Anda di halaman 1dari 16

School Psychology Review,

2005, Volume 34, No. 4, pp. 475-489

Playground-Based Observational Systems: A Review and


Implications for Practitioners and Researchers
Stephen S. Leff and Rebecca Lakin

The Children s Hospital of Philadelphia and University of


Pennsylvania School of Medicine
Abstract. Behavioral observation systems allow for a relatively objective way to
record important academic, behavioral, and/or interactional processes. Not surprisingly, the majority of school-ba.sed observational methods have been designed
for and evaluated within the classroom setting. Although this is understandable,
the playground context during recess provides an important unstructured school
context in which to understand young children's peer relationships, play behaviors, and aggressive actions. This article provides a critical review of six playground-ba.sed observation systems and discusses implications for use by practitioners and researchers.

The schooi playground provides a key


naturalistic context in which to study children's
social competence and social conflicts. Play
behaviors are extremely useful to examine in
this context because important issues in the
development of peer relationships often occur
through play, including companionship, intimacy, and inclusion/exclusion (Leff, Costigan,
& Power, 2004). Further, school recess has
been associated with the development of social competence and social cognitive abilities,
as well with the development of aggressive and
coercive behaviors (Pellegrini & Bjorklund,
1996). For instance, research has demonstrated
that many children are involved in low-level
conflicts in school every day (Nansel et al..
2(X)I), and that the majority of these aggressive behaviors occur within unstructured
school settings (see Leff. Costigan, et al..
2004). The school playground during recess

also can be an opportune setting in which to


promote children's prosocial skills (Leff,
Costigan. et al.. 2004) and where social skills
and conflict management abilities can be
taught, reinforced, and/or monitored (Leff,
Power, & Goldstein, 2004). Thus, understanding children's behaviors on the school playground during recess has important implications for both research and practice.
Many types of measures can be used to
assess children's play and aggressive behavior on the playground, including discipline referrals; nursing logs of injuries; and self-,
teacher-, and peer-report measures. Each methodology has certain strengths and limitations
(Leff, Power et al., 2004). For instance, selfreport measures can provide specific information about a child's experience of being victimized, but are also somewhat subjective and
may underestimate rates of aggression (Ladd

Acknowledgements: This research was supported by a NIMH grant to thefirstauthor, K23-MH01728.


Correspondence regarding this article should be mailed to Stephen S. Leff, PhD, The Children's Hospitai
of Philadelphia. Depi. of Pediatric Psychology, Rm. 1480,3405 Civic Center Blvd., Philadelphia, PA 19! 04;
E-mail: leff@email.chop.edu
Copyright 2005 by the National Association of School Psychologists, ISSN 0279-6015
475

School Psychology Review, 2005, Volume 34, No. 4

& Kochenderfer-Ladd. 2002). Nursing reports


and discipline referrals are relatively easy to
collect and tabulate, but have questionable reliability and validity, and are not typically standardized and systematized across schools
(Brener. Krug. Dahlberg. & Powell, 1997).
Teacher reports, such as standardized rating
scales, are also relatively easy to administer
and interpret, and many have established psychometric properties (Leff. Kupersmidt,
Patterson, & Power, 1999). However, they are
often insensitive to the playground context
(Leff, Power, et al.. 2004). and especially to
victimization (Leff et al., 1999). Finally, peer
report measures, which are widely accepted by
researchers and have strong reliability and validity (Terry & Coie, 1991), are very time-consuming and labor-intensive, and require extensive parental permission procedures.

observation systems searching the following


key terms {playground, recess, observation,
rough and tumble play, and aggression) in three
commonly used psychological, educational,
and medical databases (Psychlnfo, ERIC, and
Medline). In addition, review papers and book
chapters related to children's play, aggression
and bullying, and outcome measures involved
in aggression prevention and intervention programs were examined to identify relevant articles (e.g., Leff. Power. Manz. Costigan, &
Nabors, 2001; Leff. Power, et al.. 2004;
Pellegrini. 2002; Samples & Aber, 1998). To
maximize the focus and utility of the review,
dissertations, articles that described observation systems used only in nonplayground settings (e.g., classroom or laboratory) or articles
describing systems that have not been used
since 1985 were eliminated.

Observational methods provide an alternative means for understanding the play and
aggressive behaviors of children on the playground during school recess. Observational
methods enable researchers and practitioners to identify behavioral patterns within
the naturalistic contexts at the time during
which they actually occur, and. thus, the
social validity of these systems is often a
strength (Hintze. Volpe. & Shapiro, 2002).
Direct observations of playground behaviors
tend to also be more objective than peer,
teacher, or parent reports. To inform schoolbased practitioners and researchers about playground observational systems, this review has
the following goals: (a) to systematically review existing playground-based observation
systems; (b) to discuss the utility of established
observation systems for research and practice;
and (c) to highlight key issues and implications
for use of these systems by practitioners and
researchers.

The above criteria identified 80 articles


describing 45 separate behavioral observation
systems. The authors systematically reviewed
these 45 systems to determine if they met criteria established by Hintze et al, (2002) in their
recent review on best practice in the systematic direct observation of student behavior. The
inclusion criteria included: (a) the goal of the
system is to measure specific, targeted behaviors; (b) the behaviors being observed are operationally defined a priori; (c) observations
are conducted using standardized procedures
and are highly objective in nature; (d) the times
and places of the observations are carefully
selected; and (e) the scoring and .summation
of data are standardized. In addition, we added
a sixth criterion that the system must have a
manual that was available for our review. Of
the 45 systems, 10 were eliminated because
they did not meet one of Hintze et al.'s five
criteria. We then contacted the primary authors
of each of the remaining 35 systems to ask
whether there was a manual available for our
review. Authors from 34 of the 35 observation
systems responded to our email inquiries, and
provided additional detail. Of the 34 remaining observation systems. 17 had a manual, 16
of which were available for review. The 16
available observation manuals and the corresponding articles were reviewed by three research team members. We determined that 6

Selection of Playground or Play-Based


Observational Systems
School-based behavioral observation
systems to study children's recess and/or play
behavior were chosen for review in the current manuscript following a multistep process.
First, the primary authors conducted an initial
review of existing playground and play-based
476

Playground-Based Observations

observation systems were described in sufficient detail that they could be reviewed systematically by our research team.
We funher divided the six systems into
two categories: (a) long-standing, well-established systems in which there are many empirically based publications supporting their
use; and (b) new systems that have not yet been
subjected to as extensive empirical scrutiny.
The three well-established systems are reviewed in considerable detail, and the newer
systems are reviewed briefly to highlight their
potential research and/or clinical applications
in the future. Table I provides a summary of
key characteristics of each of these six observation systems.
Review of WeU'Established Systems
Interpersonal Process Code {IPC;
Rusby, Estes. & Dishion, 1991)
Brief overview. The IPC is used for observations of peer {and family) interactions
across different school and home settings. It
was developed through many years of research
at the Oregon Social Le;irning Center (OSLC)
and is based upon prior piayground observational systems. Typically, the IPC is coded live
on a hand-held electronic event recorder. The
manual is free and is available on-line at
www.oslc.org/obs/ipc.html.
A number of studies have utilized the
IPC or adapted versions. For instance, the IPC
was used as a pre- and postintervention outcome measure (Stoolmiller, Eddy, & Reid,
2000) to examine the association between
parenting style and children's negative behaviors (Knutson. DeGarmo, & Reid. 2004), and to
determine target children's rates of victimization
by their peers during the recess period
(Goodman, Stormshak. & Dishion, 2001). These
samples were drawn from urban public elementary schools, and were composed of primarily
European American participants from lower to
middle socioeconomic status classes.
Codes employed. The IPC consists of
three dimensions that are coded concurrently:
Activity, Content, and Affect Valence. A focal
child event recording system is used, wherein

a specific target child is identified and the frequency of each Activity, Content, and Affect
Valence code is tabulated for a prespccified
period of time. In addition to the frequency of
each behavior, the sequence and duration of
each particular behavior is also recorded. The
Activity code refers to the actual setting in
which the observation system is employed (settings are designated as being either the playground, home, laboratory, therapy session, or
classroom), as well as the behavior cKcurring
within each setting. Within the playground setting, the Activity dimension has four independent codes: Free Play (e.g.. involved in an
unstructured play activity), Participation (in
games). Parallel Play (e.g.. engaged in an activity in close proximity to a peer), and Alone.
The Content Code dimension refers to interactional behaviors initiated either by the target child or by the child or adult with whom
the target child is interacting. Within the Content Code, 14 behavioral codes are employed
that are designated as being a cross between a
behavior category (Verbal, Non-Verbal, and
Physical) and the social impact of the behaviors (Positive, Neutral, and Negative). For example. Cooperative Behavior would be coded
as a nonverbal behavior with a positive impact,
wi^eteaa Negative Physical Behavior (e.g.. hnting, kicking, pinching) would be coded as a
physical behavior with a negative impact. The
Affect Code dimension is defined as the emotional and nonverbal displays of the target child
(Happy, Caring, Neutral. Distress. Aversive,
and Sad). The Affect Codes are determined by
the target child's facial expression, body language, and/or tone of voice. A list of all behavior codes, definitions, examples, and decision rules are contained within the IPC manual
(Rusby etal, 1991).
Infonnation on training coders. The
OSLC website provides information regarding
coder training procedures schedules
{www.oslc.org/training). A handheld computer
is used for coding, and ihe IPC appears to be a
relatively complex system to learn, as it takes
coders approximately 20 hours per week for
12 weeks to achieve reliability (J. M. Eddy,
Personal Communication, June 17, 2(K)5).

477

School Psychology Review, 2005, Volume 34, No. 4

a
S

"tj

o 3
c ou
>,
ra
^-

"H.
a.

o
U

vali

cli
ra

c
ra

ra

1/:

3
C

tuden ts;
)iscrii
ariab
ontex

X
IU

ra

atio

>.,

3
O

.2

pph ble
ucal

o 'E

:ure

(U

i/j

i<

ct)

1/1
1/1

gu
JS

o
c3 3
u 3
1/1

S
oa

00

ra
1/3

.: 3O
u.
-^
OD

T3

ra

w S UJ >

X
U
u
.p

o
a
o
X

.a

a.
E

M-T;

o
y

=y

S.'

CS

ra

'

UJ

478

Acti

>

ntei

"ra

tu

av lab

terpersi

"

aygrou
lassroolm;

n
.2

othe:
Affe
non\

Com

^*

ra

00

a:

a.

a.

c3

=^

ra

> ra -s M
ra xi ta. y

^0

2 s p

o
U X

2 _' ^

ra o

I1%II
U

CW Q-t

^*

Playpround-Based Observations

-2 'C
7
C
3
c

o "O
4j
jC 3
^ c

2
.E
E
^

c
g
.^
c

11

>> y

e afj

i-

o
>

c 'a

-a

>> to
W ft)

> -

^ O

60

60

. .b

> c

o a o D.<

'

^ ? c
, Ci. 6

E E ">

O T3
5

<u c
T3 .E

v> in

" -a

tj
3

1^
60

I U o

II

y . s

k.

U- U

= s o

m3

f I
o
-a

a.
o

ll
-

>

0^

o o c
ti. tt. w

U5

nn '

""

CL U5

Si f i^ S ^ S

CL

a. u o U 5 M
60

I
o *

LU

00
>.
CO

'o
2

: .E E g
^

>

2 . i ' CO 2

111

r> s ^-s

'i/j

D. C

K
O

12^i>11II

HI

rva

1.^
1 =

si
<

X)

E=

U
V

s
60
> i

CO

a!

CO

E u s:

:,- i

feb S 2i ^ 5 i*
>.::5 Ji y 5 .E

CO CO " 3

a. 3: u O

479

School Psychology Review, 2005, Volume 34, No. 4

Psychometric properties. The Content


codes of the IPC have demonstrated adequate
interobserver agreement across different studies and across playground and laboratory context, with kappas averaging approximately .70
{Dishion. Duncan, Eddy, Fagot, & Fetrow.
1994; Eddy & Reid, 2005; Reid, Eddy, Fetrow.
& Stoolmiller. 1999). Typically, the frequency
counts are recalculated per minute for these
analyses. By contrast, interobserver agreement
for the Affect codes has been somewhat marginal, with kappas averaging ,60 for laboratory settings and .48 for playground settings
(Dishion et aL. 1994; Eddy & Reid, 2005; Reid
et al., 1999). The Activity codes of the IPC
appear to have adequate interobserver agreement as a recent investigation indicated strong
percent agreement among raters within the
playground setting (average = 95%). although
kappa statistics were not calculated (Eddy &
Reid. 2005). The use of this system with multiple populations, in conjunction with other
methodological approaches, also suggests that
this system has promising convergent validity
(Dishion etal., 1994). For example, codes related to aggression and victimization within the
IPC have been highly correlated with observer
impression scores of aggression (Pearson rs
ranging from .51 to .67) and moderately associated with peer nomination and teacher
ratings of aggression and behavior (rs ranging from .13 to .29) in the playground setting (Eddy & Reid. 2(K),'i). U is unclear whether
the discriminant validity for differentiating different groups of children has been systematically investigated.
Critique. The primary strength of tbe
IPC is that it can be used within multiple school
and home settings. Further, tbe long history of
research and training from scientists at the
OSLC provides a strong track record for tbe
IPC. The detailed coding manual available online facilitates the understanding of this system by researchers interested in understanding complex social interactions among children
across diverse settings. However, the variable
levels of interobserver agreement for tbe Affect codes (especially on tbe playground) and
the complexity of tbe system make this system challenging for direct clinical application.
480

ADHD School Ohservation Code


(ADHD-SOC; Gadow, Sprafkin, &
Nolan, 1996)
Brief overview. The ADHD-SOC grew
out of observational researcb examining
children's television viewing patterns (Gadow
& Sprafkin. 1989). The ADHD-SOC was designed to provide information about the social
and antisocial behaviors of children within classroom, playground, and lunchroom settings. The
ADHD-SOC has been used in many studies to
examine medication effects for children diagnosed with Attention Deficit Hyperactivity
Disorder (ADHD), Oppositional Defiant Disorder (ODD), and Conduct Disorder (CD).
Although the ADHD-SOC can be used
with any child, it has particular application for
observing behaviors of children with ADHD,
ODD, and/or CD. It was designed for kindergarten-sixtb graders, particularly in classroom
settings. Many studies suggest its utility for
diverse samples of children in classrooms (e.g.,
Gadow. Nolan. Sverd, Sprafkin. & Schwartz,
2002: Gadow, Sverd, Sprafkin, Nolan. &
Grossman, 1999), and playground/lunchroom
settings (e.g., Nolan & Gadow, 1994; Spratlcin,
Gadow, & Grayson, 1988). Volpe, DiPema,
Hintze. and Shapiro (2(X)5) review classroom
applications of the ADHD-SOC.
Codes employed. The codes utilized
for the playground and lunchroom settings include: Appropriate Social Behavior (e.g., cooperative play, helping another). Phy.sical
Aggression (e.g., bitting, pushing), Nonphysical Aggression (e.g., kicking a chair,
responding sarcastically), Noncompliance
(e.g., not following adult instructions), and
Verbal Aggression (e.g., cursing). A focal child
partial interval recording system is used in
which each bebavior is coded for its presence
or absence during consecutive 15-second intervals. The codes and coding decision rules
are well-defined and described in the ADHDSOC manual, which is available from Cbeckmate Plus (www.checkmateplus.com). Additional codes include: Solicitauon (trying to
get an adult's attention). Specific Aggressive
Initiations and Retaliations towards children
and/or adults, Play Aggression (play fight-

Playground-Based Observations

ing), and Immature Behavior (e.g., child tantrums).


Information on training coders. The
ADHD-SOC manual (Gaduw et al.. 1996) describes training prcKedures for new coders and
methods for calculating rales of agreement. In
addition, the manual provides user-friendly
coding sheets that can be used for initial training and then adapted for clinical and research
applications. No particular professional background is required lo learn the ADHD-SOC,
although authors suggest that it is ideal for
school psychologists, special education teachers, or school social workers. Typically, training is accomplished in 20-25 hours (Gadow et
a!., 1996).
Psychometric properties. The ADHDSOC appears to be a well-established measure
that has been evaluated extensively for its reliability and validity. Adequate interobserver
agreement has been found across school settings, with kappas ranging from .60 to .94
(Gadow. Nolan. Sprafkin. & Sverd. 1995;
Nolan & Gadow. 1994). In addition, the
ADHD-SOC has been shown to be sensitive
to medication effects across school settings
(Gadow et al., 1995: Gadow. Nolan. & Sverd.
1992). Although the most dramatic findings for
medication effects were found within the classroom, the ADHD-SOC also showed significant
decreases in some playground and lunchr(M)m
negative behaviors (e.g., physical aggression,
noncompliance) across studies (Gadow et al.,
1995; Gadow et al., 1992). Convergent validity studies have demonstrated that the rate of
occurrence of some negativistic behaviors
coded by the ADHD-SOC are moderately to
highly correlated across the playground and
lunchroom context with teacher ratings on behavior rating scales (Nolan & Gadow, 1994).
Highest correlations with teacher reports were
found for observed nonphysical aggression In
the lunchroom (rs range from .30 to .73 across
rating scales). Discriminant validity .studies
indicate that the ADHD-SOC codes within the
classroom can distinguish between children
with behavioral disorders and learning disabilities (Sprafkin & Gadow, 1987) as well as between children with and without ADHD

(Gadow, Paolicelli, et al., 1992). Less clear


findings regarding discriminant validity have
emerged for ADHD-SOC codes within the
lunchroom and playground settings. For instance, one investigation showed that the
ADHD-SOC ctxles differentiated children with
ADHD from children without ADHD on multiple coding categories in the classroom setting, whereas only Noncompliance <S'\\'fcrcnU-

ated the two groups in the lunchroom setting


(Gadow. Paolicelli, et al.. 1992).
Critique. The strengths of the ADHDSOC include that the system is extremely
straightforward (e.g., using only five target
bebaviors in playground/lunchroom setting),
can be relatively easily learned and implemented by various school personnel, and the
manual is user-friendly, comprehensive, and
affordable. Second, it has well-established reliability and validity, as summarized in the
manual. Third, the system has direct application for monitoring medication effectiveness
for children with ADHD and related externalizing disorders across school settings. There
are two primary limitations. First, some studies suggest that the multiple ADHD-SOC ccxles
in the lunchroom (and possibly) playground
setting do not as readily discriminate children
with ADHD from those without ADHD as they
do in the classroom setting. Second, the
ADHD-SOC was designed for assessing behavior of children with extenializing disorders,
and thus may not be applicable for use with all
children, especially those with internalizing
disorders.
Peer Social Behavior of the Systematic
Screening for Behavior Disorders (PSB
of the SSBD; Walker & Severson, 1991)
Brief overview. The PSB is a playground-based observation system that serves
(along with the Academic Engaged Time observation) as the third stage of the Systematic
Screening for Behavior Disorders (SSBD). The
SSBD is composed of a three-stage screening
process to identify elementary age children at
risk for serious externalizing and internalizing
disorders. At Stage 1. teachers rank the lop
three "externalizers" (e.g., children who exhibit
461

School Psychology Review, 2005, Volume 34, No. 4

aggressive, impulsive, or antisocial behaviors)


and top three "intemalizers" (e.g., children who
exhibit excessive shyness, depression, or social isolation) in their classrooms. At Stage 2,
teachers complete a Critical Events Checklist
(e.g.. steals, sets fires, has tantrums) and Combined Frequency Index for Adaptive and Maladaptive Behavior on each child identified in
Stage 1 (see Walker et al., 1988, 1990). Normative data obtained from a national standardization sample of 4,500 children (Walker et al.,
1990) are then used to determine which of the
rated children are to be observed in the classroom (using the Academic Engaged Time observational system: Walker & Severson. 1991)
and on the playground (using the PSB). The
SSBD has been used across multiple state.s both
in research and clinical applications with elementary school children (e.g., Kamps et al.,
2003; Walker et al., 1990), and also with an
adapted version for preschoolers (e.g., Feil,
Walker. & Severson, 1995).
Codes employed. The codes utilized
for the PSB include: Social Engagement (e.g.,
verbal or nonverbal social interaction between
students), Participation (e.g., playing a game
or activity). Parallel Play (e.g., doing an activity next to, but not interacting with, another
child), Alone (e.g., not participating in a game
or activity), and Not Codeable. Social Engagement and Participation are further designated
as being positive (e.g., complements) or negative (e.g., name calling) in quality. A focal child
partial interval recording system is used in
which each behavior is coded for its presence
or absence during consecutive lO-sccond intervals for 12-I5-minute periods (depending
on the nature of the study). Definitions, specific examples, and detailed coding decision
rules are contained within the PSB training
manual (Walker & Severson, 1991).
Infonnation on training coders. The
manual and related materials are extremely
thorough and user-friendly, and they include
definitions, examples, decision rules, quizzes,
coding sheets, and training videos, all of which
are avaiiabie from Sopris West. In a large-scale
study of the SSBD. Walker and colleagues
(1994) trained coders in 3 hours and these cod482

ers achieved adequate interobserver agreement


across codes (88% agreement). Finally, the
manual indicates that diverse school personnel can be trained to use the PSB.
Psychometric properties. The SSBD
is a well-established and widely used screening tool, and the PSB has consistently been
used as an integral part of this sequential
screening process. On occasion, the PSB has
also been used by itself. The SSBD manual
provides some reliability and validity data for
the PSB outside the context of the SSBD.
Interobserver agreetnent ratios were consistently in the acceptable range (total percent
agreement ranging from .78 to .90 for the PSB
codes) during initial testing and refinement
(Walker, Hops. & Greenwood, 1984: Walker
& Severson, 1991). There is also some evidence of convergent validity for the PSB behavior codes as they correlated at a low lo
moderate level (rs ranged from. 12 to .27) with
teacher ratings of adaptive and maladaptive
behaviors. The SSBD (e.g., all three stages
considered together) has been subjected to extensive reliability and validity studies and replication investigations. For instance, two largescale studies suggest that the psychometric
properties of the assessment tools that comprise Stages 1, 2, and 3 demonstrate high levels of reliability (interobserver agreement, intemal consistency, and test-retest reliability
where appropriate), adequate convergent validity (correlations between SSBD measures
and parent ratings of children's externalizing
behavior), and strong discriminant validity for
differentiating between students identified as
severely emotionally disturbed from their
classmates (Walker et al., 1990) and
externalizers from internalizers and normal
control children (Walker & Severson, 1991).
Critique. The PSB has great utility as a
component of the larger SSBD screening assessment procedure. Its main strength i.s its use
in conjunction with Stage ! and 2 procedures
to identify high-risk externalizing and internalizing elementary and preschool students. It
has been viewed as quite acceptable and costefficient in ciinical applications across the
country (e.g.. Walker et al., 1994), and is bol-

Playground-Based Observations

stered by strong psychometric properties that


have been replicated across several large-scale
and national studies. One limitation ofthe PSB
is that it has not been thoroughly evaluated as
a stand-alone observation system. Thus, further research is needed to test the reliability of
the PSB as a stand-alone observational tool.
Second, the PSB has been used primarily as a
screening tool for at risk-children with externalizing and internalizing difficulties. Thus, it
is unclear what utility it has for assessing children with other problem patterns, such as
ADHD or children who have nol already been
prcscreened with SSBD Stage 1 and 2 measures. Nevertheless, the SSBD sy.stem appears
to have great utility as a cost- and titne-efficient universal screening method and as a research tool.

Brief Review of New Observational


Systems

horts of predominately European American


children enrolled in one elementary school in
a low socioeconomic status, mixed industrialresidential neighborhood.
Adequate
interobserver agreement (kappa) and intraclass
correlations (ICC) were found for many ofthe
behavioral codes, and initial convergent validity and predictive/discriminant validity studies are quite promising from validation data
collected from the same students in third to
fourth grade. In sum, strengths ofthe C-POC
include well-defined behaviors, coding flow
chart sheets, and promising initial reliability
and validity .studies. A limitation is that the
system has been used with predominately European American kindcrgartners and first graders from only one school.
Ostrov Early Childhood Play Project
Ohservation System (ECPPOS; Ostrov,
2005)

Brief overview. The ECPPOS is a focal child continuous event recording system,
in which a target child's behavior is coded any
Brief overview. The C-POC was time he or she exhibits target behaviors during
adapted from Weiss, Dodge. Bates, and Fettit's the observation period (which ranges from 3(1992) playground observation system, and is
10 minutes depending upon the study). Codes
a ftxal child partial interval system that codes include aggressive (e.g.. Relational. Verbal,
for the presence or absence of seven different and Physical) and play behaviors (e.g.. Coopchild behaviors across consecutive 10-second
erative. Parallel Play. Solitary Play). The sysintervals for 5-minute periods. Primary codes tem is designed for observing preschoolers ( 3 include Negative Interaction (e.g., physical or 6-year-olds) across several different contexts
verbal aggression). Rough Play (e.g., play- (classroom play period, recess period, or confighting). Positive Interaction (e.g., conversa- trived color task).
tion, cooperative play). Parallel Play. Solitary
To date, two studies have been conducted
Focused Play (e.g., playing alone on swings),
using the ECPPOS with predotninatcly EuroSolitary Unfocu.sed Play (e.g., wailing to play), pean American middle class samples (Ostrov
and Other. Two victitnization behaviors are & Keating. 2004; Ostrov, Woods. Jansen,
also coded: Victim of Physical Aggression, and Casas. & Crick, 2004). Findings indicate that
Victim of Verbal Aggression. It takes less than adequate interobserver agreement for the
6 hours to train coders and a Coding Flow Chart
ECPPOS codes has been established across
Sheet helps facilitate learning the system.
both free play conditions and the .structured
The C-POC has been used in one pub- coloring task, with average ICCs ranging from
lished investigation (Snyder et al.. 2003), and
.74 to ,95 per code (Ostrov & Keating, 2004).
one manuscript currently under review
In addition, there is initial evidence for the
(Schrepferman. Snyder, Prichard, & Suarez, convergent validity, cross-situational stability,
2005). Both write-ups present different aspects and short-term stability of the aggressive beof the findings from extensive reliability and
havior codes (Ostrov & Keating. 2004). For
validity tests conducted with observations durinstance, behavioral observations and teacher
ing kindergarten and first grade for three coreports of Relational Aggression were modcr-

The Child-Peer Ohservation Code (CPOC; Snyder et al., 1998)

483

School Psycholojiy Review, 2005, Volume 34, No. 4

ately correlated for both girls (r = .54) and boys


(r = .48), whereas observations and teacher
reports of Physical Aggression were moderately correlated for boys (r = .40), but not for
girls (r = .03). There also were some crosssituational (stmctured and free play contexts)
associations for observations of Physical Aggression for boys (r = .64), and for observations of Physical, Relational, and Verbal Aggression for girls (rs ranging from .45 to .48).
The strengths of the ECPPOS include
that it can be used across multiple school settings, the codes are well-defined, training procedures are well-articulated in the manual, and
initial reliability and validity data are promising. Further, it is one of the only systems to
code Relational Aggression, such as gossiping and group exclusion (Crick & Grotpeter,
1995). Limitations of the ECPPOS are that it
has been used in only a limited number of studies with predominately European American
middle class preschoolers, and that more extensive psychometric studies are warranted. As
such, its applicability and transportability to
older children and more diverse samples is
unknown and should be investigated in future
studies.
The Student Interaction In Specific
Settings (SISS) Tool (Cushing &
Horner, 2003)
Brief overview. The SISS is a schoolwide observation system that provides data on
the rate of negative behaviors exhibited by students across three unstructured school settings
(e.g., playground, hallways, cafeteria) and the
subsequent social consequences associated
with those behaviors. Although the level of data
collection is at the individual child level, the
system ultimately provides school-level data
that are related to overall school elimate.
Within each school setting, a 3-by-6-meter area
is designated, and then all students entering
the observation area within a 5-minute period
are observed to see wbether any child within
that area exhibits any codeable behaviors
within the following three dimensions: (a) Rule
Violations/Low Intensity Behaviors (e.g.. Running, Littering); (b) Moderate Intensity Behaviors (e.g., Taunting/Teasing, Verbal Disruption,
484

Profanity, Physical Disruption); and (c) Higb


Intensity Behaviors (e.g.. Verbal Abuse, Physical Aggression). After each codeable bebavior, a social consequence is coded after 5 seconds, which includes: (a) Positive (the response
is desirable and is likely to reinforce the behavior continuing), (b) Negative (the response
is aversive and is likely to make the behavior
less likely to occur), and (c) No Response. More
detailed information on how to code social
consequences is provided in the manual
(Cushing & Horner, 2003). At the end of each
observation, coders tabulate the frequency of
each behavior per observation minute and the
number of times each behavior was followed
by a positive, negative, or no response.
The two studies by Cushing, Homer, and
Barrier (2003) provide initial reliability and
validity data for the SISS. In Study 1, the average kappa between two observers was .85
for Low Intensity Bebaviors and Moderate Intensity Behaviors in elementary schools and
.88 for Low and Moderate Intensity Behaviors
in middle schools. To determine how many
observation sessions are required to reach a
stable pattern, tbe authors conducted a number of conditional probability analyses related
to main viu-iables of interest. On the basis of
these analyses, the authors suggest that for
some schools, it only took 4-5 days of observations to obtain an accurate assessment. In
Study 2. a number of correlations were found
in the expected directions between Low and
Moderate Intensity Behaviors on the SISS and
subscales of a eommonly used school climate
survey completed by teachers (rs ranging from
.52 to .62), but only with limited indices on a
broader school-wide evaluation tool and on
office referrals.
The SISS is different from the other five
systems reviewed in that the ultimate goal is
not to learn information about a particular
child's functioning, but instead to provide information about a schools' overall social environment. Strengths of the SISS are its focus
upon overall school social climate, its application across multiple unstructured school settings, and that it appears relatively straightforward to learn and utilize. Limitations of the
SISS are that it focuses solely on inappropri-

Playground-Based Observations

ate behaviors and that more reliability and validity studies are needed to strengthen empirical support for this innovative system.
Summary and Implications
Although educators and researchers have
recognized the importance of examining
children's ptay and aggressive behaviors in
unstructured settings including the playground during recess (see Leff. Costigan. et
al.. 2004), we found only six systems that
met our criteria for potential applicability
to school-based research and practice, which
included the availability ofa manual for training and use of the coding system. We found
that thtee ofthe systems were well-established.
und Ihal the remaining three were newer systems that met our criteria for review, but had
not been subject to as extensive empirical scrutiny. Our review helped to elucidate the following considerations for research and clinical applications.
Examining Behaviors Across Multiple
School Contexts
Four of the six observational systems
reviewed have applicability across multiple
school contexts, including structured settings
(e.g., classroom, simulated play setting) and
unstructured settings (e.g.. playground and
lunchroom). For example, the IPC and
ECPPOS have been used in classrooms, playgrounds, and laboratory settings, whereas the
ADHD-SOC has application in playground,
lunchroom, and classroom settings. Further, the
SISS has been used within a number of unstructured school contexts. There are three related advantages of observing children's play
behavior and aggression across structured and
unstructured school settings. First, observers
can determine whether similar behaviors are
pervasive across different settings. Second, if
behaviors are not pervasive across settings,
observers can identify the primary settings in
which particular behaviors do occur. Third,
observers can identify behaviors that may be
less prevalent in one setting versus the other.
Conducting observations across tnultipie settings, including playground settings, may pro-

vide important information for adapting intervention plans for specific children.
Including Both Antisocial and Prosocial
Behaviors
Almost all of the programs reviewed included a range of codes including both antisocial behaviors (e.g., aggression, noncompliance) as well as pro.social behaviors (e.g.,
cooperative play, parallel play, rough play).
Given that the practical utility of these systems may be to tnonitor the effectiveness of
aggression prevention and health protnotion
programs in the schools, it is important to include indices of both negative and positive
.social behaviors.
Carefully Considering the Clinical
Utility and Transportahility ofa System
Several of the eoding systems reviewed
have direct clinical application (see Clinical
Implications section). As such, these systems
tend to have clear and descriptive manuals,
well-defined behaviors, easy to use coding
sheets, and well thought out decision rules.
These are important criteria for which to assess when deciding upon the clinical, or research utility, of a system.
Research Implications
Our review ofthe six observational systems highlights the need for more psychometric studies, especially with more ethnically and
geographically diverse samples for many ofthe
most promising playground-based observational systems. In particular, systems should
begin investigating the discriminant and predictive validity for using the primary codes
within their systems. In addition, sitiiilar to the
PSB and ADHD-SOC, we recommend that
future descriptions of other observation systems include more detailed information on the
training procedures and ease of use. as well as
the coder's fidelity in learning and following
the directions of the systems. Finally, it also
important that more school-wide systems be
developed like the SISS, so that the effects of
school-wide prevention progratns can be more
easily monitored.

485

School Psychology Review, 2005, Volume 34, No, 4

Clinical Implications
A number of the programs reviewed have
direct implicatiotis for use in the schools by
school psychologists and other school staff. For
instance, the ADHD-SOC can be used to monitor the effectiveness of medication or psychosocial interventions with students with ADHD
and comorbid externalizing disorders. Furthermore, the ADHD-SOC manual include.s information on how to adapt coding sheets to fit
one's particular needs, which makes the
ADHD-SOC a feasible system for use across
classroom and playground/lunchroom settings.
Although both the C-POC and ECPPOS are
relatively new systems that require further validation and study, they also appear to have direct clinical utility for assessing the often complex social interactions and behaviors of
preschoolers. For screening purpo.ses, the PSB
appears to be a useful assessment tool, especially when used in conjunction with Stage I
and 2 screening tools of the SSBD, for identifying pre.school and elementary students who
are at-risk for externalizing and internalizing
disorders. Finally, although the SISS requires
more research, it holds great promise for
school-wide applications across many unstructured school contexts in which aggression frequently occurs.
Study Limitations
Our methodology to identify and review
playground-based observational systems, although systematic, has several limitations.
First, it was not possible to directly cotiipare
systems to one another because of different
primary target behaviors and different methodologies and statistics employed across studies of the systems. Second, it is possible, especially with the well-established systems, that
our review missed some relevant studie.s published about the systems over the past 20 years.
As such, it should be noted that conclusions
are ba.sed upon the specific articles and coding manuals reviewed, and brief communications, where appropriate, with authors. Third,
because of our extensive inclusion criteria, the
systems reviewed are likely not representative
of all playground-based observational systems.
486

For instance, systems were excluded from review if they were created only for a specific
study or if they had not been subjected to extensive empirical scrutiny. Finally, the observational systems of several pioneers in the field
(e.g., Anthony Pellegrini, Debra Pepler) were
not reviewed, as these researchers appear to
have adapted their observational procedures
and target codes several times to fit the purposes of a particular study and its hypotheses.
Thus, it would have been challenging to describe a "unified" system emanating from their
work, despite their considerable impact on the
field.
Challenges
There are also a number of inherent challenges associated with use of playground-ba.sed
observational system.s. These include: (a) accurately coding low base-rate behaviors and/
or behaviors that are often rather subtle in nature (e.g., affect, relational aggression); (b)
applying the systems on a regular basi.s. given
that they can be time-consuming and labor-intensive; and (c) the investment of time and
expense required for training in the use of the
systems.
In sum, although there has not been as
much empirical research to develop playground-based observational systems as there
has been for classroom-based systems, the use
of playground-based observation system holds
great promise for both research and clinical
applications. School psychologists can play an
integral role in applying these systems for assessment and progress monitoring as well as
training other school staff in their use across
unstructured school contexts. Finally, playground-based observational systems can be
used as part of a multimethod assessment battery, corroborating parent and teacher reports
for problem behaviors, and identifying specific
target behaviors for progress monitoring in
interventions.
References
Brener. N.. Krug. E, G.. Dahlberg. L. L.. & Powell. K. E.
(1997). Nurses' logs as an evaluation tool for sehoolbased violence prevention programs. Journal nf School
Health. 67. 171-174.

Playground-Based Observatioriii

Crick, N. R.. & Grotpeter, J. K, (1995), Relational aggression, gender, and social-psyehological adjustment.
Child Development. Aft. 710-722.
Cushing, L. S,, & Homer. R. (2(K)3). Sliiilenr intenirtinn
in xpecific scning.s f.SISS I measure, coiling procedures
and definitions. Unpubiished coding manual. (Available from Peabody College at Vanderbilt University,
Box 328, Nashville. TN 37203)
Cushing. L., Horner. R,. & Barrier. H, (2003). Validation
and congnienl validity of a direct observation lool lo
assess student social climate. Journal of Positive Behavior. 5. 225-237.
Dishion. T.. Duncan. T, H,, Eddy. J. M., Fagot. B. 1,. &
Fclrow. R, A, (1994), The world of parents and peers:
Coercive exchanges and children's social adaption.
Social Devehpmcni. 3. 255-26S.
Edily, J. M,. & Reid. J, B, (2(H15), Liukiitfi ihe inierests of
fiimities ami icachers (LIFT): buerpfrsoniil pmcess
CIHIC reliability ami validity score.s. Unpubiished raw
data.
Feil. E. G.. Walker. H. M., & Scverson. H. (1995). The
Early Screening Projecl for young children with behavior problems. Journal of Emotional & Behavioral
Disorders. 3. 194-202,
Gadow. K. D,. Nolan. E. E.. Sprafltin. J.. & Sverd. J.
(1995), School observations nf children wilh ADHD
and comorbid lie disorder: Effects of niethylphenidate
treatment. Journal i>f Developmental and Behavioral
Pediatrics. 16. 167-176.
Gad.)w. K. D.. Nolan. E. E.. & Sverd. J. (1992). Methylphenidate in hyperactive boys with cnmorbid tic disorder: li. Behavioral effects In sch(K>l settings. Journal of the Americun Academy of Child and Ailolescenr
P.vu/a(^^..^/. 462-471.
Uadow, K. D.. Nolan. E., Sverd. J., Sprafkin. J., &
Schwanz. J. (20()2). Anxiety and depressit)n symptoms
and response lo methylphenidate in children with Attention-Deficii Hyperactivity Disorder iind Tic Disorder. Journal of Clinical Psychopharmacology, 22. 267274.
Gadow. K, D,. Paolicelli. L, M.. Nolan. E. E,, Schwartz,
J.. Sprafkin. J., & Sverd. J. (1992). Methylphcnidate
in aggressive hyperactive boys; 11. Indirect etTects of
medication on peer behavior Journal of Child and
Adolescent Psychopharmacology. 2. 49-61.
Gadow. K. D,. & Spraftin. J, (1989), Field experiments
of television violence with children: Evidence for an
environmental hazard? Pediatrics. W 3.19-405.
Gadow. K, D,, Spralkin. J,. & Nolan. E, E, (1996), ADHD
School Observation Code. Stony Brook. NY: Checkmate Plus.
Gadow, K, D,. Sverd. J.. Sprafliin. J., Nolan. E. E.. &
Grossman. S. (1999). Long-term methylphenidate
therapy in children with comorbid attetition-deficit
hyperactivity disorder and chronic multiple tic disorder. Archives of General P.sychiatry. 56. 330-336.
Goodman. M. R,. Stormshak. E., & Dishion, T, J. (2001).
The significance of peer victim I/.al ion at iwo points in
development. Applied Developmental Psychology. 22.
507-526.
Hintze, J. M,. Volpe. R, J.. & Shapiro. E. S. (2(XI2). Best
practices in the systematic direct observation of student behavior. In A. Thomas & J. Grimes (Eds.), Best

practices in school psychology-IV (pp.993-1006).


Bethesda. MD: National AsstKialion of School Psychologists.
Kamps. D., Wills. H.. Greenwnocl. C . Thortie. S..
La/o. J,. Crockett. J,. Akers, J,, & Swaggart, B.
(2003). Curriculum influences on growth in early
reading fluency for studetits with academic and
behavioral risks. Journal of Emotional A Behavioral Di.wider.s. II. 211-224.
Knutson. J. F . DeGarmo. D. S.. & Reid, J. B, (2004). Social disadvantage and neglectful parenting as precursors to the devclopmciii of iinnsociiil child behavior
Testing a theoretical model. Aggressive Behavior. M).
187-205.
Ladd. G. W,. & Kwhenderfer-Ladd. B, (2002), Identifying victims of peer aggression from early lo middle
childh(Kid: Analysis nf cross-inforniani duta ftir concordance, estimation nf relatinnal adjiisiment. prevalence of victimization, and characleristics of idcnlified victims. Psycholoiiical Assessment. 14. 74-96.
Leff. S. S.. Costigan. T. E.. & Power, T. J. (2004). Using
participatory-action research to develop a playgroundbased prevention program. Journal of School Psvcholo}>y.42.i-2\.
Leff. S. S.. Kupersmidl. J, B,. Patterson. C . & Power, T.
J, (199Q), Factors influencing teacher predictions of
peer bullying and victimization. School P.'iychology
Review, 2H. 505-5 \1.
Leff. S. S,. Power, T J.. & Goldstein. A. (2004). Outcome
measures tn assess the effectiveness nf bullying prevention programs in the schools. In, D, L, Espelage &
S, M, Swearer (Eds.). Bullying in American .scluxtls:
A social-ecological perspective on prevention and intervention (pp. 269-294), Mahwah, NJ: Lawrence
Erlbaum Associates.
Leff. S. S,. Power. T. J.. Manz., P. H., Costigan. T. E.. &
Nabt)rs, L, A. (2001). Schonl-based aggression prevention programs for young children: Current status
and implications for violence prevention. School Psychology Review. JO. 343-360,
Nansel. T. R.. Overpeck. M.. Pilla. R. S,. Ruan. W. J..
Simons-Moiion. B,. & Scheidt. P (2(HII), Bullying
behaviors among U,S. youth: Prevalence and ;issnciatinn with psychological adjustment. Journal of the
American Medical Association, 2iS.S, 2094-21 (K),
Nolan. E. E., & Gadow, K. D, (1994), Relation between
ratings and observations nf stimulant drug response in
hyperactive children. Journal of Clinical Child Psychology. 23. 7K-90.
Ostrov. J. M. (2005). Early Childhood Play Project
(ECPP) Ohservational coding manual. Unpublished
coding manual. Department of Psychology, llniversity
nf Buffalo. NY,
Ostrnv. J, M,, & Keating, C, F, (20fW), Gender difference
in preschool aggression during Ircc play and structured
interactitins: An observational study. Social Development. 13. 256-277.
Ostrnv. J. M.. Woods. K,. Janscn. E.. Casas. J,. & Crick.
N. R. (2004), An observational study of delivered and
received aggression, gender, and social-psychological
adjustment in preschool: "This White Crayon Doesn't
Work." Early Childhood Research Quarterly. 19.355371.
487

School Psychology Review, 2005, Volume 34, No. 4

Pellegrini. A. D. (2002). Rough and tumble play from


childhood through adolescence: Development and possible functions. In R K. Smith & C. H. Hart (Eds.).
Blackwell handbook of childhood social developmetU
(pp. 438-453). Maiden, MA: Blackwel! Publishing.
Pellegrini, A.. & Bjorklund, D. (1996). The place of recess in school: Issues in ihe role of recess in children's
education and development. Journal of Research in

Child Education, ll.i-n.


Reid. J.. Eddy, J., Fetrow, R. A.. & St(H>lmiller. M. (1999).
Description and immediate impacts of a preventive
intervention for conduct problems. American Journal
of Community Psychology, 27, 483-517.
Rusby. J.. Estes. A., & Dishion. T. (1991). The Interpersonal Process Code. Unpublished coding manual.
(Available from Oregon Social Learning Center, 207
East 5"' Avenue. Ste. 202, Eugene. OR)
Samples, F., & Aber, L. (1998). Evaluations of schoolbased violence prevention programs. In D. S. Elliot,
B. A. Hamburg, & K. A. Williams (Eds.), Violence in
American schools (pp. 217-252). New York: Cambridge University Press.
Schrepferman. L., Snyder. J., Prichard, J.. & Suarez,
M. (2005). A system for the observations of
children's social interactions with peers: Reliability and validity with natural raters. Manuscript
submitted for publication.
Snyder, J., Kilgore, K.. Schrepferman. L. M.. Brooker, M..
Suarez, M., & Prichard, J. (1998). Child-Peer Observation Code. Unpublished technical manual, Wichita
State University, KS.
Snyder. J., Brooker. M.. Patrick. M. R., Snyder, A.,
Schrepferman. L., & Sioolmiller, M. (2003). Observed
peer victimizaiion during early elementary school:
Continuity, growth, and relation to risk for child antisocial and depressive behavior. Child Development,
74. 1881-1898.
Sprafkin, J., & Gadow. K. (1987). An observational study
of emotionally disturbed and learning-disabled children in school settings. Journal of Abnormal Child
Psychology, 15, 393-408.
Sprafkin. J., Gadow. K., & Grayson. P (1988). Effects of
cartooas on emotionally disturbed children's social

behavior in school settings. Journal of Child Psychology APsychiatry. 29, 91-99.


Stoolmiller. M.. Eddy, J. M.. & Reid. J. B. (2000) Detecting and describing prevenlative intervention effects in
a universal school-ba.sed randomized trial targeting
deiinqueni and violent behavior. Journal of Consulting and Clinical Psychology, 68. 296-306.
Terry, R., & Coie, J. D. (1991). A comparison of methods
for defining sociometric status among children. Developmental Psychology. 27. 867-880.
Voipe. R. J.. DiPerna. J. C . Hintze, J. M.. & Shapiro. E. S.
(2005). Observing students in classroom settings: A
review of seven coding schemes. School Psychology
Review, 34. 454-473.
Walker, H. M., & Hops, H., & Greenwood, C. (1984). The
CORBBH research and development model: Programmatic issues and strategies. In S. Paine, T. Bellamy. &
B. Wilcox (Eds.), Human services that work (pp. 5778). Baltimore: Paul H. Brookes.
Walker, H. M., & Severson, H. H. (1991). Systematic
Screening for Behavior Disorders. Longmont, CO:
Sopris West.
Walker, H. M., Severson. H. H.. Nicholson. F., Kehle. T.,
Jenson, W. R., & Clark. E. (1994). Replication of the
Systematic Screening for Behavior Disorders procedure for the identification of children. Journal of Emotional and Behavioral Disorders. 2, 66-11.
Walker, H.. Severson. H.. Stiller. B.. Williams, G.. Haring,
N.. Shinn. M., & Todis, B. (1988), Systematic screening of pupils in the elementary age range at risk for
behavior disorders: Development and trial testing of a
multiple gating model. Remedial and Special Education. 9, %-\ A.

Walker, H. M., Severson, H. H.. Todis, B. J.. BlockPedego, A. E.. Williams, G. J., Haring, N. G., &
Barckley, M. (1990). Systematic Screening for
Behavior Disorders (SSBD): Further validation,
replication, and normative data. Remedial and Special Education. II, 32-46.
Weiss. B., Dodge, K., Bates, J., & Pettii, G. (1992). Some
consequences of early harsh discipline and a maladaptive social information processing style. Child Development. 63. \'i2\-l2'i5.
.
II.

Stephen S. Leff, PhD. is an Assistant Professor of Clinical Psychology in Pediatrics at


the University of Pennsylvania School of Medicine and a Psychologist at The Children's
Hospital of Philadelphia. He directs two NIMH-futided intervention programs: the Friend
to Friend Program, a group intervention program for inner-city relationally aggressive
girls, and the Preventing Relational Aggression in Schools Everyday (PRAISE) Program, a classroom-based relational aggression intervention. His publications and research
interests include aggression prevention in the urban schools, gender and social cognitive
differences in the expression of aggression, and partnership-based approaches to measurement and intervention development.

488

Playground-Based Observations

Rebecca Lakin, PhD, is a Psychology Fellow in the Leadership Education in


Neurodeveloptiiental Disabilities (LEND) Program at the Children's Hospital of Philadelphia. As part of this MCHB-t'unded program, she contributes to several interdisciplinaty research projects and (raining activities, including Dr. Stephen Leff .s Friend to
Friend and PRAISE Programs. She is also developing the NlCHD-funded Youth Empowerment through Service (YES) Program, a youth community service intervention
to promote positive identity developmeni in urban. African American youth. Her research interests include resilience and prevention programming in urban schools and
adolescent cultural and civic identity development.
f

48')

Anda mungkin juga menyukai