You are on page 1of 11

Child Psychiatry Hum Dev

DOI 10.1007/s10578-013-0368-8

ORIGINAL ARTICLE

The Peer Aggressive and Reactive Behavior Questionnaire


(Parb-Q): Measurement Invariance Across Italian
and Brazilian Children, Gender and Age
Juliane Callegaro Borsa • Bruno Figueiredo Damásio •

Denise Ruschel Bandeira • Paola Gremigni

 Springer Science+Business Media New York 2013

Abstract This study examines measurement invariance, Keywords Aggressive behavior  Reactions to aggression 
reliability and scores differences of the Peer Aggressive Children  Measurement invariance  Cross-cultural
and Reactive Behaviors Questionnaire (PARB-Q) across comparisons
Italian and Brazilian children, gender and age. Participants
were 587 Italian and 727 Brazilian children, aged
7–13 years from 12 elementary schools. The PARB-Q is a
brief self-report instrument composed by two scales that Introduction
assess aggressive behavior and reactions to peer aggres-
sion. Multigroup confirmatory factor analyses indicated In recent years, aggression in schools has become a central
full measurement invariance of the PARB-Q across groups issue of concern in many countries around the world,
based on country, gender and age, providing support for because of high prevalence indices and immediate and long-
the unidimensionality of the first scale (direct peer term negative consequences for all involved [1]. In parallel,
aggression, PA) and a 3-factor model of the second scale the way that children react to aggression is becoming an
(reactive aggression, RA; seeking teacher support, STS; important focus of research, because researchers are inter-
internalizing reaction, IR). Reliability indices were good ested in identifying effective strategies to reduce the impact
for all factors. Italian children reported a higher frequency of aggression on the victims [2]. However, research on
of PA and a lower frequency of IR than the Brazilian aggressive behavior and research on reactions to victim-
children. Boys scored higher than girls on PA and RA, ization have been poorly integrated with each other. We
while girls scored higher than boys on STS and IR. believed that studying together these aspects can provide a
Younger children reported a lower frequency of PA and a more complete picture of peer relationships.
higher frequency of STS than older children. Results Collecting data in different cultures to compare results is
provide support for structure validity and reliability of the vital to many practices in applied psychological research.
PARB-Q in two countries and information on differences For this reason aggression in schools has been studied and
related to gender, age and culture in peer relationships in compared between a variety of different countries around the
elementary school. world [3]. Nevertheless, in general, the differences on the
prevalence of aggressive behavior that may be found
between countries can be attributed not only to the cultural
characteristics of each country, but also to methodological
differences in data collection [4]. Currently, there is a need of
J. C. Borsa  B. F. Damásio  D. R. Bandeira
Institute of Psychology, Universidade Federal do Rio psychometrically sound instruments to be cross-culturally
Grande do Sul, Porto Alegre, Brazil used, aiming to better comprehend the phenomenon we are
studying. It requires that the measurement instruments are
P. Gremigni (&)
cross-cultural valid, namaly that test scores from different
Department of Psychology, University of Bologna,
Viale Berti Pichat, 5, 40127 Bologna, Italy countries measure the same construct of interest on the same
e-mail: paola.gremigni2@unibo.it metric [5]. This is related to the concept of cross-group

123
Child Psychiatry Hum Dev

measurement invariance (MI) that is important as a pre- consequences that affect both mental health and school
requisite for meaningful cross-group comparison [6]. performance. They include psychosomatic symptoms, anx-
iety, depression, low levels of self-esteem, social with-
Aggressive Behavior drawal, loneliness, peer social problems, conduct problems,
poor school adjustment, and high rates of absenteeism from
Aggressive behavior is defined as a physical or verbal school [22–26].
behavior aiming to cause physical or psychological harm to
a person or a group of people [7]. Aggression will be Reactions to Peers’ Aggression
viewed here as a form of social behavior involving direct
interactions between children [8]. Although aggressive Children who are attacked by peers may use a variety of
behavior can occur in different contexts, the school is the different strategies in response to aggression and to defend
predominant setting in which it occurs [9]. themselves [25, 27–29]. The classification of these strate-
Aggression is a heterogeneous phenomenon that encom- gies, like that of aggressive behavior, is empirically derived
passes a wide variety of behaviors and definitions; there- more than theory based. This has produced a variety of
fore, attempts have been made to subtype it into more classifications, which sometimes include the same strategies
homogeneous categories. Connor [10] underlines that most in different categories. For example, youth responses to
aggression subtypes are empirically derived from statistical scenarios depicting aggressive provocation by a peer have
techniques, and that, as such, aggressive behavior classifi- been recently classified in three statistically derived cate-
cation is evidence-based as opposed to theory based. Into gories: aggression, avoidance, and assertion [30]. Aggres-
this framework, we tried to distinguish between aggression sion includes physical, verbal and relational aggression;
displaced or acted-out by a child upon a peer and reaction avoidance includes behaviors such as going to an adult rather
to aggression inflicted to a child by a peer, focusing on than confronting the aggressor directly; and assertion
elementary school children in community settings who are includes asking someone to stop doing something. In
not clinically referred. response to physical provocations, such as being shoved or
Acted-out aggression is intended here as an unprovoked, having a property damaged, youth was found to react
deliberate attack to a peer. Like proactive aggression, it is assertively, fighting back or telling an adult more than with
based on a social learning model with aggression resulting other strategies [31]. Nevertheless, assertive and aggressive
from learned behaviors [7]. On the other hand, reaction to behaviors in practice are difficult to be clearly distinguished
aggression implies the child’s experience of being attacked [32]. Responding aggressively can be included in the broader
by a peer and may include different actions. In both cases, category of reactive aggression to a provocation that is based
we focused on direct forms of aggression such as physical on the anger-frustration theory [7]. Counter-aggression has
aggression (e.g., hitting, kicking or punching other kids, been recently reported as the most common strategy used
and destroying others’ property) and overt verbal aggres- against bullying in USA school samples [2], while telling an
sion (e.g., yelling at others or name calling). Although adult at school was less used. Counter-aggression, telling an
several studies focused on both forms of direct and indirect adult and withdrawal were observed as those used more than
aggression [11], in this study we preferred to focus only on any other strategy by children in response to peer victim-
direct aggression, because indirect/relational aggression ization in British primary school samples [29]. In this clas-
appears later in childhood than either physical or verbal sification, telling an adult was viewed as a strategy based on
aggression, and does not peak until around age 11 [12]. On seeking social support rather than on avoidance. Other fac-
the other hand, large-scale longitudinal studies show that tor-analytically derived strategies were: seeking social sup-
for around 10 % of children the early high level of physical port, self-reliance/problem-solving, and distancing that were
aggression, which has its peak between 2 and 4 years, is regarded as approach strategies; and internalizing and
maintained until 11 years or older [13]. externalizing that were regarded as avoidance strategies
A reason for studying aggressive behavior in children is [33]. Internalizing included responses such as crying,
that it can negatively impact socialization among peers, since whining, pouting, and sulking, while externalizing included
physical/verbal aggression is associated with low social aggressive reactions such as taking it out on others, yelling,
preference, rejection [14–16], and high involvement in cli- throwing or hitting something.
ques with aggressive peers [17]. It also predicts maladjust- Different strategies adopted by children to react to
ment and disorders in childhood, adolescence and early peers’ aggression may have an impact on various out-
adulthood, including difficulties in academic achievement, comes. Studies on reactive aggression have linked this type
delinquency, and drug use [18–21]. of aggression to overt displays of anger under stress, hostile
On the other hand, children who are constantly exposed to attribution, deficits in social problem solving, rejection and
acts of aggression by peers show a sequela of negative victimization by peers [34–38]. Over time, negative peer

123
Child Psychiatry Hum Dev

group experiences may contribute to the development of aggressiveness of Italian children compared to those in
loneliness and social anxiety [39]. Brazil, based on the data of prevalence reported in the lit-
Seeking teacher support is instead associated with both erature [48, 49].
positive and negative outcomes. It has been found to protect A reason for Brazilian researchers to be involved in this
against internalizing difficulties [40, 41], but also to cross-cultural study was also the fact that, in the Brazilian
decrease emotional adjustement [42]. Finally, children who context, research on peer aggressive behavior is recent and
react to victimization with internalizing behavior were has not received adequate attention [52]. Currently, most
found to be prone to sadness, anxiety, fear, shame, depres- studies performed with Brazilian children address generic
sion, and low attentional regulation [43, 44]. violence or bullying behavior [53, 54], yet reaction to peer
aggression has not been extensively studied. To our
The Current Investigation knowledge there has been just one recently published study
[55] that describes an empirical categorization of strategies
Prior research shows that there is wide geographical varia- reported by a small sample of 16 children aged 9–10 years
tion in the percentage of children reporting of aggression, in response to victimization in schools. Seeking social
with aggressive acts involving 9–54 % of 11 years old support was the most popular strategy reported by the
children from 35 nations [3]. Cultural reasons may be at the interviewed children, followed by various forms of with-
origin of such variation, as aggressive behavior may be more drawal and active coping, while counter-aggression was the
culturally sanctioned in some countries than in others. For least popular response, with no gender differences.
example, while in Western countries children’s proactive To conduct reliable comparisons between the two coun-
aggression is more tolerated than reactive aggression [45], in tries involved in this study there was the need to use mea-
Chinese populations both forms of aggression are not tol- surement tools that are valid cross-culturally, namely that
erated as viewed as challenging of social harmony [46]. measure the same constructs in the same way in both cul-
Italy is a suitable cultural context for exploring aggressive tures [5].
behavior since physical aggression, in particular, occurs Although there are several instruments that evaluate
frequently in Italian childhood [47]. Moreover, physical child aggression and reactions to victimization [56], most
bullying is more prevalent in Italy than in other European of them have some limitations. The vast majority of tools
and Western countries (e.g., Norway, England, Spain, or are excessively long and complex to be filled in autono-
Japan) with almost 40 % of Italian primary school children mously by primary school children. Moreover, most tools
reporting being bullied and 20 % reporting bullying others reflects the division between the two strands of research on
at least sometimes; rates approximately twice as high as aggression and the strategies adopted in response to peer
among English children [48]. victimization, respectively [56].
Among Brazilian children, 26.5 % of fifth graders in The peer aggressive and reactive behaviors questionnaire
primary schools in metropolitan areas reported to have (PARB-Q) has been recently developed in Italy to assess
suffered threats by peers [49]. Amongst the fifth graders the both aggressive behavior and reactions to peer aggression in
violence manifested mainly through physical threats, whilst elementary school [57]. It is a brief and easy to administer
among eighth graders verbal aggression emerged most self-report instrument that has shown good psychometric
[49]. Internalizing and aggressive behavior were the pre- characteristics in the Italian context [57]. Thus, it appears to
dominant behavior problems among students from first to be a promising measure to be used also in other countries,
fourth grade of elementary schools in Porto Alegre with a but first it should prove cross-cultural MI.
prevalence of 18 and 16 %, respectively [50]. MI examines ‘‘the extent to which items or subtests have
We decided to study the phenomena of aggression and equal meaning across groups of examinees’’ [58]. As such,
reaction to aggression among peers by comparing the the concept of MI applies not only to cross-country com-
behaviors of Italian and Brazilian children, since Italy and parisons but to any group of respondents, such as those
Brazil are geographically distant, diverse in ethnical and based on gender and age. In this way it provides a contri-
racial composition, but culturally similar for several aspects. bution to the construct validity of an instrument. It is well
For example, they have languages of Latin origin, a similar known that gender differences in physical/verbal aggres-
religious background, and comparable educational systems. sion were substantial [59], and gender differences were
Both societies are also characterised by a mixture of diverse also found in the strategies adopted to cope with peers’
values and traditions, such as group-oriented attitudes aggression [60, 61]. Although there is no strong evidence
toward family members, neighbors and peers, and concern for specific age influences on aggression [62], some studies
for individual autonomy [51]. Therefore, we can expect showed younger children to react with more physical and
somewhat similar behavior among children of the two verbal aggression compared to older ones [60, 61, 63].
countries, although it is reasonable to expect also a greater Nevertheless, gender and age differences in aggression

123
Child Psychiatry Hum Dev

have mostly been based on studies utilizing a mean dif- gender. Brazilian children were 378 males (52 %) and 349
ference method. From a measurement perspective, this female (48 %), and Italian children were 236 males
method is problematic unless an aggression measure pos- (48.7 %) and 249 females (51.3 %). No gender differences
sesses comparable validity across gender and age. were found on mean age [F(1,1214) = 2.09, p = .15].
This study aimed to provide evidence of the construct
comparability of the PARB-Q across Brazilian and Italian Measures
children and across groups of respondents based on gender
and age. Reliability of the PARB-Q as internal consistency The peer aggressive and reactive behavior questionnaire
was also tested within any of the country samples. (PARBQ) [57] is a self-report instrument consisting of two
The second aim of the present study was to compare separate scales, the 8-item peer aggression scale (PA) and
children’s aggressive behavior and reactions to peer the 12-item reaction to peer aggression scale (RPA), which
aggression across country, gender and age. aim to investigate direct aggressive behavior and reactions
This study may provide some information on whether to peer aggression of children in school, respectively.
children aggression and reaction to victimization are asso- Despite all the items are answered on a 4-point Likert scale,
ciated with age and gender in a similar fashion across dif- the formulation of questions and answers varies for the two
ferent cultures. Investigation of this issue would help us scales. Items in the PA scale are preceded by the following
understand the role of context in the development of request: ‘‘usually, how many times it happens that you …’’,
aggression and design appropriate school interventions for and the response may vary from 1—‘‘it doesn’t ever happen
children from different backgrounds who experience psy- to me’’ to 4—‘‘it happens to me all the days’’. Items in the
chological distress as a results of their aggressive behaviors RPA scale are preceded by the following request: ‘‘usually,
or of being victimized by peers. when a classmate takes one of the actions described below
what do you do?’’, and the response may vary from 1—‘‘I
never do so’’ to 4—‘‘I do so all the times’’.
Materials and Methods The PARBQ has been developed in Italy as a brief self-
report tool to be used in school [57]. The dimensions to be
Participants measured were selected not only on the basis of the liter-
ature, but mainly based on the results of a focus group with
Participants comprised 1,314 children in grades two primary school teachers. In particular, teachers required a
through five, aged 7–13 years with a mean age of brief and easy to administer tool to detect certain behaviors
9.39 years (SD = 1.22 years) and a male-to-female ratio that they recurrently observe in everyday school life. In
of 1.03:1 overall. Participants were from five public addition, teachers were interested to learn the point of view
schools in Italy (n = 587) and seven public and private of children and their self-assessment of the recurrence of
schools in Brazil (n = 727). Schools were in low to med- these behaviors. In the original study [57], the two scales of
ium class metropolitan areas in Bologna, in the north-east the PARB-Q were analyzed separately due to the different
of Italy, and Porto Alegre, in the state of Rio Grande do wording of questions and answers used to measure two
Sul, in Southern Brazil. The majority of the children in the different types of behavior (i.e., aggressive behavior and
samples lived with both parents. The first graders were reactions to be attacked by peers).
excluded because, in both countries, they are generally not The PA scale evaluates direct aggressive behavior
yet fully able to read and complete a self-report measure. toward peers and consists of five items measuring physical
Significant country-group differences were found in (e.g., ‘‘how often do you kick or hit your classmates?’’) and
mean age [F (1,1312) = 273.54, p \ .0005], the Brazilian verbal aggression (e.g., ‘‘how often do you say bad things
children (mean = 9.82 years, SD = 1.14) being older than to your classmates?’’), plus three control items that were
the Italians (mean = 8.79 years, SD = 1.24). Children not considered for scoring.
were divided in two age groups: 682 (51.9 %) aged The RPA scale consists of 12 items aimed to investigate
7–9 years and 632 (48.1 %) aged 10–13 years. Significant three reactions to peer aggression that are commonly
differences were found on age distribution by country reported by children. The first reaction, named Reactive
[Chi2(1) = 139.46, p \ .0005], with 37.7 % of Brazilian Aggression (RA), includes aggressive responses directed to
children aged 7–9 years versus 62.7 % of the Italians, and the aggressor or to objects belonging to the aggressor (e.g.,
70 % of the Brazilians aged 10–13 years versus 30 % of the ‘‘when a classmate hits or pushes you, you yell at him/
Italians. her’’). The second behavior, Seeking Teacher Support
No significant differences were found on gender distri- (STS), consists of items that refer to telling the teacher
bution [Chi2(1) = 1.294, p = .26], though a noteworthy what has happened to obtain help (e.g., ‘‘when your
proportion of Italian children (17 %) did not report their classmate yells at you, you tell the teacher’’). The last

123
Child Psychiatry Hum Dev

behavior, Internalizing Reaction (IR), aims to evaluate to The research project was approved by the Federal Uni-
what extent a child presents a passive and internalized versity of Rio Grande do Sul and University of Bologna
response to peer aggression (e.g., ‘‘when your classmate Ethics Committees. Elementary schools in the metropolitan
tells you bad things, you do cry or sulk’’). Among various areas of Porto Alegre and Bologna were contacted and data
potential strategies adopted by children in response to collection was held at those schools that first agreed to par-
aggression by peers, we selected those that were most ticipate in the survey after approval by their school boards.
commonly observed by teachers in school and also repor- Parents were informed about the study and signed a consent
ted in the literature [2]. Moreover, both the internalized form. All children agreed to complete the questionnaire, and
(IR) and the externalized (RA) responses were chosen for there was a negligible number of unanswered items.
their association with negative outcomes [39, 44], while
searching the teacher’s help has been selected for its Data Analyses
potentially, albeit contradictory, positive effects [41, 42].
The PARB-Q presented adequate psychometric properties Statistical analyses were performed using SPSS 18.0,
for both the hypothesized one-factor PA scale and the 3-factor except the MGCFAs that were conducted with EQS 6.1
RPA scale. Goodness-of-fit indices of confirmatory factor [66]. All data were checked for univariate and multivariate
analyses provided support to accept the predicted models. normality.
Those indices were as follows: s-bv2/df = .32; SRMR = The confirmatory factor analysis (CFA) framework
.01; RMSEA = .001 (95 % CI = .00-.03); CFI = .99; provided a means to test the factor structure of the PARB-Q
TLI = .99 for PA, and s-bv2/df = 1.99; SRMR = .05; as an important contribution to construct-validation.
RMSEA = .06 (95 % CI = .05-.08); CFI = .93; TLI = Multi-group confirmatory factor analyses (MGCFA) is an
.92 for RPA. Internal consistency was good for both scales and extremely powerful procedure for detecting a range of
all factors, with Cronbach’s alphas ranging between .75 and cross-group differences in a measurement model [67]. To
.86. Test–retest reliability over a 6-week period was accept- cross-validate the PARB-Q in the Brazilian and Italian
able, with all ICC values above .80. contexts, we conducted a set of MGCFAs in the baseline
Criterion validity was evaluated in the original Italian models for each scale. The MGCFAs tested across-group MI
study [57], by means of correlation with the aggressive based on country, age, and gender. Groups were formed,
behavior scale of the Childhood Social Adjustment Capacity respectively, by 717 Brazilians and 587 Italians; 682
Indicators (CSACI, child self-report form) [64] and the skill younger (aged 7–9 years) and 632 older children (aged
alternatives to aggression subscale of the Teacher Skill- 10–13 years); and 600 females and 614 males. Four differ-
streaming Checklist (TSC) [65]. Results showed significant ent parameters were evaluated: Model 1 (unconstrained
positive correlations between PA, RA, IR and CSACI, sig- model/configural invariance) assessed whether the structure
nificant negative correlations between PA, IR and TSC, and of the scale (the number of factors and the items per factor)
significant positive correlation between STS and TSC, as was plausible for both the Italian and Brazilian groups. If this
expected. model does not hold, then the structure of the instrument
The Brazilian version of PARB-Q used in the present cannot be considered equal for the groups. Model 2 (equal
study was obtained by translating it carefully from Italian factor loadings/metric invariance) analyzed if factor load-
to Portuguese by the members of our international research ings of the items were equal across groups. Model 3 (equal
team. The questionnaire was then back translated to ensure intercepts/scalar invariance) investigated whether the initial
comparability with the Italian version. level of the latent variable was equal among the different
groups. Finally, Model 4 (equal factor covariance/structural
Procedures invariance) assessed to what extent the latent variance and
covariance of the latent variable(s) were equal across groups
The PARB-Q was administered to both Italian and Bra- [68]. The assessment levels of the models were ordered
zilian children using similar procedures. Data collection hierarchically. Each constrained model was nested within a
was performed by means of a collective administration of less restricted one [67]. The goodness-of-fit of the uncon-
the questionnaire in classroom. Instructions for completing strained model was evaluated using the Chi square/degree-
the questionnaire were read aloud, to assure children’s of-freedom ratio (v2/df), the standardized root mean square
comprehension. Completion took about 15 min. All par- residual (SRMR), the comparative fit index (CFI), the
ticipants were informed that the questionnaire was not a Tucker-Lewis index (TLI) and the root mean square error of
school assignment, that there were no right or wrong approximation (RMSEA). These indexes were chosen
answers and no time limit, and that every child had to because we sought to include fit indexes from several cate-
respond on its own and sincerely all questions. gories such as absolute fit (v2/df; SRMR), parsimonious fit

123
Child Psychiatry Hum Dev

(RMSEA), and comparative fit (CFI, TLI) indexes. Absolute between groups. Scalar invariance (Model 3) was tested to
fit indexes assesses model fit by evaluating the discrepancy further emphasize the evaluation of MI. The CFI difference
between the implied model covariance’s and the observed test showed that the intercepts were the same for the
sample covariance. The parsimonious fit, RMSEA, is an evaluated groups. At this point, all group comparisons
error of approximation index, which assesses the extent to could be safely conducted, since metric and scalar invari-
which a model fits reasonably well in the population. The ance were achieved [66, 67]. The last model (Model 4,
RMSEA index presents two advantages: (1) the 90 % con- structural invariance), in turn, demonstrated that the sub-
fidence interval indicates the precision of the RMSEA point jects presented the same level of the latent trait for almost
estimate; and (2) RMSEA compensates for the effect of all subgroups, except for gender in the PA scale
model complexity by conveying discrepancy in fit (d) per (DCFI = .045). This result suggests that, in this study
each df in the model [69]. Lastly, the comparative fit indexes sample, the levels of explained variance of the PA scale
CFI and TLI evaluates the adequacy of the model by com- were different for boys and girls. It is important to note that
paring the predicted model with a baseline model, which is a the structural invariance is not a measurement problem, but
null or independence model in which the covariances among is related to the population heterogeneity [66]. For the RPA
all input indicators are fixed to zero. These fit indexes imply scale, Model 4 demonstrated that the level of latent trait
a different evaluative logic, in which the presented model is was equal among all groups, and that the patterns of cor-
compared to the worst possible model [66]. Besides the fact relations among factors were also invariant. Given that
that absolute, parsimonious and comparative fit indexes are configural, metric, scalar and structural invariances were
complementary in their information, all these indexes were held, mean difference tests can be reliably conducted.
chosen because of their positive performances on Monte-
Carlo simulation studies [for a review, see 66]. Reliability of the PARB-Q
The v2/df may be less than 2 or 3; the SRMR should be
close to 0; the CFI and TLI must be higher than .90 or .95 or In the Brazilian sample, Cronbach’s alpha reliability was
close to it, and RMSEA values less than .05 indicate a good .81 for PA, and .85, .81 and .77 for RA, STS, and IR,
fit, whereas those between .05 and .08 indicate a reasonable respectively. In the Italian sample, alpha reliability was .81
fit [66]. Measurement invariance of the constrained models for PA, and .86, .84 and .82 for RA, STS, and IR,
was evaluated using the CFI difference test (DCFI) [24]. respectively. Alpha coefficients were above the expected
Significant differences observed between the goodness- limit of .70 for all scales and factors among both samples.
of-fit indices of the models (DCFI [ .01) indicate that MI
could not be achieved in the evaluated parameter. Cross-Group Comparisons
Reliability as internal consistency was assessed for each
factor of the PARB-Q in both the Italian and Brazilian The results of MANOVA showed that all the interactions
samples using Cronbach’s alpha. A limit of .70 was between the fixed factors were not statistically significant,
acceptable, while a good degree requires .80 [69]. indicating that the effects of country, gender and age were
Multivariate analysis of variance (MANOVA) was independent of each other. The three main effects were all
conducted for all the PARB-Q dimensions using country, significant at a level of p \ .0005. Results of univariate
gender, and age-group as fixed factors. analyses (Table 2) showed that in the PA scale Italian
children scored significantly higher than the Brazilians,
boys sored significantly higher than girls, and older chil-
Results dren scored significantly higher than the youngest. In RA
there were no country or age differences, whereas boys
Measurement Invariance of the PARB-Q scored significantly higher than girls. In STS there were no
country differences, while girls scored significantly higher
As it can be seen in Table 1, the unconstrained models than boys, and younger children scored significantly higher
presented excellent goodness-of-fit indexes, suggesting than the oldest. In IR Italian children scored significantly
acceptability for the factor structure of both the PA and higher than the Brazilians, and girls sored significantly
RPA scales for all investigated groups. With support for the higher than boys, with no age differences.
configural invariance, the second model (equal factor
loadings/metric invariance) was conducted to evaluate to
what extent the items were answered similarly (unbiased) Discussion
for Brazilian and Italian children, younger and older chil-
dren, and male and females. The results of this model The primary purpose of this study was to assess measure-
showed that there were no response biases on any item ment invariance (MI) of the PARB-Q across groups based

123
Child Psychiatry Hum Dev

Table 1 Results of multigroup confirmatory factor analysis for the PARB-Q (PA and RPA scales) across country, age, and gender
Group Goodness-of-fit indices
v2/df SRMR RMSEA (90 % IC) TLI CFI DCFI

Country Brazil vs. Italy PA measurement invariance


Model 1: Equal number of factors 2.074 0.012 .029 [.010–.046] 0.989 0.995 –
Model 2: Equal factor loadings 2.389 0.019 .033 [.018–.047] 0.986 0.99 0.005
Model 3: Equal intercepts 2.41 0.02 .033 [.020–.045] 0.986 0.987 0.003
Model 4: Equal latent variance 2.259 0.02 .031 [.017–.045] 0.987 0.99 0.003
RPA measurement invariance
Model 1: Equal number of factors 2.876 0.039 .038 [.033–.043] 0.959 0.968 –
Model 2: Equal factor loadings 2.759 0.04 .037 [.032–.042] 0.961 0.968 0
Model 3: Equal intercepts 2.81 0.04 .037 [.032–.042] 0.96 0.964 0.004
Model 4: Equal latent covariance 2.753 0.048 .037 [.032–.041] 0.962 0.966 0.002
Age 7–9 years vs. 10–13 years PA measurement invariance
Model 1: Equal number of factors 1.866 0.011 .026 [.004–.044] 0.991 0.996 –
Model 2: Equal factor loadings 1.676 0.014 .023 [.000–.038] 0.993 0.995 0.001
Model 3: Equal intercepts 2.15 0.016 .030 [.017–.042] 0.988 0.989 0.006
Model 4: Equal latent variance 1.583 0.016 .032 [.020–.044] 0.986 0.986 0.003
RPA measurement invariance
Model 1: Equal number of factors 2.709 0.035 .036 [.031–.041] 0.962 0.971 –
Model 2: Equal factor loadings 2.608 0.036 .035 [.030–.040] 0.964 0.97 0.001
Model 3: Equal intercepts 2.554 0.036 .034 [.030–.039] 0.966 0.969 0.001
Model 4: Equal latent covariance 2.802 0.039 .037 [.033–.042] 0.958 0.961 0.008
Gender male vs. female PA measurement invariance
Model 1: Equal number of factors 0.91 0.01 .000 [.000–.029] 1.001 1 –
Model 2: Equal factor loadings 1.46 0.02 .020 [.000–.037] 0.994 0.996 0.004
Model 3: Equal intercepts 2.017 0.017 .029 [.015–.043] 0.988 0.989 0.007
Model 4: Equal latent variance 5.56 0.039 .061 [.050–.063] 0.952 0.944 0.045
RPA measurement invariance
Model 1: Equal number of factors 2.591 0.036 .036 [.031–.042] 0.96 0.969 –
Model 2: Equal factor loadings 2.585 0.038 .036 [.031–.041] 0.96 0.966 0.003
Model 3: Equal intercepts 2.844 0.037 .039 [.034–.044] 0.953 0.957 0.009
Model 4: Equal latent covariance 3.016 0.051 .041 [.036–.046] 0.949 0.951 0.006
2
PA peer aggression scale, RPA reactions to peer aggression scale, v /df Chi square to degrees of freedom ratio, SRMR standardized root mean
square residual, RMSEA root mean square of error of approximation, TLI Tucker-Lewis index, CFI comparative fit index

on country, gender and age. MI was achieved for both Across-group comparisons indicated that boys reported
scales of the PARB-Q, with excellent goodness-of-fit higher scores on direct aggression and aggressive reaction
indexes across groups. Other tools measuring aggression than girls. This result is consistent with well-known find-
have demonstrated at least partial MI by gender in Western ings that direct physical/verbal aggression tends to be more
[70] as well as Eastern countries [71]; while for some other common in boys than girls, from early in life to adulthood
tools MI across gender was not supported [72]. To our [45, 73–76]. For example, in Brazil [77], the percentage of
knowledge, MI of aggression tools across age groups has aggression was twice larger among boys (17.5 %) than in
been implicitly invoked but not explicitly tested. girls (8.9 %). Gender difference has also been found to
Alpha values of the PARB-Q indicated acceptable to hold across cultures, such as Italian, Spanish and English
good values of reliability of all dimensions. These findings [78] or Italian, Brazilian, Canadian and Chinese [79].
suggest that the PARB-Q is a psychometrically sound Overt aggression and reactive aggression also varied
measure to be used when evaluating aggression and reac- with age, such that older children exhibited higher levels of
tions to peer aggression in males and females children, these forms of aggression in comparison to younger chil-
aged 7–13 years, in both the Italian and Brazilian contexts. dren. It was in accordance to findings of a recent study [80]

123
Child Psychiatry Hum Dev

0.05

0.06
0.01
0.02
where older children (aged 12 years) exhibited higher

g2
levels of proactive-overt aggression in comparison to
younger children (aged 9 years).

7.33*
63.51*

74.64*

20.51*

PA peer aggression scale, RPA reactions to peers aggression, RA reactive aggression, STS seeking teacher support, IR internalizing reaction, SD standard deviation, g2 effect size
In the present study Italian children had higher scores in
aggression than the Brazilians, although there were no
F

country differences on reactive aggression. Italian children


were found to score higher in aggression than the Brazil-
Mean (SD)

7.31 (2.60)

8.72 (3.50)
8.74 (2.74)
5.89 (2.36)
(n = 600)

ians in a previous study [79], where gender differences


were also larger among Italian than Brazilian children.
Girls

Results of this study showed that boys and girls differed


also in other types of reaction to peer aggression. Indepen-
dent of country, seeking teacher support was more fre-
8.60 (3.18)

10.78 (4.61)
8.27 (2.93)
5.31 (2.38)
Mean (SD)
(n = 614)

quently reported by girls than boys, in agreement with other


findings from the literature [81, 82]. Seeking support may
Boys

prove to be an effective means for dealing with situations of


victimization. However, its effectiveness is different
The total sample used in this analysis includes 1,214 participants due to the lack of information on the gender of 100 Italian children

according to gender. The practice of seeking teacher support


0.001

0.001
0.01

0.03

can be evaluated as effective for girls, but not for boys. As


g2

indicated by previous studies, boys seeking teacher support


tend to have high incidences of future victimization as well
12.99*

31.92*
1.05

1.54

as loneliness and rejection experiences [83, 84].


Table 2 Descriptive statistics and results of univariate ANOVA for the PARB-Q across country, age and gender

A cross-cultural study with Italian and English children


F

found results similar to those of the present study, with


girls being more confident than boys in seeking teachers’
Mean (SD)

8.16 (3.00)

9.96 (4.33)
7.96 (2.81)
5.60 (2.37)
(n = 565)

help in both countries [85]. Thus, this seems to be a


strategy typically feminine, suggesting that the impact of
Old

social and cultural context on this behavior may not be a


straightforward issue.
Mean (SD)

7.70 (2.93)

9.41 (4.12)
8.98 (2.79)
5.50 (2.42)

The practice of seeking teacher support also varied with


(n = 649)

age, being more frequently reported by older than younger


Young

children. This result is consistent with findings from the


literature showing that this strategy decreases in older
children, probably because of their achieved greater
0.001
0.001
0.02

0.01

autonomy in managing relationships with peers [60, 86].


g2

The internalizing reaction was also more common among


girls than boys, in accordance with the literature where girls
22.38*

15.26*

were found to be more likely than boys to respond to a


0.85
0.01

provocation with helplessness or internalizing behavior [60,


F

61, 81, 82]. These results might be due to the fact that girls
are traditionally expected to behave in a more passive and
Mean (SD)

8.24 (2.89)

9.49 (3.81)
8.73 (2.68)
5.29 (2.22)
(n = 487)

controlled manner than boys in social situations [87] and


this may affect also their reaction to aggression. The inter-
Italy

nalized strategy was also more reported by younger than


older children, in accordance with results from the literature
[60, 61]. Smith et al. [61] hypothesized that a change of
Mean (SD)

7.78 (3.02)

9.94 (4.47)
8.35 (2.94)
5.80 (2.47)
(n = 727)

strategies with age may occur as a result of learning from


Brazil

experience to cope with peer harassment.


Finally, significant country differences were found in
internalized behavior in response to aggression, where
PARB-Q-RPA
PARB-Q-PA

Brazilian children had higher scores than the Italians. We


* p \ .01

can try to read this result together with that concerning the
STS

higher scores in direct aggression reported by Italian


RA

IR

children over the Brazilians. These findings might be partly

123
Child Psychiatry Hum Dev

understood as a consequence of the cultural endorsement of behaviors questionnaire (PARB-Q) across Italy and Brazil,
social assertiveness [88] that might have recently affected across gender and age (children aged 7–9 years versus
Italy more than Brazil. As social assertiveness and com- 10–13 years). The second objective was to compare the
petitiveness are increasingly required, especially in scores of aggressive and reactive behaviors across country,
Northern Italy, children’s shy-sensitive behavior may gender and age. The PARB-Q is a brief self-report instru-
decline, whereas aggressive behavior may increase. This ment developed in Italy to assess both aggressive behavior
may occurs because the former is increasingly perceived as and reactions to peer aggression. Multigroup confirmatory
negative, indicative of emotional fragility, and therefore factor analyses indicated full MI based on country, gender
rejected, while the latter is seen as potentially related to and age for the one-dimensional peer aggression (PA) scale
social assertiveness, and, as such it is not totally rejected. and for the three-factor model of the reaction to peer
On the other hand, in Brazil, given emphasis on group aggression (RPA) scale. RPA factors were: reactive
affiliation and interpersonal harmony [79], aggression may aggression (RA), seeking teacher support (STS), and inter-
be more strongly discouraged than in Italy, whereas shy- nalizing reaction (IR). Cronbach’s alpha indicated good
sensitive behavior, perceived as associated with self- reliability for all scales and factors. Comparisons across
restraint, may be less discouraged. groups confirmed findings of previous research on the higher
Similarities and differences were found between the prevalence of aggression among boys than girls and among
Brazilian and Italian samples. Further studies investigating older than younger children. STS and IR were more pre-
more in-depth those aspects could increase understanding of valent among girls than boys and among younger than older
cross-cultural differences in children’s relations with peers. children, as was found in the literature. Independent of the
The major limitation of this study was the lack of access influence of gender and age, Italian children reported higher
to socioeconomic and cultural characteristics, as some of levels of PA and lower level of IR than Brazilian children that
those variables have been found to predict different levels of were regarded as potentially related to cultural differences
aggressive behavior among children [89]. Future research between countries. Results provide support for the structural
that investigate other context variables will enable more validity and reliability of the PARB-Q in the Italian and
in-depth assessment and interpretation of the phenomena Brazilian contexts and added information on influences of
under study. Another limitation of this study was the lack of culture, gender and age on peer relationships in school.
other criterion variables that could have add evidence to the
cross-cultural validity, particularly criterion validity of the
PARB-Q. This was mainly due to the lack of other validated
tools in either Italy and Brazil to be used as criteria. References
Findings of this study provide support for the structure
validity and reliability of the PARB-Q to assess both child 1. Nansel TR, Craig W, Overpeck MD, Saluja G, Ruan WJ, The
aggression and reaction to aggression in two countries. Health Behaviour in School-aged Children Bullying Analyses
This brief and easy to use questionnaire may be used for Working Group (2004) Cross-national consistency in the rela-
tionship between bullying behaviors and psychosocial adjust-
understanding the predictors and the trajectory of peer ment. Arch Pediat Adol Med 158:730–736
relationships in elementary school. It may also be used to 2. Black S, Weinles D, Washington E (2010) Victim strategies to
evaluate the effectiveness of interventions aiming to reduce stop bullying. Youth Violence Juv Justice 8:138–147
aggressive interactions and improve children’s social 3. Currie C, Roberts C, Morgan A, Smith R, Settertobulte W,
Samdal O et al. (2004) Young people’s health in context. Health
functioning by focusing on the skills required to manage Behaviour in School-aged Children (HBSC) study: international
aggressive situations. report from the 2001/2002 survey. WHO Regional Office for
Europe, Copenhagen
4. Taverna L, Bornstein MH, Putnick DL, Axia G (2011) Adaptive
behaviors in young children: a unique cultural comparison in
Summary Italy. J Cross Cult Psychol 42:445–465
5. Wu AD, Li Z, Zumbo BD (2007) Decoding the meaning of
Aggressive behavior is a common problem in childhood. factorial invariance and updating the practice of multi-group
When persistent, it can have a negative impact on social, confirmatory factor analysis: a demonstration with TIMSS data.
PARE 12: retrieved from http://pareonline.net/pdf/v12n3.pdf
emotional and psychological adjustment of all children 6. Bollen KA (1989) Structural equations with latent variables.
involved. It is important to develop and adapt psychological Wiley, Oxford
measures that can reliably evaluate child aggressive behav- 7. Coie JD, Dodge KA (1998) Aggression and antisocial behavior.
ior as well as reactions to peer aggression to be used across In: Damon W, Eisenberg N (eds) Handbook of child psychology:
social, emotional, and personality development. Wiley, Toronto,
different countries. This study aimed to evaluate the reli- pp 779–862
ability and structural validity, based on measurement 8. Baron RA, Richardson DR (2004) Human aggression, 2nd edn.
invariance (MI), of the peer aggressive and reactive Plenum Press, New York

123
Child Psychiatry Hum Dev

9. Thomas DE, Bierman KL, The Conduct Problems Prevention 30. Dirks MA, Treat TA, Weersing VR (2011) The latent structure of
Research Group (2006) The impact of classroom aggression on youth responses to peer provocation. J Psychopathol Behav
the development of aggressive behavior problems in children. Assess 33:58–68
Dev Psychopathol 18:471–487 31. Dirks MA, Treat TA, Weersing VR (2007) The situation speci-
10. Connor DF (2004) Aggression and antisocial behavior in children ficity of youth responses to peer provocation. J Clin Child Ado-
and adolescent: rearearch and treatment. Guilfor, New York, p 10 lesc Psychol 36:621–628
11. Björkqvist K (2001) Different names, same issue. Social Dev 32. Ostrov JM, Pilat MM, Crick NR (2006) Assertion strategies and
10:272–275 aggression during early childhood: a short-term longitudinal
12. Björkqvist K, Österman K, Kaukiainen A (1992) The development study. Early Child Res Q 21:403–416
of direct and indirect aggressive strategies in males and females. 33. Causey DL, Dubow EF (1992) Development of a self-report
In: Björkqvist K, Niemelä P (eds) Of mice and women. Aspects of coping measure for elementary school children. J Clin Child
female aggression. Academic Press, San Diego, pp 51–64 Psychol 21:47–59
13. Archer J, Côté S (2005) Sex differences in aggressive behavior: a 34. Hubbard JA, Dodge KA, Cillessen AHN, Coie JD, Schwartz D
developmental and evolutionary perspective. In: Tremblay R, (2001) The dyadic nature of social information processing in boys’
Hartup WW, Archer J (eds) Developmental origins of aggression. reactive and proactive aggression. J Pers Soc Psychol 80:268–280
Guilford, New York, pp 425–443 35. Hubbard JA, Smithmyer CM, Ramsden SR, Parker EH, Flanagan
14. LaFontana K, Cillessen A (2002) Children’s perceptions of KD, Dearing KF et al (2002) Observational, physiological, and
popular and unpopular peers: a multimethod assessment. Dev self-report measures of children’s anger: relations to reactive
Psychol 38:635–647 versus proactive aggression. Child Dev 73:1101–1118
15. Graham S, Juvonen J (2002) Ethnicity, peer harassment, and 36. Poulin F, Boivin M (2000) Proactive and reactive aggression:
adjustment in middle school: an exploratory study. J Early evidence of a two-factor model. Psychol Assess 12:115–122
Adolesc 22:173–199 37. Poulin F, Boivin M (2000) The role of proactive and reactive
16. Lee BR, Thompson R (2009) Examining externalizing behavior aggression in the formation and development of friendships in
trajectories of youth in group homes: Is there evidence for peer boys. Dev Psychol 36:1–8
contagion? J Abnorm Child Psychol 37:31–44 38. Schwartz D, Dodge KA, Coie JD, Hubbard JA, Cillessen AH,
17. Bagwell CL, Coie JD, Terry RA, Lochman JE (2000) Peer clique Lemerise EA et al (1998) Social cognitive and behavioral cor-
participation and social status in preadolescence. Merrill Palmer relates of aggression and victimization in boys’ play groups.
Q 46:280–305 J Abnorm Child Psychol 26:431–440
18. Barke ED, Vitaro F, Lacourse E, Fontaine NMG, Carbonneau R, 39. Raine A, Dodge K, Loeber R, Gatzke-Kopp L, Lynam D,
Tremblay RE (2010) Testing the developmental distinctiveness of Reynolds C et al (2006) The reactive–proactive aggression
male proactive and reactive aggression with a nested longitudinal questionnaire: differential correlates of reactive and proactive
experimental intervention. Aggress Behav 36:127–140 aggression in adolescent boys. Aggress Behav 32:159–171
19. Emond A, Ormel J, Veenstra R, Oldehinkel AJ (2007) Preschool 40. Davidson LM, Demaray MK (2007) Social support as a moder-
behavioral and social-cognitive problems as predictors of (pre)ado- ator between victimization and internalizing–externalizing dis-
lescent disruptive behavior. Child Psychiatry Hum Dev 38:221–236 tress from bullying. School Psychol Rev 36:383–405
20. Farrell AD, Sullivan TN, Esposito LE, Meyer AL, Valois RF 41. Schmidt ME, Bagwell CL (2007) The protective role of friend-
(2005) A latent growth curve analysis of the structure of aggression, ships in overtly and relationally victimized boys and girls. Merrill
drug use, and delinquent behaviors and their interrelations over Palmer Q 53:439–460
time in urban and rural adolescents. J Res Adolesc 15:179–204 42. Visconti KJ, Troop-Gordon W (2010) Prospective relations
21. Pepler DJ, Craig WM (2005) Aggressive girls on troubled tra- between children’s responses to peer victimization and their
jectories: a developmental perspective. In: Pepler DJ, Madsen socioemotional adjustment. J Appl Dev Psychol 31:261–272
KC, Webster C, Levene KS (eds) The development and treatment 43. Eisenberg N, Shepard SA, Fabes RA, Murphy BC, Guthrie IK
of girlhood aggression. Lawrence Earlbaum, Mahwah, pp 3–28 (1998) Shyness and children’s emotionality, regulation, and
22. Arseneault L, Bowes L, Shakoor S (2010) Bullying victimization coping: contemporaneous, longitudinal, and across context rela-
in youths and mental health problems: ‘much ado about nothing’? tions. Child Dev 69:767–790
Psychol Med 40:717–729 44. Eisenberg N, Cumberland A, Spinrad TL, Fabes RA, Shepard SA,
23. Hanish LD, Guerra NG (2002) A longitudinal analysis of patterns Reiser M et al (2001) The relations of regulation and emotionality
of adjustment following peer victimization. Dev Psychopathol to children’s externalizing and internalizing problem behavior.
14:69–91 Child Dev 72:1112–1134
24. Kochenderfer BJ, Ladd GW (1996) Peer victimization: cause or 45. Card NA, Little TD (2006) Proactive and reactive aggression in
consequences of school maladjustment? Child Dev 67:1305–1317 childhood and adolescence: a meta-analysis of differential rela-
25. Kochenderfer BJ, Ladd GW (1997) Victimized children’s tions with psychosocial adjustment. Int J Behav Dev 30:466–480
responses to peers’ aggression: behaviors associated with reduced 46. Xu Y, Zhang Z (2008) Distinguishing proactive and reactive
versus continued victimization. Dev Psychopathol 9:59–73 aggression in Chinese children. J Abnorm Child Psychol
26. Ladd GW, Kochenderfer-Ladd B (2002) Identifying victims 36:539–552
of peer aggression from early to middle childhood: analysis of 47. Fonzi A, Genta ML, Menesini E, Bacchini D, Bonino S, Costa-
cross-informant data for concordance, estimation of relational bile A (1999) The Latin countries: Italy. In: Catalano R, Junger-
adjustment, prevalence of victimization, and characteristics of Tas J, Morita Y, Olweus D, Slee P, Smith PK (eds) The nature of
identified victims. Psychol Assess 14:74–96 school bullying: a cross-national perspective. Routledge, London,
27. Fields L, Prinz RJ (1997) Coping and adjustment during child- pp 140–156
hood and adolescence. Clin Psychol Rev 17:937–976 48. Genta ML, Menesini E, Fonzi A, Costabile A, Smith PK (1996)
28. Losoya S, Eisenberg N, Fabes RA (1998) Developmental issues Bullies and victims in schools in central and southern Italy. Eur J
in the study of coping. Int J Behav Dev 22:287–313 Psychol Educ 11:97–110
29. Tapper K, Boulton MJ (2005) Victim and peer group responses to 49. Francisco MV, Libório RMC (2009) Um estudo sobre bullying
different forms of aggression among primary school children. entre escolares do ensino fundamental. Psicol Reflex Crit
Aggress Behav 31:238–253 22:200–207

123
Child Psychiatry Hum Dev

50. Borsa CJ, Nunes MLT (2011) Prevalência de problemas de 70. Bryant FB, Smith BD (2001) Refining the architecture of
comportamento em uma amostra de crianças em idade escolar da aggression: a measurement model for the Buss-Perry aggression
cidade de Porto Alegre. Aletheia 34:32–46 questionnaire. J Res Person 35:138–167
51. Oyserman D, Coon HM, Kemmelmeier M (2002) Rethinking 71. Ang RP (2007) Factor structure of the 12-item aggression ques-
individualism and collectivism: evaluation of theoretical tionnaire: further evidence from Asian adolescent samples.
assumptions and meta-analyses. Psychol Bul 128:3–72 J Adolesc 30:671–685
52. Gonçalves LAO, Sposito MP (2002) Public initiatives of scholar 72. Sangwon K, Seock-Ho K, Kamphaus RW (2010) Is aggression
violence reduction in Brazil. Cad Pesq 115:101–138 the same for boys and girls? assessing measurement invariance
53. Frizzo MN, Bisol LW, Lara DR (2012) Bullying victimization is with confirmatory factor analysis and item response theory.
associated with dysfunctional emotional traits and affective School Psychol Q 25:45–61
temperaments J Affect Disord doi:10.1016/j.jad.2012.11.046 73. Kistner J, Counts-Allan C, Dunkel S, Drew CH, David-Ferdon C,
54. Lopes Neto AA, Saavedra LH (2003) Diga não para o Bullying. Lopez C (2010) Sex differences in relational and overt aggression
Abrapia, Rio de Janeiro in the late elementary school years. Aggress Behav 36:282–291
55. de Lima Braga L, Lisboa C (2010) Estratégias de coping para 74. Lim SH, Ang RP (2009) Relationship between boys’ normative
lidar com o processo de bullying: um estudo qualitativo. Inter- beliefs about aggression and their physical, verbal, and indirect
Am J Psychol 44:321–331 aggressive behaviors. Adolesc 44:635–650
56. Borsa JC, Bandeira DR (2011) The use of psychological mea- 75. Archer J (2004) Sex differences in aggression in real-world set-
sures of child aggressive behavior: analysis of Brazilian scientific tings: a meta-analytic review. Rev Gen Psychol 8:291–322
production. Aval Psi 12:193–203 76. Archer J (2009) Does sexual selection explain human sex dif-
57. Gremigni P, Damásio BF, Borsa JC (2013) Development and ferences in aggression? Behav Brain Sci 32:249–311
validation of a questionnaire to evaluate overt aggression and 77. Malta DC, de Souza ER, da Silva MMA, Silva CDS, de Andre-
reactions to peer aggression. Psicol Reflex Crit 26: in press azzi MAR, Crespo C et al (2010) Vivência de violência entre
58. French BF, Finch WH (2006) Confirmatory factor analytic pro- escolares brasileiros: resultados da Pesquisa Nacional de Saúde
cedures for the determination of measurement invariance. Struct do Escolar. Cienc Saude Colet 15:3065–3076
Equ Modeling 13:378–402 78. Monks CP, Palermiti A, Ortega R, Costabile A (2011) A cross-
59. Card NA, Stucky BD, Sawalani GM, Little TD (2008) Direct and national comparison of aggressors, victims and defenders in pre-
indirect aggression during childhood and adolescence: a meta- schools in England, Spain and Italy. Span J Psychol 14:133–144
analytic review of gender differences, intercorrelations, and 79. Chen X, De Oliveira AM, Coco AL, Zappulla C, Kaspar V,
relations to maladjustment. Child Dev 79:1467–8624 Schneider B et al (2004) Loneliness and social adaptation in Bra-
60. Kristensen SM, Smith PK (2003) The use of coping strategies by zilian, Canadian, Chinese and Italian children: a multi-national
Danish children classed as bullies, victims, bully/victims, and not comparative study. J Child Psychol Psychiatry 45:1373–1384
involved, in 82 responses to different (hypothetical) types of 80. Preddy TM, Fite PJ (2012) The impact of aggression subtypes
bullying. Scand J Psyc 44:479–488 and friendship quality on child symptoms of depression. Child
61. Smith P, Shu S, Madsen K (2001) Characteristics of victims of Indic Res. doi:10.1007/s12187-012-9143-9
school bullying: developmental changes in coping strategies and 81. Craig W, Pepler D, Blais J (2007) Responding to bullying: what
skills. In: Juvonen J, Graham S (eds) Peer harassment in school: works? School Psychol 28:465–477
the plight of the vulnerable and victimized. Guilford, New York, 82. Hunter SC, Boyle JM (2004) Appraisal and coping strategy use in
pp 332–351 victims of school bullying. Br J Educ Psychol 74:83–107
62. Hyde JS (2005) The gender similarities hypothesis. Am Psychol 83. Houbre B, Tarquinio C, Lanfranchi JB (2010) Expression of self-
60:581–592 concept and adjustment against repeated aggressions: the case of
63. Zakriski AL, Wright JC, Underwood MK (2005) Gender simi- a longitudinal study on school bullying. Eur J Psychol Educ
larities and differences in children’s social behavior: finding 25:105–123
personality in contextualized patterns of adaptation. J Pers Soc 84. Shelley D, Craig WM (2010) Attributions and coping styles in
Psychol 88:844–855 reducing victimization. Can J School Psychol 25:84–100
64. Caprara GV, Pastorelli C, Barbaranelli C, Vallone R (1992) In- 85. Menesini E, Eslea M, Smith PK, Genta ML, Giannetti E, Fonzi A et al
dicatori della capacità di adattamento sociale in età evolutiva. (1997) Cross-national comparison of children’s attitudes towards
Organizzazioni Speciali, Firenze bully/victim problems in school. Aggress Behav 23:245–257
65. McGinnis E, Goldstein AP (1997) Skillstreaming the elementary 86. Kochenderfer-Ladd B (2004) Peer victimization: the role of emo-
school child: new strategies and perspectives for teaching pro- tions in adaptive and maladaptive coping. Soc Dev 13:329–349
social skills, 2nd edn. Research Press, Champaign 87. Ruble DN, Martin CL (1998) Gender development. In: Damon W
66. Bentler PM, Wu E (2005) EQS 6.1 for windows user’s guide. (ed) Handbook of child psychology. Wiley, New York,
Multivariate Software Inc, Encino pp 933–1016
67. Cheung GW, Rensvold RB (2002) Evaluating goodness-of-fit 88. Larson RW (1999) The uses of loneliness in adolescence. In:
indexes for testing measurement invariance. Struct Equ Modeling Rotenberg KJ, Hymel S (eds) Loneliness in childhood and ado-
9:233–255 lescence. Cambridge University Press, New York, pp 244–262
68. Brown TA (2006) Confirmatory factor analysis for applied 89. Nicole S, Caprara GV (2005) Assessing aggressive behavior in
research. Guilford, New York children. Eur J Psychol Assess 21:255–264
69. Nunnally JC (1978) Psychometric theory, 2nd edn. McGraw-Hill,
New York

123

You might also like