutkusand Trabasso (1974) have expressed two sig-nificant concerns, particularly with regardto the Mueller-Lyer transitivity procedure.(See also Bryant & Trabasso, 1971.) First,they note that the Mueller-Lyer illusion pro-duces a gratuitous perceptual conflict thatmay distract subjects who could otherwisereason transitively. Second, they argue thatthe standard procedure of requiring correctverbal explanations for transitive judgmentscan lead to misclassification; subjects whocan reason transitively may often be unableto give lucid explanations of their reasoning


processes. Requiring explanations, they ar-gued, can bias the results in favor of subjectswith greater verbal facility. Lutkus and Tra-basso used a procedure designed to eliminatethese problems and to insure that childrenhad learned all necessary preinference in-formation (e.g., that A > B). They testedinstitutionalized retarded adolescents at twoMA levels; both groups were apparently un-screened for organic impairment. Overall,the retarded subjects averaged 48 in IQ, 6.0years in MA, and 15.6 years in CA. Theseyoungsters were compared with nonretardedchildren from an earlier study (Bryant &Trabasso, 1971) who were thought to be"of comparable MA" (p. 602). Results dif-fered substantially from those of McManis(1969c, 1970) and Gruen (1973). Comparedwith the nonretarded, retarded subjects didrequire more trials to learn and rememberthe necessary preinference information. Butonce retention of this information was in-sured, the transitivity performance of theretarded and nonretarded groups did notdiffer significantly at either MA level.

Number concept development. We havefound one comparative study that involvedmultiple number concepts (Cohn-Jones &Seim, 1978). The investigators apparentlydid not screen for organic impairment. Re-tarded and nonretarded samples were bothdrawn from public schools. The retardedchildren averaged 72.7 in IQ, 6.8 years inMA, and 10,3 years in CA. In the nonre-tarded group the mean IQ was 102.2; themean MA, 6.8 years; and the mean CA, 6.5years. All subjects were given the DodwellNumber Concept Test (Dodwell, 1960). Itis designed to assess abilities in several areas,including conservation of discontinuousquantity, seriation, and serial correspon-dence. The component measures are com-bined to yield an overall number conceptscore and overall explanation scores. Resultsrevealed no significant difference betweenthe two groups on either score.

Studies of Conservation With EtiologyUncontrolled

We now review research in the most fre-quently studied Piagetian conceptual do-main, conservation. Because the studies are

so numerous, it is useful to divide them intosubgroups; because of the concerns we ex-pressed earlier about the importance of con-trolling etiology we have grouped the studiesaccording to whether they exercised suchcontrol. Cultural-familial retardation ac-counts for about 75% of all cases of mentalretardation (Zigler, 1967); most of the re-maining 25% have some form of organic im-pairment. Consequently, experimenters whodo not systematically exclude organicallyimpaired individuals can be expected to havea substantial number of them in any sizeableretarded sample. In dividing studies into twogroups, we assigned to the "etiology uncon-trolled" group those which reported no effortto eliminate organic cases. Also included inthis group are studies where only partial con-trol seems to have been exerted; for example,statements indicating that cases of "grosssensory or motor defect" were eliminated orthat the retarded sample was "primarily"familial seem to imply the inclusion of atleast some organically impaired children.

First we examine research in the etiologyuncontrolled category. The study by DeVries(1970, 1973a, 1973b) included measures ofconservation of mass (two tasks), number,length, and continuous quantity (liquid). Inaddition, constancy of generic identity andsex identity were assessed. DeVries's high-IQ group was significantly superior to herCA-matched average-IQ group on all thesemeasures except one conservation of masstask. Yet the MA-matched average- andlow-IQ groups differed significantly on onlyone of the seven measures. On that measure,sex identity, the retarded group performedsignificantly better than the nonretardedgroup.

Kahn's (1976) investigation of moral andcognitive development included measures ofconservation of mass and weight. Scores onthe two measures were combined to form acomposite "cognitive functioning" measure.On this measure, Kahn's nonretarded sam-ple outperformed an MA-matched retardedgroup, most of whom were organically im-paired.

Three studies by McManis included con-servation assessments among MA-matchedretarded and nonretarded groups. In one,McManis (1969e) sampled institutionalized


retarded subjects ranging from 47 to 73 inIQ, from 5.0 to 11.8 years in MA, and from7.7 to 21.1 years in CA (no means reported).The nonretarded public school sample rangedfrom 85 to 115 in IQ, 5.3 to 11.9 years inMA, and 5.3 to 10.9 years in CA. Conser-vation of mass, weight, and volume was as-sessed. Children were asked to make (a) con-servation predictions (e.g., "will there be asmuch clay in the hotdog as in the ball?",(b) conservation judgments (e.g., "do theyboth have the same amount of clay?"), and(c) explanations of each judgment. Thesethree responses were combined to obtain ascore for each type of conservation. Weightconservation scores showed significant ef-fects of MA but did not differentiate MA-matched retarded and nonretarded subjects.There were significant retarded-nonretardedgroup differences on mass and volume con-servation scores, with the nonretarded groupsuperior on mass and the retarded group su-perior on volume.

McManis (1969c) assessed conservationof weight (clay) and of length (sticks) amongMA-matched retarded and nonretardedgroups, using a procedure like that inMcManis (1969d). He did not report sig-nificance tests, but inspection of his data re-veals strong MA effects on both measures.However, our own chi-square analysis ofMcManis's (1969c) frequency data showsthat the negligible retarded-nonretardeddifferences did not approach significance foreither type of conservation.

In McManis (1970), MA-matched re-tarded and nonretarded subjects were givena two-part length conservation task. Part 1was a relatively standard length conservationprocedure. Part 2, patterned after Smed-slund (1961), was designed to assess (a) sub-jects' susceptibility to inducement and ex-tinction of conservation responses and (b)their ability to retain the induced conser-vation for a 1-week period. Although theMA effect was not tested, the data reportedwere strongly MA related. However, therewas no reliable difference between retardedand nonretarded subjects.

The study by Stevenson et al. (1968), inwhich children were group-administered avariety of tasks by means of films and test

booklets, included groups averaging about70 and 100 in IQ, The two groups were dis-cribed as comparable in MA, but they werenot precisely MA matched. The procedureincluded a conservation of volume task; sub-jects were asked to judge the relative amountsof liquid displaced by two lumps of clay withequal mass but dissimilar shapes. Retardedsubjects were significantly inferior to thenonretarded.

A conservation study by Goodnow andBethon (1966) included a group of "dull"and "average" public school boys, matchedperson-for-person on MA. The dull grouphad a median IQ of 81, a median MA of 8.6years, and a median CA of 11.0 years. Forthe average group the IQ median was 110;the MA median, 8.7 years; and the CA me-dian, 7.8 years. The procedure included mea-sures of conservation of mass, weight, andarea. The mass and weight assessments werefairly standard procedures involving lumpsof clay. The area measure was more unusual:Subjects compared two fields, one withhouses clustered at the side, the other withhouses scattered about. In addition, twomeasures of volume conservation called forchildren to make judgments about clay ballsdisplacing water. Summing across all con-servation measures except mass, the inves-tigators found significant differences be-tween two groups not relevant to our review:high-IQ and average-IQ 11-year-olds. How-ever, summing over all conservation mea-sures, they found no significant differencebetween the MA-matched groups of averageand low IQ. When the measures were ana-lyzed separately, however, three group dif-ferences were significant. On the weightmeasure and the two volume measures, thenonretarded group was significantly supe-rior.

Brown (1973) studied conservation ofnumber and continuous quantity amongMA-matched children at three IQ levels.Her bright group averaged 149 in IQ, 6.5years in MA, and 4.5 years in CA. Her av-erage group had a mean IQ of 102, a meanMA of 6.3 years, and a mean CA of 6.6years; Her retarded group averaged 74 inIQ, 6.6 years in MA, and 8.4 years inCA. Subjects were drawn from public and


university-supported nursery and primaryschools in England. On both types of con-servation Brown found the bright group sig-nificantly inferior to both the average andretarded groups. The retarded and averagegroups, however, did not differ significantlyon either measure. Brown (1973) interpretedthis pattern as evidence that "the youngchild with a high IQ is held back by sheerlack of experience that the additional CAprovides" (1973, p. 378). This interpretationis reminiscent of the unconventional differ-ence position advanced by Kohlberg (1968).

Brekke and Williams (1974) administeredthree conservation of weight measures toretarded and nonretarded children. The non-retarded were from public schools, and theretarded were a mix of public school andinstitutionalized individuals. Group meanswere not reported, but IQs ranged from 24to 86 in the retarded group and from 81 to160 in the nonretarded group; thus, thegroups overlapped in IQ. The retarded groupranged from 3.1 to 12.2 years in MA andfrom 6.5 to 27.4 years in CA. The nonre-tarded group ranged from 6.4 to 8.6 yearsin MA and from 5.5 to 14.9 years in CA.In the 13-step testing sequence, equal-sizedballs of clay were taken through varioustransformations in shape. Subjects wereasked to judge the weight equivalence of theclay balls across the successive transforma-tions; they were also asked to explain theirjudgments (see Furth, 1964, for proceduraldetails). Responses to all conservation ques-tions were combined into one overall con-servation of weight score. Scores were higherfor nonretarded subjects than for either theinstitutional or the public school retardedgroup. This difference was not theoreticallymeaningful, however; the groups had notbeen matched for MA, and the nonretardedgroup had much higher MAs than both re-tarded groups. To correct for this problem,Brekke and Williams carried out an analysisof covariance (ANCOVA), with MA as co-variate. The group differences remained sig-nificant. It should be noted, however, thatANCOVAS may pose both conceptual and sta-tistical problems when Piagetian measuresare involved (see Kappauf, 1976; Weisz,1976). One problem is that the assumption

of a linear relationship between variate andcovariate (Winer, 1971) may be violated ifthe dependent variable is actually a stagelikephenomenon.

Cardozo and Allen (1975) assessed con-servation with the Goldschmid-Bentler(1968) Concept Assessment ConservationTest. They compared institutionalized re-tarded children with nonretarded public andprivate school children. The retarded groupaveraged 63 in IQ, 7.3 years in MA, and12.2 years in CA. The nonretarded groupaveraged 107 in IQ, 7.5 years in MA, and6.9 years in CA. The test used was designedto measure conservation concepts in six dif-ferent areas: mass, number, weight, contin-uous and discontinuous quantity, and two-dimensional space. Scores were summedacross all six types of conservation. On thiscomposite measure retarded subjects weresignificantly inferior to the nonretarded con-trast group.

Achenbach (1973) assessed one of the ear-liest manifestations of conservation: chil-dren's identity concepts. Identity of color,number, length, and continuous quantitywere inferred from children's surprise re-actions when these properties were unex-pectedly changed. In the test of color iden-tity, for example, five white marbles wereinserted into a hole in a box; the box wasthen turned around, and five black marblesfell out. Children who showed surprise hadpresumably expected the marbles to have afixed identity with respect to color. The sam-ple included nonretarded nursery and ele-mentary school children averaging 116 inIQ, 6,3 years in MA, and 5,5 years in CA.The retarded group included both institu-tional and public school subjects; one fourthsuffered from Down's syndrome, but the re-mainder showed no signs of organic impair-ment. The Down's syndrome youngsters andthe other retarded subjects did not differ infrequency of surprise reactions, so their datawere analyzed together. The combined re-tarded groups averaged 47 in IQ, 6.2 yearsin MA, and 17.3 years in CA. To fashiona carefully controlled group comparison,Achenbach (1973) formed 49 subject pairsin which a retarded and nonretarded childwere matched for MA, sex, race, and order


of presentation of the tasks. There were sig-nificant differences among the four tasks inthe frequency of surprise reactions theyevoked; but only one of the four tasks re-vealed a reliable retarded-nonretarded dif-ference. On the continuous quantity task,retarded subjects showed surprise more fre-quently than the nonretarded (p = .058).2

Studies of Conservation, With EtiologyControlled

We now examine conservation studies us-ing retarded samples that were reportedlyscreened for organic impairment. In onesuch study, Achenbach (1969) assessed con-servation of length, area, and "volume" (ac-tually, continuous and discontinuous quan-tity in Piagetian terms). He used fourdifferent tasks for each property. Each in-volved an optical illusion that created a dis-crepancy between the actual and apparentmagnitude of a stimulus. In one conservationof area task, for example, the child founda cream colored disk that fit a circular recessin a red piece of plexiglass. Painted arrowspointing to the center of the recess made itlook relatively small. Then the child placedthe same disk into a second piece of plex-iglass; the arrows pointing outward made therecess and the disk seem larger. The childwas then asked, (a) which recess lookedlarger, (b) whether the circle would actuallyfit back in the original recess, and (c) whyit would or would not fit. The retarded sub-jects averaged 52 in IQ, 6.1 years in MA,and 15.8 years in CA; some were drawn frompublic schools and others came from stateregional institutions. The nonretarded con-trast group came from public schools and aprivate nursery school; they averaged 128 inIQ, 6.8 years in MA, and 5.3 years in CA.

From these two samples 48 retarded-non-retarded subject pairs were formed; pairmembers were matched for MA, sex, race,and task order. Most of the 12 tasks showedno significant effects of MA or CA. Onlyone, a discontinuous quantity task, revealedsignificant IQ-group differences; on it, nineretarded and one nonretarded subject failedto show a conservation response. Achenbach(1969) also scored subjects' explanations oftheir responses to the 12 items. For 11 of the

12 items, these explanations differentiatedsignificantly between nonretarded subjectsof low CA and MA and nonretarded subjectsof higher CA and MA, but no significantretarded-nonretarded difference was foundon any item.

Keasey and Charles (1967) assessed con-servation of mass. Their institutionalized re-tarded and public school nonretarded sub-jects were matched person-for-person onMA. IQs were not reported. Retarded sub-jects ranged from 5.3 to 10.9 years in MAand from 13.0 to 28.7 years in CA. In thenonretarded group the MA range was 5.3-10.9 years, and the CA range was 5.6-10.9years. Subjects judged whether two identicalplasticene balls had the same amount ofmass even after transformations in shape.These judgments and their explanationswere scored. Both judgments and explana-tions were significantly related to MA, butneither differentiated significantly betweenretarded and nonretarded subjects.

Gruen and Vore (1972) used conventionalmeasures to study conservation of number,weight, and continuous quantity. Their sam-ple included retarded children averaging 70in IQ, 6.9 years in MA, and 10.2 years inCA. An MA-matched nonretarded groupaveraged 108 in IQ, 7.2 years in MA, and6.7 years in CA. All subjects came frompublic schools. The retarded and nonre-tarded groups were divided into three

2 The Achenbach (1973) study provides two examplesof means by which problems resulting from motivationalcharacteristics of retarded children may be circum-vented. The use of brief, minimally verbal techniquesfor assessing conservation concepts could reasonably beexpected to minimize the influence of group differencesin motivation to persist (and group differences in verbalfacility as well). In addition, Achenbach assessed sub-jects' IQ and MA scores by means of an "optimizing"procedure. The Stanford-Binet Intelligence Test wasadministered in a manner calculated to encourage mo-tivation to succeed, with each series of failures followedby simpler problems aimed at producing a success ex-perience, before the examiner proceeded to the nextitems, Evidence (Zigler, Abelson, & Seitz, 1973; Zigler& Butterfield, 1968) suggests that such a procedure maycorrect mean underestimates of IQ as high as 10 pointsin children of lower sociocconomic status (SES). Sincethe organically intact retarded are almost always fromthe lower SES, the maximizing procedure may contrib-ute importantly to precision in the formation of exper-imental groups in any study having a bearing on thesimilar structure hypothesis.


subgroups, at MAs of 5, 7, and 9 years. Theinvestigators used both conservation-judg-ment-alone and judgment-plus-explanationas dependent variables. These two measures,assessed across the three types of conserva-tion, yielded numerous significant differ-ences between MA levels and between CA-matched retarded and nonretarded subjects.None of the dependent variables differen-tiated significantly between retarded sub-jects and their MA-peers at any of the threeMA levels.

Taylor and Achenbach's (1975) study ofmoral and cognitive development includedtwo conservation tasks. One was a surprisetask (from Achenbach, 1973) designed togauge subjects' color identity concepts. Thesecond was an illusion task (from Achen-bach, 1969) designed to gauge conservationof continuous quantity. The continuousquantity task differentiated significantly be-tween MA levels, but the color identity taskdid not. The MA-matched retarded and non-retarded groups did not differ significantlyon either measure.

In Kahn's (1976) sequel to the Taylor-Achenbach study, more traditional measureswere used to assess conservation of mass andweight. Scores on both measures were com-bined to form a cognitive score. Kahn's sam-ple included three MA-matched groups:nonretarded, mildly retarded, and moder-ately retarded; mean IQs were 101, 66, and45, respectively. The mildly retarded groupwas screened for organic impairment; themoderately retarded group included organ-ically impaired individuals. Scores of thenonretarded group were significantly supe-rior to both the moderately and mildly re-tarded group.3

Status of Evidence on the SimilarStructure Hypothesis

We have now surveyed Piagetian evidencebearing on the similar structure hypothesis.We examined 30 published experiments in-volving a total of 104 comparisons betweenretarded and nonretarded groups with MAcontrolled. Some 18 different conceptualdomains were addressed; these provide goodcoverage of the preoperational and concreteoperational cognitive processes described by

Piaget. An uncritical overview of the box-score totals reveals the following: of the 104group comparisons, only 4 (4%) supportedKohlberg's (1968) unconventional differ-ence position by revealing significantly su-perior performance among the retardedgroup; some 25 comparisons (24%) sup-ported the conventional difference position(e.g., Milgram, 1973) by revealing signifi-cantly superior performance among the non-retarded group; and 75 comparisons (72%)supported the similar structure hypothesisof the developmental position (Zigler, 1969)by revealing no reliable difference betweenretarded and nonretarded groups.

Support for the developmental positionseems stronger when one critically examinesthe methodology of these studies. Earlier westressed the importance of sampling homo-geneous groups of retarded subjects by ex-cluding the organically impaired. Of the 104comparisons, we find 33 that satisfied thiscondition. Of these 33, only three compari-sons (10%) are inconsistent with the similarstructure hypothesis. Therefore the simplebox-score evidence seems to support the de-velopmental position and its similar struc-ture hypothesis.

Similar Structure Prediction, Power, andthe Null Hypothesis

This body of evidence, though, deservesmore intense analysis than these simple tal-lies provide. Several trends may have biasedthe results. On the one hand, more of thecomparisons would probably have been sig-nificant if statistical power had been greater.If the studies reviewed had used larger sam-ples, or, perhaps, more precise measurement,some of the nonsignificant group differencesmight have attained significance. The issueof what is "enough" power is far from re-solved; it can be argued from several phil-osophical and empirical positions (Hays,1973). Evidence does indicate that psychol-ogists tend to accept relatively low power intheir experimental designs (Greenwald,1975). This tendency may have obscured

3 This information was provided by James V. Kahnin correspondence with Herman Spitz and with us. Weare grateful for this supplement to data reported in Kahn(1976).


trends in the research reviewed, resulting ina spuriously high number of comparisonssupporting the similar structure hypothesis.

On the other hand, evidence may havebeen influenced in the opposite direction bythe very nature of the similar structure hy-pothesis. It is, as we noted earlier, essentiallya null hypothesis; that is, it predicts no dif-ference between target groups. Within ourdiscipline, there is a tradition of prejudiceagainst the null hypothesis, a prejudice thatmay work against the similar structure hy-pothesis in reviews like this one. We aretaught early in our statistics courses that thenull hypothesis may be disconfirmed butnever confirmed. We learn to regard findingsof no difference as failure, and we discoverthat such findings often remain unpublished.For some time, quantitative social scientistshave engaged in a lively debate about thenull hypothesis. Statisticians of the Fisherand Yule school regard acceptance of thenull hypothesis as inappropriate; statisti-cians of the Neyrnan and Pearson schoolargue that if the null hypothesis can betested, it also can be accepted as valid (fordiscussion of the issues involved see, e.g.,Binder, 1963, and Grant, 1962).

Greenwald (1975) has argued that the tra-dition of discrimination against the null mayimpede scientific progress. There may be in-appropriately few publications on problemsfor which the null hypothesis is a reasonablealternative; a substantial proportion of thesewill incorrectly reject the hypothesis of nodifference. According to this reasoning,findings supporting the similar structure hy-pothesis may be underrepresented in pub-lished reports. This may have beenparticularly true in the years prior to a cleararticulation of the developmental positionand its no-difference prediction (Zigler,1967, 1969). In those years there was noclear theoretical explanation for findings ofno difference; this may have undermined thechances that such findings would be pub-lished. It is difficult to gauge or correct forthe existence of biases for or against the sim-ilar structure hypothesis. The existence ofsuch biases, though, does suggest the needfor a more critical analysis than the box-score method.

Combining the Evidence

Fortunately, the need for a more rigorousanalysis can be met. The task is to use theresults of multiple studies to fashion a sta-tistically sound overall test of the similarstructure hypothesis. A variety of methodsexist (see Rosenthal, 1978). The most pre-cise methods involve pooling exact signifi-cance levels and their corresponding Z scores.These methods cannot be used in the absenceof very complete statistical data. Such data(e.g., exact significance levels) were not re-ported in many of the published studies wereviewed; in many cases they were not evenavailable from the authors. Consequently,we selected an alternative approach that al-lowed us to use all of the studies reviewedwhile maintaining statistical rigor.4 We useda counting method that relies on this fact:If a null hypothesis is true, differences be-tween any two groups across independentstudies should approximate a normal distri-bution. This fact means that the proportionof significant differences in either directionfound by chance should equal whatever al-pha level is selected. A chi-square statisticcan thus be used to determine whether thearray of differences approximates a normaldistribution; it should if the null hypothesisis true. The only requirement for this testis knowledge of the number and direction ofsignificant differences.

All 104 comparisons satisfied this require-ment. For our puposes, we chose an alphalevel of .05 for both tails of the distribution.The expectation would thus be for 10% of

4 In an effort to obtain the information needed forpooling of exact significance levels, we mailed requeststo the authors, then used follow-up letters and phonecalls when necessary. We reached nearly all the authors,but many had discarded or misplaced the data weneeded. We also requested information on the directionof all nonsignificant differences between MA-matchedretarded and nonretarded groups. Such informationwould have permitted a four-cell chi-square analysis(i.e., significant and nonsignificant differences favoringretarded and nonretarded groups) that would have al-lowed us to check for trends in directionality of differ-ences. Even this information was unavailable for almost20% of the total comparisons reviewed above. We havesatisfied ourselves that the analysis we have selected isthe most precise one that allows us to include all relevantcomparisons.


the findings to be significant (5% on eachtail); 90% should not be significant. In otherwords, if the similar structure hypothesis istrue, the comparisons should be distributedas follows: 5% should show the retarded sig-nificantly superior to the nonretarded, 90%should show no significant retarded-nonre-tarded difference, and 5% should show theretarded significantly inferior to the nonre-tarded. Chi-square statistics were computedcomparing the actual to the expected pro-portions; the statistics were calculated sep-arately for etiology controlled and un-controlled comparisons.5 The 104 groupcomparisons are shown, by category, in Ta-ble 1. We classified studies as etiology con-trolled if their authors explicitly stated thatorganically impaired persons were excludedfrom the retarded sample. We labeled theremainder, etiology uncontrolled. A tabu-lation of actual and expected frequencies forboth categories is shown in Table 2.

The distribution of the etiology uncon-trolled comparisons was significantly differ-ent from that expected by chance, x2(2) =101.53, /j<.001; a greater number ofcomparisons than expected showed differ-ences favoring nonretarded subjects (seeTable 1). The distribution of the etiologycontrolled comparisons was not significantlydifferent from that expected by chance,X2(2) = 2.76, p > .25. That is, the distribu-tion of comparisons excluding organicallyimpaired retarded persons is not signifi-cantly different from that expected if thesimilar structure hypothesis is true.6

Conclusions, Criticisms, and Thoughts onFuture Research

Our analysis suggests that when investi-gators have systematically excluded organ-ically impaired subjects, their findings havegenerally supported the similar structurehypothesis. When such screening has notbeen employed, however, findings have sup-ported the conventional difference position.The similar structure hypothesis has beenclearly articulated as applicable only to non-organically impaired children (cf. Weisz &Zigler, 1979; Zigler 1967, 1969); thus, theevidence must be regarded as substantially

in support of the hypothesis. An earlier re-view of Piagetian research on the other ma-jor facet of the developmental position (i.e.,the similar sequence hypothesis) revealedthat it too was supported by the findings (seeWeisz & Zigler, 1979). In summary then,a substantial body of Piagetian data supportsboth major tenets of the developmental po-sition.

If in the future the developmental positionis supported by other bodies of research notreviewed here or by Weisz and Zigler (1979),significant implications would follow for theways we construe and diagnose mental re-tardation. One implication would be thatmentally retarded persons who do not sufferfrom specific organic deficits should not beconstrued as fundamentally abnormal. In-stead, they might best be described as man-

5 The expected cell frequencies for these analyses arelower than those frequently recommended as a limit onthe applicability of the chi-square statistic (e.g., Hays,1973). Recent research (Camilli & Hopkins, 1976; Ros-coe & Byars, 1971), however, has shown that the chi-square statistic is very robust in the face of violationsof these limits. Indeed, the research suggests that ex-pected cell frequencies as low as 1 or 2 produce validstatistics and significance levels as long as total samplesize is 20 or more. Our analyses fulfill these conditions.We felt that the analysis was further justified becauserecent research (Cooper & Rosenthal, 1980) suggestedthat literature reviews such as ours, using statisticalprocedures for combining studies, provide more rigorousconclusions than those employing subjective judgments.

6 Spitz (in press), in a critique of the developmentalposition, has argued that four of the studies included inthis table are questionable as regards the diagnostic cat-egories in which we have placed them (i.e., whether ornot they include organically impaired retarded persons).Although we agree with Spitz that the sample descrip-tions given in some of the studies do not permit completecertainty, we have adopted what seems to us a reason-able policy of classifying studies as having excluded or-ganically impaired persons if their authors stated ex-plicitly that this procedure was followed. Nonetheless,as a check on whether the experiments questioned bySpitz might have distorted the pattern of results, werecalculated the chi-square analyses depicted in Table2, excluding the four studies regarded by Spitz as doubt-ful (Achenbach, 1969; Cardozo & Allen, 1975; Good-now & Bethon, 1966; Stevenson et al., 1968). The pat-tern of results was similar to that shown in Table 2. Thetable for comparisons in which organic subjects werenot systematically excluded was again significant, x2(2) =57.35, p < .001; the table for comparisons in which or-ganic subjects were excluded was again nonsignificant,X 2 (2 )= 1 .91,p> .25.


Table 1Directional Comparisons of MA-Matched Retarded and Nonretarded Persons onVarious Piagetian Tasks

Direction of difference

Study MR > NMR MR s; NMR MR < NMR

Etiology controlled

Achenbach (1969)

Gruen (1973)

Gruen & Vore (1972)

Kahn (1976)

Keasey & Charles (1967)

Taylor & Achenbach (1975)

Conservation of length X4, area X 4, continuousquantity X 2,discontinuous quantity

Transitivity (no memoryaid)

Conservation of number X3, weight X 3,continuous quantity X 3

Moral judgement

Conservation of mass

Conservation of continuousquantity

Color identity, relativethinking, role taking X2, seriation, moraljudgment

Conservation ofdiscontinuous quantity3

Transitivity (memory aid)

Conservation of mass andweight (combined)

Etiology uncontrolled

Achenbach (1973)

Brekke & Williams (1974)

Brown (1973)

Cardozo & Allen (1975)

Cohn-Jones & Seim (1978)

DeVries (1970, 1973a,1973b)

Conservation ofcontinuousquantity

Sex identity

Goodnow & Bethon (1966)

Granich (1940)'

Conservation of number,color, length

Conservation of number,continuous quantity

Conservation of mass,weight, two-dimensionalspace, continuous anddiscontinuous quantity(combined)

Multiple number concepts,number conceptjudgments

Conservation of mass X 2,number, length,continuous quantity

Generic identity,classification, classinclusion, transitivity,relative thinking X 2,dream concepts, magic

Conservation of mass, area

Animism, artificialism,realism

Conservation of weight

Role taking

Conservation of weight,volume X 2


Table 1 (Continued)


Study MR > NMR MR * NMR MR < NMR

Kahn (1976)

Lane & Kinder (1939)°

Lutkus & Trabasso (1974)

McManis(1968)

McManis (1969a)

McManis (1969c)

McManis (1969e)

McManis (19690"

McManis (1970)

Prothro (1943)

Rubin & Orr (1974)

Russell, Dennis, & Ash(1940)

Smeets (1973, 1974)

Smith (1977a)

Smith (1977b)

Stevenson, Hale, Klein, &Miller (1968)"

Moral judgmentConservation of mass and

weight (combined)

Relative thinking X 2

Transitivity X 2

Additive composition

Gross quantity

Conservation of weight,length

Conservation of Conservation of weightvolume

Conservation of length

Animism

Intensive and extensivequanti ty

Transitivity X 2

Conservation of mass

Relative thinking X 2

Transitivity, seriation

Relative thinking,classification, causality

Animism

Perspective taking X 2

Life attributions X 2, lifetrait attributions X 2

Logical contradiction

Perceptual decentering

Probability X 2Conservation of volume

Note. MA = mental age; MR = mentally retarded subjects; NMR = nonretarded subjects; the symbol £ indicatesthat there was no reliable group difference." This result was only true for initial conservation response. Conservation explanation scores on this task did notdifferentiate significantly between retarded and nonretarded groups. b The directional designation of each com-parison from this study in the table relies on the judgments of the respective authors because no significance testswere reported.c Lane and Kinder actually compared two groups of retarded persons, matched for MA but differingsignificantly in IQ. d In this study, MA matching is only approximate and MA differences between retarded andnonretarded groups were not actually tested for significance.

ifesting a developmental lag. According tothis view, they pass through a normal se-quence of cognitive developmental stages.Their rate of progress through that sequenceis relatively slow, and their upper limit ofcognitive development is relatively low.However, when a nonorganically impairedretarded person and a nonretarded personare at the same level of development, theirreasoning derives from similar underlyingcognitive structure.

If this perspective is valid, then the di-agnosis of mental retardation might best beconditioned on the magnitude of develop-mental lag. Prognostic expectations woulddepend on rate of cognitive development.According to this reasoning, the psychomet-ric MA may be used to make rough esti-mates of relative cognitive deficiency at agiven point in time. The psychometric IQhas been described elsewhere as an index oflifetime learning rate (Harter, 1965; Weir,


Table 2Directional Comparisons of MA-Matched Retarded and Nonretarded Persons SummedAcross Various Piagetian Tasks

Group


MR > NMR MR s NMR MR < NMR

Etiology controlled"Actual totalExpected total

Etiology uncontrolled15

Actual totalExpected total

01,65

43.55

3029.7

4563.9

31.65

223.55

Note. MA = mental age; MR = mentally retarded subjects; NMR = nonretarded subjects; the symbol = indicatesthat there was no reliable group difference.V (2) = 2.76, p> .25. V (2) = 101.53, p < .001.

1967); it may provide a useful approxima-tion of the relative pace of development.

This view of mental retardation might alsocarry implications for mainstreaming pro-grams. Such programs often yield classesthat are relatively homogeneous with respectto physical development but heterogeneousin cognitive maturity. Such an arrangementmay pose problems for both teachers andpupils (Zigler & Meunchow, 1979). Thefindings reviewed here and in Weisz andZigler (1979) raise a question as to whetherMA should be a basis for mainstreaming insome cases. One example may be classeswhere a low teacher-pupil ratio precludesone-to-one instruction and where the subjectmatter is cognitively demanding for the ma-jority of the class. In such classes retardedchildren similar to the majority in CA arenot likely to gain much intellectually; theyalso may suffer some loss of self-esteem. Incases like this, mainstreaming retarded chil-dren with nonretarded children of similarMA would offer certain advantages. Therewould be disadvantages as well. For exam-ple, retarded and nonretarded children wouldcontinue to differ in significant ways thatwould need to be taken into account in ed-ucational planning; differences would cer-tainly exist in physical development, andperhaps also in the ability to utilize cognitivestructures in learning situations. However,mainstreaming based on MA would at leastallow teachers to employ similar ability ex-pectations for their retarded and their non-retarded students7; it might also prevent re-

tarded children from being repeatedly placedat a competitive disadvantage.

All these speculations, however, are basedon the assumption that the developmentalposition is valid. Such an assumption is pre-mature. Research bearing on the develop-mental position needs to be improved in anumber of ways, Investigators have paid lit-tle attention to personality factors that canundermine performance in retarded children(see e.g., Weisz, in press-b; Zigler, 1971).Few of the studies reviewed here includedany explicit effort to maximize motivationor self-confidence prior to the Piagetiantasks. Moreover, a substantial number of theretarded subjects resided in institutions; anumber of these were adults, some middle-aged. Research has demonstrated significantdepressing effects of institutionalization onlearning activity (see Denny, 1964; Harter,1967), quality of language behavior (seeLyle, 1959), and level of abstraction in vo-cabulary (see Badt, 1958). In addition tothese environmental effects, one can imagine

7 Of course this position assumes that teachers wouldemploy similar ability expectations. Recent evidencesuggests that this may not be the case. Adults attributethe failure of a mentally retarded child to low ability,but they attribute the failure of an MA-matched non-retarded child to low effort (Weisz, in press-a). Thispattern may represent an attributional bias; adults mayovergeneralize their perception of the retarded child as"low in ability." Or it may be that most adults hold animplicit difference position like that described in thisarticle. In either case, adults, particularly potential par-ents and teachers of the retarded, may benefit from ev-idence such as that presented in the present review.


motivational problems arising in the re-tarded adults who were asked to work withmodeling clay or to engage in other Piage-tian tasks designed for young children.

Much of the evidence can be faulted forimprecise MA matching and for a failure tocontrol for organic impairment. Studies thatevidenced such control yielded a markedlydifferent pattern of findings than studies thatdid not. This difference in pattern might beseen as support for Zigler's (1967) "twogroup" approach. That approach takes theview that organically impaired retarded per-sons are cognitively different from those notsuffering from specific organic deficits. Fol-lowing this line of reasoning one could arguethat much of the support found for the con-ventional difference position is an artifactof organic impairment in the retarded sam-ples. The data reviewed do not clearly war-rant such a conclusion. The mean IQs in thesamples that included organically impairedsubjects were generally lower than the meanIQs of the screened samples. Thus, as Spitz(in press) has noted, IQ level and organicityare generally confounded across the studiesreviewed. To generate more definitive evi-dence on the two-group approach in the fu-ture, investigators will need to accomplisha difficult task: disentangling organicity andIQ level by structuring designs in which thetwo factors are orthogonal.

Perhaps the most important task for thefuture is to survey non-Piagetian research.Evidence from the Genevan tradition has avigorous cognitive developmental emphasis.It is thus the logical starting point for re-views on the status of the developmental ver-sus difference controversy. But the contro-versy also addresses processes of learningand reasoning that lie outside the range ofPiagetian theory. Some of these processes(discussed in Zigler, 1967, 1969) have stim-ulated substantial research by now. Theywarrant critical analyses of the sort under-taken here. Such analyses are needed if weare to consolidate the gains in knowledgestimulated by the developmental versus dif-ference debate. For the time being, any an-swer to the question raised at the outset ofthis article must be carefully qualified. Whenretarded persons suffering from organic im-pairment are excluded, retarded and non-

retarded persons of the same developmentallevel do seem to use the same cognitive pro-cesses, at least in the cognitive domains ad-dressed by Piaget. The returns from otherdomains are not yet in.

Reference Note

1. Kuhn, D. The development of role-taking ability.Unpublished manuscript, Columbia University, 1972.

References

Achenbach, T. M. Conservation of illusion-distortedidentity: Its relation to MA and CA in normals andretardates. Child Development, 1969, 40, 663-679.

Achenbach. T. M. Comparison of Stanford-Binet per-formance of nonretarded and retarded persons matchedfor MA and sex. American Journal of Menial De-ficiency, 1970,74,488-494.

Achenbach, T. M. Stanford-Binet Short-Form perfor-mance of retarded and nonretarded persons matchedfor MA. American Journal of Mental Deficiency,1971, 76, 30-32.

Achenbach, T. M. Surprise and memory as indices ofconcrete operational development. Psychological Re-ports, 1973 33, 47-57.

Badt, M. I. Levels of abstraction in vocabulary defini-tions of mentally retarded school children. AmericanJournal of Mental Deficiency, 1958, 63, 241-246.

Balla, D., Styfco, S. J., & Zigler, E. Use of the oppo-sition concept and outerdirectedness in intellectuallyaverage, familial retarded, and organically retardedchildren. American Journal of Mental Deficiency,1971, 75, 663-680.

Balla, D., & Zigler, E. Discrimination and switchinglearning in normal, familial retarded, and organicretarded children. Journal of Abnormal and SocialPsychology, 1964, 61, 664-669.

Binder, A. Further consideration on testing the nullhypothesis and the strategy and tactics of investigat-ing theoretical models. Psychological Review, 1963,70, 107-115.

Blount, W. R. Concept usage research with the mentallyretarded. Psychological Bulletin, 1968, 69, 281-294.

Brekke, B., & Williams, J. D. Conservation of weightwith the mentally retarded. Journal of Genetic Psy-chology, 1974, 125, 225-231.

Brown, A. L. Conservation of number and continuousquantity in normal, bright, and retarded children.Child Development, 1973, 44, 376-379.

Bryant, P. E., & Trabasso, T. Transitive inferences andmemory in young children. Nature, 1971, 232, 456-458.

Camilli, G., & Hopkins, K. D. Applicability of chi-square to 2 X 2 contingency tables with small expectedcell frequencies. Psychological Bulletin, 1976, 85,163-167.

Cardozo, C. W., & Allen, R. M. Contribution of visualperceptual maturation to the ability to conserve.American Journal of Mental Deficiency, 1975, 79,701-704.

Cohn-Jones, L., & Seim, R. Perceptual and intellectual


factors affecting number concept development in re-tarded and nonretarded children. American Journalof Mental Deficiency, 1978, 83, 9-15.

Cole, M., & Bruner, J. Cultural differences and infer-ences about psychological processes. American Psy-chologist, 1971, 26, 867-876.

Cooper, H. M., & Rosenthal, R. Statistical versus tra-ditional procedures for summarizing research find-ings. Psychological Bulletin, 1980 87, 442-449.

Cromwell, R. L. A social learning approach to mentalretardation. In N. R. Ellis (Ed.), Handbook of mentaldeficiency. New York: McGraw-Hill, 1963.

Cruikshank, W. M. The brain-injured child in home,school, and community. Syracuse, N.Y.: SyracuseUniversity Press, 1967.

Denny, M. R. Research in learning and performance.In J. A. Stevens & R. Heber (Eds.), Menial retar-dation: A review of research. Chicago: University ofChicago Press, 1964.

DeVries, R. The development of role-taking as reflectedby behavior of bright, average, and retarded childrenin a social guessing game. Child Development, 1970,41, 759-770.

DeVries, R. Performance on Piaget-type tasks ofhigh-1Q, average-IQ, and low-IQ children. Chicago: Uni-versity of Illinois at Chicago Circle, 1973. (ERICDocument Reproduction Service No. ED086 374/PS007 129) (a)

DeVries, R. The two intelligences of bright, average,and retarded children. Chicago: University of Illinoisat Chicago Circle, 1973. (ERIC Document Repro-duction Service No. ED 079/102/SE 016 419) (b)

Dodwell, P. Children's understanding of number andrelated concepts. Canadian Journal of Psychology,1960, 14, 191-205.

Elkind, D., Koegler, R., & Go, E. Studies in perceptualdevelopment: II. Part-whole perception. Child Devel-opment, 1964, 35, 81-90.

Elkind, D., Koegler, R. R., Go, E., & Van Doorninck,W. Effects of perceptual training on unmatched sam-ples of brain-injured and familial retarded children.Journal of Abnormal Psychology, 1965,70, 107-110.

Ellis, N. R. A behavioral research strategy in mentalretardation: Defense and critique. American Journalof Mental Deficiency, 1969, 73, 557-566.

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

Furth, H. G. Conservation of Weight in deaf and hear-ing children. Child Development, 1964, 35, 143-150.

Gibson, B. J. An attributional analysis of performanceoutcomes and alleviation of learned helplessness onmotor performance tasks: A comparative study ofeducable mentally retarded and nonretarded boys.Unpublished doctoral dissertation, University of Al-berta, 1980.

Goldschmid, M., & Bentler, P. M. The dimension andmeasurement of conservation. Child Development,1968, 39, 787-802.

Goodnow, J. J., & Bethon, G. Piaget's tasks: The effectsof schooling and intelligence. Child Development,1966, 37, 573-582.

Granich, L. A qualitative analysis of concepts in men-tally deficient schoolboys. Archives of Psychology,1940, 251, 1-47.

Grant, D. A. Testing the nul l hypothesis and the strategyand tactics of investigating theoretical models. Psy-chological Review, 1962, 69, 54-61.

Greenwald, A. G. Consequences of prejudice against thenull hypothesis. Psychological Bulletin, 1975, 82, 1-20.

Gruen, G. E. Memory, IQ, and transitive inference innormals and retardates. Developmental Psychology,1973, P, 436.

Gruen, G. E., & Vore, D. A. Development of conser-vation in normal and retarded children. Developmen-tal Psychology, 1972, 6, 146-157.

Harter, S. Discrimination learning set in children as afunction of IQ and MA. Journal of ExperimentalChild Psychology, 1965, 2, 31-43.

Harter, S. Mental age, IQ, and motivational factors inthe discrimination learning set performance of normaland retarded children. Journal of ExperimentalChild Psychology, 1967, 5, 123-141.

Harter, S., Brown, L., & Zigler, E. The discriminationlearning of normal and retarded children as a functionof penalty conditions and etiology of retarded. ChildDevelopment, 1971, 42, 517-536.

Hays, W. L. Statistics for the social sciences (2nd ed.).New York: Holt, Rinehart & Winston, 1973.

Heber, R. T. Personality. In H. A. Stevens & R. T.Heber (Eds.), Mental retardation: A review of re-search. Chicago: University of Chicago Press, 1964.

Inhelder, B. The diagnosis of reasoning in the mentallyretarded. New York: Day, 1968. (Originally pub-lished, 1943).

Kahn, J. V. Moral and cognitive development in mod-erately retarded, mildly retarded, and nonretardedindividuals. American Journal of Mental Deficiency,1976, 81, 209-214.

Kamin, L. J. The science and politics of IQ. Potomac,Md.: Erlbaum, 1975.

Kappauf, W. E. Critique of the use of covariance ad-justments for CA and MA in comparative studies ofretarded and nonretarded persons. American Journalof Mental Deficiency, 1976, 81, 240-247.

Keasey, C. T., & Charles, D. C. Conservation of sub-stance in normal and retarded children. Journal ofGenetic Psychology, 1967, 111, 271-279.

Kohlberg, L. Early education: A cognitive-developmen-tal view. Child Development, 1968, 39, 1013-1062.

Kohlberg, L. Stage and sequence: The cognitive-devel-opmental approach to socialization. In D. Goslin(Ed.), Handbook of socialization theory and re-search. Chicago: Rand McNally, 1969.

Labov, W. The logical non-standard English. In F.Williams (Ed.), Language and poverty. Chicago:Markham Press, 1970.

Lane, E. B., & Kinder, E. F. Relativism in the thinkingof subnormal subjects as measured by certain of Pia-get's tests. Journal of Genetic Psychology, 1939, 54,107-118.

Luria, A. R. Psychological studies of mental deficiencyin the Soviet Union. In N. R. Ellie (Ed.), Handbookof mental deficiency. New York: McGraw-Hill, 1963.


Lutkus, A., & Trafaasso, T. Transitive inferences bypreoperational retarded adolescents. American Jour-nal of Mental Deficiency, 1974 78, 599-606.

Lyle, L. The effect of an institutional environment uponthe verbal development of imbecile children. I. Verbalintelligence. Journal of Mental Deficiency Research,1959, 3, 122-128.

Mahaney, E. J., & Stephens, B. Two-year gains in moraljudgment by retarded and nonretarded persons.American Journal of Mental Deficiency, 1974, 79,134-141.

McManis, D. L. Additive composition by normal andretarded children of matched MA. Psychonomic Sci-ence, 1968, 11,207.

McManis, D. L. Comparisons of gross, intensive, andextensive quantities by normals and retardates. ChildDevelopment, 1969, 40, 237-244 (a)

McManis, D. L. Comparisons of gross, intensive, andextensive quantities by retardates. Journal of GeneticPsychology, 1969, US, 229-236. (b)

McManis, D. L. Conservation and transitivity of weightand length by normals and retardates. DevelopmentalPsychology, 1969, /, 373-382. (c)

McManis, D. L, Conservation of identity and equiva-lence of quantity by retardates. Journal of GeneticPsychology, 1969, 115, 63-69. (d)

McManis, D. L. Conservation of mass, weight, and vol-ume by normal and retarded children. AmericanJournal of Mental Deficiency, 1969, 73, 762-767.(e)

McManis, D. L. Relative th inking by normals and re-tardates. Developmental Psychology, 1969, /, 69. (0

McManis, D. Conservation, seriation, and transitivityperformance by retarded and average individuals.American Journal of Mental Deficiency, 1970, 74,784-791.

Milgram, N. A. The rationale and irrational in Zigler'smotivational approach to mental retardation. Amer-ican Journal of Mental Deficiency, 1969, 73, 527-532.

Milgram, N. A. Cognition and language in mental re-tardation: Distinctions and implications. In D. K.Routh (Ed.), The experimental psychology of mentalretardation. Chicago: Adline, 1973.

Piaget, J. Judgment and reasoning in the child. NewYork: Harcourt, Brace, 1928.

Piaget, J. The child's conception of the world. NewYork: Harcourt, Brace, 1929.

Piaget, J. The child's conception of physical causality.London: Kegan Paul, Trench, & Trubner, 1930.

Piaget, J. The child's conception of number. New York:Humanities Press, 1952.

Piaget, J. Cognitive development in children: The Piagetpapers. In R. E. Ripple & V. N. Rockcastle (Eds.),Piaget rediscovered; A report of the Conference onCognitive Studies and Curriculum Development. Ith-aca, N.Y.: Cornell University, School of Education,1964.

Piaget, J. Piaget's theory. In P. H. Mussen (Ed.), Car-michael's manual of child psychology (3rd ed.). NewYork: Wiley, 1970.

Prothro, E. T. Egocentricity and abstraction in children

and in adult aments. American Journal of Psychol-ogy, 1943, 56, 66-77.

Reitan, R. M. Psychological assessment of deficits as-sociated with brain lesions in subjects. In J. L. Khanna(Ed.), Brain damage and mental retardation: A psy-chological evaluation. Springfield, 111.: Charles CThomas, 1973.

Roscoe, J. T., & Byars, J. A. An investigation of therestraints with respect to sample size commonly im-posed on the use of the chi-square statistic. Journalof the American Statistical Association, 1971, 66,755-759.

Rosenthal, R. Combining results of independent studies.Psychological Bulletin, 1978, 85, 185-193.

Rubin, K. H., & Orr, P. R. Spatial egocentrism in non-retarded and retarded children. American Journal ofMental Deficiency, 1974, 79, 95-97.

Russell, R. W., Dennis, W., & Ash, F. E. Studies inanimism: III Animism in feebleminded subjects.Journal of Genetic Psychology, 1940, 57, 57-63.

Sarason, S. B., & Doris, J. Psychological problems inmental deficiency. New York: Harper & Row, 1969.

Smedslund, J. The acquisition of conservation of sub-stance and weight in children: VI. Practice on con-tinuous versus discontinuous material in conflict sit-uations without external reinforcement. ScandinavianJournal of Psychology, 1961, 2, 203-210.

Smeets P. M. The animism controversy revisited: Aprobability analysis. Journal of Genetic Psychology,1973, 123, 219-225.

Smeets, P. M. The influence of MA and CA on theattribution of life and life traits to animate and in-animate objects. Journal of Genetic Psychology,1974, 124, 17-27.

Smith, J. D. EMR and nonretarded children's reactionsto contradiction. American Journal of Mental Defi-ciency, 1977, 82, 94-94. (a)

Smith, J. D. Perceptual decentering in EMR and non-retarded children. American Journal of Mental De-ficiency, 1977, 81, 499-501. (b)

Spitz, H. H. Critique of the developmental position. InE. Zigler & D. Balla (Eds.), Mental retardation: Thedevelopmental-difference controversy. Hillsdale, N.J.:Lawrence Erlbaum, in press.

Stephens, B. The development of reasoning, moraljudgment, and moral conduct in retardates and nor-mals: Phase II. Philadelphia, Pa.; Temple University,1972.

Stephens, B., & McLaughlin, J. A. Two-year gains inreasoning by retarded and nonretarded persons.American Journal of Mental Deficiency, 1974, 79,116-126.

Stephens, B., et al. Symposium; Developmental gainsin the reasoning, moral judgment, and moral conductof retarded and nonretarded persons. American Jour-nal of Mental Deficiency, 1974, 79, 113-161.

Stevenson, H. W. Social reinforcement of children'sbehavior. In L. P. Lipsitt & C. Spiker (Eds.), Ad-vances in child development (Vol. 2) New York: Ac-ademic Press, 1965.

Stevenson, J. W., Hale, G. A., Klein, R., & Miller, L.Interrelation and correlations in children's learningand problem solving. Monographs of the Society for

178 JOHN R, WEISZ AND KEITH OWEN YEATES

Research in Child Development, 1968, 33(1, SerialNo. 123).

Taylor, J. J., & Achenbach, T. M. Moral and cognitivedevelopment in retarded and nonretarded children.American Journal of Mental Deficiency, 1975, 80,43-50.

Tulkin, S. R., & Konner, M. J. Alternative conceptionsof intellectual functioning. Human Development,1973, 16, 33-52.

Weir, M. Mental retardation. Science, 1967, 157, 576-577.

Weisz, J. R. Studying cognitive development in retardedand nonretarded groups: The role of theory. AmericanJournal of Mental Deficiency, 1976, 81, 235-239.

Weisz, J. R. A follow-up developmental study of hy-pothesis behavior among mentally retarded and non-retarded children. Journal of Experimental ChildPsychology, 1977, 34, 108-122.

Weisz, J. R. Perceived control and learned helplessnessamong mentally retarded and nonretarded children:A developmental analysis. Developmental Psychol-ogy, 1979, 15, 311-319.

Weisz, J. R. Learned helplessness in black and whitechildren identified by their schools as retarded andnonretarded: Performance deterioration in responseto failure. Developmental Psychology, 1981, 4, 385-390.

Weisz, J. R. Effects of the "mentally retarded" labelon adults' attributions for child failure. Journal ofAbnormal Psychology, in press, (a)

Weisz, J. R. Learned helplessness and the retardedchild. In E. Zigler & D. Balla (Eds.), Mental retar-dation: The developmental-difference controversy.Hillsdale, N.J.: Lawrence Erlbaum, in press, (b)

Weisz, J. R., & Achenbach, T. M. Effects of IQ andmental age on hypothesis behavior in normal and re-tarded children. Developmental Psychology, 1975,11, 304-310.

Weisz, J. R., Yeates, K. O., & Zigler, E. Piagetianevidence and the developmental-difference contro-versy. In E. Zigler & D. Balla (Eds.), Mental retar-dation: The developmental-difference controversy,Hillsdale, N.J.: Lawrence Erlbaum, in press.

Weisz, J. R., & Zigler, E. Cognitive development inretarded and nonretarded persons: Piagetian tests ofthe similar sequence hypothesis. Psychological Bul-letin, 1979, 86, 831-851.

Winer, B. J. Statistical principles in experimental de-sign (2nd ed.). New York: McGraw-Hill, 1971.

Zigler, E. Familial mental retardation: A continuingdilemma. Science, 1967, 755, 292-298.

Zigler, E. Developmental versus difference theories ofmental retardation and the problem of motivation.American Journal of Mental Deficiency, 1969, 73,536-556.

Zigler, E. The retarded child as a whole person. InH. E. Adams & W. K. Boardman (Eds.), Advancesin experimental clinical psychology. New York: Per-gamon Press, 1971.

Zigler, E., Abelson, W., & Seitz, V. Motivational fac-tors in the performance of economically disadvan-taged children on the Peabody Picture VocabularyTest. Child Development, 1973, 44, 294-303.

Zigler, E., & Butterfield, E. C. Motivational aspects ofchanges in IQ test performance of culturally deprivednursery school children. Child Development, 1968,39, 1-14.

Zigler, E., & Harter, S. Socialization of the mentallyretarded. In D. A. Goslin (Ed.), Handbook of so-cialization theory and research. Chicago: RandMcNally, 1969.

Zigler, E., & Meunchow, S. Mainstreaming: The proofis in the implementation. American Psychologist,1979, 34, 993-996.

Received February 2, 1981 •

Documents

Cognitive Development in Retarded and Nonretarded Persons ... · Cognitive Development in Retarded and Nonretarded Persons: ... 1977; Weisz & Achenbach, 1975). The emphasis on pro-cess