Early Childhood Research Quarterly, 4, 245-260 (1989)
Elementary School Children's Rough-and-Tumble Play A.D. Pellegrini University of Georgia
The intent of this study was to describe elementary school children's roughand-tumble play (R&T) vis-a-vis aggression, and to provide exploratory data on the functional significance of R&T. Children in grades K, 2, and 4 were observed during their playground school recess periods. Results suggested that the incidence of R&T varied according to the sex of the child and playground location; it accounted for I 1°70of children's play behaviors. Aggression rarely occurred (0.3%) and did not vary by age, sex, or playground location. These descriptive results are discussed in terms of providing ecologically valid data on children's R&T vis-a-vis aggression. R&T was also found to lead to games-with-rules and to be positively correlated, for boys, with measures of social competence. These functional results are discussed in terms of the social skills function of R&T.
In recent years researchers from a number of fields such as anthropology, education, and psychology have systematically examined children's play. These examinations were often motivated by the theoretical stances of Piaget (1962) and Vygotsky (1967), both o f whom saw play as serving an important role in preschoolers' social-cognitive development. As such, much o f the subsequent research on children's play has been confined to the preschool period (Rubin, Fein, & Vandenberg, 1983). The play o f elementary school-age children has been sorely neglected. Further, aspects of children's play studied are often related to the cognitive (e.g., McCune-Nicholich, 1981) and social-cognitive dimensions (e.g., PeUegrini, 1987; Rubin, Maioni, & Hornung, 1976) of these theories. In the present study we will examine a This work was partly supported while the author was a Sarah H. Moss Fellow at Sheffield University. The author acknowledges the principal, Tom Davis, teachers, and children of the Timothy Road School in Athens, Georgia; C. Coombs, C. Davis, and J. Perlmutter for help observing and testing children and data coding; Peter Smith for helpful comments and materials; and Rex Forehand for his reading of an earlier version of the paper. Correspondence and requests for reprints should be sent to A. Pellegrini, Institute for Behavioral Research, University of Georgia, Athens, GA 30602.
245
246
Pellegrini
heretofore infrequently examined aspect of elementary school children's play, rough-and-tumble play (R&T). The term R& T was used by Harlow (1962) to describe the quasi-agonistic behavior of rhesus monkeys. The behavioral components of R&T are typically playful and nonaggressive (Blurton Jones, 1972; Humphreys & Smith, 1987; Pellegrini, 1987). Examples of R&T include play fighting and chasing. To date, most of the work in the area of R&T has involved defining the behavioral components of the category vis-a-vis other forms of social interaction, particularly aggression. This work has provided a behavioral foundation describing preschoolers' R&T. Systematic study of elementary school children's R&T, however, is necessary for least three reasons. First, there is a paucity of data for this period. Second, preliminary observations, done primarily in England (Humphrey & Smith, 1984), suggest that R&T accounts for almost 15°70 of children's playground behaviors. This percentage is similar to the occurrence of preschool children's pretense play, 12070 to 32070, and functional play, 17070 to 34070 (Rubin et al., 1983). The third reason for needed examinations of R&T is that, like pretense play, it may provide insight into children's social-cognitive development (Pellegrini, 1987; Smith, 1982). Human ethologists have suggested that R&T may serve important physical and social developmental functions (e.g., Smith, 1982). Like studies of other forms of play, most of the studies of children's R&T have investigated preschool children (e.g., Rubin et al., 1983; Smith & Lewis, 1985). The reason for this may be because of the decreasing importance assigned to play in elementary schools (Glickman, 1984). Where elementary school children's R&T has been studied, the children were typically observed on the playground. The reasons for this are fairly straightforward: playground time, or recess, is one of the few times that children are "allowed" to play in school. Furthermore, spacious outdoor environments seem to facilitate R&T (e.g., Humphreys & Smith, 1984; Smith & Connolly, 1980). In short, there seems to be a paucity of research describing the R&T play of elementary school children. A few studies have examined the role of R&T in children's development (e.g., Smith, 1982; Smith & Lewis, 1985). The results of these functional examinations of R&T have been equivocal, however. Some researchers suggest that R&T and aggression are related during the elementary school years (e.g., Ladd, 1983; Neil, 1976). Ethologists, on the other hand, see R&T and aggression as separate behavioral phenomena (e.g., Blurton Jones, 1972; Humphreys & Smith, 1987; Smith, 1982), with R&T serving a positive role in children's social development. This confusion may in part be due to definitional problems. For example, some researchers (e.g., Ladd, 1983; Potts, Huston, & Wright, 1986) have defined R&T as including both fighting and mock fighting.
Rough-and-TumblePlay in School
247
This confounding of R&T with aggression may be partly responsible for the negative relations between R&T and measures of social competence, such as sociometric status. R&T and aggression can be easily confused because of their general similarity (Parke & Slaby, 1983). When preschool children's specific motor patterns are observed, however, two distinct factors seem to emerge: (1) aggression: fixate, frown, hit, push, take, and grab; and (2) R&T: laugh, run, jump, open beat, wrestle, chase, and flee (Blurton Jones, 1972; Smith, 1982). R&T is also characterized by positive affect (i.e., laugh), partners' reciprocating roles (e.g., chaser-chased), and sustained social interaction. Aggression is characterized by negative affect (i.e., frown, fixate), unilateral roles (e.g., aggressor-victim roles are not exchanged), and children separating after an aggressive act. Indeed, when R&T and aggression are analyzed separately they each have very different behavioral patterns and correlates. For example, popular children tend to engage in R&T, and they do so with their friends and others of similar dominance status (Humphreys & Smith, 1987). Given the definitional uncertainty of R&T vis-a-vis aggression, it seems necessary to describe the occurrence of each in elementary school children. The first objective of the study was to describe the occurrence of both R&T and aggression according to grade, sex, and location on the playground. This objective extends the previous line of R&T research, which provided behavioral definitions of R&T and aggression. By describing the contexts in which R&T and aggression occur, we provide further insight into their distinctiveness and their possible functional significance (Boulton & Smith, 1989; Hole & Einon, 1984). Further, examining the contexts in which R&T behaviors occur defines R&T in ecologically valid terms (Boulton & Smith, 1989). Separate bodies of research suggest that R&T and aggression should vary differentially according to grade level, sex, and location on the playground. Regarding age, R&T, like pretense play, follows an inverted-U developmental function, accounting for 5% of preschoolers' free play, increasing to 13.3% for 7-year-olds, decreasing to 9% at 9 years and to 4.6% at 11 years (Humphreys & Smith, 1984). Aggression, a relatively infrequent behavior, is stable (Olweus, 1979; Parke & Slaby, 1983). Therefore, age effects are not expected for aggression, while they are expected for R&T. The effect of sex on both R&T and aggression is one of the more robust findings in developmental psychology: boys do more of both than girls do (Maccoby & Jacklin, 1974; Pellegrini, 1987). For this reason, we hypothesize that boys will engage in more R&T and aggression than girls will. Physical environmental variables affect R&T and aggression, although this literature is extremely limited. R&T tends occur more frequently on soft, grassy surfaces than on hard, asphalt surfaces (Humphreys & Smith, 1984). The literature on physical environmental effects on aggression is
248
Pellegrini
generally limited to discussions of the effects of temperature and crowding (Parke & Slaby, 1983). Thus descriptive data on relations between playground locations and children's R&T and aggression is greatly needed. We expect, however, R&T to occur on soft surfaces more frequently than on hard surfaces. In summary, the first objective of the study was to examine the differential effects of age, sex, and physical contexts on children's R&T and aggression. This objective should provide further insight into the differences between R&T and aggression, as well as examining R&T in ecologically valid terms. A second and related objective of the present study was to examine the extent to which R&T moved into aggressive behavior. Neil (1976), for example, has found that R&T and real fighting co-occur among middle school boys. If this is the case, one would expect R&T events to lead into aggression. For this objective we will examine, through sequential lag analyses (Bakeman & Gottman, 1986), the extent to which R&T episodes lead directly into aggression. The third objective of the study was to examine the functional significance, or beneficial consequences (Hinde, 1982), of R&T. Theoretical discussions of R&T vis-fi-vis aggression have hypothesized that R&T should be positively related to children's social competence, because in R&T children practice a number of skills that are necessary for successful social interaction (Pellegrini, 1987; Smith, 1982). We hypothesize that during the elementary school years R&T should lead to children's engagement in games-with-rules. This hypothesis is based on the suggestion that the physical skills gained in R&T (Smith, 1982) are then used in playground gameswith-rules (Pellegrini, 1987). This hypothesis is feasible to the extent that children tend to engage in more games-with-rules and less R&T as they move from early to middle childhood (Rubin et al., 1983). This hypothesis will be tested by examining the probability of R&T events moving directly into games-with-rules. It has also been argued that, because children engage in different reciprocal roles in R&T (e.g., chaser-chased, aggressor-aggressee), they should be adept at solving social problems (Pellegrini, 1987). That is, play that involves children's playing and negotiating different reciprocal roles should be related to their social problem-solving flexibility (Spivack & Shure, 1974). In aggressive behavior there is no such reciprocity. Thus we would expect aggressive children to be poor social problem solvers. It has also been argued that R&T serves a social affiliation function for children to the extent that R&T play partners tend to be friends (Humphreys & Smith, 1987; Smith & Lewis, 1985). It should follow, then, that R&T is positively related to popularity, while aggression is negatively related to popularity. R&T should be beneficial to children's social competence to the extent that it provides children with practice in discriminating between real aggres-
Rough-and-Tumble Play in School
249
sion and mock aggression (Pellegrini, 1987). Given the similarities between the two behavioral categories, children, especially aggressive children, must learn to attend carefully to details such as frown versus smile and open-hand versus closed-hand in order to discriminate between real and mock aggression. Dodge and his colleagues (Dodge & Frame, 1982) have shown that aggressive children tend to attribute aggressive intent to ambiguous provocation situations because, in part, they processed fewer relevant social cues. It should follow, then, that engaging in aggressive behavior would be negatively related to one's ability to discriminate mock from real fighting. In summary, the intent of the present study was to describe the contexts in which R&T and aggression occurred during the elementary school years. These data are needed to the extent that the two categories have often been confused. The functional significance of R&T is also explored.
ME THOD Subjects All children attended a public elementary school. Their participation in the study was determined by their returning parental consent forms. Children from two of the four classrooms at grades K, 2, and 4 participated in the study; those who had repeated a grade were not included. Of the children participating in the study, 35 were kindergartners (males = 18, females = 17; M = 64 months of age), 30 second graders (males = 16, females = 14; M = 90 months), and 29 fourth graders (males -- 14, females -- 15; M = 117 months). Procedures Children were observed during their recess periods on the school playground from October to May. Recess periods for each grade level lasted 25 minutes. Between 120 and 150 children were on the playground during each period. Three to five adults supervised the children. The playground had three distinct components: a blacktop area (approximately 10 yards x 100 yards); a contemporary playscape (Hayward, Rothenberg, & Beasley, 1974) located in a pine forest (I00 yards x 35 yards); and a grassy area (100 yards x 20 yards) separating the blacktop from the playscape. There were a total of four observers of children's playground behavior. The behavior of individual children was coded following scan and event sampling techniques (Altman, 1974; Humphreys & Smith, 1987). More specifically, observers used scan sampling techniques to observe each child, in predetermined counterbalanced order, for 5 seconds. Each observer recorded the following information (by whispering into an audio recorder) for each scan: the target's name; the location of the behavior; the numbers of boys, girls, and adults present; target's behavior, and reactors (peers and the adults). Each child was observed at least 100 times across the school year.
250
Pellegrini
The behaviors recorded were based on the previous ethological lists of Smith and his colleages (Humphreys & Smith, 1987). These behaviors are listed in the Appendix. The information from each taped scan was recorded on coding sheets. The behaviors making up the R&T category, following Humphreys and Smith (1987), included: tease, hit and kick at, poke, pounce, sneak up, carry child, play fight, pile on, chase, hold, push. Aggressive behavior was defined according to both topographical dimensions--hit with closed hand, frown, take, grab, push, fixate, swear at, insult (Blurton Jones, 1972; Humphreys & Smith, 1986)--and outcome dimensions--whether children separated after the act (Parke & Slaby, 1983). Children's R&T and aggression were observed according to an event sampling procedure. Observers described these behaviors into a tape recorder as they occurred. These events were subsequently transformed into time intervals of 5 seconds. The observer also noted behaviors preceding and succeeding the targeted behavior, who initiated the behavior, reactors, who was dominant and submissive, and whether the participants stayed together or separated after the target behavior occurred. Reliability was established by having the observers simultaneously conducting 50 scans and 10 event samples. To prevent reliability decay, checks and retraining occurred every month. Agreement for the scans was 87°7o and 81°70 for the event sampling.
Measures Sociometry. Children's sociometric status was determined following Dodge (e.g., Coie & Dodge, 1983; Dodge, Schlundt, Schocken, & Delugach, 1983). In this procedure each child was seated at a table before individual photos of all his or her classmates. Each child was then asked to nominate three peers whom he or she liked the most and three peers whom he or she liked the least. Following Coie and Dodge (1983), each child received a social preference score (n liked most, n liked least) and a social impact score (n liked, n liked least). The frequency of positive nominations and negative nominations was also scored. These measures of social preferences and impact were used because counting only number of "likes most" or "likes least" as measures of popularity and rejection, respectively, results in confounds of each (Coie & Dodge, 1983; Peery, 1979). Social Problem Solving. Spivack and Shure's Interpersonal Cognitive Problem Solving (ICPS) procedure was administered individually to all children. In this procedure an experimenter presented each child with five separate pictures of a child trying to get a toy from a peer and five separate pictures of a child trying to avoid being reprimanded by his or her mother. Each child was asked to generate as many different solutions as possible to each social problem. Experimenters followed scripts for each picture, and
Rough-and-Tumble Play in School
251
the number of prompts was standardized. Children's score on the ICPS was the variety of strategies they generated.
Antisocial Behavior Questionnaire. The Children's Behavior Questionnaire, developed by Rutter (1967), was given to teachers at each of the three levels. The questionnaire is comprised of 26 items on which a child is rated: 0 (doesn't apply), 1 (sometimes applies), or 2 (certainly applies). The questionnaire yielded an antisocial factor. Each child's mean score on this factor was calculated. Ability to Discriminate Between R& T and Aggression. Small groups o f children were shown a videotape containing 11 discrete incidents of children engaging in either R&T or aggressive behavior (Smith & Lewis, 1985). The first incident served as a practice item. Children were then shown the remaining 10 incidents and told to m a r k on a standardized answer sheet a " s a d face" if it was real fighting and a " h a p p y face" if it was play fighting. Children's score was the number of correct answers. Children as young as 4 years have been able to discriminate reliably between R&T and aggression on this film (Smith & Lewis, 1985). RESULTS The first objective of the study was to describe the occurrence o f elementary school children's R&T and aggression on the school playground during recess period. The frequency of occurrence of children's R&T and aggression were analyzed with separate grade (3: K, 2, 4) x sex (2) x location (3: grass, blacktop, playscape) ANOVAS. The means for these analyses are presented in Table 1. The ANOVA results are summarized in Table 2. The results for the R&T analysis suggest that boys engage in R&T more frequently than girls do. R&T typically occurred in soft, grassy surfaces and on the playscape more frequently than on the blacktop. The grade x location interaction for R&T further suggests that location differentially affected the R&T of kindergarten and f0urth-grade children. In kindergarten, significantly more R&T occurred on the playscape than on either the blacktop or in the grassy area. Fourth graders engaged in significantly more R&T on the grass than on either the playscape or the blacktop. The extent to which children's aggression varied by grade, sex, and location was also analyzed with a grade (3) x sex (2) x location (3) ANOVA. No significant main or interactive effects were observed for frequency o f occurrence of aggression or percentage of occurrence (i.e., proportion to all observed behaviors) of aggression. Aggressive behaviors accounted for only 0.3°7o o f children's playground behavior. When percentage of R&T behavior (i.e., proportion to all behavior observed) was examined with a grade × sex ANOVA, a significant sex effect
Table 1.
Means o f Children's R&T and Aggression" Girls
Boys R&T
Aggression
R&T
Aggression
2.52 .17 9.64
.17 .0 .11
.60 .60 2.73
.0 .0 .13
12.33
.28
3.93
.13
6.80 2.06 3.73
.06 .0 .13
.64 .14 3.57
.28 .0 .0
12.59
.19
4.35
.28
Grass Blacktop Playscape
17.75 .41 5.41
.0 .0 .41
9.60 .40 4.70
.10 .0 .2
Total
23.23
.41
Kindergarten Grass Blacktop Playscape Total
2nd Grade Grass Blacktop Playscape Total
4th Grade
14.7
.3
a Based on a total of 10,613 observations.
Table 2. A N O V A Summaries for Grade x Sex x Location Effects on Frequency o f R&T and Aggression DF
F
p
Contrasts
.07 .05 .001 .08 .001
ns B< G G = P G + P > BT a ns K: P > B T + G a 4: G > B T + P a ns ns
R&T Grade (G) Sex (S) Location (L) LxS GxL
2 1 2 1 4
GxS GxSxL
2 2
2.74 3.82 4.92 3.11 4.92 .001 .13
.99 .87
Aggression Grade (G) Sex (S) Location (L) LxS Gx L
2 1 2 1 4
.21 .10 2.61 .11 1.08
.81 .79 .07 .74 .3
GxS
2
.15
.86
GxSxL
2
.15
.86
ns ns ns ns ns ns ns
a B = b o y s , G = girls, K = kindergarten, 2 = 2 n d grade, 4 = 4th grade, P = playscape, G = grassy area, BT = blacktop.
252
Rough-and-Tumble Play in School
253
Table 3. Correlations Between R&T, Aggression, and Measures of Social Competence R&T
Social p r e f e r e n c e Social i m p a c t Likes m o s t L i k e s least ICPS a Antisocial Film
Frequency
R&T 07o
Boys
Girls
Boys
Girls
.30* .13 .22 - .0002 .30* - .04 .21
- . 10 .01 - .20 .08 - .03 .10 .17
.36** -.01 .18 - . 13 .31" .02 .06
- . 18 -.01 - . 16 .14 .01 .36** .17
Aggression Frequency
Social p r e f e r e n c e Social i m p a c t Likes most L i k e s least ICPS a Antisocial Film
Boys
Girls
Boys
A g g r e s s i o n 070 Girls
-.09 -.25 -.30* -.07 .16 .07 - .32*
-.10 -.01 -.006 .04 - .03 .10 - .03
-.05 -.25 -.28* -.10 .15 .01 - .21
.01 -.09 -.04 .06 .22 .01 .01
a Interpersonal Cognitive Problem Solving procedure. * p < .05. ** p < .01.
was also observed, F (1,92)= 4.72, p < .03; boys ( M = . 13) engaged in proportionally more R&T than girls did ( M = .08). These percentage data indicate that approximately 11°70 o f elementary school children's behavior on the playground is spent in R&T. In conclusion, R&T was affected by personal and contextual variables. Aggression was not affected by these variables, probably as a result o f its low occurrence. The second objective was to examine the extent to which R&T play episodes lead to aggressive episodes. In accord with procedures outlined by Bakeman & Gottman (1986), transitional probabilities o f R&T leading to aggression (lag 1), a two-state model, were calculated by converting the frequency o f occurrence to Z-scores. If Z's exceed + 1.96, the difference between observed and expected probabilities reached .05. The Z-score, .06, for the transition from R&T to aggression was not significant. The third objective was to examine the functional significance o f R&T. We examined this question in two ways. First, we examined the extent to which R&T events led to games-with-rules. Second, correlation coefficients between R&T and aggression and measures o f social competence were calculated. Regarding the transitional probability o f R&T leading to games-withrules (lag 1), a significant Z-score (5.63) was observed. These correlation coefficients are presented in Table 3.
254
Pellegrini
Generally, the results suggest that, for boys, R&T is positively correlated with measures of social competence, whereas aggression is negatively correlated with these measures. Girls' R&T was positively related to the antisocial measure. Further, frequency of aggression and R&T was not significantly intercorrelated for either boys (r = - . 13, p < .40) or girls (r = .01, p < .9). DISCUSSION
This study was concerned with describing elementary school children's R&T and possible benefits of engaging in this form of play. Such descriptive data are needed because of previous definitional problems of R&T vis-A-vis aggression. By describing the contexts in which R&T occur, we are extending previous behavioral descriptions in ecologically valid terms. Our results suggested that R&T and aggression occurred with different frequency. R&T accounted for 11°70 of children's playground behaviors, while aggression accounted for 0.3o70. Neither of these behaviors varied significantly with grade level. These results for R&T are consistent with the Humphreys and Smith (1984) results, which also found that in England R&T accounted for about 10°70 of elementary school children's playground behavior. The low occurrence of aggression (0.3070) is also consistent with the literature on aggression in school settings (Parke & Slaby, 1983). The reason for this may be partly due to the fact that aggression, particularly boys' aggression, was discouraged by teachers (Serbin, Connor, Burchardt, & Citron, 1979; Smith & Green, 1974). That school officials both on and off the playground may have been responsible for the low frequency of playground aggression is further supported by the fact that they notified children that fighting on the playground would result in suspension from school. Another related factor that represses the occurrence of aggression is low social density (McGrew, 1972; Parke & Slaby, 1983). In the present study children played on a spacious playground (approximately 44 square yards per child). Low social density and the school policy discouraging aggression were probably responsible for the low levels of aggression. The low levels of aggression, in turn, may have been responsible for the nonsignificant sex difference in observed aggression. The sex difference observed for R&T in this study replicates many other reports. Boys engaging in more R&T than girls do has been observed in England (e.g., Smith & Connolly, 1972), in America (e.g., DiPietro, 1981), and cross-culturally (e.g., Whiting & Edwards, 1973). The occurrence of R&T, not aggression, varied significantly according to children's location on the playground. Again, the low occurrence of aggression may have been responsible for the absence of location effects. That more R&T occurred on soft, grassy areas than on blacktop is commonsensical and replicates observations of English school children (Humphreys &
Rough-and-TumblePlay in School
255
Smith, 1987). Children engaged in R&T in soft areas probably because it minimized the possibility of injury. More R&T was observed on the contemporary playscape than on blacktop. This finding is similar to the observations of Haywood, Rothenberg, and Beasley (1974) in their comparative study of different playgrounds. In their observations of the behavior of children from many age groups on different types of playgrounds, they found that more play fighting occurred on the contemporary playscapes than on blacktop playgrounds. We found that the R&T on the playscape often involved children chasing each other around, through, and under the playscape structures. These location effects on children's R&T are important for at least two reasons. First, they fill a gap in the developmental and educational literature describing children's outdoor play. To date, the descriptions of elementary school children's behavior in different outdoor play areas have often not included R&T or have used observation systems more appropriate for preschool children (e.g., Frost & Sunderlin, 1985). The second point of interest about the location effects relates to the generalizability of our results. That R&T varied significantly by location should caution researchers from generalizing their results to other, nonsimilar, contexts. Cross-study comparisons of R&T should also be interpreted with these context effects in mind. The lack of age effects on children's R&T is inconsistent with the one other observation of elementary school children (Humphreys & Smith, 1984, 1987). This inconsistency may be due to nationality differences; our sample was American and that of Humphreys and Smith was English. The inconsistency may also have been due to the different playgrounds on which the children were observed. In the present study children had free access to a contemporary playscape, a grassy wooded area, and a large blacktop area. In the Humphreys and Smith study (1986) younger children (to 9 years of age) played on a blacktop area, while older children (I0 to 12 years of age) played on a separate area comprised of both blacktop and grassy areas and fields. Thus age and play environment seem to have been confounded. In order to establish clear age effects on R&T during the elementary school years, more studies are needed in which children have access to similar play environments. In summary, the first objective of this study was to describe both R&T and aggression with the intent of shedding light on their interrelation. Where R&T accounted for about 11% of children's behavior and covaried systematically according to sex and context, aggression rarely occurred and did not vary according to these variables. In light of these different patterns of occurrence and the nonsignificant correlation between R&T and aggression, it must be concluded that R&T and aggression represent two distinct categories during the elementary school years. This conclusion is corroborated by topographical (e.g., Blurton Jones, 1972) and outcome differences (Humphreys & Smith, 1987) between children's R&T and aggression.
256
Pellegrini
The second objective is related to the first: We examined the probability of R&T moving into aggression. The motivation behind this question was Neil's (1976) research suggesting that children's R&T turns from friendly encounters to aggression. This question has been indirectly answered by Smith and his colleagues in their findings that social interaction continues after the initiation of R&T (Humphreys & Smith, 1984). We tested the probability of R&T events moving into aggressive events by using sequential lag analyses, at lag 1, with a two-state model. The probability of R&T moving into aggression was not significant. Taken together with the results from objective one, our data suggest that observers should consider R&T and aggression as distinct categories. The third objective of the study was to examine the functional significance of R&T in elementary school children. We defined functional significance in a broad sense, wherein a behavior has beneficial consequences (Hinde, 1982). We hypothesized that R&T would be beneficial in children's development of social competence. First, it was hypothesized that R&T would lead to children's engaging in cooperative games-with-rules. To test this hypothesis, we examined the transitional probability, at lag 1, of R&T leading into games-with-rules; the probability was significant. The similarity of the motor patterns involved in both suggests that R&T provides practice for gameswith-rules and that the two often merge. For example, chasing behavior, a component of R&T, and tag both involve running and dodging. Chasing behaviors, in turn, often turned into games of tag. Thus R&T may be functional to the extent that it provides physical skills practice and results in sustained cooperative games-with-rules. We also expected that R&T would be related to other aspects of social competence. Granted that the correlational data to be discussed do not provide a causal explanation for the role of R&T in children's development of social competence, but they do provide exploratory data on this neglected area of child development. The correlational data suggest that boys' R&T is positively correlated with social problem solving and social preference, while girls' R&T is positively correlated with their being considered antisocial by their teachers. The social problem-solving instrument used in this study measured children's ability to provide a variety of responses to hypothetical social problems (Spivack & Shure, 1974). A high score on this measure suggests that children have and use a large repertoire of strategies to negotiate and solve problems; it is a measure of behavioral flexibility. Relations between play generally (e.g., Fagen, 1984) and R&T specifically (Humphreys & Smith, 1984) and social flexibility have been suggested because the R&T of elementary school boys often involves social skills, such as negotiation, multiparty alliances, and redefining situations (Humphreys & Smith, 1984), which are related to flexibility in social problem solving. Thus R&T may serve the
Rough-and-Tumble Playin School
257
function of helping boys develop and practice a varied repertoire of social problem-solving skills. The positive relation between boys' R&T and social preference supports the notion that R&T serves a social skills development function (Smith & Lewis, 1985). Our results, which replicate and extend the Smith and Lewis findings with preschoolers, suggest that popular boys often engage in R&T. They do so to have fun and to interact cooperatively with peers. Popular boys are better at the latter than are rejected or withdrawn boys. As such, popular boys are more likely to engage in R&T. In R&T they practice and develop social skills that are related to their popularity. It may be, as Humphreys and Smith (1987) suggest, that R&T serves a social affiliation function in that it tends to occur among popular boys who are also friends. Boys' aggressive behavior was negatively correlated with popularity (i.e., the number of "likes most" nominations). Boys' aggressive behavior, in turn, was independent of R&T. Aggressive boys may not engage in R&T because they often attribute aggressive intent to such episodes (Smith & Lewis, 1985). Our data support this conclusion in that boys' ability to discriminate R&T from real fighting was negatively correlated with aggression. Thus popular boys would be more likely to engage in R&T with each other because they realize that certain boys, for example, rejected children, may respond to R&T overtures with aggression. To summarize the results from objective three thus far, R&T seems to serve a social skills function for boys, whereas aggression does not. Aggressive behavior, not R&T, was negatively correlated with popularity. This finding replicates the work of Smith and his colleagues (Humphreys & Smith, 1987; Smith & Lewis, 1985), but it is counter to another body of research (e.g., Ladd, 1983). This inconsistency, again, may be due to the confounding of R&T with aggression. Girls' engagement in R&T was positively correlated with teachers' antisocial ratings. As noted above, girls do not engage in R&T as frequently as boys do. Indeed, girls who engage in R&T or similar behaviors are often discouraged from doing so by adults (Fagot, 1978) and are negatively labeled by adults (Huston, 1983). As such, the girls who did engage in R&T may have been labeled by teachers as antisocial because they engaged in a maleoriented activity. Boys who engaged in R&T may not have been labeled antisocial because teachers saw R&T as a sex-role appropriate activity. In summary, results from objective two suggest that R&T may play a positive role in the development of boys' social competence. The physical and social skills used in R&T become useful for boys while interacting with their peers. The positive correlates of R&T are contrasted with the negative correlates of aggression. This difference corroborates objective one data, which suggested that R&T and aggression are distinct behavioral forms. As such, researchers should be aware of the differences between R&T and aggression.
258
Pellegrini
REFERENCES Altman, J. (1974). Observational study of behavior: Sampling methods. Behavior, 49, 227-265. Bakeman, R., & Gottman, J. (1986). Observing interaction: An introduction to sequential analysis. New York: Cambridge University Press. Blurton Jones, N. (1972). Categories of child-child interaction. In N. Blurton Jones (Ed.), Ethological studies o f chiM behavior (pp. 97-129). London: Cambridge University Press. Boulton, M., & Smith, P.C. (1989). Issues in the study of children's rough-and-tumble play. In M. Bloch & A. Pellegrini (Eds.), The ecological context o f children's play (pp. 57-83). Norwood, N J: Ablex. Coie, J., & Dodge, K. (1983). Continuities and changes in children's social status: A five-year longitudinal study. Merrill-Pahner Quarterly, 29, 261-287. DiPietro, J. (1981). Rough-and-tumble play: A function of gender. Developmental Psycholog.v, 17, 50-58. Dodge, K., & Frame, C. (1982). Social cognitive biases and deficits in aggressive boys. Child Development, 53, 620-635. Dodge, K., Schlundt, P., Schocken, I., & Delugach, J. (1983). Social competence and children's sociometric status: The role of peer group entry strategies. Merrill-Palmer Quarterly, 29, 309-336. Fagen, R. (1984). Play and behavioral flexibility. In P.K. Smith (Ed.), Play in animals and humans (pp. 159-174). London: Blackwell. Fagot, B. (1978). The influence of sex of child and parental reactions to toddler children. Child Development, 49, 554-558. Frost, J., & Sunderlin, S. (Eds.) (1985). When children play. Wheaton, MD: Association for Childhood Education International. Glickman, C. (1984). Play in public school settings: A philosophical question. In T. Yawkey & A. Pellegrini (Eds.), Child's play (pp. 255-272). Hillsdale, N J: Erlbaum. Harlow, H. (1962). The heterosexual affective system in monkeys. American Psychologist, 17, 1-9. Haywood, G., Rothenberg, M., & Beasley, R. (1974). Children's play and urban playground environments: A comparison of traditional, contemporary, and adventure types. Environment and Behavior, 6, 131-168. Hinde, R. (1982). Ethology. London: Fontana. Hole, G., & Einon, D. (1984). Play in rodents. In P. Smith (Ed.), Play in animals and humans (pp. 119-146). London: Blackwell. Humphreys, A., & Smith, P. (1984). Rough-and-tumble in preschool and playground. In P. Smith (Ed.), Play in animals and humans (pp. 291-270). London: Blackwell. Humphreys, A., & Smith, P. (1987). Rough-and-tumble play, friendship and dominance in school children: Evidence for continuity and change with age. Child Development, 58, 201-212. Huston, A. (1983). Sex typing. In E. Hetherington (Ed.), Handbook o f child psycholog.v: Socialization, personality, and social development (Vol. 4, pp. 387-468). New York: Wiley. Ladd, G. (1983). Social networks of popular, average, and rejected children in school settings. MerrilI-Pahner Quarterly, 29, 283-307. Maccoby, E., & Jacklin, C. (1974). Thepsvchology o f sex differences. Stanford, CA: Stanford University Press. McCune-Nicholich, L. (1981). Toward symbolic functioning: Structure of early pretend games and potential parallel with language. Child Development, 52, 785-797. McGrew, W. (1972). An ethological study o f children's behavior. London: Methuen.
Rough-and-TumblePlay in School
259
Neil, S. (1976). Aggressive and non-aggressive fighting in twelve-to-thirteen year old preadolescent boys. Journal of Child Psychology and Psvchiatr.v, 17, 213-220. Olweus, D. (1979). Stability and aggressive reaction patterns in males: A review. Psychological Bulletin, 86, 852-875. Parke, R., & Slaby, R. (1983). The development of aggression. In E. Hetherington fed.), Handbook of child psychology: Socialization, personalit.v, and social development (Vol. 4, pp. 547-642). New York: Wiley. Peery, J. (1979). Popular, amiable, isolated, rejected: A reconceptualization of sociometric status in preschool children. Child Development, 50, 1231-1234. Pellegrini, A. (1987). Rough-and-tumble play: Developmental and educational significance. Educational Psychologist, 22, 23-43. Piaget, J. (1962). Play, dreams, and imitations. New York: Norton. Potts, R., Huston, A., & Wright, J. (1986). The effects of television form and violent content on boys attention and social behavior. Journal of Experimental Child Psychology, 41, 1-17. Rubin, K., Fein, G., & Vandenberg, B. (1983). Play. In E.M. Hetherington fEd.), Handbook of child psychology: Socialization, personality, and social development (Vol. 4, pp. 693-779). New York: Wiley. Rubin, K., Maioni, T., & Hornung, M. (1976). Free play in middle and lower class preschoolers: Parten and Piaget revisited. Child Development, 47, 414--419. Rutter, M. (1967). A children's behavior questionnaire for completion by teachers: Preliminary findings. Journal of Child Psychology and Psychiatry, 8, 1-11. Serbin, L., Connor, J., Burchardt, C., & Citron, C. (1979). Effects of peer presence on sextyping of children's play behavior. Journal of Experimental Child Psychology, 27, 303-309. Shure, M. (1985). Inter-personal problem solving: A cognitive approach to behavior. In R. Hinde, A.N. Perret-Clermont, & J.S. Hinde fEds.), Social relationships and cognitive development (pp. 191-207). Oxford: Clarendon Press. Smith, P. (1982). Does play matter? Functional and evolutionary aspects of animal and human play. The Behavioral and Brain Sciences, 5, 139-184. Smith, P., & Connolly, K. (1980). Patterns of play and social interaction in pre-school children. In N. Blurton Jones fEd.), Ethological studies in child behavior (pp. 65-96). New York: Cambridge University Press. Smith, P., & Green, M. (1974). Aggressive behavior in English nurseries and play groups: Sex differences and response of adults. Child Development, 45, 211-214. Smith, P., & Lewis, K. (1985). Rough-and-tumble play, fighting, and chasing in nursery school children. Ethology and Sociobiology, 6, 175-181. Spivack, G., & Shure, M. (1974). Social adjustment of young children. San Francisco: JosseyBass. Vygotsky, K. (1967). Play and its role in the mental development of the child. Soviet Psychology, 12, 62-76. Whiting, B., & Edwards, C. (1973). A cross-cultural analysis of sex-differences in the behavior of children age three through I 1. Journal of Social Psychology, 91, 171-188.
260
Pellegrini
A P P E N D I X : I N V E N T O R Y OF CHILDREN'S PLA YGROUND B E HAVI OR Passive/noninteractive sit stand lie eat watch person look at place sedentary wait turn Passive interact talk/adult talk/peer contact/comfort hug groom walk/look dress Observer directed Attend to observer Adult organizer Official game Aggressive hit with closed hand frown fixate take grab push swear at insult
Distress cry/sob R&T tease hit/kick poke pounce chase hold/grab pull/push carry sneak-up play fight Locomotion walk run skip/hop walk/follow balance jump slide swing climb roll/spin piggy-back dance
Games-with-rules jump rope compete tag clap/song ballgame catch follow leader
Object play throw object active quiet small motor large motor active