Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Developmental Psychology

1978, Vol. 14, No. 5,517-523

A Longitudinal Study of Social Participation in


Preschool Children: Solitary and Parallel Play
Reexamined
PETER K. SMITH
University of Sheffield
Sheffield, England

Longitudinal observations were made over a 9-month period of the social participa-
tion of 48 preschool children. Group play increased and solitary play decreased
during the period, while parallelplay did not vary much in overall occurrence. Some
2-year-olds went through successive stages of predominantly solitary, then parallel,
then group play, but many others did not. Some 3- and 4-year-olds alternated
between periods of predominantly group play and periods of predominantly solitary
play. The results are discussed in relation to the relative maturity of solitary and
parallel play and the usefulness of a social participation index at this age range.

Mildred Parten's (Fatten, 1932; Parten & ery school. Nor has the utility of a social
Newhall, 1943) study of the social participa- participation index been much examined.
tion of preschool children has become one of Particularly puzzling is the nonsignificant
the classics of tBfc 1930s era of child psy- correlation of .12 that Parten obtained be-
chology. Parten introduced six categories of tween social participation and nursery ex-
participation in play behavior. In particular, perience. This finding might suggest that a
she used the term parallel play to indicate a longitudinal study of social participation
limited degree of participation, in which would give different results from the cross-
children play near each other, with similar sectional data she presented.
materials, but do not engage in substantial Parten's own cross-sectional data show
interaction. Parten regarded parallel play as some decline in solitary behavior with age
"more socialized" than solitary or onlooker and also some decline in parallel play, with
behavior but less socialized than associative an increase in both associative and coopera-
or cooperative group play. By appropriately tive group play. Somewhat similar cross-
weighting her categories, she obtained sectional data were obtained by Barnes
a composite "social participation" index, (1971), these data also being from one nurs-
which correlated .61 with age. ery group. However, the trends for solitary
The importance and nature of parallel play and parallel play are small, and for parallel
has been subjected to surprisingly little criti- play nonlinear with age, in both studies.
cal Scrutiny, despite the fact that Parten's Some support for the utility of a social
observations were based on only one nurs- participation index comes from work by
Smith and Connolly (1972) and Smith
The facilities on which this research is based were
supported financially by Grant 1414/2 from the Social (1973). These studies employed principal-
Science Research Council, London. components analyses to find main dimen-
The author is grateful to Robert Hinde and Rosemary sions of individual difference in children,
Roper for their comments on an earlier draft of the based on frequencies of observed behaviors.
manuscript. In both cases a main component of "social
Requests for reprints should be sent to Peter K.
Smith, Department of Psychology, University of maturity" was obtained. Smith and Con-
Sheffield, Sheffield, S10 2TN, England. nolly (1972) found group play to be loaded
Copyright 1978 by the American Psychological Association, Inc. OOI2-1649/78/l405-05l7$00.75

517
518 PETER K. SMITH

positively, parallel play negligibly, and sol- in an intermediate position. However, they
itary play negatively on this component. do not particularly support the contention of
Smith (1973) in a separate study found that a Moore et al. (1974) and of Rubin et al. (1976)
social participation index (weighting group that parallel play is less mature than solitary
play as 1, parallel play as 0, solitary play as play. Roper and Hinde's first factor shows
-1) correlated .81 with age and was highly no appreciable age correlation, although
loaded on the main social maturity compo- negative correlations with age were ob-
nent. These studies were based on 4 day tained for the second and third factors.
nurseries in England. From these contradictory findings, one
Some recent work, however, has sug- consensual implication might be drawn.
gested that parallel play should not necessar- This is that whether a child is occupied at a
ily be regarded as an intermediate form of task or not, is conceptually and empirically
socialization between solitary and group largely separate from whether he or she is in
play; and this in turn would question the use proximity or interaction with other children
of a social participation index, as used by or not. There is no dispute that a preschool
Fatten (1932) and Smith (1973). Moore, child can be engaged alone on a constructive
Everton, and Brophy (1974) suggest that sol- task. However, the fact of being engaged in
itary play need not be an indicator of poor task activity may not tell us anything about
social adjustment. This is based on the ob- that child's social maturity. Indeed, as Jen-
servation that much solitary play is occupied nings (1975) has indicated, people and ob-
in goal-directed activities, large muscle ject orientation may correspond to fairly
play, or educational play; less than 16.5% separate abilities in preschool children.
was classified as onlooking, daydreaming, Similarly, the existence of Roper and
sulking, or seeking the teacher. However, Hinde's separate third factor (unoccupied or
Moore et al. were studying kindergarten not) contrasts with their first two factors
rather than nursery school children, and in relating to social abilities (proximity to and
the latter the proportion of solitary behavior interaction with others). From this view-
that is unoccupied is considerably higher point, Parten's original scheme can be seen
(see Table 1). Rubin, Maioni, and Hornung to have compounded purely social participa-
(1976) found that about 48% of solitary play tion categories with task-related categories
in preschool was onlooker or unoccupied such as "unoccupied." It may well be pref-
behavior. Nevertheless, Rubin et al. also erable to limit social participation levels to
argue that parallel and not solitary play is solitary, parallel, and group (and sublevels
"the least mature level of a social cognitive of these categories). This has been done in
play hierarchy for 3- and 4-year-olds" (p. the present review and research. With the
418). data in Table 1, "solitary" includes "unoc-
The principal-components analysis jus- cupied" and "onlooker" of Parten's catego-
tification of a social participation index has ries, on the assumption that the vast ma-
also been reevaluated in recent work by jority of such instances involve neither close
Roper and Hinde (in press). These authors proximity with similar materials (parallel)
observed 3- and 4-year-olds in two English nor substantial interaction (group).
nursery classes. Although a principal- Even given this simplification, the evi-
components analysis gave a first component dence for the intermediate position of paral-
somewhat similar to that of Smith (1973), a lel play, so far as social abilities are con-
further rotated factor-analytic solution re- cerned, remains poorly substantiated at
vealed what these authors interpret as a best. Critically lacking is any longitudinal
breakdown of this first component into three evidence that children do in fact proceed
subfactors. These were (a) group play ver- through sequential stages of solitary, paral-
sus parallel play, or whether given proximity lel, and group behavior, as is suggested in
children were interacting; (b) solitary play, many texts of early childhood education and
or whether a child was in proximity to others play. The present report provides longitudi-
or not; and (c) unoccupied behavior. These nal data on social participation and suggests
findings cast doubt on the utility of a linear some reasons for the discrepant findings in
social participation index with parallel play the earlier research referred to above.
SOCIAL PARTICIPATION IN PRESCHOOL CHILDREN 519

Method changing objects, or in the organization of a game. This


category subsumes Parten's categories of associative
and cooperative play.
Subjects
Parallel. The focal child has one or more other children
Two groups of preschool children were observed who are in close proximity to him or her and are en-
over a 9-month period, from October 1971 to June 1972, gaged in similar behaviors. However, these compan-
at a playgroup in Sheffield, England. Each group met ions do not interact with the focal child or they do so
for two mornings each week during school terms. The only fleetingly; their presence does not substantially
two groups met in the same premises, with the same affect his or her behavior. This category corresponds to
play-group staff and equipment, but at different times; Parten's category of parallel play.
subjects for the two groups were from the same popula-
tion. Thus results for the two groups provide, in Lyk- Solitary. The child has no companion in group activity
ken's (1968) terminology, an "inbuilt" literal replica- or parallel behavior. This category subsumes Parten's
tion. categories of solitary play and onlooker activity and
The two groups were started at the end of September would usually include her category of unoccupied be-
1971 by recruiting children from the local neigh- havior.
borhood. Social background was varied, mainly from
skilled working or middle classes. The great majority of Adult. The child is in group or parallel activity with an
the children had not had any previous nursery experi- adult.
ence at other nurseries or play groups.
Each group contained 12 boys and 12 girls, matched These four categories were comprehensive and
for age, with the age range on entry from 28 to 47 mutually exclusive; one category was scored at the end
months (M = 35 months) in the first group (Group 1) of each sample period.
and from 28 to 48 months (M = 36 months) in the second
group (Group 2). Half of the children were over 33
months (older), and half were 33 months or under Nature of Task Activity Measures
(younger).
Occupied. The child is engaged in active use of some
object or apparatus in the environment, or in gross
Procedure motor activity, or in substantial social interaction.
Using time-sampling methods, the children's behav-
Unoccupied. The child is not occupied as defined above.
ior was observed each morning session through the Generally, this includes wandering around, just watch-
9-month period by one observer. The main purpose of ing other children,just holding, or repetitively fumbling
the study was to examine the effects of varied an object.
amounts of floor space, and play equipment, on the
children's behavior (Smith, 1974; Smith & Connolly,
1977). An independent factor in the design was time Interobserver Agreements
through the project. The observational data were ob-
tained in six consecutive time blocks, of about 5 weeks Simultaneous recording with a second observer, ex-
each; each time block contained environmental varia- perienced in child watching and given the above defini-
tions as noted above, but separate time blocks were tions but no specific further training, gave interobserver
equivalent in this respect. agreements (no. agreements/(no. agreements + mean
At each session, children's behavior was observed no. disagreements) of .88 for group, .76 for parallel, .85
and recorded in short focal-child samples of 40-sec du- for solitary, and .94 for adult. Judgment as to whether a
ration. Four such samples were obtained for each child child was occupied or unoccupied gave a concordance
at each session, with at least a 10-min separation (usu- of .94.
ally more) between samples on any one child. Supple- These concordances were based on 100 simultaneous
mentary samples were obtained at the next session in samples by the two observers. As a check on the time
cases of absence (not very frequent; mean attendance stability of the principal observer, 50 samples were
was 91% in Group 1 and 90% in Group 2). Each time made prior to the start of the main study and another 50
block contained 10 sessions for each group, which at the midpoint. Concordances were substantially the
therefore yielded 40 independent samples per child. same at these two occasions.
At the end of each sample period of 40 sec, the nature
of the child's social participation and activity choice for
the majority of the sample period was recorded by the Results
observer. The categories relevant to this article are
described below, In analyzing the results, samples in which
the child was with an adult were ignored, as
generally this involved some constraint on
Nature of Social Behavior Measures the child's behavior. For the remaining
Group. The focal child has one or more other children samples, the percentages of solitary, parallel
who interact substantially with him or her in the nature and group behavior are shown in Table 1, in
of the activity, either visually, verbally, through ex- comparison with results from Parten (1932),
520 PETER K. SMITH

Table 1: Percentages of Solitary, Parallel, and Group Activity, and of Unoccupied/Onlooker as a


Percentage of Solitary, in Previous and Present Studies on Nursery-School-Aged Children
Rubin et al. Present
Barnes (1976) study
Activity Parten
measure (1932) (1971) MC LC Group 1 Group 2
Solitary 29 51 32 35 39 35
Parallel 32 24 29 37 23 27
Group 38 25 39 28 37 38
Unoccupied/solitary 40 47 54 43 37 42
Note. MC = middle class; LC = lower class.

Barnes (1971), and Rubin et al. (1976). The Individual Sequences


figures are similar, although not supporting Results for each child in each time block
Barnes' claim for a decrease in social par- were considered separately and assigned a
ticipation in the 45 years since Parten's data code of G (group), P (parallel), or S (soli-
were gathered.
tary), according to which category had the
When unoccupied was scored, the child
highest total. A few cases of ties (10 out of
was almost invariably solitary, although a 288) were resolved in favor of the "higher"
few cases of parallel behavior did occur. category. Each code thus shows the most
Unoccupied as a percentage of solitary be- common form of social behavior for a par-
havior is also not greatly different from the ticular child during one 5-week time block.
other nursery school studies, all being much The child sequences were put into six dif-
larger than the 16.5% maximum figure that ferent types, shown in Table 3. Children are
Moore et al. (1974) obtained for kindergar- coded by group (1 and 2), older or younger
ten children. age (O or Y), boy or girl (B and G), and a
number from 1 to 6.
The types and subtypes are listed in order
Overall Time Changes
of age means. Type A children started in G.
The successive totals for each category in There is little difference between Al chil-
each time block are shown for both groups of dren who always stayed in G and A2 chil-
children in Table 2. There is a consistent dren who alternated G with some blocks of
increase in group behavior and decrease in S.
solitary and unoccupied behavior, while Type B children started in S (Bl and B2)
parallel behavior shows no significant varia- or P (B3) and made a transition to G at some
tions. However, these overall trends ob- point. The 11 B1 children went direct from S
scure the actual changes in the nature of to G, whereas the 6 B2 children went
primary social behavior in individual chil- through from S to G with P as an inter-
dren. mediary stage.
Table 2: Category Totals in Successive Time Blocks and Significance Levels
from Analysis of Variance
Time Sequence
Measure and group
Group
1 192 263 294 317 322 390 .001
2 159 183 290 366 432 378 .001
Parallel
1 197 188 192 157 179 185 ns
2 261 223 199 212 220 198 ns
Solitary
1 356 342 308 317 318 235 .001
2 360 376 288 243 205 222 .001
Unoccupied
1 169 132 78 87 99 65 .001
2 184 162 86 74 73 67 .001
SOCIAL PARTICIPATION IN PRESCHOOL CHILDREN 521

Table 3: Types of Sequence of Predominant Social Activity for Individual Children over
Six Successive Time Blocks

Type and M age M unoccupied/ Time sequence


subtype (in mo.) solitary (%) Children T, T2 T3 T4 T5 T6
A (« = 13)"
Al 40.6 31.7 10B2, 1OB4, 1OG1 G G G G G G
1OG4, 1OG6, 20B1 G G G G G G
2OB4, 2OB5, 20G1 G G G G G G
A2 39.5 33.3 1OG3, 2OB2 G S G G G G
1YG3 G G G S G G
10G2 G S G S G G
B (n = 24)"
Bl 34.6 31.3 1YB4, 2YB3 S S G G G G
20B3 s S S G G G
1OB6, 1OG5 S S S S G G
1OB3, 1YG4 s s s S S G
1YB1 s G s G G G
20G3 s S s G G S
2OB6 s s s S G S
1YG5 s s s G G P
B2 34.3 48.5 2OG2, 2YG4 s p G G G G
IYB5 s s P G G G
1YG2 s s S P G G
2YB4 s s S S P G
1OB5 s s p S G G
B3 33.3 67.9 2YG2, 2YG5, 2YG6 p p p G G G
2YB2 p G p G G P
2OG3 p S G G G G
c
2OG6, 2YG1 p s S S G S
C(n = ll)
Cl 33.2 40.8 1OB1, 1YB2, 1YB3 s s s S S S
1YB6, 1YG1, 1YG6 s s s S S S
2OG4, 2YB1, 2YB6 s s s s S s
C2 31.0 31.5 2YG3 s s p s s s
2YB5 p s p p s p
Note. Children are coded, in order, by group ( l a n d 2), olderor younger (O and Y), boy orgirl (Band O). and a number from 1 to 6. Codes to which the
children were assigned in successive time blocks on the basis of predominant social behavior are as follows: G (group), S (solitary), and P (parallel).
Children IOG6 and 2YB3 left shortly before the study ended. Their T, data were obtained during Ts.
" M age = 40.2 mo.; mean unoccupied/solitary = 32.2%; starts in G.
" M age = 34.2 mo.; mean unoccupied/solitary = 46.3%; makes transition to G.
' M age = 32.8 mo.; mean unoccupied/solitary = 39,1%; makes no transition to G.

Type C children failed to make a transitionchildren (34-48 months) and all 24 younger
children (28-33 months). The older children
to G. They either stayed in S all the time (Cl)
or varied between S and P (C2). most often stay in G, next most often stay in
Type A children are significantly older S, and next most often transfer between the
than Type B children, [7(13, 24) = 56.0,p < two states. Transitions to P are very rare.
.01, and Type C children, [7(13, 11) = 21.0, The younger children most often stay in S,
p < .01. Type B and C children do not dif- next most often stay in G, and are approxi-
fer significantly in age, C/(24, 11) = 105.0, mately equally likely to move from S to P, P
ns. to G, or S to G.
Unoccupied as a percentage of solitary As parallel is overall the least likely qf the
behavior does not differ significantly be- three social behavior categories (see Table
tween any of the three main types on 1), it is possible that its apparent relative
Mann-Whitney tests. Individual children unimportance (see Table 3 and Figure 1) is
showed a lot of variation in this statistic. exaggerated when coding each time block as
S, P, or G. Perhaps even for Type Bl chil-
Transition Probabilities Between States
dren, for example, there is an increase in
parallel before the child shows predomi-
Figure 1 shows the transition probabilities nantly group activity, and a decrease after-
between S, P, and G, from one time block to wards, even though parallel is never the most
the next, summed separately for all 24 older likely category for a substantial time period.
522 PETER K. SMITH

older younger

.01

.06
Figure 1. Transition probabilities between states of social behavior for older and
younger preschool children.

These possibilities were examined for the 11 Eleven children started predominantly in
Type Bl children. solitary behavior, and at some point they
First, the two successive S states before changed directly to predominantly group
the transition to G were examined to see behavior. No substantial period of predomi-
whether parallel showed an increase. Out of nantly parallel behavior intervened, nor was
10 cases, 4 showed an increase and 6 a de- there evidence for any nonrandom change in
crease. parallel behavior during the transition pe-
Second, the two successive G states after riod. One sequence for social involvement
the transition from S were examined to see therefore seems to be to go fairly directly
whether parallel showed a decrease. Out of from playing alone to playing with others.
eight cases, 4 showed a decrease and 4 an This seems to be a preferred sequence for
increase. older children (see Figure 1). It is also fol-
In both cases, there is no evidence for any lowed by some younger children: 5 of the 11
systematic relation between the amount of were under 33 months old on entry.
parallel activity and the transition from S to However, six children did show a transi-
G. tion from solitary, through parallel, to group
behavior. Four of the six were under 33
Discussion months old on entry. This seems to be an
alternative sequence for the 2-year-olds or
The results point clearly to differences in younger three-year-olds, and it corresponds
the sequences of social participation in chil- to the "classic" view of social participation,
dren over time, only partly attributable to although it is not the most common in this
chronological age. Thirteen children, mainly sample.
older (11 were over 3 years old) went more Five children went directly from parallel
or less directly into predominantly group to group behavior, and three more did so with
play with companions. Of these, 9 remained bouts of solitary behavior intervening. Fi-
this way continuously, but 4 alternated sub- nally, 11 children either stayed in predomi-
stantial time blocks of predominantly soli- nantly solitary behavior, or (two children)
tary behavior. These children were no alternated with bouts of parallel behavior.
younger than the others, and their solitary Although older children generally showed
play was no more or less occupied. Proba- less solitary behavior, the percentage of un-
bly, these bouts of solitary behavior re- occupied activity in solitary behavior was
flected changing friendship preferences, not significantly different. Variations be-
without indicating immaturity on the part of tween individual children outweighed varia-
the children concerned. tions between types. This supports the no-
SOCIAL PARTICIPATION IN PRESCHOOL CHILDREN 523

tion that the extent to which a child may press; Rubin et al., 1976) all the children
occupy him- or herself in an activity is, to were 3 years old or over.
some extent at least, independent of his or It would be worthwhile to investigate fur-
her social participation level. ther the different sequences shown by chil-
In summary, parallel behavior is found dren as they become able to play a lot with
throughout the preschool period, in 2- to other children. Why do some 2-year-olds
5-year-old children. It does decrease with seem to rely on parallel play and others not?
age, but not very substantially (cf. Barnes, This will require more detailed longitudinal
1971; Parten, 1932). However it is usually studies at this age range as well as further
only found as a predominant behavior in knowledge of factors outside the immediate
2-year-olds, or younger 3-year-olds, and in nursery situation.
such circumstances it does often precede a
period of predominantly group behavior. REFERENCES
Most 3- and 4-year-olds move from mainly Barnes, K. E. Preschool play norms: A replication.
solitary to mainly group behavior directly. Developmental Psychology, 1971, 5, 99-103.
Thus a phase of predominantly parallel be- Jennings, K. D. People versus object orientation, social
havior is a sign of a younger or perhaps a less behavior, and intellectual abilities in preschool chil-
dren. Developmental Psychology, 1975,//, 511-519.
mature child. To this extent, Rubin et al. Lykken, D. T. Statistical significance in psychological
(1976) are right to question the social matu- research. Psychological Bulletin, 1968,70, 151-159.
rity of parallel play. Moore, N. V., Everton, C. M., & Brophy, J. E. Solitary
Solitary behavior differs in that it does play: Some functional reconsiderations. Develop-
decrease noticeably with age through the mental Psychology, 1974, 10, 830-834.
Parten, M. B. Social participation among preschool
preschool period (cf. Barnes, 1971; Parten, children. Journal of Abnormal and Social Psychol-
1932). Although less frequent in older chil- ogy, 1932, 27, 243-269.
dren, however, some 3- and 4-year-olds do Parten, M., & Newhall, S. M. Social behavior of pre-
show periods of predominantly solitary be- school children. In R. G. Barker, J. S. Kounin, & H.
F. Wright (Eds.), Child behavior and development,
havior, alternating with predominantly New York: McGraw-Hill, 1943.
group behavior. In this sense, solitary play Roper, R., & Hinde, R. A. Social behavior in a nursery
can be seen perhaps as a mature coping be- school: Consistency and complexity. Child Devel-
havior, in children who would not now show opment, in press.
predominantly parallel play. However, sol- Rubin, K. H., Maioni, T. L., & Hornung, M. Free play
behaviors in middle- and lower-class preschoolers:
itary play seems to be an "option" for these Parten and Piaget revisited. Child Development,
older children, in a way in which it is not for 1976,47, 414-419.
the younger ones (see Table 3). Smith, P. K. Temporal clusters and individual differ-
The results do question the usefulness of a ences in the behavior of preschool children. In R. P.
Michael & J. H. Crook (Eds.), Comparative ecology
social participation index in which parallel is and behaviour of primates. New York and London:
weighted intermediately between solitary Academic Press, 1973.
and group. At best, it would be a reasonable Smith, P. K. Social and situational determinants of fear
approximation for 2-year-olds and younger in the playgroup. In M. Lewis & L. A. Rosenblum
3-year-olds. For older 3-year-olds and (Eds.), The origins of fear. New York: Wiley, 1974.
Smith, P. K., & Connolly, K. Patterns of play and social
4-year-olds, it is a much less good approxi- interaction in preschool children. I n N , Blurton Jones
mation. It is probably not coincidental that (Ed.), Ethological studies of child behaviour. Cam-
the earlier studies using social participation bridge, England: Cambridge University Press, 1972.
indexes (Parten, 1932; Smith, 1973; Smith & Smith, P. K., & Connolly, K. J. Social and aggressive
behavior in preschool children as a function of crowd-
Connolly, 1972) had many 2-year-olds in the ing. Social Science Information, 1977,16, 601-620.
sample, whereas in the more critical studies
(Moore et al., 1974; Roper & Hinde, in (Received August 29, 1977)

You might also like