Professional Documents
Culture Documents
1998 CoplanRubin PPBS
1998 CoplanRubin PPBS
See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Exploring and Assessing Nonsocial Play in the
Preschool: The Development and Validation of
the Preschool Play Behavior Scale
Robert J. Coplan, Carleton University and Kenneth H. Rubin,
University of Maryland
Abstract
Recently, researchers have investigated the different structural forms of young chil-
dren’s nonsocial play behaviors. The primary goal of this research was to develop and
validate the Preschool Play Behavior Scale (PPBS), a teacher rating scale designed to
assess the multiple forms of young children’s solitary behaviors. In this regard, two
studies were conducted. In Study 1, 39 preschool children were observed, and their
social and nonsocial behaviors recorded over a three month period. Teachers then com-
pleted the PPBS. Results supported the construct validity of the PPBS; teacher-rated
and observed nonsocial and social behaviors were significantly associated in predicted
ways. In Study 2, parental (e.g., demographic, child temperament) and teacher (e.g.,
PPBS, child behavior problems) ratings were collected for 337 preschoolers. Results
established the psychometric properties of the PPBS (e.g., inter-rater reliability, factor
structure, internal consistency, and convergent validity).
Children who rarely interact with peers during free play activities are typically
viewed by parents, teachers and researchers as shy or socially withdrawn. Socially-
noninteractive children are generally considered at risk for later maladjustment diffi-
culties in later childhood and adolescence (see Kupersmidt, Coie, & Dodge, 1990;
Parker, Rubin, Price, & DeRosiers, 1995, for recent reviews). More specifically, from
early through late childhood, social withdrawal (as defined, in part, by the lack of
social interaction) is associated concurrently and predictively with the constructs of
felt insecurity, negative self-perceptions, dependency, and internalizing difficulties in
adolescence (e.g., Rubin et al., 1995; Rubin & Mills, 1988).
Recently, researchers have begun to investigate the possibility that young chil-
dren’s behavioral solitude represents a heterogeneous construct. To date, three sub-
This research was supported by a Social Science and Humanities Research Council of Canada Doctoral
Fellowship to Robert J. Coplan and an Ontario Mental Health Senior Research Fellowship to Kenneth
H. Rubin. The authors wish to thank Steve Udvari, Amy Rubin, Kerri Hogg, Kevin McNichol, Kelly
Lemon, and Bev Watson for their help in the collection and coding of data. We also wish to thank all of
the preschool teachers, children, and families that participated in this study. Finally, the suggestions from
several anonymous reviewers were very helpful in preparing this manuscript for publication. Requests for
reprints should be addressed to Robert J. Coplan, Department of Psychology, Carleton University, 1125
Colonel By Drive, Ottawa, Ontario, K1S 5B6, Canada.
© Blackwell Publishers Ltd. 1998. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street,
Malden, MA 02148, USA.
14679507, 1998, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1467-9507.00052 by Cochrane Portugal, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Preschool Play Behavior Scale 73
categories of nonsocial behaviors have been identified, each apparently with its own
underlying psychological mechanisms. Solitary-passive behavior includes the quies-
cent exploration of objects and/or constructive activity while playing alone. Rubin
(1982) argued that in early childhood, such behavior is reinforced positively by
teachers, parents and peers, and that its display is associated with competent prob-
lem solving as well as peer acceptance. Generally, solitary-passive play among
preschoolers is not associated with indices of maladaptation (Coplan, Rubin, Fox,
Calkins, & Stewart, 1994). Further, it has been argued that children displaying a
high frequency of solitary-passive play are disinterested in social interaction, having
low approach and low avoidance motivations (Asendorpf, 1991; Rubin &
Asendorpf, 1993). Rubin, Coplan, Fox and Calkins (1995) recently found that
solitary-passive behavior was the preferred play modality among an extreme group
of socially noninteractive preschoolers who were also emotionally well-regulated.
Reticent behavior consists of prolonged looking at the partner without accompa-
nying play and being unoccupied (Asendorpf, 1991; Coplan et al., 1994). Reticent
behavior is believed to reflect social fear and anxiety in a social context. Among
preschoolers, this subgroup of nonsocial behavior has been associated with the overt
demonstration of anxious behaviors and hovering behaviors during free play, mater-
nal ratings of shyness (Coplan et al., 1994), and right frontal EEG activation asym-
metries (a measure of emotion dysregulation) (Fox et al., 1995). Reticence is
theorized to reflect an approach-avoidance conflict in young children (Asendorpf,
1990); thus, children may retreat to reticent behavior as an attempt to cope with the
fearfulness of social situations.
Solitary-active behavior is characterized by repeated sensorimotor actions with or
without objects and/or by solitary dramatizing (in the presence of a social group)
(Rubin, 1982). This cluster of behaviors has been associated with maternal ratings of
impulsivity among preschoolers (Coplan et al., 1994), and is postulated to correlate
with peer rejection from as early as the preschool years and to be related to external-
izing problems (in particular aggressiveness) and immaturity throughout the child-
hood years (Rubin, 1982; Rubin & Mills, 1988). However, solitary-active behavior
occurs only about 4% of the time during free play in novel settings (Coplan et al.,
1994), making it difficult to observe and study.
Further empirical support for the presence of multiple and distinct subgroups of
socially withdrawn behaviors has been reported by several researchers (e.g., Hinde,
Tamplin, & Barrett, 1993; Levy-Shiff & Hoffman, 1989; Younger & Daniels, 1992;
Harrist et al., in press). Thus, it appears as if social withdrawal in early childhood
is a heterogeneous construct, and that only some forms of solitude may prove to
be associated contemporaneously and predictively with socio-emotional maladapta-
tion.
Method
Subjects. The participants in this study were 39 preschool children (23 females and
16 males) between the ages of 37 and 59 months (M 5 49.18, SD 5 5.58). The chil-
dren were primarily of middle class background, living with their families in a mid-
sized Ontario community. The sample was primarily but not exclusively Caucasian.
The children were attending a university-based early childhood education program.
Procedure
Observational measures. Each child was observed from behind a one-way mirror, in
random order, for a series of twelve ten-second intervals (totalling to two minutes)
during free play. No child was observed for more than two minutes on any given
day. Over a period of approximately three months, each child was observed on ten
separate occasions, yielding a total of 120 ten-second coding intervals (20 minutes of
free play observations) per child. Observations did not commence until at least four
weeks into the semester, in order to allow the children to familiarize themselves with
their new environment.
Children’s free play behaviors were coded using Rubin’s (1989) Play Observation
Scale. Ten second intervals were coded for both social participation (onlooker, unoc-
cupied, solitary, parallel, group) and cognitive quality of play (functional, dramatic,
exploratory, constructive). In the instances where multiple behaviors were observed
in the same ten-second interval, the predominant behavior (in terms of time
observed) was coded. The resulting raw frequencies were proportionalized, dividing
by the total number of coding intervals per child. Following the procedure outlined in
Coplan et al. (1994), the following aggregate variables were created, (1) reticent
behavior (computed by summing the proportion of coding intervals spent in unoccu-
pied and/or onlooking behaviors); (2) solitary-passive behavior (solitary-exploratory
plus constructive behaviors); (3) solitary-active behavior (solitary-functional plus
© Blackwell Publishers Ltd. 1998 Social Development, 7, 1, 1998
14679507, 1998, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1467-9507.00052 by Cochrane Portugal, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
76 Robert J. Coplan and Kenneth H. Rubin
dramatic behavior); (4) social play (all forms of group interaction plus peer
conversation) and (5) rough-play (playful mock fighting and rough-and-tumble
activities).
Reliability. Interobserver reliability data were obtained by two observers who
simultaneously coded sequences of 2-minute intervals on randomly selected children
over a period of three weeks (before the start of the actual data collection). A total
of 720 coding intervals (120 minutes) was collected, representing approximately 20%
of the amount of data that was collected for the study itself. For a complete variable
matrix, Cohen’s Kappa was calculated at K 5 .83. Data were subsequently collected
by one of the two reliable coders.
PPBS item selection and test construction. An original pool of Preschool Play
Behavior Scale (PPBS) items was created based on the coding manual descriptions
of various Play Observation Scale (POS) assessed behaviors (Rubin, 1989).
Thereafter, this preliminary item pool was circulated among five trained and reliable
POS coders and four preschool teachers. Based on discussions and feedback from
this pool of experts, items were re-worded, dropped, and added. The resulting com-
plete original item pool is displayed in Table 1. Items were arranged in a five-point
Likert (1932) format, with response choices designed to reflect frequency of occur-
rence (1 5 never, 2 5 hardly ever, 3 5 sometimes, 4 5 often, 5 5 very often).
Four months into the preschool year, three teachers completed the PPBS for the
children in the sample. Wherever possible, two teachers completed the scale for the
same children, in order to allow for an examination of inter-teacher reliability. The
reliability data were analyzed in Study 2.
Results
The first set of analyses concerned the preliminary validation of the PPBS as a
teacher rating scale of preschoolers’ non-social and social free play behaviors. To
begin with, summary variables were computed by summing the appropriate items
for reticent behavior (4 items), solitary-passive behavior (4 items), solitary-active
behavior (4 items), social play (6 items), and rough-play (2 items).
In order to assess the construct validity of PPBS summary variables, the relations
between teacher-rated and observed child behaviors were examined. To begin with, a
series of correlations was computed between teacher-rated and observed reticent,
solitary-passive, solitary-active, social play and rough-play behaviors. Results for
each of the three teachers are displayed in Table 2. Overall, moderate to high corre-
lations were found between observed and teacher-rated behaviors. Of the fifteen
computed correlations (all correlations 1-tailed), all but two were significant at the
.05 level.
In order to assess the discriminant validity of the teachers’ ratings of the various
forms of nonsocial play, a series of correlations was computed between teacher rat-
ings of reticent, solitary-passive, and solitary-active behaviors and observations of
the other non-corresponding nonsocial behaviors. These correlations are displayed in
Table 3. Overall, the pattern of correlations did not indicate a strong relation
between teacher ratings of one form of solitary activity and behavioral observations
of the others. Specifically, only 2 of the 18 computed correlations were significantly
and positively associated at the .05 level.
Reticent behavior:
1. takes on the role of onlooker or spectator
2. wanders around aimlessly
3. watches or listens to other children without trying to join in
4. remains alone and unoccupied, perhaps staring off into space
Solitary-passive behavior:
1. plays by himself/herself, examining a toy or object
2. plays alone, building things with blocks and/or other toys
3. plays by himself/herself, drawing, painting pictures, or doing puzzles
4. plays alone, exploring toys or objects, trying to figure out how they work
Solitary-active behavior:
1. plays by himself/herself, engaging in simple motor activities (e.g., running, ringing
bells/buzzers)
2. engages in pretend play by himself/herself
3. plays ‘make-believe’, but not with other children
4. plays alone in an active fashion, enjoying an activity solely for the physical
sensation it creates
Social play:
1. talks to other children during play
2. plays ‘make-believe’ with other children
3. engages in group play
4. plays in groups with (not just beside) other children
5. engages in active conversations with other children during play
6. engages in pretend play with other children
Rough-play:
1. plays ‘rough-and-tumble’ with other children
2. engages in playful/mock fighting with other children
Discussion
Results from this study provided some preliminary support for the validity of the
Preschool Play Behavior Scale (PPBS) as an assessment tool for young children’s
nonsocial play behaviors. It seems clear that teachers were capable of distinguishing
between the various subtypes of young children’s nonsocial play behaviors: Teacher
ratings of nonsocial behaviors (i.e., reticent, solitary-passive, solitary-active), social
play, and rough-play were generally significantly correlated with observed indices of
the same behaviors. Moreover, there was an overall lack of significant intercorrela-
tions among teacher-rated and observed non-corresponding forms of nonsocial
behaviors. This suggests an acceptable level of discriminant validity for the ‘nonso-
cial’ items of the PPBS. The results concerning solitary-active behaviors, however,
must be interpreted with some caution. On the average, solitary-active behavior was
observed to occur about 3% of the time during free play (M 5 .031, SD 5 .036).
© Blackwell Publishers Ltd. 1998 Social Development, 7, 1, 1998
14679507, 1998, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1467-9507.00052 by Cochrane Portugal, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
78 Robert J. Coplan and Kenneth H. Rubin
Table 2. Correlations between Teacher-rated and Observed
Forms of Social and Nonsocial Behaviors
reticent solpas
(observed) (observed)
reticent solpas
t1 (n 5 20) .50* t1 .37*
t2 (n 5 21) .46* t2 .321
t3 (n 5 20) .54** t3 .46*
solact socplay
(observed) (observed)
solact socplay
t1 .56** t1 .60**
t2 .57** t2 .45*
t3 .02 t3 .62**
rough
(observed)
rough
t1 .65**
t2 .44*
t3 .55**
This low frequency of occurrence and restricted range may have impacted upon the
computed correlations.
Method
Subjects. The subjects were 337 preschoolers (173 males, 164 females) from eleven
early childhood education centers, preschools, and day care facilities in three mid-
sized southwestern Ontario communities. Subjects ranged in age from 33 to 68
months (mean age 5 51.56 mos., SD 5 9.07). Thirty-seven of the children from a
university-based early child education center had also participated in Study 1,
approximately six months earlier. The children were primarily of middle class back-
ground. Approximately 25% of parents in the sample had completed high school,
about 20% had gone on to attend community college, and around 40% had com-
pleted a univerisity degree. Approximately 72% of the sample was Caucasian.
Measures
Teacher ratings. Four months into the school semester, teachers completed the
PPBS for all children in the sample. Inter-teacher reliability data were collected in
one school, for two small samples of children (n 5 12, and n 5 9). However, the
additional reliability data from another school in Study 1 were analyzed herein for
another three pairs of teachers (n 5 20, n 5 11, and n 5 11).
Teachers also completed the Preschool Behavior Questionnaire (PBQ, Behar &
Springfield, 1974) for all children in the sample. The psychometric properties of the
PBQ have been well established (Behar 1977; Hoge, Meginbir, Khan, & Weatherall,
1985; Martin, 1988; Rubin & Clark, 1983). The original factor analysis of the PBQ
(Behar & Springfield, 1974) yielded a three factor solution: Anxious–Fearful,
Hostile–Aggressive, and Hyperactive–Distractible. Results from several recent stud-
ies have suggested a broader two factor solution of Internalizing and Externalizing
Results
Factor and item analysis of PPBS items. In order to examine the various subscales of
the PPBS, the data were subjected to exploratory factor analysis, using principal
components extraction and varimax rotation. Results indicated a five factor solution
(i.e., five factors had Eigenvalues greater than 1.0). Together, these five factors
accounted for 75.8 percent of the variance. The factor loadings confirmed, almost
completely, the theoretical associations that had been assumed between various
items. A complete list of factor Eigenvalues and individual factor item loadings are
displayed in Appendix A.
The first factor, social play, consisted of the six items referring to group and socio-
dramatic play, and peer conversation. The second factor, solitary-passive behavior,
comprised the four items regarding solitary-constructive and -exploratory behaviors.
In addition, the two items designed to assess solitary-functional behaviors cross-
loaded on this factor. The third factor, reticent behavior, contained the four items
concerning unoccupied and onlooking behaviors. The fourth factor, rough-play, con-
sisted of the two items assessing rough-and-tumble play. The fifth and final factor,
solitary-active behavior, was expected to comprise the four scale items designed to
assess solitary-functional and -dramatic behaviors. Although the two solitary-
dramatic items loaded highly on this factor, with factor loadings of .80 and .86, the
two solitary-functional items displayed only moderate (.42) and low (.30) factor
loadings.
Factor scores were computed by summing the appropriate items for each factor.
With regard to the solitary-active factor, the decision was made to discard the two
solitary-functional items because of their cross loadings. Thus, in all subsequent
analyses, the solitary-active factor consisted of the two items assessing solitary-dra-
matic behaviors.3 Although a two-item factor is not ideal, Tabachnick and Fidell
(1989) have argued that two-item factors can be cautiously interpreted if the two
items are highly correlated with each other and relatively uncorrelated with other
variables.4 The correlations between the five computed subscales of the PPBS are
presented in Table 4. For the most part, the factors were moderately inter-related.
In order to examine potential sex differences in the factor structure of the PPBS,
the data were factor analyzed again using principal components extraction and vari-
max rotation, but this time separately for males and females. Although this proce-
dure reduced the sample size considerably, the goal of these analyses was to search
for large discrepancies in the factor structure as a function of gender. Overall, the
factor structure and item loadings for both males and females alone were consistent
with the full sample results.
© Blackwell Publishers Ltd. 1998 Social Development, 7, 1, 1998
14679507, 1998, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1467-9507.00052 by Cochrane Portugal, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Preschool Play Behavior Scale 81
Table 4. Intercorrelations between the Five Subscales of the PPBS
1 2 3 4
1. reticent —
2. solitary-passive .51*** —
3. solitary-active .40*** .40*** —
4. social play –.64*** –.39*** –.16** —
5. rough-play –.14** –.11* –.03 .26***
*** p ,.001
** p , .01
* p ,.05
In order to examine the internal consistency of the newly created five factors,
Cronbach’s Alpha was computed for each of the subscales. Nunnally (1978) pro-
vides a widely accepted rule of thumb that a should be at least .70 for a scale to
demonstrate internal consistency. Alpha values for the subscales were: social play
.96; reticent .85; solitary-passive .79; rough-play .92; and solitary-active .76.
Reliability and Stability. Reliability correlations were calculated for each PPBS sub-
scale for five pairs of teachers (three pairs in Study 1, two pairs in Study 2). All cor-
relations are displayed in Table 5. Overall, the PPBS subscales displayed moderate
to high reliability correlation coefficients, despite small sample sizes. Reticent-wary,
social play, and rough-play all had relatively high reliability correlations, ranging
from r 5 .54 (p , .05) to r 5 .89 (p , .001). The solitary-passive factor had moder-
ate to high reliability correlations, ranging from r 5 .33 (p ,.05) to r 5 .79 (p , .01).
Solitary-active behavior was less consistently reliable, with a wide range of reliability
correlations from r 5 .10 (n.s.) to r 5 .83 (p , .01).
The short term stability of each of the five PPBS subscales was assessed by corre-
lating two teachers’ ratings (from Study 1) with the same two teachers’ ratings
(n1 5 18, n2 5 19) of the same children six months later (Study 2). Overall, the PPBS
subscales displayed moderate to high stability over a six month period. Stability cor-
relations for the two teachers were as follows: reticent, r 5 .65, r 5 .39; social play,
*** p,.001
** p,.01
* p,.05
Discussion
The major goal of the present study was to establish the psychometric properties of
the Preschool Play Behavior Scale (PPBS). Specifically, the factor structure, internal
consistency, interrater reliability, short-term stability, and the convergent and dis-
criminant validity of the scale were examined. Results provided support for the use
of the PPBS as a reliable and valid accompaniment or alternative to the time con-
suming methodology of behavioral observations in the assessment of social and
nonsocial play forms in young children. Moreover, results from this study provided
insight into the nature of children’s nonsocial play behaviors in the familiar setting
of the preschool classroom. These findings extend the understanding of behavioral
solitude beyond the context of the laboratory playroom (Asendorpf, 1991; Coplan et
al., 1994).
The PPBS was demonstrated to comprise five factors: (1) social play; (2) reticent
behavior; (3) solitary-passive behavior; (4) solitary-active behavior; and (5) rough-
play. The results for the five factors are discussed in turn.
The social play factor consisted of items pertaining to group play, sociodramatic
play, and peer conversation. There does not currently exist a well-established teacher
rating scale designed to assess this specific construct. As mentioned previously, the
majority of teacher rating scales designed to assess children’s behaviors in the
preschool focus on externalizing and internalizing problems, and do not provide
items pertaining to social play or social competence (e.g., PBQ, Behar & Springfield,
1974; Revised Behavior Problem Checklist, Quay, 1983; Conners Teacher Rating
Scale, Conners, 1969, 1973). Some measures do assess social competence—the
Preschool Socio-Affective Profile (PSAP, LaFreniere et al., 1992), or prosocial
behavior—the newly developed Child Behavior Scale (CBS, Ladd & Profilet, in
press). However, the included items in these scales involve general social compe-
tence, as opposed to the specific behavioral construct of social play (i.e., PSAP:
© Blackwell Publishers Ltd. 1998 Social Development, 7, 1, 1998
14679507, 1998, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1467-9507.00052 by Cochrane Portugal, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
84 Robert J. Coplan and Kenneth H. Rubin
‘negotiates solutions to conflicts’; ‘takes other children’s viewpoints into account’;
‘helps in everyday tasks’; CBS: ‘helps other children; ‘kind toward peers’; ‘coopera-
tive with peers’).
Although social play may be one component of social competence, these two con-
structs do not overlap completely. Moreover, it has been demonstrated recently that
not all sociable children are socially competent (e.g., Rubin et al., 1995). Social play,
in and of itself, is an important and unique factor in children’s social and emotional
development (see Coplan & Rubin, in press, for a recent review), and can now be
assessed directly using the PPBS.
The reticent behavior subscale comprised items referring to unoccupied and
onlooking behaviors. There is growing empirical support for the notion that reticent
behavior in a novel setting is characteristic of anxiety in the face of unfamiliar social
stimuli (e.g., Asendorpf, 1991; Coplan et al., 1994). However, the results from the
present study provide some of the first information regarding reticent behavior in a
clearly familiar environment. Reticent behavior during free play was positively asso-
ciated with maternal ratings of shyness and negative emotionality, and negatively
associated with sociability (although the magnitude of these significant correlations
was modest). However, a strong relation was found between the display of reticent
behavior during free play and internalizing problems. These findings, in concert with
the results of previous studies (Asendorpf, 1991; Coplan et al., 1994; Fox et al.,
1995) provide converging evidence that reticent behavior (both in a familiar and
unfamiliar social context) may be a marker variable for social fear, anxiety, and
internalizing problems.
The solitary-passive behavior subscale contained items dealing with solitary-
constructive and -exploratory behaviors. The solitary-passive factor was found to
have moderate to high inter-teacher reliability and stability. Overall, there was sub-
stantial empirical evidence to support the hypothesis that teachers were able to dis-
tinguish between this form of behavioral solitude and reticent behaviors. However,
two items designed to assess solitary-functional behaviors loaded on both this factor
and the solitary-active factor (these items were subsequently dropped). This finding
is discussed in more detail in a later section.
Solitary-passive behavior was moderately but significantly associated with
teacher-rated internalizing problems. It seems important to acknowledge that all
three forms of nonsocial play were significantly associated with internalizing prob-
lems in the present study, although the relation between the reticent factor and inter-
nalizing problems was significantly stronger than the others. It is thus possible that
teachers generally rated all children who engaged in any form of solitary activity as
having more internalizing problems. The PBQ internalizing subscale does contain a
few items referring to non-specific solitary activities (i.e., ‘tends to do things on
her/his own, rather solitary’). Notwithstanding, results from a series of partial corre-
lations suggested that the relation between solitary-passive and -active behaviors and
internalizing problems was largely due to shared variance with the reticent subscale.
Consistent with past findings (Coplan et al., 1994; Rubin, 1982), solitary-passive
behaviors were not associated with concurrent indices of maladaptation among
preschool-aged children. It is indeed difficult to draw strong conclusions about the
construct of solitary-passive behavior in the presence of a series of non-significant
relations with other variables. However, many of the variables assessed in this study
were potential indices of maladjustment (i.e., shyness, negative emotionality, inter-
nalizing and externalizing problems). Given the fact that solitary-passive behavior is
© Blackwell Publishers Ltd. 1998 Social Development, 7, 1, 1998
14679507, 1998, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1467-9507.00052 by Cochrane Portugal, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Preschool Play Behavior Scale 85
not theorized to be associated with indices of maladjustment, these lack of relations
are not surprising. It has been suggested that solitary-passive behavior is representa-
tive of social disinterest (Coplan et al., 1994; Rubin & Asendorpf, 1993). Future
investigations are required to further explore the relations between solitary-passive
behavior and conceptually relevant constructs (i.e., social motivation, object-
preference vs. person-preference).
The results concerning the solitary-active factor were somewhat mixed, and thus
must be interpreted with some caution. To begin with, results from the factor analy-
sis revealed the unexpected finding that the two items designed to assess solitary-
functional behaviors loaded on both the solitary-active and solitary-passive factors.
These items were subsequently discarded, but the question remains as to why this
cross-loading occurred.
Some teachers may have misinterpreted or misconstrued the two solitary-func-
tional items. This may have led to spurious associations. In retrospect, it is not sur-
prising that teachers appeared to be more accurate and reliable in their assessments
of solitary-dramatic play, as compared to solitary-functional activities. Solitary-
dramatizing can be a very salient activity, particularly if the child is enacting aggres-
sive themes. Functional activities, in comparison, with their repetitive nature, may
not ‘stand out’ as much to teachers. As a result, teachers may group functional activ-
ities together with other ‘quieter’ solitary activities (i.e., construction and explora-
tion).
It is important to recall that solitary-active behavior only occurs about three per-
cent of the time during free play. Thus, it is understandable that teachers may have
some difficulties in accurately assessing this form of behavior. Bearing this in mind,
conclusions concerning solitary-active behavior must be tentative. There was
evidence to suggest that children who engage in a comparatively high frequency of
solitary-active behaviors are not shy (there was a significant negative correlation
between solitary-active behaviors and maternal-rated shyness). As well, the signifi-
cant (although modest) association with temperamental activity level was consistent
with the results of previous studies (e.g., Coplan et al., 1994). Thus, it does seem
clear that solitary-active behaviors should not be included among behavioral indices
of shy, anxious behaviors. However, there are important issues with this construct
that remain unresolved.
To begin with, additional items are required for the solitary-active subscale in
order to improve inter-teacher reliability and stability. Moreover, fundamental ques-
tions need to be answered concerning the theoretical and conceptual relation
between solitary-functional and -dramatic activities. Unfortunately, the paucity of
solitary-active behaviors during indoor free play makes this a somewhat difficult
task. It may be the case that solitary-active behavior simply does not occur often
enough during indoor free play at the preschool for teachers to accurately assess it.
Perhaps solitary-active behaviors would be observed more frequently on play-
grounds, since outdoor contexts may be more conducive to functional play activities.
This hypothesis merits further empirical investigation.
The rough-play subscale consisted of items concerning rough-and-tumble play and
mock fighting. Not surprisingly, rough play was positively associated with tempera-
mental activity level, and negatively related to attention span and shyness. As well,
rough play was strongly associated with externalizing problems. At first glance, the
strong relation between rough-play and externalizing problems seems inconsistent
with past findings, given that rough-and-tumble play has typically been viewed as an
© Blackwell Publishers Ltd. 1998 Social Development, 7, 1, 1998
14679507, 1998, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1467-9507.00052 by Cochrane Portugal, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
86 Robert J. Coplan and Kenneth H. Rubin
index of social competence, particularly among popular children (e.g., Pellegrini,
1988). However, there does exist some empirical evidence to suggest the co-
occurrence of rough-and-tumble play and aggression/peer rejection. For example, in
a factor analytic study of 12- and 13-year-old British boys, Neill (1976) found that
rough-and-tumble and aggression co-occurred. Among school aged children, it
appears as if both popular and rejected children engage in rough-and-tumble-play
(Smith, 1989). As well, Hart et al. (1992) reported that both sociometrically most-
preferred and least-preferred preschoolers engaged in rough play.
Pellegrini (1994) distinguished between two forms of rough-and-tumble behavior;
play fighting and chase. He found that play fighting was associated with aggression,
particularly among rejected boys. A closer look at the items used to assess rough-
play in the PPBS (i.e., ‘engages in playful/mock fighting with other children’; ‘plays
rough-and-tumble with other children’) reveals that it was the behavior of play fight-
ing that was most likely being assessed by teachers. Since rough-and-tumble play can
be a precursor to aggression (Hinde, 1982), and thus would likely be discouraged in
the classroom, it was not surprising, in hindsight, that teachers might ‘stereotype’
this kind of behavior and associate it with aggression.
Sex Differences
As far as sex differences were concerned, the results from this study were character-
ized by a marked lack of sex differences. Overall, the factor structure of the PPBS
was very similar for subsamples of males and females. As well, there were no major
differences between males and females in terms of the PPBS factors of reticent, soli-
tary-passive, and solitary-active behavior, social play, and rough-play. Given the
fact that boys exhibit more rough-and-tumble play (e.g., DiPietro, 1981; Maccoby &
Jacklin, 1974; Whiting & Edwards, 1973) and aggression (e.g., Crowell, 1987;
Cummings, Iannotti, & Zahn-Waxler, 1989; Hyde, 1986) than girls, it was not sur-
prising to find that boys were rated by teachers as significantly higher than girls on
the rough-play subscale.
Appendix A
Factor loadings for the five factor solution to the PPBS (n5337)
Appendix B
The Preschool Play Behavior Scale
The following scale examines various behaviors that children may engage in during indoor free
play. Please rate the child on each item and COMPARE HIM/HER TO OTHER CHIL-
DREN OF THE SAME AGE IN THE CLASS. Although it is true that children’s behaviors
may be quite variable, please try to make a general evaluation of the child’s ‘everyday’ behav-
iors.
1. Talks to other children during play.
1 2 3 4 5
never hardly ever sometimes often very often
2. Plays by himself/herself, examining an object or toy.
1 2 3 4 5
never hardly ever sometimes often very often
3. Plays ‘rough-and-tumble’ with other children.
1 2 3 4 5
never hardly ever sometimes often very often
4. Takes on the role of onlooker or spectator.
1 2 3 4 5
never hardly ever sometimes often very often
5. Plays ‘make-believe’ with other children.
1 2 3 4 5
never hardly ever sometimes often very often
6. Engages in group play.
1 2 3 4 5
never hardly ever sometimes often very often
7. Engages in pretend play by himself/herself.
1 2 3 4 5
never hardly ever sometimes often very often
8. Plays alone, building things with blocks and/or other toys.
1 2 3 4 5
never hardly ever sometimes often very often
9. Wanders around aimlessly.
1 2 3 4 5
never hardly ever sometimes often very often
Notes
1. With regard to peer ratings, although young children can give reliable and valid ratings of their
classmates’ aggressive behaviours (e.g., Ladd & Mars, 1986, Younger Schwartzman & Ledingham, 1985),
young children have much more difficulty in accurately and reliably assessing socially withdrawn behav-
iours (Younger, Gentile, & Burgess, 1993). Socially withdrawn behaviours are not negatively salient to the
preschool peer group. It appears as though children do not possess a well developed schema for social
withdrawal until the mid-childhood years (Younger & Boyko, 1987; Younger et al., 1993). Clearly, it does
not seem as if young children would be very successful in assessing the frequency of their peer’s engage-
ment in various forms of non-social activities (e.g., solitary-constructive vs. solitary-functional).
2. The 29 PBQ items were factor analyzed using principal components extraction and varimax rota-
tion. A two factor solution was generated. The first factor, externalizing problems, had an Eigenvalue of
9.7 and accounted for 33.5 percent of the variance in the data. The second factor, internalizing problems,
had an Eigenvalue of 3.5 and accounted for 11.9 percent of the variance in the data. The externalizing fac-
tor contained items dealing with aggression and hyperactivity/distractibility. The internalizing factor con-
tained items concerning anxiety, fear, and worrying. Four items did not load highly on either factor and
were thus excluded from future analysis.
3. Having reached this decision, the analyses from Study 1 were recomputed using the two-item version
of the solitary-active subscale. Teacher ratings of the new solitary-active factor were moderately corre-
lated with observed solitary-active behaviors (r 5 .57, p , .01 for teacher1; r 5 .44, p , .05 for teacher2, r
5 20, n.s. for teacher3). As well, for all three teachers, there were no significant correlations between
teacher ratings of solitary-active behaviors and observed reticent and solitary-passive behaviors.
4. We thank an anonymous reviewer for this suggestion.