Journal of Psychopathology and Behavioral Assessment (2018) 40:725–735

https://doi.org/10.1007/s10862-018-9678-2

Confirmatory Factor Structure and Psychometric Properties

of the Multidimensional Peer Victimization Scale
Meridith L. Eastman 1 & Ashlee A. Moore 1 & Jennifer Cecilione 1 & John M. Hettema 1 & Roxann Roberson-Nay 1

Published online: 14 April 2018

# Springer Science+Business Media, LLC, part of Springer Nature 2018

Abstract
The Multidimensional Peer Victimization Scale (MPVS; Mynard and Joseph, British Journal of Educational Psychology, 67, 51-
54 2000) is a 16-item self-report scale that captures peer victimization across four dimensions: physical victimization, verbal
victimization, social manipulation, and attacks on property. Performance of the scale has not been evaluated among older
adolescents. We examined the factor structure, internal consistency reliability, and performance of the scale in two separate
epidemiological U.S. samples representing different age groups: 9–14 year olds (N = 610) and 15–17 year olds (N = 524). The
four-factor structure of the scale was affirmed in both samples, however; there was not metric invariance by gender in the younger
age group. The scale and its subscales were found to have good internal consistency. Expected relationships between the MPVS
and measures of irritability, anxiety, and depression were affirmed. Results support continued use of the MPVS among child and
adolescent samples.

Keywords Peer victimization . Confirmatory factor analysis . Reliability . Measurement invariance

Introduction Furthermore, other-report methods do not take into account the

subjective nature of victimization. The self-perception of vic-
Between 20 and 40% of children experience peer victimization timization, rather than reputation among peers or Bobjective^
at least once during their school careers (Shetgiri et al. 2013). victimization, may be what puts one at risk for the negative
Abundant evidence has accumulated indicating that peer vic- sequelae of peer victimization (Graham and Juvonen 1998;
timization results in negative psychological sequelae including Gromann et al. 2013). Self-reported accounts, therefore, may
internalizing and externalizing outcomes (see Arseneault et al. be more relevant assessments of peer victimization.
2010 for a review). Given the high prevalence of peer One important differentiating characteristic of self-report
victimization and its known deleterious effects on mental instruments is how they capture types of victimization.
health, attention has been increasingly placed on the Summaries of self-report instruments to assess peer victimiza-
measurement of the phenomenon to capture its breadth and tion are available elsewhere (e.g., Betts et al. 2015; Crothers
depth. Previous work by Betts et al. (2015) reviewed different and Levinson 2004). While many assessments aim to capture
approaches to measuring victimization (i.e., observation, par- at least some mixture of physical and non-physical victimiza-
ent/teacher/peer-report, and self-report) and noted that self- tion, others focus exclusively on one type of victimization
report measurement avoids potential underestimation of vic- (e.g., Social Bullying Involvement Scale; Fitzpatrick and
tims’ experiences that can occur particularly with more covert Bussey 2011). The Multidimensional Peer Victimization
types of victimization which are less visible to parents, Scale (MPVS; Mynard and Joseph 2000) uniquely contains
teachers, peers, and outside observers (Demaray et al. 2013). subscales that encompass four different types of peer victim-
ization: physical victimization, verbal victimization, social
manipulation, and attacks on property. This latter category,
in particular, is understudied in the literature (Crothers and
* Meridith L. Eastman
[email protected] Levinson 2004; Betts et al. 2015). As such, the MPVS is
ideally suited for research that aims to disentangle the effects
1 of varying types of peer victimization.
Virginia Institute for Psychiatric and Behavioral Genetics, Virginia
Commonwealth University, 800 E. Leigh St., Suite 101, Biotech The MPVS was developed for use among children aged
One, P.O. Box 980126, Richmond, VA 23298, USA 11–16 in England, and its factor structure was affirmed in a
726 J Psychopathol Behav Assess (2018) 40:725–735

sample of 7–12 year olds in Nigeria (Balogun and Olapegba and depression symptoms, such that the experience of peer
2007) using principal components analysis. Betts et al. (2015) victimization results in depression symptoms while display
conducted a confirmatory analysis on 11–15 year olds in the of symptoms of depression (e.g., withdrawal, crying) may
UK and affirmed the factor structure in a mixed-gender sam- elicit victimization from peers (Cook et al. 2010). Therefore,
ple. However, there were significant differences in the relative it was thought that scores on a depression measure would be
strength of item factor loadings (i.e., measurement non-invari- significantly positively correlated with scores on the MPVS.
ance) by gender. Additionally, they assessed convergent va- Although we are unaware of studies that examine the relation-
lidity using a self-esteem scale and found a significant nega- ship of irritability per se and peer victimization, peer victimi-
tive association between peer victimization and self-esteem. zation could cause one to become irritable, while displays of
Despite this valuable previous work, several questions irritability (e.g., anger) may elicit peer victimization. Hence, a
about the MPVS remain unanswered. Of note, the scale has significant positive correlation between irritability and MPVS
been tested exclusively with children and younger adoles- scores was expected.
cents. How the scale performs among older adolescents re- Because extraversion is associated with sociability (Plomin
mains unexplored. Researchers who aim to use the MPVS to 1976), either a null or a negative correlation between extra-
assess developmental trends in peer victimization and among version scores the MPVS was expected.
types of peer victimization must first be assured that the scale We anticipated that any measurement invariance by gender
is applicable to multiple age groups. Theoretical work identified in the child sample (and consistent with that found
(Björkqvist et al. 1992) and age-sensitive analyses (e.g., by Betts et al. 2015) would be exacerbated in the adolescent
Rivers and Smith 1994) suggest the nature of peer victimiza- sample because of the gender intensification hypothesis,
tion changes across development, such that more overt types which suggests increased pressure during adolescence to con-
of victimization are replaced by—or are at least accompanied form to culturally-sanctioned gender roles (Hill and Lynch
by—attacks that are subtler in nature. Hence, it would be 1983). That is, differences between boys and girls in their
useful to analyze the properties of the MPVS for future assess- understanding and interpretation of peer victimization may
ment of different types of peer victimization that occur become stronger with age. As a consequence of this differen-
throughout development. Furthermore, although recent trajec- tiation, we suspected the fit of the four-factor model of the
tory research suggests grade-related declines in peer victimi- MPVS may be worse in the adolescent sample as compared
zation throughout schooling (Ladd et al. 2017), mixture to the child sample.
modeling efforts have identified classes of Bmoderate^ or
Bhigh chronic victims^ (Boivin et al. 2010; Goldbaum et al.
2003; Sumter et al. 2012) who continue to experience peer Method
victimization as they age despite lower overall normative
rates. The persistence of peer victimization across all ages Participants
underscores the need to assess psychometric properties of peer
victimization measurement tools for use across developmental Participants in the present analysis were part of two related
timepoints. As such, the aim of this paper was to conduct studies: The Twin Study of Negative Valence Emotional
confirmatory factor analysis and assess performance of the Constructs (hereafter referred to as the Juvenile Anxiety
MPVS in two separate samples representing different age Study or JAS; Carney et al. 2016) and Genetic Contributions
groups (i.e., children and adolescents). of Negative Valence Systems to Internalizing Pathways study
To assess validity of the peer victimization construct mea- (hereafter referred to as the Adolescent/Young Adult Twin
sured by the MPVS, we identified a set of constructs that we Study or AYATS). Both JAS and AYATS recruited twin pairs
expected to correlate (either positively or negatively, as de- through the Mid-Atlantic Twin Registry (MATR), a database of
scribed next) to peer victimization. Given that peer victimiza- twins and their family members who have expressed interest in
tion is one type of dysfunction with peers, we anticipated a research participation that is housed at Virginia Commonwealth
significant positive association between MPVS and a measure University (Lilley and Silberg 2013). An unselected sample
of difficulty with peer relationships. The relationship between was recruited from MATR according to registry protocols to
anxiety and peer victimization appears to be bidirectional ensure generalizability of genetic and environmental estimates
(Silberg et al. 2016); the experience of peer victimization and to reproduce an epidemiological distribution of psychopa-
may result in symptoms of anxiety (Arseneault et al. 2010) thology. Detailed descriptions of participant recruitment pro-
while displaying symptoms of anxiety may elicit maltreatment cesses, power, and demographics for the parent studies are lo-
from peers (Cook et al. 2010). Therefore, it was expected that cated elsewhere (Carney et al. 2016 for JAS; Cecilione et al.
there would be a significant positive correlation between mea- 2018 for AYATS). The analytic sample for the present investi-
sures of anxiety and MPVS scores. As with anxiety, there is a gation was comprised of 292 twin pairs and 26 half pairs
possible bidirectional relationship between peer victimization (N = 610, Mage = 11.40 years[SD = 1.52], 52.8% female,
J Psychopathol Behav Assess (2018) 40:725–735 727

100% Caucasian) from JAS and 256 twin pairs and 12 half Index of Peer Relations The Index of Peer Relations (IPR) is a
pairs (N = 524, Mage = 16.04 years [SD = .65], 57.5% female, 25-item scale that assesses the extent to which a respondent
87.0% Caucasian, 7.9% Black, and 5.2% Hispanic) from has problems with peers (Hudson et al. 1993). Statements
AYATS who completed the MPVS at the first study assessment, about the respondent’s relationship with his or her peer
as described below. The AYATS sample was limited to partic- group are endorsed using a 7-point Likert scale ranging
ipants under age 18 to exclude college aged emerging adults from BNone of the time^ to BAll of the time^. Select items
and to ensure generalizability of results to mid- to late adoles- are reverse-coded, such that higher scores indicate greater
cence. Written informed consent was obtained from adults (par- difficulty with the peer group. Reliability and validity of the
ents or guardians of minors) and assent was obtained from scale have been established (Hudson 1990). The internal
minors whose parents or guardians provided written consent. consistency reliability of the IPR in the adolescent sample
The Virginia Commonwealth University Institutional Review was α = .95.
Board reviewed and approved of all study procedures and ma-
terials. The study procedures were performed in accordance Screen for Child Anxiety Related Disorders The Screen for
with the ethical standards as laid down in the 1964 Childhood Anxiety Related Disorders (SCARED; Birmaher
Declaration of Helsinki and its later amendments. et al. 1997) is a 41- item scale to screen for childhood anxiety
disorders, including generalized anxiety disorder, separation
Procedures anxiety disorder, panic disorder, and social phobia. Statements
about feelings, fears, and worries are assessed with response
During the first assessment for each studies, participants were options: 0 = BNot true or hardly ever true^, 1 = BSomewhat
asked to complete a self-report questionnaire in the Research true or sometimes true^, or 2 = BVery true or often true^.
Electronic Data Capture (REDCap) system hosted at Virginia Summed scores greater than or equal to 25 may indicate the
Commonwealth University (Harris et al. 2009) that included presence of an anxiety disorder. Previous studies have found
the MPVS (Mynard and Joseph 2000). The MPVS was added the scale to be reliable and valid (Birmaher et al. 1997;
after the start of JAS and was therefore not completed by all Birmaher et al. 1999). Internal consistency reliability for the
participants; 610 JAS participants completed the MPVS dur- SCARED in the child sample was α = .91.
ing their first visit. Technical difficulties prevented a portion of
AYATS participants from completing all items of the MPVS Short Mood and Feelings Questionnaire The Short Mood and
and these partial scores were not included in the analysis. 524 Feelings Questionnaire (SMFQ) is 13-item scale used to mea-
AYATS participants aged 15–17 completed the MPVS. sure symptoms of depression among children and adolescents
Participants in both studies also completed the Short Mood (Angold et al. 1995). Each item of the scale is rated on a 3-
and Feelings Questionnaire (SMFQ) (Angold et al. 1995), the point Likert scale: 0 = BTrue^, 1 = BSometimes^ and 2 = BNot
Affective Reactivity Index (ARI) (Stringaris et al. 2012), and True^. The validity of this scale to assess depressive symp-
versions of the neuroticism and extraversion subscales of the toms has been demonstrated in a clinical sample of children
Eysenck Personality Questionnaire (EPQ)—Junior for the (Angold et al. 1995), a sample of detained adolescents (Kuo
child sample (Eysenck 1975) and adult short form for adoles- et al. 2005), a community sample of children and adolescents
cent sample (Eysenck et al. 1985). In addition, younger par- (Thapar and McGuffin 1998), and a school-based sample of
ticipants completed the Screen for Child Anxiety Related adolescents (Rhew et al. 2010). The internal consistency reli-
Disorders (SCARED; Birmaher et al. 1997) and adolescents ability of the scale in the child sample was α = .83 and in the
completed the Index of Peer Relations (IPR) (Hudson et al. adolescent sample was α = .84.
1993).
Affective Reactivity Index The Affective Reactivity Index
Measures (ARI) is a 7-item unidimensional scale used to measure irrita-
bility (Stringaris et al. 2012). Items ask the respondent to
The Multidimensional Peer Victimization Scale The assess how well each describes his or her feelings or behavior
Multidimensional Peer Victimization Scale (MPVS; Mynard in the past six months. Items are rated on a 3-point Likert scale
and Joseph 2000) is a 16-item questionnaire that asks Bhow corresponding to the responses BNot true^, BSomewhat True^,
often during the last school year has another pupil done these and BCertainly True^. The total score is the sum of the first six
things to you?^. Response options are 0 = BNot at all^, 1 = B items, yielding possible scores of 0–12. The 7th item of the
Once^, and 2 = BMore than once^. Four factors (consisting of scale assesses irritability-related impairment and is scored sep-
four items each) comprise the MPVS to address multiple di- arately. The ARI has been found to be reliable and valid in
mensions of victimization: physical victimization, verbal vic- samples of children with and at risk for bipolar disorder, chil-
timization, social manipulation, and attacks on property. The dren with severe mood dysregulation, and typically develop-
items of the MPVS are provided as an appendix. ing children (Stringaris et al. 2012). The internal consistency
728 J Psychopathol Behav Assess (2018) 40:725–735

reliability for ARI in the child and adolescent samples was by Browne and Cudeck (1993) are less than or equal to .05 for
α = .82 and α = .86, respectively. close fit, .05–.08 for reasonable fit, and greater than or equal to
.10 for poor fit. Additionally, because prior research suggests
Eysenck Personality Questionnaire, Extraversion Subscale that the scale may perform differently in boys and girls (Betts
Adolescents answered eight items from the short form et al. 2015), we tested measurement invariance by gender
Eysenck Personality Questionnaire (EPQ-SF) to assess extra- using a multiple group CFA in each sample separately.
version (Eysenck et al. 1985). Respondents were asked to Configural invariance ensures that the basic model structure,
answer Byes^ (1)/ Bno^ (0) to questions about their personal- with no parameter constraints, is the same between groups and
ity. Internal consistency reliability for the extraversion sub- suggests that the different groups of respondents being tested
scale was α = .80. conceptualize the constructs of the scale in the same way
Children answered 23 items from the Junior Eysenck (Cheung and Rensvold 2002; Marsh 1994; Riordan and
Personality Questionnaire (JEPQ) to measure extraversion. Vandenberg 1994).
Respondents were asked a series of Byes^ (1)/Bno^ (0) ques- Cronbach’s alphas for internal consistency reliability were
tions about their personality. Select items were reverse-coded calculated for the total and factor sums of the MPVS using SAS
and higher scores indicate higher levels of extraversion. 9.4 (SAS Institute, Cary NC). Cronbach’s alpha ranges from 0
Internal consistency reliability for the extraversion subscale to 1, with higher numbers indicating that items within the scale
in JAS was α = .64. or subscale are highly related to one another and, therefore,
measure the same concept (Tavakol and Dennick 2011).
Analytic Strategy Spearman’s correlation coefficient, accommodating the
non-normal distribution of the MPVS, was calculated in
Because previous work suggests that the nature of peer vic- SAS 9.4 (SAS Institute, Cary NC) to assess theoretical re-
timization changes over the course of development (Rivers lationships between MPVS and its subscales and the afore-
and Smith 1994), psychometric properties of the MPVS were mentioned measures of peer relations, anxiety, depression,
examined in the child (JAS) and adolescent (AYATS) samples irritability, and extraversion. We used an alpha of .05 to
separately. First, we computed descriptive statistics (i.e., mean determine whether the correlation between the two scales
and standard deviation) of the total score of the MPVS and its was significantly different from zero, and the sign of the
four factors (i.e., physical victimization, verbal victimization, coefficient to determine whether the correlation was posi-
social manipulation, and attacks on property) in each sample tive or negative.
separately. The majority of children and adolescents did not
experience peer victimization. The distribution of the MPVS
in both samples was, therefore, positively skewed, thus vio- Results
lating the assumption of normality required for parametric
tests of mean differences. To compare mean levels of victim- Descriptive Statistics
ization for each factor of the MPVS across age groups ac-
counting for the non-normality of the data, we conducted Comparison of mean levels of overall victimization across
Wilcoxon Rank Sum Tests. We then conducted one-way anal- studies, as measured by the total MPVS score, showed signif-
ysis of variance (ANOVA), a test robust to violations of nor- icantly lower levels of victimization in the older age group
mality, to examine differences in means by gender. SAS 9.4 (M = 6.31, SD = 6.52) as compared to the younger one
(SAS Institute Inc 2013, Cary NC) was used to calculate de- (M = 8.27, SD = 7.06; Z = −5.25, p < .0001). Examination of
scriptive statistics and to test for mean differences by study subscales revealed significant differences in mean levels of
and by gender. physical victimization (Z = −7.95, p < .0001), social manipu-
Next, we conducted confirmatory factor analyses (CFA) for lation (Z = −5.01, p < .0001), and attacks on property (Z =
ordered categorical (ordinal) data in Mplus 8 (Muthén and −9.34, p < .0001) between the two studies, with the older
Muthén 1998-2017) in the child and adolescent samples sep- age group demonstrating lower means than the younger age
arately. This analysis is appropriate in later stages of scale group. While the mean level of verbal victimization was
development and tests of psychometric properties when the higher in the older age group (M = 3.03, SD = 2.82), this
number of factors has been previously established (e.g., was not significantly different from that of the younger age
Balogun and Olapegba 2007; Betts et al. 2015). We, therefore, group (M = 2.64, SD = 2.47).
assessed the fit of the data to the aforementioned four-factor In both age groups, boys had significantly higher mean
structure of the MPVS, using chi-square tests and model fit levels of physical victimization and verbal victimization than
indices (specifically, CFI, TLI, and RMSEA). Hu and Bentler girls (see Table 1). Whereas boys also had significantly higher
(1999) suggest that CFI and TLI scores greater than or equal to mean levels of property victimization in the younger age
.95 demonstrate good fit. Cut-offs for RMSEA recommended group, this was not the case in the adolescent sample, in which
J Psychopathol Behav Assess (2018) 40:725–735 729

Table 1 Comparison of mean Multidimensional Peer Victimization Table 2 Goodness-of-fit statistics for 4-factor confirmatory factor
Scale subscale and total scores by gender in JAS and AYATS analysis of the Multidimensional Peer Victimization scale in JAS and
AYATS
Mean (SD) ANOVA F
Study Χ2 df p CFI TLI RMSEA (95% CI)
JAS
Boys Girls N = 610 JAS 217.05 98 <.0001 .98 .98 .05 (.04, .05)
N = 288 N = 322 AYATS 270.40 98 <.0001 .98 .92 .06 (.05, .07)
Physical victimization 1.68 (2.36) .85 (1.67) 25.98***
Verbal victimization 3.13 (2.44) 2.23 (2.43) 20.57*** df = degrees of freedom; CFI=Comparative Fit Index; TLI = Tucker-
Lewis Index; RMSEA = Root Mean Square Error of Approximation;
Social manipulation 2.12 (2.24) 2.12 (2.26) 0.00
CI=Confidence interval
Attacks on property 2.54 (2.33) 2.07 (2.12) 6.93**
MPVS total 9.47 (7.37) 7.27 (6.34) 15.07***
confirmed configural invariance between these groups, but
not metric invariance (see Table 3), suggesting that factor
AYATS
loadings are not the same between groups. Lack of metric
Boys Girls N = 524
N = 226 N = 298 invariance by gender in in this age group precluded tests of
Physical victimization .69 (1.63) .34 (1.16) 8.21** scalar invariance by gender in this sample and between the
Verbal victimization 3.50 (2.98) 2.67 (2.65) 11.08*** child and adolescent samples (Bollen 1989). The CFAs con-
Social manipulation 1.38 (2.07) 1.79 (2.38) 4.32* ducted separately for adolescent boys and girls also showed
Attacks on property 1.32 (1.75) 1.06 (1.60) 3.21 excellent fit for both genders. Furthermore, multiple group
MPVS total 6.89 (6.79) 5.87 (6.29) 2.66 CFA confirmed configural, metric, and scalar invariance (see
Table 3).
SD = Standard deviation
*p < .05
**p < .01 Reliability
***p < .0001
Cronbach’s α for the total MPVS score in the child sample
there was no significant difference in property victimization was .78, which is an indication of good internal consistency
by gender. Social manipulation was higher among girls in the reliability. Internal consistency reliability was also high in
adolescent sample; however, there were no sex differences in each of the subscales: α = .89 for physical victimization,
social manipulation in the younger age group. α = .83 for verbal victimization, α = .78 for social manipula-
tion, and α = .76 for attacks on property. Cronbach’s α for the
total MPVS in the adolescent sample, was considerably higher
Confirmatory Factor Analysis at .89. The subscales demonstrated good internal consistency
reliability: α = .82 for physical victimization, α = .83 for ver-
Confirmatory factor analysis (CFA) showed that the MPVS per- bal victimization, α = .82 for social manipulation, and α = .67
formed excellently in both age groups examined separately (see for attacks on property.
Table 2). Both studies have a significant Χ2 test (p < .0001), but
this is not uncommon with larger sample sizes (Schumacker and
Lomax 1996). CFI, TLI, and RMSEA goodness-of-fit statistics Correlations with Other Scales
suggest a slightly better fit for the younger sample as compared to
the older sample. The standardized loadings (represented by ar- As shown in Table 4, we found the expected positive relation-
rows extending to the right from each factor to the abbreviated ships between the MPVS and the IPR (adolescents only),
item) for each factor are provided for each study in Figs. 1 and 2. SCARED (children only), SMFQ (children and adolescents)
All factor loadings in both studies were significant at p < .001. In and ARI (children and adolescents). Among children, we
the child sample, the factor loadings ranged from a loading of .65 found the expected null or negative correlations between the
for BSwore at me^ on verbal victimization to .91 for BPunched MPVS and its subscales and the extraversion subscale of the
me^ on physical victimization. In the adolescent sample, the EPQ. Among adolescents, we found no significant correla-
factor loadings ranged from loadings of .78 for BBeat me up^ tions between extraversion and subscale scores for physical
on physical victimization and for BStole something from me^ on victimization, verbal victimization, and property victimiza-
attacks on property to .96 for BHurt me physically in some way^ tion. There were, however, significant, positive associations
on physical victimization. between extraversion and total MPVS score and the social
The CFAs conducted for younger boys and girls separately manipulation subscale, albeit small (r = .09, p < .05 and
showed excellent fit for both groups, and multiple group CFA r = .11, p < .05 respectively).
730 J Psychopathol Behav Assess (2018) 40:725–735

Fig. 1 Standardized parameter

estimates for 4-factor
punch
confirmatory factor analysis of
the Multidimensional Peer
Victimization Scale in JAS (N =
610). Note: See Appendix for .91
complete scale items kick
.93
1.00
physical
.88 hurt

.79
beat

.61 call
.54 name

.85
appear
.82
1.00 verbal
.86
reason

.65

swore

.68 .73

trouble

.79
against
.84
1.00 social
.71
refused
.82
.66
no

talk
.68
took

.76
break
.84
property
1.00 .81
stole

.84

damage
J Psychopathol Behav Assess (2018) 40:725–735 731

Fig. 2 Standardized parameter

estimates for 4-factor
punch
confirmatory factor analysis of
the Multidimensional Peer
Victimization Scale in AYATS
(N = 629). Note: See Appendix .94
for complete scale items kick
.93
1.00
physical
.96 hurt

.78
beat

.66 call
.55 name

.87
appear
.91
1.00 verbal
.88
reason

.81

swore

.73 .73

trouble

.87
against
.86
1.00 social
.82
refused
.92
.74
no

talk
.78
took

.83
break
.89
property
1.00 .78
stole

.86

damage
732 J Psychopathol Behav Assess (2018) 40:725–735

Table 3 Goodness-of-fit statistics

for confirmatory factor analysis of Study Χ2 df p CFI TLI RMSEA (95% CI)
the Multidimensional Peer
Victimization scale in JAS and JAS
AYATS by gender Boys 143.27 98 <.0001 .99 .98 .04(.03–05)
Girls 147.80 98 <.0001 .99 .98 .04(.03–.05)
Configural invariance model 329.34 196 <.0001 .98 .98 .05(.04–.06)
Metric invariance model* 362.63 208 <.0001 .98 .98 .05(.04–.06)
AYATS
Boys 171.56 98 <.0001 .98 .98 .06(.04, .07)
Girls 171.44 98 <.0001 .98 .98 .05(.04, .06)
Configural invariance model 342.99 196 <.0001 .98 .98 .05 (.04,.06)
Metric invariance model 357.82 208 <.0001 .98 .98 .05(.04,.06)
Scalar invariance model 367.74 220 <.0001 .98 .98 .05(.04,.06)

df = degrees of freedom; CFI=Comparative Fit Index; TLI = Tucker-Lewis Index; RMSEA = Root Mean Square
Error of Approximation; CI=Confidence interval;
*p-value of comparison between configural and metric invariance models <.01

Discussion years of schooling (Ladd et al. 2017). Unlike the other types
of peer victimization, verbal victimization was higher, though
We examined performance of the MPVS in two samples not significantly so, in the older age group. This perhaps rep-
representing different age groups: one sample encompassing resents an age-related shift in norms regarding Bacceptable^
ages 9–14, and the other encompassing ages 15–17. While forms of peer victimization. While overt physical forms of
some psychometric properties of the MPVS have been ex- victimization become less common as children age
plored in previous research with children and younger adoles- (Björkqvist et al. 1992), verbal insults continue to be a salient
cents (Balogun and Olapegba 2007; Betts et al. 2015), this is form of victimization. The lack of difference between older
the first investigation to examine properties of the scale in and younger groups in mean levels of social manipulation is
individuals from mid- to late adolescence. Mixture modeling interesting, given that social manipulation is thought to require
(e.g., Boivin et al. 2010; Goldbaum et al. 2003; Sumter et al. more social sophistication than is typical among younger chil-
2012) indicates that subgroups of adolescents continue to be dren (Björkqvist et al. 1992).
victimized by their peers despite overall normative declines in In the child sample, comparisons by gender show that boys
peer victimization, underscoring the need to assess peer vic- experienced more physical victimization, verbal victimiza-
timization across developmental timepoints. tion, attacks on property, and overall victimization than girls;
On the whole, the younger age group reported more vic- however, there was no difference in levels of social manipu-
timization than the older age group (with the exception of lation between boys and girls. These findings are consistent
verbal victimization). This is consistent with longitudinal re- with a meta-analysis by Casper and Card (2017), which found
search, which shows declines in peer victimization across that while physical victimization is more common among

Table 4 Correlations between the

Multidimensional Peer Scale Spearman’s rho
Victimization Scale and subscales
and other validated measures IPR SCARED SMFQ ARI EPQ extraversion

AYATS JAS JAS AYATS JAS AYATS JAS AYATS

Physical victimization .12** .16** .32** .20** .21** .22** −.12** .04
Verbal victimization .30** .24** .34** .31** .29** .29** −.05 .05
Social manipulation .25** .31** .33** .28** .25** .27** −.05 .11*
Attacks on property .26** .23** .27** .24** .22** .19** −.05 .08
MPVS total .32** .31** .40** .34** .31** .32** −.08 .09*

ARI = Affective Reactivity Index; SCARED = Screen for Child Anxiety Related Disorders; SMFQ = Short Mood
and Feelings Questionnaire; EPQ = Eysenck Personality Questionnaire
*p < .05
**p < .01
J Psychopathol Behav Assess (2018) 40:725–735 733

boys, rates of relational victimization (i.e., social manipula- of electronic victimization represents an important advance in
tion) do not differ by gender. In contrast, we found that ado- bullying measurement, particularly as phone and web-based
lescent girls reported more social victimization than adoles- communications have increasingly become tools for peer vic-
cent boys, whereas the pattern was reversed for physical and timization (Wang et al. 2009), we note that the MPVS as it was
verbal victimization. Overall victimization did not differ by initially conceptualized does not exclude victimization that oc-
gender among adolescents. curs through electronic means. For example, name-calling (item
CFA results indicated excellent performance and configural 1 of the MPVS) may occur in a face-to-face context or via
invariance by gender the child sample. This suggests that the phone or web-based communication. As such, the MPVS in
MPVS reproduces the same pattern of factors for boys and girls its original conception remains worthy of use, particularly in
in younger age groups. Lack of metric invariance precluded light of findings regarding its performance and reliability.
formal testing of stronger forms of invariance (i.e., scalar) in Although our statistical analyses took into account the non-
this sample. Post hoc analysis (data not shown) found the factor independence of data from twins, twin pairs may not be rep-
loading for the physical victimization item BBeat me up,^ to be resentative of the general child and adolescent population with
much greater for girls than for boys when compared with gen- regard to peer victimization. Some evidence (e.g., Lamarche
der differences for other items. Betts et al. (2015) also found et al. 2006) suggests that the presence of a prosocial co-twin
factor loading differences by gender for the physical victimiza- (as with a prosocial friend or non-twin sibling) may reduce the
tion factor. Qualitative research may be informative in investi- likelihood of being targeted for peer victimization, particularly
gating how and why this and other loadings might differ be- for individuals with reactive aggression tendencies. On the
tween genders in the younger age group. other hand, research also indicates no differences between
CFA results for the adolescent sample also indicated excel- twin and non-twin samples in behavior problems and person-
lent performance of the 4-factor model. Furthermore, ality characteristics (Pulkkinen et al. 2003), suggesting that
configural, metric, and scalar measurement invariance held the psychometric properties of the MPVS should not differ
across genders in the adolescent, indicating that the structure between twin and singleton samples. Additionally, the JAS
of factors for the MPVS is the same for boys and girls (i.e., and AYATS samples were each restricted in racial and ethnic
configural invariance), that the factor loadings across groups diversity to ensure homogeneity required for genetic analysis.
are statistically equivalent (i.e., metric invariance), and that the Further research on the performance of the MPVS in more
thresholds of the scale are invariant between the two genders diverse samples is needed.
(Bowen and Masa 2015). The finding of measurement invari- Despite these limitations, this investigation represents a
ance by gender was counter to our expectations based on the significant contribution. First, we tested performance of the
gender intensification process (Hill and Lynch 1983) and pro- MPVS against a series of measures with established psycho-
vides reassurance that the MPVS may be used in mixed gen- metric properties. Second, the use of the same measurements
der samples of adolescents. in a sample of children and a sample of adolescents allowed
We found the MPVS—total scale and subscales—to be for tests of performance of the MPVS in a broader age range
highly reliable in terms of internal consistency reliability in than has been previously possible, which is essential for in-
both samples. Additionally, we found expected correlations vestigators who which to use the MPVS to study developmen-
with measures of peer relations, anxiety, depression, and irri- tal trends in types of peer victimization. In sum, our findings
tability, and extraversion. While there is some research on the support continued use of the MPVS in mixed gender samples
relationship between personality traits and peer victimization in childhood and adolescence.
(e.g., Jensen-Campbell et al. 2002; Mynard and Joseph 1997),
type of victimization is often not taken into account. The find- Acknowledgements This project was supported by National Institute of
Mental Health grants R01MH098055 (J.M.H.), R01MH101518 (R.R.-
ing of a significant correlation between extraversion and
N.), T32MH020030 (M.L.E.), as well as UL1TR000058 from the
social manipulation in adolescents is consistent with the National Center for Research Resources (for REDCap). We are grateful
work of Tackett et al. (2014) which suggests a relationship for the contributions of the twins and their families who participated in the
between sociability and involvement in socially sophisticated studies providing data for this article.
forms of aggression (i.e., social manipulation). Our finding of Funding This study was funded by National Institute of Mental Health
grants R01MH098055 and R01MH101518, T32MH020030, as well as
a small negative correlation between extraversion and physi- UL1TR000058 from the National Center for Research Resources (for
cal victimization in the child sample suggests that the person- REDCap).
ality trait may be protective against less sophisticated social
behaviors, particularly at younger ages. Compliance with Ethical Standards
Betts et al. (2015) proposed a revision to the
Multidimensional Peer Victimization Scale (MPVS-R), con- Conflict of Interest Meridith L. Eastman, Ashlee A. Moore, Jennifer
taining a fifth subscale intending to capture electronic victimi- Cecilione, John M. Hettema, and Roxann Roberson-Nay declare that they
have no conflict of interest.
zation (i.e., cyberbullying) in isolation. While the specification
734 J Psychopathol Behav Assess (2018) 40:725–735

Ethical Approval This article does not contain any studies with animals of Genetic Psychology, 176, 93–109. https://doi.org/10.1080/
performed by any of the authors. 00221325.2015.1007915.
Birmaher, B., Khetarpal, S., Brent, D., Cully, M., Balach, L., Kaufman, J.,
Informed Consent Informed consent was obtained from all individual & Neer, S. M. (1997). The screen for child anxiety related emotional
participants included in the study. disorders (SCARED): Scale construction and psychometric charac-
teristics. Journal of the American Academy of Child and Adolescent
Psychiatry, 36, 545–553. https://doi.org/10.1097/00004583-
Experiment Participants All procedures performed in studies involving
199704000-00018.
human participants were in accordance with the ethical standards of the
Birmaher, B., Brent, D. A., Chiappetta, L., Bridge, J., Monga, S., &
institutional and/or national research committee and with the 1964
Baugher, M. (1999). Psychometric properties of the screen for child
Helsinki declaration and its later amendments or comparable ethical
anxiety related emotional disorders (SCARED): A replication study.
standards.
Journal of the American Academy of Child and Adolescent
Psychiatry, 38, 1230–1236. https://doi.org/10.1097/00004583-
Appendix 199910000-00011.
Björkqvist, K., Lagerspetz, K. M., & Kaukiainen, A. (1992). Do girls
manipulate and boys fight? Developmental trends in regard to direct
Table 5 Multidimensional Peer Victimization Scale (Mynard and and indirect aggression. Aggressive Behavior, 18, 117–127. https://doi.
Joseph 2000) org/10.1002/1098-2337(1992)18:2<117::AID-AB2480180205>3.0.
CO;2-3.
Below is a list of things that some children do to other children. How Boivin, M., Petitclerc, A., Feng, B., & Barker, E. D. (2010). The devel-
often during the last school year has another pupil done these things to opmental trajectories of peer victimization in middle to late child-
you? hood and the changing nature of their behavioral correlates. Merrill-
Palmer Quarterly, 56, 231–260. https://doi.org/10.1353/mpq.0.
1. Called me names
0050.
2. Tried to get me into trouble with my friends
Bollen, K. A. (1989). Structural equations with latent variables. New
3. Took something of mine without permission
York, NY: Wiley. https://doi.org/10.1002/9781118619179.
4. Made fun of me because of my appearance
Bowen, N. K., & Masa, R. D. (2015). Conducting measurement invari-
5. Made fun of me for some reason
ance tests with ordinal data: A guide for social work researchers.
6. Punched me
Journal of the Society for Social Work and Research, 6(2), 229–249.
7. Kicked me
https://doi.org/10.1086/681607.
8. Hurt me physically in some way
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing
9. Beat me up
model fit. Sage Focus Editions, 154, 136–136.
10. Tried to break something of mine
Carney, D. M., Moroney, E., Machlin, L., Hahn, S., Savage, J. E., Lee,
11. Tried to make my friends turn against me
M., … & Hettema, J. M. (2016). The twin study of negative valence
12. Stole something from me
emotional constructs. Twin Research and Human Genetics, 19, 456–
13. Refused to talk to me
464. https://doi.org/10.1017/thg.2016.59.
14. Made others not talk to me
Casper, D. M., & Card, N. A. (2017). Overt and relational victimization:
15. Deliberately damaged some property of mine
A meta-analytic review of their overlap and associations with so-
16. Swore at me
cial–psychological adjustment. Child Development, 88(2), 466–483.
Response options: 0 = Not at all, 1 = Once, 2 = More than once Cecilione, J., Rappaport, L., Hahn, S., Anderson, A., Hazlett, L.,
Burchett, J., … Roberson-Nay, R. (2018). Genetic and environmen-
Bold denotes variable names used in confirmatory factor analysis (see tal contributions of negative valence systems to internalizing path-
Figs. 1 and 2) ways. Twin Research and Human Genetics, 21(1), 12–23. https://
doi.org/10.1017/thg.2017.72.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit index-
es for testing measurement invariance. Structural Equation Modeling, 9,
233–255. https://doi.org/10.1207/S15328007SEM0902_5.
References Cook, C. R., Williams, K. R., Guerra, N. G., Kim, T. E., & Sadek, S.
(2010). Predictors of bullying and victimization in childhood and
adolescence: A meta-analytic investigation. School Psychology
Angold, A., Costello, E. J., Messer, S. C., Pickles, A., Winder, F., & Quarterly, 35, 65–83. https://doi.org/10.1037/a0020149.
Silver, D. (1995). Development of a short questionnaire for use Crothers, L. M., & Levinson, E. M. (2004). Assessment of bullying: A
in epidemiological studies of depression in children and adoles- review of methods. Journal of Counseling and Development, 82,
cents. International Journal of Methods in Psychiatric Research, 496–503. https://doi.org/10.1002/j.1556-6678.2004.tb00338.x.
5, 237–249. Demaray, M. K., Malecki, C. K., Secord, S. M., & Lyell, K. M. (2013).
Arseneault, L., Bowes, L., & Shakoor, S. (2010). Bullying victimization Agreement among students’, teachers’, and parents’ perceptions of
in youths and mental health problems: ‘Much ado about nothing’? victimization by bullying. Children and Youth Services Review, 35,
Psychological Medicine, 40, 717–729. https://doi.org/10.1017/ 2091–2100. https://doi.org/10.1016/j.childyouth.2013.10.018.
S0033291709991383. Eysenck, H. J. (1975). Manual of the Eysenck personality questionnaire.
Balogun, S. K., & Olapegba, P. O. (2007). Cultural validation of the San Diego, CA: Digits.
multidimensional peer victimization scale in Nigerian children. Eysenck, S. B., Eysenck, H. J., & Barrett, P. (1985). A revised version of
Journal of Cross-Cultural Psychology, 38, 573–580. https://doi. the psychoticism scale. Personality and Individual Differences, 6,
org/10.1177/0022022107305239. 21–29. https://doi.org/10.1016/0191-8869(85)90026-1.
Betts, L. R., Houston, J. E., & Steer, O. L. (2015). Development of the Fitzpatrick, S., & Bussey, K. (2011). The development of the social bul-
multidimensional peer victimization scale–revised (MPVS-R) and lying involvement scales. Aggressive Behavior, 37, 177–192. https://
the multidimensional peer bullying scale (MPVS-RB). The Journal doi.org/10.1002/ab.20379.
J Psychopathol Behav Assess (2018) 40:725–735 735

Goldbaum, S., Craig, W. M., Pepler, D., & Connolly, J. (2003). Mynard, H., & Joseph, S. (2000). Development of the multidimensional
Developmental trajectories of victimization: Identifying risk and peer-victimization scale. Aggressive Behavior, 26, 169–178. https://
protective factors. Journal of Applied School Psychology, 19, 139– doi.org/10.1002/(SICI)1098-2337(2000)26:2<169::AID-AB3>3.0.
156. https://doi.org/10.1300/J008v19n02_09. CO;2-A.
Graham, S., & Juvonen, J. (1998). Self-blame and peer victimization in Plomin, R. (1976). Extraversion: Sociability and impulsivity? Journal of
middle school: An attributional analysis. Developmental Personality Assessment, 40(1), 24–30. https://doi.org/10.1207/
Psychology, 34, 587–599. https://doi.org/10.1037/0012-1649.34.3. s15327752jpa4001_6.
587. Pulkkinen, L., Vaalamo, I., Hietala, R., Kaprio, J., & Rose, R. J. (2003).
Gromann, P. M., Goossens, F. A., Olthof, T., Pronk, J., & Krabbendam, L. Peer reports of adaptive behavior in twins and singletons: Is twinship
(2013). Self- perception but not peer reputation of bullying victim- a risk or an advantage? Twin Research and Human Genetics, 6(2),
ization is associated with non- clinical psychotic experiences in ad- 106–118. https://doi.org/10.1375/136905203321536236.
olescents. Psychological Medicine, 43, 781–787. https://doi.org/10. Rhew, I. C., Simpson, K., Tracy, M., Lymp, J., McCauley, E., Tsuang, D.,
1017/S003329171200178X. & Vander Stoep, A. (2010) Criterion validity of the Short Mood and
Harris, P. A., Taylor, R., Thielke, R., Payne, J., Gonzalez, N., & Conde, J. Feelings Questionnaire and one-and two-item depression screens in
G. (2009). Research electronic data capture (REDCap)—A young adolescents. Child and Adolescent Psychiatry and Mental
metadata-driven methodology and workflow process for providing Health, 4, 8. https://doi.org/10.1186/1753-2000-4-8
translational research informatics support. Journal of Biomedical Riordan, C. M., & Vandenberg, R. J. (1994). A central question in cross-
Informatics, 42, 377–381. https://doi.org/10.1016/j.jbi.2008.08.010. cultural research: Do employees of different cultures interpret work-
Hill, J. P., & Lynch, M. E. (1983). The intensification of gender-related related measures in an equivalent manner? Journal of Management,
role expectations during early adolescence. In Girls at puberty (pp. 20, 643–671. https://doi.org/10.1177/1094428103251542.
201–228). Boston: Springer. Rivers, I., & Smith, P. K. (1994). Types of bullying behaviour and their
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in correlates. Aggressive. Behavior, 20, 359–368. https://doi.org/10.
covariance structure analysis: Conventional criteria versus new al- 1002/1098-2337(1994)20:5<359::AID-AB2480200503>3.0.CO;2-J.
ternatives. Structural Equation Modeling: A Multidisciplinary SAS Institute Inc. (2013). SAS 9.4 Cary, NC: SAS Institute Inc., Cary,
Journal, 6, 1–55. https://doi.org/10.1080/10705519909540118. NC, USA.
Hudson, W. (1990). A short-form scale to measure peer relations dys-
Schumacker, R. E., & Lomax, R. G. (1996). A beginner’s guide to struc-
function. Journal of Social. Service Research, 13, 57–69. https://doi.
tural equation modeling. Mahwah, NJ: Erlbaum. https://doi.org/10.
org/10.1300/J079v13n04_04.
1080/10705519609540025.
Hudson, W. W., Nurius, P. S., Daley, J. G., & Newsome, R. D. (1993).
Shetgiri, R., Lin, H., & Flores, G. (2013). Trends in risk and protective
Index of peer relations (IPR). Tallahassee, FL: Walmyr Publishing
factors for child bullying perpetration in the United States. Child
Company.
Psychiatry and Human Development, 44, 89–104. https://doi.org/
Jensen-Campbell, L. A., Adams, R., Perry, D. G., Workman, K. A.,
10.1007/s10578-012-0312-3.
Furdella, J. Q., & Egan, S. K. (2002). Agreeableness, extraversion,
and peer relations in early adolescence: Winning friends and Silberg, J. L., Copeland, W., Linker, J., Moore, A. A., Roberson-Nay, R., &
deflecting aggression. Journal of Research in Personality, 36, York, T. P. (2016). Psychiatric outcomes of bullying victimization: A
224–251. https://doi.org/10.1006/jrpe.2002.2348. study of discordant monozygotic twins. Psychological Medicine, 46,
Kuo, E., Stoep Vander, A., & Steward, D. G. (2005). Using the short 1875–1883. https://doi.org/10.1017/S0033291716000362.
mood and feelings.Questionnaire to detect depression in detained Stringaris, A., Goodman, R., Ferdinando, S., Razdan, V., Muhrer, E.,
adolescents. Assessment, 12, 374–383. https://doi.org/10.1177/ Leibenluft, E., & Brotman, M. A. (2012). The affective reactivity
1073191105279984. index: A concise irritability scale for clinical and research settings.
Ladd, G. W., Ettekal, I., & Kochenderfer-Ladd, B. (2017). Peer victimi- Journal of Child Psychology and Psychiatry, 53, 1109–1117. https://
zation trajectories from kindergarten through high school: doi.org/10.1111/j.1469-7610.2012.02561.
Differential pathways for children’s school engagement and Sumter, S. R., Baumgartner, S. E., Valkenburg, P. M., & Peter, J. (2012).
achievement? Journal of Educational Psychology. Advance online Developmental trajectories of peer victimization: Off-line and online
publication., 109, 826–841. https://doi.org/10.1037/edu0000177. experiences during adolescence. Journal of Adolescent Health, 50,
Lamarche, V., Brendgen, M., Boivin, M., Vitaro, F., Perusse, D., & 607–613. https://doi.org/10.1016/j.jadohealth.2011.10.251.
Dionne, G. (2006). Do friendships and sibling relationships provide Tackett, J. L., Kushner, S. C., Herzhoff, K., Smack, A. J., & Reardon, K.
protection against peer victimization in a similar way? Social W. (2014). Viewing relational aggression through multiple lenses:
Development, 15(3), 373–393. https://doi.org/10.1111/j.1467-9507. Temperament, personality, and personality pathology. Development
2006.00347.x. and Psychopathology, 26(3), 863–877. https://doi.org/10.1017/
Lilley, E. C., & Silberg, J. L. (2013). The mid-Atlantic twin registry, S0954579414000443.
revisited. Twin Research and Human Genetics, 16, 424–428. Tavakol, M., & Dennick, R. (2011). Making sense of Cronbach’s alpha.
https://doi.org/10.1017/thg.2012.125. International Journal of. Medical Education, 2, 53–55. https://doi.
Marsh, H. W. (1994). Confirmatory factor analysis models of factorial org/10.5116/ijme.4dfb.8dfd.
invariance: A multifaceted approach. Structural Equation Modeling, Thapar, A., & McGuffin, P. (1998). Validity of the shortened mood and
1(1), 5–34. https://doi.org/10.1080/10705519409539960. feelings questionnaire in a community sample of children and ado-
Muthén, L. K., & Muthén, B. O. (1998-2017). Mplus User’s Guide lescents: A preliminary research note. Psychiatry Research, 81,
(Eighth ed.). Los Angeles, CA: Muthén & Muthén. 259–268. https://doi.org/10.1016/S0165-1781(98)00073-0.
Mynard, H., & Joseph, S. (1997). Bully/victim problems and their asso- Wang, J., Iannotti, R. J., & Nansel, T. R. (2009). School bullying among
ciation with Eysenck's personality dimensions in 8 to 13 year-olds. adolescents in the. United States: Physical, verbal, relational, and
British Journal of Educational Psychology, 67, 51–54. https://doi. cyber. Journal of Adolescent Health, 45, 368–375. https://doi.org/
org/10.1111/j.2044-8279.1997.tb01226.x. 10.1016/j.jadohealth.2009.03.021
Reproduced with permission of copyright owner. Further reproduction
prohibited without permission.