Children’s Understanding of Social Rules and Social Status

Many thanks to the staff and pupils of six London primary schools for making this research possible, Lucía Summers for help with reliability coding, and Derek Moore for statistical advice. We are also very grateful to James Blair, Brian Clifford, Patrick Nobes, and four anonymous reviewers for their comments on an earlier draft of this article. Correspondence should be addressed to Dr. Gavin Nobes, School of Psychology, University of East London, Romford Road, London, E15 4LZ, UK. E-mail: g.nobes@uel.ac.uk. ME R R I L L PA L M E R QU A RT E R LY, VO L . 49, NO. 1

) investigated children's conceptions of rules by interviewing them about their games. He found that young children said that game rules were invented by adults (or God) and that it would be wrong to try to change them. This finding led Piaget to characterize young children as having "heteronomous" conceptions of rules: They believe all rules to be sacred, universal, and unalterable. According to this account, young children do not understand that the conventional rules (such as prohibitions of talking in class and wearing inappropriate clothes) that regulate social institutions are invented through con-sensus, differ from place to place, and are alterable. Piaget proposed that, to young children, conventions are indistinguishable from morals (for example, prohibitions of hitting and stealing), which are universal, independent of consensus, and immutable.
Piaget sought to explain his findings in social psychological and cognitive developmental terms. He argued that children have a "morality of constraint" that stems from their "unilateral respect" for their elders. As a result, he claimed, children accept adults' rules without question. Piaget thought that children are unaware of the social processes that lead to the invention or change of conventions because, owing both to their inferior social status and their egocentrism, they cannot participate in the negotiation of these rules. It is only when children become less egocentric that they are able to negotiate rules with mutually respecting peers. For Piaget, it is through this negotiation with equals that autonomy emerges as children begin to realize the nature of social conventions and to make the moral/conventional distinction. In the case of game rules, he believed, this first occurs at about 12 years.
Recent research has demonstrated that children's understanding of rules and of authority is more sophisticated than Piaget believed. The "domain theorists" (e.g., Nucci & Nucci, 1982;Nucci & Turiel, 1993;Smetana, 1981;Tisak & Jankowski, 1996;Turiel, 1983) have shown that, even at the age of 3 years (Smetana & Braeges, 1990), children's understanding of rules is differentiated: Children distinguish between domains of rules, such as morals and conventions. These researchers have told children stories involving transgressions of moral and conventional rules and asked them to judge the actions by a number of criteria that distinguish the two domains. They have reported, for example, that children recognize that conventions are relative and contingent on rules and authority, whereas morals are universal and noncontingent. Further support for the domain theorists' account comes from observations of children's social interactions (e.g., Nucci & Turiel, 1978) and studies of children's judgments of actual transgressions (e.g., Smetana, Schlagman, & Walsh Adams, 1993) and justifications of judgments (e.g., Tisak & Turiel, 1988).
Recent research has also demonstrated that children do not have the unitary orientation toward adults that Piaget proposed, since their understanding of authority is also differentiated. As with their conceptions of social rules, children's judgments of the legitimacy of authority are influenced by the domain of directives (Laupa & Turiel, 1986). The knowledge or competence of individuals who issue directives is also an important factor in these judgments (e.g., Laupa, 1991): 1 Most authorities encountered by children in hierarchical organizations combine several authority attributes because they are adults who have social position, competence, and experience.
Younger children (but not necessarily adolescents) consider the directives of knowledgeable individuals to be more legitimate than those of incompetent individuals. Furthermore, at least for older children (Laupa, 1991), judgments of authority are strongly influenced by the social position-such as school principal or parent-of the people who give directives (e.g., Laupa & Turiel, 1986). And children understand that the jurisdiction of authorities extends only to certain contexts: Children judge teachers' authority to be more legitimate within school than at home, and parents' authority to be less legitimate at school, or at a friend's house, than at home (Laupa & Turiel, 1993;Laupa, Turiel, & Cowan, 1995;Tisak, Crane-Ross, Tisak, & Maynard, 2000).
Contrary to Piaget's claim, children acknowledge that legitimate authorities are not necessarily adult. Children accept peers in authority positions at school when the position has been delegated by a teacher (Laupa, 1991;Laupa & Turiel, 1986). They even recognize the authority of a younger sibling when this authority has been delegated by adults (Laupa, 1995). However, some studies point to adulthood being influential, especially in the judgments of younger children (Laupa, 1995;Laupa & Turiel, 1986).
In this study we investigated how children's conceptions of rules and authority interact to influence children's social judgments. It was not our aim to investigate further which of the components of authority, such as social position, competence, experience, and adulthood, are important in children's social judgments. Instead, we explored the possibility that children's conceptions of morals and conventions are influenced by whether the inventors, transgressors, and changers of these rules are adults, who have relatively high authority status, or children, whose authority status is relatively low. 1 For this reason, "adult status" refers here to adults who have at least one authority attribute, and "child status" to children who have none.
It seems likely on intuitive grounds that children's conceptions of rules are influenced by adult/child status. For example, children might be expected to say that adults can usually transgress and alter rules more easily than can children. Also, children are likely to think that rules that are invented by adults are generally less violable, contingent, and alterable than rules that are invented by children. That is, depending on the status of the individuals involved in inventing, transgressing, 2 These figures relate to familiar rules and are similar to those for unfamiliar rules.  show that he asked some children whether they (the children) could change the rules, while others were invited to comment on changes proposed by the adult researcher. But Piaget does not report whether his interviewees responded differently to these different questions, nor on which responses he based his conclusions. and altering, children might judge rules to be more, or less, absolute. These would be rational and realistic judgments based on children's everyday interactions with adult authorities in social organizations. It is even possible that differences in status between rule inventors and transgressors will lead children to say that some transgressions and alterations of conventions are as impermissible as transgressions and alterations of morals: That is, in these circumstances, they might not make the moral/conventional distinction.
There is some evidence that children's conceptions of rules are influenced in this way. Davidson, Turiel, and Black (1983) asked children, "Is it all right to change this rule?" and found that 18% said that morals could be changed, compared with 89% who said that conventions could be changed. In contrast, when the question was, "Could the children get this rule changed?" the proportion of children who replied that they could was 7% for both morals and conventional rules. 2 By the criterion of alterability, then, and when the attempted changes were made by children, the respondents no longer distinguished between morals and conventions.
To investigate the influence of adult/child status, it is necessary to compare children's judgments of rules that are invented by adults and children and transgressed and altered by adults and children. This has not been done in previous research because investigators have focused almost exclusively on children's conceptions of adults' rules and authority (even if it is delegated to children) in adult-led social organizations such as schools and families. Children have not been asked about rules that children invent for themselves (Nobes, 1999). Furthermore, children have been asked about situations and actions in which the subjects of authority, and the transgressors of rules, have almost invariably been children, rather than adults. Piaget, too, asked children only about the origins and alterability of elders' rules. 3 In this study we systematically varied the adult/child status of the inventors, transgressors and changers of conventional rules. In addition to adult-invented and child-invented conventions, morals and cultural conventions-the familiar, culture-wide conventions that govern, 4 Previous researchers have not distinguished between conventions that are newly invented and local-used here to compare the influence of rule invention by adults and by children-and cultural conventions. In order to compare our findings with theirs, it was therefore necessary to include all three kinds of conventions: adult invented, child invented, and cultural. for example, the ways in which we dress and eat-were also included. 4 This design enabled us to assess the influence on children's judgments of the adult/child status of the inventors, transgressors, and changers of rules, to compare this with the influence of social domain, and to explore the possibility that children conceive of cultural and newly invented conventions differently.
It was predicted that children would distinguish between morals and conventions by judging morals to be less alterable and their transgression less permissible than all three kinds of convention. And, in the absence of a rule, or of an authority's knowledge, moral transgressions would still be judged to be wrong, whereas conventional transgressions would not. These are the criterion judgments of alterability, permissibility, rule contingency, and authority contingency, respectively.
It was also predicted that children would be sensitive to the status of the people who invented, transgressed, or changed the rules. Thus, children were expected to judge rules less absolute-less violable, alterable, rule contingent, and authority contingent-when transgressed or altered by adults than by children. In addition, it was predicted that adult-invented conventions would be considered more absolute than child-invented conventions. It was also expected that children's judgments of transgressions and alterations of cultural and adult-invented conventions would be similar, because both are adults' rules.
Since older children tend to have a more differentiated understanding of social rules, authority, and status than younger children (e.g., Laupa, 1995;Shantz, 1982;Tisak et al., 2000;Tisak & Turiel, 1988), it was expected that older children would distinguish between actions more clearly than younger children on the grounds of both social domain and the status of the inventors, changers, and transgressors.
Another aim of the study was to investigate the influence of story content on children's judgments. It is likely that transgressions and alterations of rules of the same source (moral, cultural conventional, adult-invented conventional, or child-invented conventional) are judged differently because, for example, the consequences of transgressions of some rules are considered more serious than of others. If this is the case then, depending on researchers' choice of rules, there is 5 Use of multiple stories of each story type also reduced the likelihood of the content of individual stories (as opposed to rule source and agent status) influencing the criterion judgment scores of individual children or of the sample. In addition, since each child's criterion judgment scores for each story type were derived from judgments of two or three stories, variance and hence statistical power were increased. potential for differences in judgments being wrongly attributed to differences in domains, or in status, when actually they reflect differences in story contents. For this reason, children were asked about several stories involving transgressions and alterations of each source of rule. 5

Design
Each child was asked questions about stories in which rules of four different sources-moral, cultural conventional, adult-invented conventional, and child-invented conventional-were broken and changed either by adult or child agents. There were, then, eight rule-source/agentstatus story types: moral/adult, moral/child, cultural conventional /adult, cultural conventional/child, adult-invented conventional/adult, adult-invented conventional/child, child-invented conventional /adult, child-invented conventional/child. Rule source and agent status were the two within-participants IVs, with four and two levels, respectively.
There were also two between-participants IVs-age group (4-5 years, 6-7 years, and 8-9 years) and gender. The questions asked about each story elicited four criterion judgments: permissibility, authority contingency, rule contingency, and alterability. These were the DVs.

Stimuli
To ensure that each child could have three stories of each type, and that no child heard each story more than once, there were 24 stories involving rule transgressions. These rules were drawn from previous research (e.g., Blair, 1995;Nucci, 1981;Sanderson & Siegal, 1988;Smetana & Braeges, 1990;Smetana et al., 1993;Tisak & Turiel, 1988;Turiel, 1983). The six stories involving transgressions of moral rules were stealing money, pushing, not helping, telling someone else's secret, not sharing, and falsely blaming. There were also six stories involving transgressions of cultural conventional rules: going naked in public, using the wrong lavatory, removing shoes in a restaurant, male wearing nail polish, not saying "please," and swearing.
Finally, there were twelve stories in which newly invented conventional rules were transgressed: sitting in the wrong place on a bus, not wearing a club's colors, not doing a job as arranged, calling the team captain the wrong name, not counting in hide-and-seek, not replacing crayons in the right place, not swimming a required distance, not participating in a pregame ritual, talking in a quiet place, using the wrong bowl to feed the dog, playing indoors in fine weather, and wearing the wrong color shoes to school.
Each moral and cultural conventional story had two versionsone in which the agent of transgression and alteration was an adult, and in the other, a child. For example, the adult (and, in parentheses, the child) version of the moral story about stealing went as follows: This is Mrs. Foster (Amanda). Mrs. Foster is a grown-up (Amanda is a girl about your age). One day Mrs. Foster (Amanda) notices someone has left a purse on the table. Mrs. Foster (Amanda) opens the purse and finds five pounds. She takes the five pounds.
Each newly invented conventional story had four versions, in which the inventors were either adult or child, and the transgressor and changer were either adult or child. For example, the adult (and child) versions of the story about being quiet in a corner of the classroom went as follows: At Lancaster Primary School, all the teachers at the school (children in a class) made a rule that in the corner of the classroom by the window there is no talking allowed by the teachers or the children. One day, Mrs. Beaton, who is a teacher (June, who is a girl in the class), was sitting in the quiet corner. She looked out of the window and said, "Look at the birds in that tree." Procedure Story versions were systematically allocated so that each child could have 24 different stories, and approximately equal numbers of each version of each story were presented within each age group.
Each child was interviewed individually in a quiet room in his or her school. The researcher first gave a brief introduction and explanation of the task and then presented the child's stories in random order. Having ensured that the child understood, each story was followed by 6 These questions were used by Smetana and Braeges (1990). questions about four criterion judgments. First, to assess judgments of the permissibility of the actions, participants were asked "Was it OK for X to have done that?" and, if the answer was "no," "Was it bad or very bad?" Then, to assess judgments of the authority contingency and rule contingency of the actions, they were asked "If no one had seen X doing that, would it have been OK?" and "What if no one had told X the rule about not doing that, would it have been OK?" 6 The next question was "If all the children in X's class [or grown-ups where X works] wanted to change the rule so that [the rule was changed locally], would that be OK?" to assess judgments of the alterability of the rules. If the answer to this question was "yes," they were asked "How would they do that?" to ensure it had been understood.
Each interview lasted 20-30 minutes. When children showed signs of reduced attention their interviews were cut short and continued during the researcher's next visit to the school, between 1 and 5 days later. All participants responded to a minimum of 2 stories of each type (i.e., a minimum total of 16 stories), and 34 (26.4%) children answered the questions to all 24 stories. All interviews were tape-recorded.

Scoring
Participants' judgments of the permissibility of each action were coded 0 (OK), 1 (bad), or 2 (very bad). Each of their judgments of rule contingency, authority contingency, and rule alterability was coded 0 (OK) or 1 (not OK). A child's criterion judgment score for a story type was his or her mean score for stories of that type. Higher scores indicated that children considered rules more absolute.

Reliability and Consistency
Coding reliability was assessed by randomly selecting a subsample of 20 (15.5%) of the interviews for recoding by a second judge. Interjudge reliability ranged from 94.8% to 100%, with a mean of 98.9%.
The overall internal consistency across all eight story types and four scales (i.e., 32 judgments) was high (Cronbach's ␣ ϭ .93). Taking each of the eight story types separately, internal consistency across the four criterion judgment scales ranged from .73 to .78. The internal consistency of each of the four scales across the eight story types ranged from .83 to .90.

Results
The children's mean scores on each of the four criterion judgments for each of the eight story types are presented in Table 1. Figure 1 shows their mean scores on judgments of source of rule, and Figure 2 on judgments of status of agent.
A mixed 4 (source of rule) x 2 (status of agent of transgression or alteration) x 2 (gender of participant) x 3 (age group of participant) multivariate analysis of variance (MANOVA) with repeated measures on source and status was carried out on responses to the four criterion judgment scales. The MANOVA revealed a significant main effect of source of rule (F(12, 112) ϭ 22.10, p Ͻ .001, partial 2 ϭ .70). Children distinguished between the rules of different sources on all four criteria: permissibility (F(3, 124) ϭ 58.25, p Ͻ .001, partial 2 ϭ .32), authority contingency (F(3, 124) ϭ 31.73, p Ͻ .001, partial 2 ϭ .21), rule contingency (F(3, 124) ϭ 88.77, p Ͻ .001, partial 2 ϭ .42), and alterability (F(3, 124) ϭ 71.63, p Ͻ .001, partial 2 ϭ .37). In this and other multiple comparisons, post hoc analyses of significant effects were conducted using Tukey HSD tests. Moral rules were judged more absolute on all four criterion judgments than any of the three sources of convention (ps Ͻ .01), except that there was no significant difference between the children's judgments of the alterability of morals and cultural conventions. Cultural conventions were judged to be more absolute on all criterion judgments than adult-or child-invented conventions, and adult-invented conventions were considered less alterable than child-invented conventions (ps Ͻ .01).
The interaction between source and agent was significant (F(12, 112) ϭ 6.83, p Ͻ .001, partial 2 ϭ .42). This was due to a difference in judgments of alterability (F(3, 124) ϭ 21.77, p Ͻ .001, partial 2 ϭ .15) and a small difference in judgments of permissibility (F(3, 124) ϭ 2.66, p ϭ .048, partial 2 ϭ .02). Only violations and changes of adultinvented conventions, and changes of morals, were judged to be less acceptable when the agents were children than when they were adults (ps Ͻ .05): alteration of adult-invented conventions by children was considered to be as unacceptable as children's alteration of cultural    conventions and morals. Adults were judged no more able to break and change child-invented conventions than their own conventions. The MANOVA also revealed that the main effect of age of respondent was significant (F(8, 240) ϭ 6.61, p Ͻ .001, partial 2 ϭ .18). Univariate analyses showed that this was due to age-related differences in judgments of rule contingency (F(2, 123) ϭ 14.04, p Ͻ .001, partial 2 ϭ .19). The 4-5-year-olds were found to judge transgressions to be less rule contingent than did either of the other age groups (ps Ͻ .003).
There was also a significant interaction between source, agent, and age (F(24, 224) ϭ 1.70, p ϭ .03, partial 2 ϭ .15). This reflected differences in judgments of alterability (F(6, 248) ϭ 2.89, p ϭ .01, partial 2 ϭ .05). The 4-5-year-olds said that adults' alterations of adultinvented and child-invented conventions, and children's alterations of child-invented conventions, were less acceptable than did the 6-9-yearolds (ps Ͻ .01). In contrast to the 6-9-year-olds, the youngest children considered adults' changes of adult-invented conventions to be as unacceptable as adults' changes of morals, and neither adults' nor children's alterations of children's conventions to be any more acceptable than their alterations of adults' conventions.
The main effect of gender, and interactions between gender and other variables, were all nonsignificant.
The influence of the content of individual stories on children's judgments was investigated by comparing the criterion judgments for each of the stories, regardless of the status of inventors and transgressors. These are shown in Table 2. One moral rule (pushing) was considered more absolute (i.e., had higher mean criterion judgment) than all the cultural conventions (ts у 2.48, ps р .02), and two other morals (inappropriate blaming and stealing money) were judged to be more  absolute than all cultural conventions except two (ts у 2.26, ps р .02). These two cultural conventions were swearing and using the wrong lavatory, both of which were considered to be more absolute than the morals of helping someone who had fallen over and sharing (ts у 2.79, ps р .01). This latter moral was not judged to be more absolute than any cultural convention.
Most of the cultural conventions were judged to be more absolute than most of the conventions that were invented by adults and children. Swearing, using the wrong lavatory, and undressing in public were all considered more absolute than all the invented conventions (ts у 2.78, ps р .01). The highest rated invented convention (sitting in a designated place) was considered less absolute than all cultural conventions (ts у 2.31, ps Ͻ .03) except saying "please" and wearing shoes in a restaurant.

Discussion
Consistent with the first hypothesis and the findings of the domain theorists, children distinguished between moral and conventional rules. Morals were considered less violable, authority contingent, rule contingent, and alterable than all three kinds of conventions, with the single exception that cultural conventions were not judged to be significantly more alterable than morals.
As well as the social domain of rules being influential in children's judgments, so too was the social status (adult or child) of the people who invented, transgressed, or changed the rules; the second hypothesis, then, was also supported. When adults were the agents, rules were judged to be considerably more alterable and slightly more violable and authority contingent than when the transgressors or changers were children. Adult-invented conventions were considered less alterable than child-invented conventions. Status was especially important when the rules were invented by adults and changed by children: These alterations were regarded as no more acceptable than changes of morals. Indeed, by the criterion of alterability, the influence of social status on the participants' judgments was greater than that of social domain.
The participants considered adults' transgressions and alterations of adult-invented conventions to be relatively acceptable on all four criterion judgments. Similarly, they said that children could break or change the child-invented conventions. Indeed, children were considered to have as much authority to break and change child-invented rules as have adults. These findings are consistent with Laupa and her colleagues' (Laupa, 1991(Laupa, , 1995Laupa & Turiel, 1986Tisak et al., 2000) view that social position-as opposed to adulthood per seis an important determinant of children's social judgments. When children invent rules, they construct contexts in which they are in positions of authority (Nobes, 1999). Within these contexts, adults have no special position that enables them legitimately to break or change the rules.
There were considerable differences between the judgments of children of different ages. Consistent with predictions, and with the findings of previous researchers (e.g., Laupa, 1995;Shantz, 1982;Tisak & Turiel, 1988;Tisak et al., 2000), older children tended to distinguish more clearly than younger children between rules according to domain. This was also found to be the case with the status of rule inventors, transgressors, and changers. In contrast to the 6-9-yearolds, the youngest children did not differentiate morals from cultural conventions, nor adult-invented conventions from child-invented conventions, by any criterion. Neither did these 4-5-year-olds significantly distinguish between the rule contingency of the four sources of rules, nor between alterations by adults and by children.
It is possible that the youngest children's failure to differentiate as well as older children was at least partly due to their not adequately understanding some of the questions. Other researchers have reported similar problems: Tisak and Turiel (1988), for example, found that their 1 st graders, who were older than our youngest group, found the rule contingency question too difficult.
The use of relatively large numbers of rules of each source in this study (i.e., 6 morals, 6 cultural conventions, and 12 invented conventions) revealed a high degree of variation in children's criterion judgments within, as well as between, rule sources. Some of the cultural conventions were considered more absolute than some morals. These findings indicate that story content has a significant influence on children's judgments. They suggest that it is important that researchers in this field consider the content of the stories they use, for it is possible that their findings might sometimes reflect the influence of story content rather than, say, the influence of domain or status. Indeed, our own conclusions might have been different-for example, we could have found a clearer moral/conventional distinction, or none at all-had we chosen to use only certain moral and conventional stories.
Contrary to expectations, cultural conventions were judged on all four criterion judgments to be more absolute than adult-invented conventions. Younger children were more likely to make this distinction than older. Since cultural conventions are invented and enforced by adults, children were not expected to distinguish them from adultinvented conventions. This finding might be another result of the children's recognition that people-whether adults or children-hold positions of authority only in certain contexts, such as when they have been involved in inventing rules for their own school, family, or peer group. The cultural conventions might, then, have been seen to be beyond the jurisdiction even of the adults in the stories because the adults did not hold legitimate positions of authority regarding the rules of the wider culture.
An alternative explanation relates to the issue of story content, since it is likely that most of the cultural conventions used in this study concerned more serious issues than the newly invented rules. Moreover, transgressions of several of the cultural conventions might be considered to be second-order events (Turiel, 1983) because they overlap with the moral domain in that they would lead to other people being upset or distressed. Indeed, the cultural conventions that children rated most absolute-swearing, taking clothes off in public, and a male using a females' lavatory-might all have been interpreted by children as acts that were likely to cause distress. Similarly, the rule of sharing, which many children considered to be less absolute than several cultural conventions, might have been interpreted as belonging in the personal, rather than the moral, domain; that is, children might have judged whether to share biscuits with others to be a matter of personal choice.
Our results suggest that judgments of alterability differ from other criterion judgments. The children distinguished between morals and conventions, but not between adult-and child-invented conventions, in terms of permissibility, rule contingency, and authority contingency; in contrast, morals and cultural conventions were not distinguished by children regarding alterability, but adult-and child-invented conventions were. Furthermore, although children's judgments of permissibility and authority contingency were only slightly influenced by the status of transgressor, children considered alterations of rules by adults to be very much more acceptable than alterations of rules by children. These differences occurred primarily because children judged alterations of adult-invented conventions by children to be unacceptable or impossible.
Perhaps the differences between children's judgments of alterability and of other criteria are to be explained in terms of children's likely experiences of breaking and changing rules: By midchildhood, most children will have broken adults' rules, and invented and changed their own. They will know that adults break and change rules, too. However, children are very rarely given the opportunity to alter adults' rules. These judgments, then, accurately and realistically reflect their experiences and knowledge of the social world.
These results concerning judgments of alterability support Piaget's (1932Piaget's ( /1968) finding that children consider changes of elders' rules by children to be unacceptable or impossible. Indeed, they are consistent with his claim that children's conceptions of rules are to be understood partly in terms of the inequality in status between adults and children. Similarly, Davidson et al. (1983) report that children judge children's alterations of conventions to be as unacceptable as children's alterations of morals. But the present data also concur with Davidson et al.'s finding that, at least from the age of 6 years, children say that conventions can be changed: Our results reveal that 6-9-year-old children understand that children can alter their own conventions, and that adults can change theirs. And, consistent with the large majority of findings reported by the domain theorists, children distinguished between transgressions (as opposed to alterations) by the criteria of permissibility and contingency according to domain. None of these latter findings would have been predicted by Piaget: They show status is an important factor in children's judgments, but not for the reasons he proposed. Children are not heteronomous, nor do they have unilateral respect for adulthood per se. Instead, the evidence from this and other studies indicates that children respect adults' status and rules because they realistically appraise adults' competence and positions of authority in social organizations.
The apparently contradictory findings of Piaget and the domain theorists might stem from the different emphases they have placed on the various criterion judgments. For Piaget, the key criterion by which to assess children's concepts of social rules was alterability. As we have reported, children's judgments by this criterion appear to differ from those by other criteria such as permissibility, authority contingency, and rule contingency. In contrast, Turiel and his colleagues have focused on these other criterion judgments. Here, then, is an explanation of Piaget's findings that led him erroneously to conclude that children are heteronomous. Had he focused on criterion judgments other than alterability, or asked about children's changes of children's rules, or adults' changes of adults' rules, he might well have come to very different conclusions about children's conceptions of rules.
Children have been asked about the alterability of morals and conventions in only a few of the domain theorists' studies. 7 Some researchers have asked whether the rules could be changed by adult authorities (Nucci and Turiel (1993) asked about ministers and the congregation of a church changing church rules) or by society (Lockhart, Abrahams, & Osherson, 1977). Like Davidson et al. (1983), Buchanan-Barrow and Barrett (1998) posed a general question ("Do you think this rule could be changed?") without specifying who made the changes. And, as in the present study and Davidson et al.'s, Smetana (1986) asked children about stories in which changes were made by children. In contrast to Davidson et al.'s and our findings, Smetana found that children considered the conventions more alterable than the morals. The explanation for this discrepancy might be found in the content of the stories used in the different studies. Smetana asked children about two morals-hitting another child and taking another child's toy-and two conventions-wearing a swimsuit to day care and not sitting in one's seat during snack. The mean alterability judgment for the morals (converted to allow comparison with our data) was .73, and for the conventions .54. The mean alterability judgments of similar stories in the present study (pushing and stealing, and wearing the wrong color clothes and sitting in the wrong seat in a bus) were .84 for morals and .53 for conventions. These means are significantly different (t (95) ϭ 5.65, p Ͻ .001). It seems likely, then, that the discrepancy between the findings of Davidson et al. (1983) and those of the present study on the one hand, and Smetana's (1986) findings on the other, reflects the content of the stories chosen by these different researchers. This comparison illustrates the point that story content can have a substantial impact on children's judgments, independent of domain and status, and can influence researchers' findings and conclusions accordingly.
Together, the findings of these and the present studies suggest that, when it is not specified who changes rules, children recognize that conventions, but not morals, can be changed. Also, they think that adult authorities such as church ministers, and a whole society, can change conventions. But the present findings indicate that children think that a 7 Of the 23 published studies of criterion judgment reported in Helwig, Tisak and Turiel's (1990) review, only four compared children's judgments of the alterability of morals and conventions and found differences in the predicted direction. The Davidson et al. (1983) study is the only one in which children were asked about children's alterations of rules. small group of adults cannot change the rules of the wider society, presumably because they have no authority to do so, and that children cannot change adults' rules.
In summary, the findings of this study indicate that children's reasoning about social rules is influenced by a number of interacting factors. These include, first, the domain of the rule that is broken or altered: Morals are more absolute than conventions. Second, the status of the inventor, transgressor, or changer of the rule: Transgressions and alterations by adults are considered more legitimate than are violations and changes by children, and alterations of adultinvented rules are less acceptable than alterations of child-invented rules. There was evidence that the adults' authority derived more from their positions of authority than from their adulthood per se. Third, the criterion by which actions are judged: Children appear to be less sensitive to domain, and more sensitive to status, when asked about alterability than when they make other criterion judgments. Fourth, the content of stories: There is wide variation in children's conceptions of rules even of the same domain or source. And fifth, children's ages: Older children distinguish more clearly than younger children both with regards to domain and to status of inventor, transgressor, or changer of rules. With age, children's understanding of the social world becomes increasingly sophisticated and differentiated.