Accepted Manuscript

Validation of the Cybervictimization Questionnaire (CYVIC) for adolescents

David Álvarez-García, José Carlos Núñez, Alejandra Barreiro, Trinidad García

PII: S0747-5632(17)30007-9

DOI: 10.1016/j.chb.2017.01.007

Reference: CHB 4702

To appear in: Computers in Human Behavior

Received Date: 29 October 2015

Revised Date: 31 December 2016

Accepted Date: 03 January 2017

Please cite this article as: David Álvarez-García, José Carlos Núñez, Alejandra Barreiro, Trinidad
García, Validation of the Cybervictimization Questionnaire (CYVIC) for adolescents, Computers in
Human Behavior (2017), doi: 10.1016/j.chb.2017.01.007

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form.
Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Highlights

- The CYVIC was tested using a large representative sample of Spanish

adolescents.

- It is a valid and reliable self-report measure of cybervictimization in adolescents.

- Four factors: Written-verbal, Visual-Sexual, Online exclusion and

Impersonation.

- Four additional indicators of Visual Cybervictimization-Teasing/Happy

slapping.

- Scores correlate with Internet Risk Behaviors, School Victimization and Self-

esteem.
 ACCEPTED MANUSCRIPT

Validation of the Cybervictimization Questionnaire (CYVIC) for adolescents

David Álvarez-García, José Carlos Núñez, Alejandra Barreiro y Trinidad García

Department of Psychology. University of Oviedo (Spain)

David Álvarez-García (Corresponding author)

Faculty of Psychology. University of Oviedo. Plaza Feijóo, s/n, 33003 Oviedo (Spain).

E-mail address: [email protected]

José Carlos Núñez

Faculty of Psychology. University of Oviedo. Plaza Feijóo, s/n, 33003 Oviedo (Spain).

E-mail address: [email protected]

Alejandra Barreiro

Faculty of Psychology. University of Oviedo. Plaza Feijóo, s/n, 33003 Oviedo (Spain).

E-mail address: [email protected]

Trinidad García

Faculty of Psychology. University of Oviedo. Plaza Feijóo, s/n, 33003 Oviedo (Spain).

E-mail address: [email protected]

 ACCEPTED MANUSCRIPT

Abstract

The goal of this study is to design a new self-report, targeting adolescents, called the

Cybervictimization Questionnaire (CYVIC), to assess the extent to which the informant

is a victim of aggression by mobile phone or Internet, and to analyze its factor and

criterion validity and reliability in a sample of adolescents of Asturias (Spain). For this

purpose, the CYVIC was applied to 3159 young people, aged 12 to 18, along with three

scales to measure Internet Risk Behaviors, Offline School Victimization, and Self-

esteem. Regarding factor validity, the model that best represents the internal structure of

the CYVIC has four factors (Impersonation, Visual-Sexual Cybervictimization, Verbal

Cybervictimization, and Online Exclusion) and four additional indicators of Visual

Cybervictimization-Teasing/Happy slapping. Regarding criterion validity, the CYVIC

score correlates positively with Internet Risk Behaviors and Offline School

Victimization, and negatively with Self-esteem, three variables that previous empirical

evidence indicates correlate with cybervictimization. The reliability of the CYVIC

factors and items are both adequate. Therefore, it is concluded that the CYVIC is a valid

and reliable self-report measure of cybervictimization in adolescents.

Key words: questionnaire, cybervictimization, adolescence, assessment.

 ACCEPTED MANUSCRIPT

Validation of the Cybervictimization Questionnaire (CYVIC) for adolescents

Abstract

The Cybervictimization Questionnaire (CYVIC) for adolescents was designed to assess

the extent to which the informant is a victim of aggression by mobile phone or Internet.

The goal of this study is to analyze its factor and criterion validity and reliability in a

sample of adolescents of Asturias (Spain). For this purpose, the CYVIC was applied to

3159 young people, aged 12 to 18, along with three scales to measure Internet Risk

Behaviors, Offline School Victimization, and Self-esteem. Regarding factor validity, the

model that best represents the internal structure of the CYVIC has four factors

(Impersonation, Visual-Sexual Cybervictimization, Written-Verbal Cybervictimization,

and Online Exclusion) and four additional indicators of Visual Cybervictimization-

Teasing/Happy slapping. Regarding criterion validity, the CYVIC scores correlate

positively with Internet Risk Behaviors and Offline School Victimization, and

negatively with Self-esteem, three variables that previous empirical evidence indicates

correlate with cybervictimization. The reliability of the CYVIC factors and items are

both adequate. Therefore, it is concluded that the CYVIC is a valid and reliable self-

report measure of cybervictimization in adolescents.

Key words: questionnaire, cybervictimization, adolescence, assessment.

1. Introduction

Adolescents' way of relating socially has changed enormously in recent years,

with the popularization of the mobile phone and the internet. The development of

electronic devices and software applications offer new possibilities to establish new

friendships or keep in touch with acquaintances. However, despite their many

1
 ACCEPTED MANUSCRIPT

advantages, their inappropriate use can also give rise to serious problems (Gámez-

Guadix, Orue, & Calvete, 2013). One of them is the possibility of using these means to

intentionally harm or injure others (Ortega-Ruiz & Núñez, 2012).

1.1. Definition and types of cybervictimization in adolescence

The term cybervictimization is usually used to refer to aggressions by means of

electronic devices, mainly mobile phones and the Internet (Del Rey, Elipe, & Ortega-

Ruiz, 2012). These attacks can take on various forms in adolescence. Nocentini et al.

(2010) distinguish four types: visual, written-verbal, online exclusion, and

impersonation. Visual cybervictimization consists of images —generally photos or

videos— that have been taken or disseminated by electronic means and that are

offensive, harmful, or injurious to the victim. Written-verbal cybervictimization refers to

being the target of annoying, threatening, or offensive calls, messages, or written

comments, through mobile phone or the Internet. Online exclusion implies not being

accepted or being expelled from a group, usually a social network or an instant

messaging program. Impersonation refers to situations in which someone impersonates

the victim over the mobile phone or Internet, to make fun of or get him/her into trouble.

Previous research, using both qualitative and quantitative approaches, supports the

relevance of this classification (Nocentini et al., 2010; Palladino, Nocentini, &

Menesini, 2015).

Each of these four types of cybervictimization can manifest in different ways.

For example, because of its seriousness, we can highlight two specific types of visual

cybervictimization, associated with different risk factors and dynamics. On the one hand

are images—mainly photos or videos—that have been taken or disseminated by

electronic means with the intention of making fun of or ridiculing the victim. In this
2
 ACCEPTED MANUSCRIPT

work, the term Visual cybervictimization—Teasing/Happy slapping will be used to refer

to this type of aggressions. Happy slapping is a particularly disturbing case of this type

of visual cybervictimization. It typically consists of recording a physical aggression of

an unsuspecting victim with a mobile phone and uploading the recording and

disseminating it over the Internet (Palasinski, 2013). On the other hand, there is the

dissemination of explicitly sexual or suggestive images of the victim, which, in this

work, is called Visual-sexual cybervictimization. These can be images the victim shares

with a trusted third party —sexting (Drouin, Ross, & Tobin, 2015; Morelli, Bianchi,

Baiocco, Pezzuti, & Chirumbolo, 2016)—who then disseminates them without the

victim's consent; or they can be images taken by the offender without the victim's

consent. Sometimes, the person who has the images can blackmail the victim,

threatening to disseminate them if the victim does not comply with their wishes —

sextortion (Flores, 2015). On other occasions, it is an ex who uploads and shares the

photos of his/her former couple on the Intertet, which is done out of spite or a desire for

revenge —revenge porn (Calvert, 2013). The Visual-sexual cybervictimization covers

such events and is part, therefore, of what has been called aggravated cases of

youth‐produced sexual images (Wolak & Finkelor, 2011). Despite of the severity of

these cases, and the concern that this situation generates to families, schools and society

in general, empiric research on this issue, as well as on the design of new assessment

tools, is limited to date (Klettke, Hallford, & Mellor, 2014).

It may also be helpful to differentiate two different expressions of verbal

cybervictimization. Oral cybervictimization usually occurs through voice calls on the

mobile phone. Written cybervictimization currently takes place primarily through

comments on social networks and text messages in instant messaging programs. It is

commonly considered to be part of the same construct (within the same factor or even
3
 ACCEPTED MANUSCRIPT

the same item). However, the use of a written means of communication may favor an

individual's disinhibition, as well as the rapid and massive spreading of the message

(Hinduja & Patchin, 2015). Oral and written cybervictimization may differ, therefore, in

the severity of their effects. Hence the utility of having specific measures of both forms

of verbal cybervictimization.

1.2. Prevalence and effects of cybervictimization in adolescence

Cybervictimization among adolescents is a serious problem that, due to its

prevalence and negative effects, warrants analysis and treatment. Although there are

differences as a function of the sample analyzed and the methodology used, it is

currently estimated that between 20 and 40% of children and adolescents have at some

time been victims of aggression through electronic means (Aboujaoude, Savage,

Starcevic, & Salame, 2015) and between 2 and 7% have been victims of severe

aggression (Garaigordobil, 2011).

Cybervictimization can have very important effects on the victim, especially in

severe cases. Mainly, it has been associated with an increase in internalizing problems

such as anxiety (Rose & Tynes, 2015), low self-esteem (Chang et al., 2013), social

anxiety (Juvonen & Gross, 2008), depressive symptomatology (Bonanno & Hymel,

2013) and suicidal ideation (Van Geel, Vedder, & Tanilon, 2014). In turn, these

internalizing problems can have negative consequences for adolescents in other areas,

such as the academic area. Cybervictimization has been associated with an increase in

concentration problems, poor performance, and truancy (Beran & Li, 2007; Tsitsika et

al., 2015). Cybervictimization has also been associated with an increase in externalizing

problems, although to a lesser extent. Some studies suggest that cybervictimization in

adolescence predicts cyberaggression some months later, although this relationship is

4
 ACCEPTED MANUSCRIPT

mediated by other variables, such as the level of rejection or of social support from

peers (Wright & Li, 2013).

2. Measures of cybervictimization in adolescence

For adequate identification and analysis of the problem, it is essential to have

assessment instruments with adequate and rigorous contents and metric properties. In

recent years, different instruments for the assessment of cybervictimization among

adolescents have been published. The most complete and current review is the one by

Berne et al. (2013) which includes assessment instruments published up to October of

2010. The absence of updated systematic reviews is a drawback to identify potential

gaps and controversies in the previous proposals. Therefore, we present below a brief

review of the instruments published in recent years (Table 1). All of them are self-

reports, with multiple-choice response formats on which the evaluated adolescent rates

the frequency with which he/she suffers the behavior described in the item. All the

reviewed questionnaires have made some contribution to the study of

cybervictimization. However, some common limitations suggest the need continue

working on the design of an instrument to overcome them.

A first limitation of previously published questionnaires is that the vast majority

of them only offer a general measure of cybervictimization, which makes it difficult to

understand the problem. Having a single cybervictimization score hinders, for example,

the analysis of possible differential effect of some variables on different types of

cybervictimization. The case of gender may be illustrative. Despite the fact that boys

and girls use mobile phones and the Internet in different ways (Fernández-Montalvo,

Peñalva, & Irazabal, 2015) and of differences in offline victimization (Álvarez-García,

Dobarro, Álvarez, Núñez, & Rodríguez, 2014), most studies find no differences
5
 ACCEPTED MANUSCRIPT

between boys and girls in cybervictimization frequency (Tokunaga, 2010). A possibility

is that, in most cases, this is due to the fact that the analyses are based on a general

cybervictimization score, which may offset the differential effects of gender. As can be

observed in Table 1, only 5 of the 16 revised questionnaires allow obtaining specific

scores for different forms of cybervictimization. All the nonspecific cybervictimization

questionnaires, with the exception of the MOOPV (Sumter, Valkenburg, Baumgartner,

Peter, & Van der Hof, 2015), offer a single measure of cybervictimization; and five of

the nine questionnaires specifically aimed at assessing cybervictimization present

unifactorial models.

A second limitation of the reviewed self-reports is that the few that do provide

specific measures of different types of cybervictimization do not offer measures of some

types that are relevant due to their severity and current social relevance. Most of them

focus on verbal cybervictimization: four of the five questionnaires that provide specific

scores of different types of cybervictimization offer, using different terms, a score of

verbal cybervictimization. However, only two of the five questionnaires include a

specific measure of online exclusion, two of the five include a specific measure of

impersonation, and any of them offers neither a specific visual cybervictimization score

that involves teasing or happy slapping, nor a visual cybervictimization score related to

sexting and sextortion.

Lastly, a third limitation of the previously published self-reports has to do with

certain methodological limitations (Table 1). One of them is that the questionnaires

specifically aimed at assessing cybervictimization among adolescents have generally

been validated with small samples: six of the nine questionnaires use samples of less

than 500 subjects to conduct either exploratory or confirmatory factor analyses. Another

limitation is that only two of the 16 questionnaires revised explicitly state that the
6
 ACCEPTED MANUSCRIPT

analyses were based on the polychoric correlation matrix, which is the most adequate

option for variables measured at the ordinal level and with nonnormal distributions

(Lloret-Segura, Ferreres-Traver, Hernández-Baeza, & Tomás-Marco, 2014). The

reviewed self-reports have a polytomic response format, and cybervictimization scores

tend to concentrate at the lowest values of the scale. A third and final methodological

limitation is that the models considered to be the most representative of its factor

structure generally do not present acceptable fit indexes. Only 6 of the 16 reviewed

questionnaires satisfy the criteria usually used to be considered a good fit (these criteria

are mentioned Section 4.4). In the case of one of these six questionnaires, it has been

necessary to include in the model the correlated measurement errors of a pair of items to

obtain good fit indexes.

7
Table 1. Questionnaires for measuring cybervictimization among adolescents.

Questionnaire Authors EFA or Sample Are Factors Fit indexes

CFA? size calculations

based on the

polychoric

correlation

matrix?

Nonspecific cybervictimization self-reports

European Cyberbullying Intervention Del Rey et al. (2015) EFA and EFA: Yes - Cyber-aggression 2=1484.15, df=208

Project Questionnaire (ECIPQ) CFA 2,820 - Cyber-victimization GFI=.986, CFI=.993, NNFI=.993

CFA: RMSEA=.030, SRMR=.080

2,859

Cyberbullying Questionnaire (CBQ) Gámez-Guadix, CFA 1,491 Not indicated - Perpetration 2=293, df=220, p<.001 NNFI=.98,

Villa-George, and - Victimization CFI=.99 RMSEA=.030(.027-.034)

Calvete (2014)

Cyberbullying Test Garaigordobil (2015) EFA and EFA: Not indicated - Cyberobserver χ2=4604.73, df=942, p<.001, χ2/df

CFA 1,513 - Cyberaggressor = 4.88, SBχ2/df=1.28

8
 CFA:1,513 - Cybervictim CFI=.91, NNFI=.90, GFI=.92

RMSEA=.056(.056-.063)

SRMR=.050

Virtual Aggression subscale Jiménez, Castillo, EFA 675 Not indicated - Perpetrated or ---

and Cisternas (2012) observed virtual

aggression.

- Virtual aggression

victimization.

Personal Experiences Checklist Hunt, Peters, and EFA and EFA: 433 Not indicated Second-order factor: χ2= 1,486.23, df=460 CFI=.88

(PECK) Rapee (2012) CFA CFA: 432 - Being bullied RMSEA=.07 (.07-.08)

First-order factors:

-Relational or verbal

bullying

-Cyber bullying

- Physical bullying

- Bullying based on

culture

9
Multidimensional Peer Victimization Betts, Houston, and CFA 371 Not indicated - Physical 2=304.08, df=160, p<.001

Scale—Revised (MPVS-R) Steer (2015) - Social manipulation CFI=.95, NFI=.90

- Verbal RMSEA=.05

- Attacks on property

- Electronic

Multidimensional Offline and Online Sumter et al. (2015) EFA and EFA: 325 Not indicated - Offline peer 2=1238.35, df=170, p<.005 CFI =

Peer Victimization Scale (MOOPV) CFA CFA: 799 victimization: .85

. Direct offline RMSEA = .10

. Indirect offline

-Online peer Adding correlated errors:

victimization: 2=613.02, df=153, p<.005 CFI =

. Direct online .94
. Indirect online RMSEA = .06

Specific cybervictimization questionnaires

Cyber Victimization Survey (CVS) Brown, Demaray, EFA 106 Not indicated - Cybervictimization ---

and Secord (2014)

Cybervictimization scale of the Menesini, Nocentini, CFA 1,092 Not indicated - Cybervictimization Male:

Cyberbullying Scale (CS) and Calussi (2011) 2=20.59, df=14, p=.11; CFI=.98,

10
 RMSEA=.030

Female:

2=19.18, df=11, p=.06

CFI=.99, RMSEA=.037

E-Victimisation Scale (E-VS) Lam and Li (2013) EFA and EFA: 231 Not indicated - Cybervictimization 2=67.90, df=5, p<.001;

CFA CFA: 234 2/df=13.58

GFI=.92, AGFI=.75

RMSR=.034

Cyberbullying Scale (CBS) Stewart, Drescher, EFA and EFA: 368 Yes - Cybervictimization CFI=.98, TLI=.98

Maack, Ebesutani, CFA CFA: 368 RMSEA=.06.

and Young (2014)

Cybervictimization scale of the Topcu and Erdur- EFA and EFA: 358 Not indicated - Cybervictimization GFI=.82, AGFI=.76, CFI=.58,

Revised Cyber-Bullying Inventory Baker (2010) CFA CFA: 339 NFI=.54, TLI=.50

(RCBI) RMSEA=.13

Adding correlated errors:

GFI=.93, AGFI=.90, CFI=.89,

NFI=.84, TLI=.86

11
 RMSEA=.06

Adolescent Victimization through Buelga, Cava, and EFA and EFA: Not indicated - Cybervictimization S-B2 =366.45, df=131, p<.001

Mobile Phone and Internet Scale Musitu (2012) CFA 1,934 through mobile phone CFI=.93

(CYBVIC) CFA: - Cybervictimization RMSEA=.03 (.03–.03)

1,483 through the Internet

Cybervictimization scale of the Çetin, Yaman, and EFA and 404 Not indicated - Cyber verbal bullying 2 = 482.33, df=200, p<.001, 2/df=

Cyber Victim and Bullying Scale Peker (2011) CFA - Hiding identity 2.41

(CVBS) - Cyber forgery NFI=.96, CFI=.96, IFI=.94,

RFI=.93, GFI =.90, NNFI= .96

RMSEA=.058

Cybervictimization scale of the Palladino et al. CFA 1,123 Not indicated - Written-verbal 2 = 127.02, df=71, p<.001

Florence CyberBullying- (2015) - Visual CFI=.92

CyberVictimization Scales - Impersonation RMSEA=.03

(FCBVSs) - Exclusion

Adding correlated errors:

2 = 105.10, df=70, p=.004

CFI=.95

RMSEA=.02

12
Online Victimization Scale (OVS) Tynes, Rose, and CFA Study 1: Not indicated - General online Study 1:

Williams (2010) 222 victimization 2 = 556.71, df=183, p<.001

Study 2: - Online sexual CFI=.93, IFI=.93, TLI=.92

254 victimization RMSEA=.096 (.087-.105)

- Individual online

racial discrimination Study 2:

- Vicarious online 2 = 570.30, df=183, p<.001

racial discrimination CFI=.94, IFI=.94, NNFI=.93

RMSEA=.087 (.079-.095)

13
 ACCEPTED MANUSCRIPT

3. Objective

The goal of this study is to analyze the factor and criterion validity and

reliability of the Cybervictimization Questionnaire (CYVIC) in a sample of adolescents

from Asturias (Spain). The CYVIC was designed based on a previous instrument (CBV;

Álvarez-García, Dobarro, & Núñez, 2015), to assess the extent to which the informant is

a victim of aggression by mobile phone or Internet. With the CYVIC, we expect to

overcome the main methodological and content limitations of the previously published

questionnaires for the assessment of cybervictimization among adolescents. This is

intended to contribute to the definition of the construct cybervictimization, its types, and

behavioral indicators, as well as to provide a valid and reliable measure for use in

applied fields and research.

Concerning this issue, as it was stated before in section 1.2, cybervictimization

may have a negative impact on the psychosocial wellbeing and academic performance

of the adolescent. Thus, it is important to have instruments that allow assessing this

problem with the greatest validity and reliability possible. The CYVIC is expected to

contribute to improving practice in educational and clinical contexts in different ways.

Firstly, it is expected to be a useful instrument to identify adolescents being victims of

cyberaggression. Collectively administered, the CYVIC would serve as a screening

measure. Individually applied, it would serve as a guide about the observable indicators

of each type of cybervictimization. Secondly, it is expected that the CYVIC will be

useful for analyzing the prevalence of the different forms of cybervictimization, as well

as the risk factors and consequences associated with them. This information is decisive

for the design of effective intervention programs. Lastly, the CYCIC is expected to be

useful for measuring the impact of intervention programs.

14
 ACCEPTED MANUSCRIPT

The initial hypothesis of this study is that the model that best represents the

response of the assessed adolescents to the CYVIC is the one purposed by Nocentini et

al. (2010), made up of four types of cybervictimization correlated to each other: written-

verbal, visual, online exclusion, and impersonation (Figure 1).

Figure 1. Theoretical model of the Cybervictimization Questionnaire (CYVIC) for

adolescents.

15
 ACCEPTED MANUSCRIPT

4. Material and methods

4.1. Participants

The responses of 3,159 teenagers from Asturias (Spain) were analyzed.

Regarding sex and age, 52.0% are boys and 48.0% are girls, aged between 12 and 18

years [M = 14.01; SD = 1.39; Skewness= 0.28 (SE= 0.04); Kurtosis= -0.56 (SE= 0.09)].

Of them, 95.1% have their own mobile phone, 84.1% surf the Internet in their free time

for non-academic tasks, 93.5% use instant messaging programs (Messenger, Whatsapp,

or others), and 77.8% participate in social networks (Tuenti, Facebook, or others) in

their free time.

The assessed adolescents belong to 19 schools, selected randomly through

stratified sampling from among the total of publicly funded schools of Asturias in which

Compulsory Secondary Education (CSE) is taught. Publicly funded schools make up

95.9% of the total number of schools in which CSE is taught in Asturias. To select the

sample, we divided the population of schools according to their ownership (public or

private-concerted) and selected in each stratum a number of schools proportional to the

population. In Spain, public schools are those in which both their management and their

financing are entirely public, and private-concerted schools are those whose

management is private, but their financing is partially public. As a result, 11 public

schools and 8 private-concerted schools were selected. Although the sample is diverse,

the selected schools were predominantly located in urban areas and from the middle

socio-economic class, as in the population from which they were extracted. In each

selected school, all the groups were assessed. Of these adolescents, 28.3% study 1st

grade, 25.4% are in 2nd grade, 24.3% are studying 3rd grade, and 22.0% study 4th grade

of CSE.
16
 ACCEPTED MANUSCRIPT

4.2. Measures

Cybervictimization Questionnaire for Adolescents (CYVIC)

The CYVIC (Appendix A), designed and tested in this study, is a self-report

composed of 19 items, each one of which presents an aggression suffered through

mobile phone or the Internet. The students should mark the frequency with which they

were the victim of each one of these situations in the past three months, on a 4-point

Likert-type scale (1 = Never, 2 = Rarely, 3 = Often, and 4 = Always).

The CYVIC was designed based on the model purposed by Nocentini et al.

(2010), and the validation of a previously published questionnaire, the

Cybervictimization Questionnaire (CBV; Álvarez-García et al., 2015). For the final

construction of the CYVIC, ten items were removed from the CBV, on statistical and

content grounds, and three new items were added, referring to visual cybervictimization

with sexual connotation, an important kind of cybervictimization that has not been

assessed in the reviewed self-reports.

The original CBV has 26 items, also based on the model of Nocentini et al.

(2010), with four types of cybervictimization. Confirmatory factor analyses yielded

very similar fit indexes for the three models compared (unifactorial, simple

multifactorial, and hierarchical multifactorial), so they did not strongly support any of

the positions on the dimensionality of the construct. We finally chose the one-factor

model, due to its more parsimonious nature, although the fit indexes were not good for

any of the three models considered a priori. To achieve an acceptable fit, we included in

the one-factor model the main correlations between the measurement errors of the items

(a total of 13 correlations). The inclusion in a model of so many correlations leads to

17
 ACCEPTED MANUSCRIPT

two problems. Firstly, the possibility that the relation between the pairs of items may be

due to variables other than those considered in the model. Secondly, there is a greater

probability that the resulting model will be unstable when replicated in other samples.

To try to solve both problems, when designing the CYVIC, we removed 10 items from

the CBV that were involved in these measurement error correlations, also taking into

account their factor loadings in the analyzed models.

To construct the CYVIC, besides eliminating these 10 items, we included 3 new

items of visual cybervictimization of a sexual nature. This was done to obtain a specific

measure of visual-sexual cybervictimization, differentiated from other types of visual

violence more related to teasing and happy slapping. They are different phenomena,

with different risk factors and concomitant variables. As shown in Section 2, none of

the reviewed questionnaires of cybervictimization among adolescents offer a specific

score of visual cybervictimization that involves teasing or happy slapping or of visual

cybervictimization related to sexting and sextortion, in spite of their severity and the

relevance of identifying and analyzing them.

Internet Risk Behaviors

An ad hoc adaptation of the Internet Risk Behaviors scale of the

Cybervictimization Risk Factors Questionnaire (Álvarez-García, Núñez, Dobarro, &

Rodríguez, 2015) was used. The scale used in the present study is a self-report measure

consisting of 8 items describing Internet risk behaviors, that is, potentially harmful

behaviors related to the use of the Internet: “I allow others to upload my photos or

videos to Internet”, “I add people I do not know personally to my social networks”,

“Sometimes, I've communicated online with strangers (for example, chatting)”, “I

usually publish personal information on my social network: what am I going to do,

18
 ACCEPTED MANUSCRIPT

where and with whom, personal or family photos or videos,..”, “Sometimes, I've made a

date with someone in person whom I only knew from Internet”, “I give my phone

number to boys or girls whom I just met or hardly know”, “I have sent a compromising

picture of me to other people via mobile phone or Internet”, and “I have exchanged

personal information with people whom I only know through the Internet”. The items

are rated on a 4-point Likert-type scale (1 = Totally False, 2 = Rather false, 3 = Mostly

true, 4 = Completely true). The internal consistency of the scale in the sample for this

study is high (Raykov’s ρ = .878).

Offline school victimization

We used the Offline School Victimization scale of the Cybervictimization Risk

Factors Questionnaire (Álvarez-García, Núñez, Dobarro, & Rodríguez, 2015). This

self-report scale is comprised of six items referring to the frequency with which the

respondent reports having suffered offline school victimization; that is, aggression in

the physical environment at school: “Some classmates reject me in games, on walks, or

in recess activities”, “My classmates avoid me when they do homework in a group in

the class”, “My classmates mock me or laugh at me”, “My classmates speak ill of me

behind my back”, “I have been insulted face-to-face by some classmate”, and “Some

classmate has hit me, either in the school or outside of the school grounds”. The items

are rated on a 4-point Likert-type scale (1 = Never, 2 = Rarely, 3 = Often, 4 = Always).

The internal consistency of the scale in the sample for this study is high (Raykov’s ρ =

.883).

Self-esteem

19
 ACCEPTED MANUSCRIPT

We used the Self-esteem scale of the Cybervictimization Risk Factors

Questionnaire (Álvarez-García, Núñez, Dobarro, & Rodríguez, 2015). This self-report

scale is made up of five items related to the respondent's self-assessment: “I'm content

about my physical appearance”, “I am proud of what I do”, “I can do things at least as

well as most of my classmates”, “I like the way I am” and “I consider myself a good

person”. The items are rated on a 4-point Likert-type scale (1 = Totally false, 2 = Rather

false, 3 = Mostly true, 4 = Completely true). The internal consistency of the scale in the

sample for this study is high (Raykov’s ρ = .829).

4.3. Procedure

Firstly, the CYVIC questionnaire was designed and the schools were selected.

After selecting the schools, permission was sought from their respective management

teams to apply the questionnaire. Each board of directors was informed of the goals and

procedures of the study, its voluntary and anonymous nature, and the confidential

treatment of the results. Once the school management had agreed to participate,

informed consent was sought from the parents or guardians of the students, given their

status as minors. The questionnaire was applied in all the centers during the second and

third trimester of the 2014-2015 academic course. Before completing the questionnaire,

students were also informed of the anonymous, confidential, and voluntary nature of

their participation. In general, the students had 20 minutes to complete the

questionnaires, although this was flexible depending on the students’ age and

characteristics. The test was applied by the research team to all the groups in each of the

schools, during the regular school schedule.

4.4. Data analysis

20
 ACCEPTED MANUSCRIPT

Initially, the presence of missing values or outliers in the data matrix was

analyzed. All the items were completed validly by more than 90% of the participants, so

none were eliminated for this reason for the final analysis of the test. Of the 3233

students who completed the CYVIC, 74 responded validly to less than 90% of the

items, so they were eliminated from the data matrix. After their data were deleted, the

number of missing values in the CYVIC matrix dropped to 125 (0.2% of the total).

We used the statistical program EQS 6.2 (Bentler, 2014) to analyze the

normality and the dimensionality of the test. As there are sufficient theoretical grounds

to hypothesize a priori about the dimensions that make up the cybervictimization

construct, we chose confirmatory factor analysis to examine the dimensionality of the

scale and, thereby, to test the theory (Izquierdo, Olea, & Abad, 2014; Rios & Wells,

2014). Multivariate kurtosis indicated the non-normality of the scale (Mardia’s

coefficient = 2002.94), so we used Arbitrary Generalized Least Squares (AGLS) as

estimation method. Given the ordinal nature of the scale, analyses were conducted using

the polychoric correlation matrix. As AGLS requires extensive sample sizes and, in

order not to eliminate more subjects and to be able to use all the available data, missing

values were treated by computing the pairwise covariance matrix.

To determine the degree of fit of the tested models, we used the chi-square (2) /

degrees of freedom (df) ratio, the Comparative Fit Index (CFI), the Bentler-Bonett Non-

Normed Fit Index (NNFI), the Standardized Root Mean Square Residual (SRMR), the

Root Mean Square Error of Approximation (RMSEA) and the Akaike Information

Criterion (AIC). Usually, it is considered that the fit is good when CFI ≥ .95, NNFI ≥

.95, SRMR ≤.08 and RMSEA ≤ .06 (Hu & Bentler, 1999), and χ2/df < 3 (Ruiz, Pardo, &

San Martín, 2010). The AIC allows comparing the models, and the lowest value is

21
 ACCEPTED MANUSCRIPT

preferable. Model fit differences were also determined using the chi-square difference

test (Bollen, 1989).

We studied the discriminant validity of the model with the best fit, analyzing the

correlation between its factors. Very high correlations (r ≥. 85) suggest possible

collinearity, redundancy between factors and, therefore, poor discriminant validity

(Brown, 2015). Given that we actually found a very high correlation between some

factors, the model was respecified post-hoc.

The fit indexes and correlations between the factors of the three respecified

models were analyzed. After identifying the model with the most adequate discriminant

validity and fit to the data, factor loadings and standard errors of each item for that

model were found. Factor loadings higher than .30 are usually considered acceptable

(Izquierdo et al., 2014).

Subsequently, the reliability of the test was analyzed. The reliability of each

subscale was analyzed, in terms of internal consistency, finding their respective Raykov

rho coefficient (composite reliability) from the polychoric correlation matrix. This

coefficient is currently proposed as an alternative to the weaknesses of the Cronbach

alpha coefficient, highlighted by authors like Sijtsma (2009). In general, the Raykov rho

coefficient is considered acceptable if its value is above .70, high if it is greater than .80,

and very high if it is higher than .90. We also found the multiple squared correlation of

each item, an indicator of the proportion of the item’s variance explained by the latent

variable and, therefore, the item’s reliability to measure this variable (Bollen, 1989).

We used the statistical software SPSS 21 (IBM Corp., 2012) to analyze the

criterion validity of the test. We calculated Spearman's rank correlation coefficient

between the score in each of the four factors of the CYVIC and three external criteria,

about whose association with cybervictimization there is prior evidence (Álvarez-

22
 ACCEPTED MANUSCRIPT

García, Núñez, Dobarro, & Rodríguez, 2015): internet risk behaviors, offline school

victimization, and self-esteem.

5. Results

5.1. Analysis of the initial theoretical model

Factor validity

We compared the fit of the data to the theoretical model, consisting of four

factors (4FM), with that of four other models that were also plausible from a theoretical

point of view (Table 2). On the one hand, a model with six factors was tested (6FM),

which, unlike the 4FM, makes the distinction between two types of visual

cybervictimization (sexual and teasing/happy slapping), and between two types of

verbal cybervictimization (oral and written). On the other hand, all the items from the

questionnaire refer to the same construct (cybervictimization). Hence, it makes sense to

also test a one-factor model (1FM), as well as hierarchical versions of the four- and six-

factor models, with the cybervictimization factor as a second-order factor (4FM2 and

6FM2). In the case of the non-hierarchical models tested (M4F and M6F), the factors

are latent variables significantly related to each other and free of measurement error;

each item (observable indicator) is only explained by one factor, and is associated with

a certain measurement error. The results (Table 3) indicate that it is not the 4FM, but the

6FM, the one that best fit indexes presents. Goodness of fit of the 6FM is better, in a

statistically significant way, than that of the 1FM (Δ2= 506.59; Δdf= 15; p<.001), the

4FM (Δ2= 95.20; Δdf= 9; p<.001), the 4FM2 (Δ2= 107.55; Δdf= 11; p<.001) and the

6FM2 (Δ2= 194.49; Δdf= 9; p<.001).

23
 ACCEPTED MANUSCRIPT

Table 2. Models proposed for the analysis of the dimensionality of the CYVIC.

Factors
Model Items
Second-order First-order

1FM - Cybervictimization All (1-19)

Impersonation
1, 12, and 18
Visual cybervictimization 2, 3, 6, 9, 10, 14,
4FM - and 15
Written-Verbal cybervictimization 5, 7, 8, 11, 17, and
19
Online exclusion
4, 13, and 16
Impersonation
1, 12, and 18
Visual cybervictimization 2, 3, 6, 9, 10, 14,
4FM2 Cybervictimization and 15
Written-Verbal cybervictimization 5, 7, 8, 11, 17, and
19
Online exclusion
4, 13, and 16
Impersonation 1, 12, and 18

Visual-Sexual cybervictimization 2, 9, and 14

Visual cybervictimization -Teasing /

3, 6, 10, and 15
6FM - Happy slapping

Verbal-written cybervictimization 8, 11, and 19

Verbal-oral cybervictimization 5, 7, and 17

Online exclusion 4, 13, and 16

Impersonation 1, 12, and 18

Visual-Sexual cybervictimization 2, 9, and 14

Visual cybervictimization -teasing /

3, 6, 10, and 15
6FM2 Cybervictimization Happy slapping

Verbal-written cybervictimization 8, 11, and 19

Verbal-Oral cybervictimization 5, 7, and 17

Online exclusion 4, 13, and 16

1FM = One-factor model; 4FM = Four-factor model; 4FM2 = Hierarchical model with four first-order

24
 ACCEPTED MANUSCRIPT

factors and one second-order factor; 6FM = Six-factor model; 6FM2 = Hierarchical model with six first-

order factors and one second-order factor.

Table 3. Goodness-of-fit indexes of the five tested models for the CYVIC, with the total

sample (N = 3159).

RMSEA
Model 2 df p 2/df CFI NNFI SRMR AIC
[CI 90%]

1FM 863.34 152 <.001 5.68 .921 .911 .159 .038(.036-.041) 559.34

4FM 451.95 146 <.001 3.10 .966 .960 .106 .026(.023-.028) 159.95

4FM2 464.30 148 <.001 3.14 .965 .959 .109 .026(.023-.029) 168.30

6FM 356.75 137 <.001 2.60 .976 .969 .090 .023(.020-.025) 82.75

6FM2 551.24 146 <.001 3.78 .955 .947 .114 .030(.027-.032) 259.24

1FM = One-factor model; 4FM = Four-factor model; 4FM2 = Hierarchical model with four first-order

factors and one second-order factor; 6FM = Six-factor model; 6FM2 = Hierarchical model with six first-

order factors and one second-order factor.

CFI = Comparative Fit Index; NNFI = Non-Normed Fit Index; SRMR = Standardized Root Mean Square

Residual; RMSEA = Root Mean Square Error of Approximation; CI = Confidence interval; AIC = Akaike

Information Criterion.

Discriminant validity

The factors of the 6FM present positive and statistically significant correlations

with each other (Table 4). The two factors referring to verbal violence (written and oral)

and, even more so, the two factors referring to visual violence (Sexual and

Teasing/Happy slapping) have very high correlations with each other (r >. 85).

25
 ACCEPTED MANUSCRIPT

Table 4. Correlation matrix between the components of the Six-factor model (N =

3159).

Visual-
Visual- Verbal- Verbal- Online
Impersonation Teasing/Happy
Sexual Written Oral Exclusion
slapping

Impersonation 1.00

Visual-Sexual .78 1.00

Visual-

Teasing/Happy .82 .98 1.00

slapping

Verbal-Written .73 .79 .72 1.00

Verbal-Oral .75 .66 .70 .86 1.00

Online Exclusion .71 .69 .73 .82 .68 1.00

5.2. Post-hoc modifications

In view of the very high correlation between the two factors of Visual

cybervictimization, we decided to extract the four items referring to Teasing/Happy

slapping from the model and retain them in the questionnaire as additional indicators.

We decided to remove them from the model and not to merge both factors for the

practical use of the questionnaire. It is interesting to have a specific measure of visual

cybervictimization of a sexual nature, distinguished from other types of visual

cybervictimization.

Thus, three models were tested post-hoc. First, we analyzed the fit indexes and

correlations between factors of a five-factor model (ph5FM), made up of the same

factors as the 6FM except for the factor of Visual Cybervictimization-Teasing/Happy

slapping. This model presents good fit indexes (Table 5). However, the two factors of

Verbal cybervictimization (Oral and Written) still present a correlation of .86. For this
26
 ACCEPTED MANUSCRIPT

reason, a second model was tested post-hoc, with four factors, in which the two verbal

factors are combined into a single factor. In this way, the four-factor model (ph4FM)

consists of Impersonation, Online exclusion, Verbal Cybervictimization, and Visual-

Sexual Cybervictimization. The fit of the 4FM is good — although worse than that of

the ph5FM (Δ2= 71.30; Δdf= 4; p<.001)— and the correlations between its factors are

lower than .85 (Figure 2). We finally tested a third post-hoc model, in which we merged

the items referring to Verbal cybervictimization and Exclusion into a single factor.

Thus, the three-factor model (ph3FM) is made up of the factors “Impersonation”,

“Visual-sexual cybervictimization”, and “Verbal cybervictimization and exclusion”.

The correlation between each pair of factors is lower than .80, but its level of fit to the

data is the worst of the three post-hoc models tested (Table 5). The fit of the ph3FM is

worse, in a statistically significant way, than that of the ph4FM (Δ2= 58.12; Δdf= 3;

p<.001) and the ph5FM (Δ2= 129.42; Δdf= 7; p<.001).

Table 5. Goodness of fit indexes of the three post-hoc models, with the total sample (N

= 3159).

RMSEA [CI
Model 2 df p 2/df CFI NNFI SRMR AIC
90%]

.019 [.016,
ph5FM 175.75 80 <.001 2.20 .982 .976 .061 15.75
.023]

.025 [.021,
ph4FM 247.05 84 <.001 2.94 .969 .961 .083 79.05
.028]

.028 [.025,
ph3FM 305.17 87 <.001 3.51 .958 .949 .092 131.17
.032]

27
 ACCEPTED MANUSCRIPT

ph5FM = Five-factor model; ph4FM = Four-factor model; ph3FM = Three-factor model.

CFI = Comparative Fit Index; NNFI = Non-Normed Fit Index; SRMR = Standardized Root Mean Square

Residual; RMSEA = Root Mean Square Error of Approximation; CI = Confidence interval; AIC = Akaike

Information Criterion.

5.3. Analysis of the post-hoc four-factor model

Factor loadings and standard errors

In the ph4FM, the standardized factor loadings of the items presented moderate

or high values (Figure 2). All the items have a standardized regression coefficient equal

to or higher than .60, except Item 5 (λ =. 58). However, some items have high standard

errors.

28
 ACCEPTED MANUSCRIPT

Figure 2. Factorial structure of the CYVIC (I = Impersonation; VS = Visual-Sexual

Cybervictimization; WV = Written-Verbal Cybervictimization; E = Online exclusion).

Reliability of items and subscales

The reliability of the CYVIC subscales and of each of their items can be

considered moderate or high (Table 6). The Raykov rho composite reliability coefficient

29
 ACCEPTED MANUSCRIPT

of each of its four factors has values between .74 and .89. The proportion of variance in

an item explained by the latent variable (R2) ranges between 34 and 77%.

Table 6. Reliability of each item and factor of the CYVIC (N = 3159).

Factor Item ρ R2

Impersonation .816

1 .610

12 .765

18 .514

Visual-Sexual Cybervictimization .795

2 .561

9 .738

14 .540

Written-Verbal Cybervictimization .890

5 .339

7 .709

8 .633

11 .533

17 .760

19 .590

Online exclusion .741

4 .357

13 .458

16 .633

30
 ACCEPTED MANUSCRIPT

ρ: Raykov's rho coefficient (composite reliability); R2: Squared multiple correlation.

Criterion validity

The four CYVIC factors correlated statistically and significantly with the three

analyzed external criteria. There was a positive correlation with participants´ reported

Internet Risk Behaviors and Offline School Victimization; and a negative one with Self-

esteem (Table 7).

Table 7. Spearman's rank correlation coefficients between the score on each factor of

the CYVIC and the scores on the Internet Risk Behaviors, Offline School Victimization,

and Self-esteem scales (N = 3159).

Internet Risk Behaviors Offline School Victimization Self-esteem

I .181* .163* -.063*

VS .191* .186* -.101*

WV .377* .369* -.184*

E .148* .326* -.140*

I = Impersonation; VS = Visual-Sexual Cybervictimization; WV = Written-Verbal

Cybervictimization; E = Online exclusion

* p ≤ .001.

6. Discussion

The Cybervictimization Questionnaire (CYVIC) for adolescents was designed to

assess the extent to which the informant is a victim of aggression by mobile phone or

Internet. The goal of the present study was to analyze its factor and criterion validity

and reliability in a sample of adolescents of Asturias (Spain). The results show that the
31
 ACCEPTED MANUSCRIPT

CYVIC offers adequate statistical guarantees for the purpose for which it was designed.

The initial hypothesis, according to which the model that best represents the structure of

the test is the one purposed by Nocentini et al. (2010), made up of four types of

cybervictimization (written-verbal, visual, online exclusion, and impersonation), was

confirmed, but with nuances.

The model that best represents the data obtained, with adequate discriminant

validity, is the one made up of four factors (written-verbal, visual-sexual, online

exclusion, and impersonation) and in which the four items of Visual Cybervictimization

- Teasing/Happy slapping are removed from the model and maintained as

complementary indicators. These results suggest that the cybervictimization construct

consists of different, significantly interrelated types. As indicated in Section 2, the

previously published questionnaires to assess cybervictimization in adolescence differ

in terms of the univariate or multifactorial nature of the construct, although most of

them show an unifactorial structure. Conversely, the results obtained in this study

support the multifactorial perspective of the construct, in the line of other questionnaires

such as CYBVIC (Buelga et al., 2012), CVBS (Çetin et al., 2011), FCBVSs (Palladino

et al., 2015), MOOPV (Sumter et al., 2015), and OVS (Tynes et al., 2010). The CYVIC,

however, differs from these multifactorial scales in the specific types of

cybervictimization that it assesses and in the observable indicators used. The results do

not justify the division of the two factors of Verbal cybervictimization and the two

factors of Visual cybervictimization, which was considered in two of the models tested

in the present study. These two divisions could be of practical interest. However, from

a psychometric perspective, the results suggest that the response of the evaluated

adolescents does not differentiate the observable indicators of these factors sufficiently

32
 ACCEPTED MANUSCRIPT

to consider two different types of visual violence and two different types of verbal

violence.

Regarding the criterion validity of the test, the CYVIC scores correlate

statistically and significantly with three external variables that previous evidence

indicates correlate with cybervictimization in adolescence. In accordance with our

expectations (Álvarez-García, Núñez, Dobarro, & Rodríguez, 2015), the CYVIC scores

correlated positively with Offline school victimization and Internet risk behaviors, and

negatively with Self-esteem. The CYVIC subscales differ from one another in the

strength of the correlation, which supports the multidimensional nature of the construct.

The reliability of the subscales, evaluated in terms of internal consistency, is

moderate-high. The reliability of the items can be considered acceptable. Some of the

items present high standard errors. Although this may be due to the influence of

exogenous variables that are not considered in this study, it may also be due, at least

partially, to the correlation between the scale factors. For example, the variance of the

scores on Item 4 can be explained by the latent variable Online exclusion, but also

partially by the variable Verbal Cybervictimization.

6.1. Implications

The present work represents a contribution to the study of cybervictimization

among adolescents. The CYVIC overcomes some methodological and content

limitations of the previously published questionnaires described in Section 2. Regarding

the methodological aspects, the CYVIC has been validated with a broad sample of

randomly selected adolescents; the analyses were based on the polychoric correlation

matrix; and the model finally identified as representative of the internal structure of the

questionnaire shows a good fit to the empirical data obtained. With regard to the
33
 ACCEPTED MANUSCRIPT

contents, the CYVIC provides a measure of four types of cybervictimization and four

additional indicators, which allows a more accurate analysis of the problem, its

prevalence, risk factors, and the effect of interventions, in comparison with most

questionnaires, which offer a single general cybervictimization score. In comparison

with the few questionnaires that provide specific measures of different types of

cybervictimization, the CYVIC offers some new features. The CYVIC provides a

specific measure of visual cybervictimization with a sexual component, related to

sexting and sextortion, separate from other forms of visual cybervictimization related to

teasing and happy slapping. None of the reviewed questionnaires offers this possibility.

In addition, the CYVIC offers a measure of online exclusion and another of

impersonation, not very common in the reviewed questionnaires. These contributions,

along with the possibility of obtaining score of verbal cybervictimization, more

commonly included in this type of questionnaires, makes the CYVIC a complete, useful

tool to identify and understand the problem.

In comparison with the CBV—the source of the final version of the CYVIC—,

the removal of 10 items and the addition of 3 items allowed us to more clearly identify

the model that best represents the internal structure of the questionnaire, to notably

improve the fit of the model to the empirical data obtained, and the internal consistency

of the factors, and, hence, to more clearly support the multidimensional nature of the

construct.

This work has various theoretical and practical implications. From the

theoretical point of view, it contributes to the conceptual delimitation of the construct:

its subtypes and observable indicators. The results support the hypothesis of the

multifactorial nature of cybervictimization, the relevance of the model proposed by

Nocentini et al. (2010), and the suitability of the observable indicators included in the
34
 ACCEPTED MANUSCRIPT

CYVIC. From the practical viewpoint, researchers, educators, and clinicians now have

available a brief assessment tool, easy to apply, encode and analyze, economical in

terms of time and cost when compared to other assessment methods, and with adequate

metric guarantees. As it has been stated in section 3, it is expected that this instrument

will be useful for the identification of adolescents being victims of cyberaggression; the

estimation of cybervictimization prevalence; the analysis of risk factors and

consequences associated with the problem; the design of intervention programs and the

analysis of their effectiveness. To date, the CYVIC has been successfully used to

determine prevalence rates and gender differences in cybervictimization (Álvarez-

García, Barreiro-Collazo, & Núñez, 2017). Likewise, the relationship found in the

present study between cybervictimization and the three external criteria analyzed

suggests the desirability for intervention programs for the prevention and treatment of

the cybervictimization in adolescence to include the prevention of internet risk

behaviors and offline school victimization, as well as the development of self-esteem,

from early stages.

6.2. Limitations and future directions

Therefore, this work is a contribution to the study of cybervictimization in

adolescence. However, it has some limitations. We note mainly three. Firstly, the

CYVIC is a self-report instrument, so the results may be affected by response bias such

as distortions or social desirability (Navarro-González, Lorenzo-Seva, & Vigil-Colet,

2016). In the future, a sincerity scale could be included, or the results could be

complemented with hetero-reports completed by peers or adults. Secondly, the

questionnaire has been tested with a broad and randomly selected sample, but it was

extracted from a specific population, limited to certain ages and geographical areas.
35
 ACCEPTED MANUSCRIPT

Although we do not expect that the sample used in this study will differ significantly

from others drawn from contexts similar to the one described in subsection 4.1, any

generalization of the findings of this study to other age groups or regions should be

made with caution (Muñiz, Elosua, & Hambleton, 2013). With a view to the future, the

validation of the test in other ages and cultural contexts would be of interest. Third and

lastly, the convergent validity of the questionnaire was not analyzed in the present

study. In future applications of the instrument, the correlation between the CYVIC

scores and those from other tests designed to measure the same variables should be

examined.

7. Conclusions

In recent years, different questionnaires for the assessment of cybervictimization

in adolescence have been published. All the reviewed questionnaires have made a

contribution to the study of cybervictimization. However, they present some significant

methodological and content-related limitations that we have attempted to overcome with

the design and validation of the CYVIC. The instrument was validated in a wide sample

of adolescents, greater than in most of the previously published questionnaires. The

polychoric correlation matrix has been used to perform the analysis, as required by the

characteristics of the analyzed data. Unlikely most of the previously published

questionnaires, the CYVIC shows a multifactorial structure, which allows obtaining

individual scores for specific types of cybervictimization. This instrument provides a

measure of verbal cybervictimization, which is common in the few multidimensional

questionnaires published to date, but it also includes a measure of social exclusion,

impersonation, visual cybervictimization-teasing/happy slapping, and visual-sexual

cybervictimization. These dimensions are not included in most of the previously

36
 ACCEPTED MANUSCRIPT

validated questionnaires, despite of their theoretical and practical relevance. Results

from the present study indicate that the CYVIC shows better goodness of fit indexes

than the most of the previously published instruments.

The validation of the CYVIC contributes to the theoretical development of the

field of study, as it helps to define cybervictimization, its types and observable

indicators. In this sense, our findings support the pertinence of the typology purposed by

Nocentini et al. (2010), although the indicators used in the present study differ from

those included by the authors in their recent questionnaire (Palladino et al., 2015). From

a practical perspective, the CYVIC offers a valid and reliable measure of

cybervictimization, suitable to be administered in both applied and research contexts.

This instrument allows professionals to detect adolescent being victims of

cyberaggression; to analyze the prevalence of different types of cybervictimization; to

define risk factors and consequences associated with the problem; to design, on the

basis of this information, effective intervention programs; and to measure the impact of

these programs. Regarding the design of intervention programs, results from the present

study suggest that, for the prevention and treatment of cybervictimization, it is

necessary to take into account, among other variables, those related to Internet risk

behaviors, offline school victimization and self-esteem.

8. References

Aboujaoude, E., Savage, M. W., Starcevic, V., & Salame, W. O. (2015). Cyberbullying:

Review of an old problem gone viral. Journal of Adolescent Health, 57(1), 10-

18. doi: 10.1016/j.jadohealth.2015.04.011

37
 ACCEPTED MANUSCRIPT

Álvarez-García, D., Barreiro-Collazo, A, & Núñez, J. C. (2017). Cyberaggression

among Adolescents: Prevalence and Gender Differences. Comunicar, 50. doi:

10.3916/C50-2017-08

Álvarez-García, D., Dobarro, A., Álvarez, L., Núñez, J. C., & Rodríguez, C. (2014). La

violencia escolar en los centros de Educación Secundaria de Asturias desde la

perspectiva del alumnado [School Violence in Secondary Schools in Asturias

from the Perspective of Students]. Educación XX1, 17(2), 337-360. doi:

10.5944/educxx1.17.2.11494

Álvarez-García, D., Dobarro, A., & Núñez, J. C. (2015). Validez y fiabilidad del

Cuestionario de Cibervictimización en estudiantes de Secundaria [Validity and

reliability of the Cybervictimization Questionnaire in Secondary students]. Aula

Abierta, 43(1), 32-38. doi: 10.1016/j.aula.2014.11.001

Álvarez-García, D., Núñez, J. C., Dobarro, A., & Rodríguez, C. (2015). Risk factors

associated with cybervictimization in adolescence. International Journal of

Clinical and Health Psychology, 15, 226-235. doi: 10.1016/j.ijchp.2015.03.002

Bentler, P. M. (2014). EQS 6.2 for Windows (Build 107)- Structural Equation Modeling

Software. Encino, CA: Multivariate Software.

Beran, T., & Li, Q. (2007). The relationship between cyberbullying and school bullying.

The Journal of Student Wellbeing, 1(2), 15–33.

Berne, S., Frisén, A., Schultze-Krumbholz, A., Scheithauer, H., Naruskov, K., Luik, P.,

Katzer, C., Erentaite, R., & Zukauskiene, R. (2013). Cyberbullying assessment

instruments: A systematic review. Aggression and Violent Behavior, 18(2), 320-

334. doi: 10.1016/j.avb.2012.11.022

Betts, L. R., Houston, J. E., & Steer, O. L. (2015). Development of the

Multidimensional Peer Victimization Scale–Revised (MPVS-R) and the

38
 ACCEPTED MANUSCRIPT

Multidimensional Peer Bullying Scale (MPVS-RB). The Journal of Genetic

Psychology, 176(2), 93-109. doi: 10.1080/00221325.2015.1007915

Bollen, K. A. (1989). Structural equations with latent variables. New York: John Wiley

& Sons.

Bonanno, R. A., & Hymel, S. (2013). Cyber bullying and internalizing difficulties:

Above and beyond the impact of traditional forms of bullying. Journal of Youth

and Adolescence, 42, 685-697. doi: 10.1007/s10964-013-9937-1

Brown, C. F., Demaray, M. K., & Secord, S. M. (2014). Cyber victimization in middle

school and relations to social emotional outcomes. Computers in Human

Behavior, 35, 12-21. doi: 10.1016/j.chb.2014.02.014

Brown, T. A. (2015). Confirmatory factor analysis for applied research (2nd ed). New

York, NY: Guilford Press.

Buelga, S., Cava, M. J., & Musitu, G. (2012). Validación de la Escala de Victimización

entre Adolescentes a través del Teléfono Móvil y de Internet [Validation of the

Scale of Victimization among Adolescents through the Mobile Phone and the

Internet]. Revista Panamericana de Salud Publica, 32(1), 36-42.

Calvert, C. (2015). Revenge Porn and Freedom of Expression: Legislative Pushback to

an Online Weapon of Emotional and Reputational Destruction. Fordham

Intellectual Property, Media and Entertainment Law Journal, 24(3), 673-702.

Çetin, B., Yaman, E., & Peker, A. (2011). Cyber Victim and Bullying Scale: A study of

validity and reliability. Computers & Education, 57, 2261-2271. doi:

10.1016/j.compedu.2011.06.014

Chang, F. C., Lee, C. M., Chiu, C. H., Hsi, W. Y., Huang, T. F., & Pan, Y. C. (2013).

Relationships among cyberbullying, school bullying, and mental health in

Taiwanese adolescents. Journal of School Health, 83(6), 454-462.

39
 ACCEPTED MANUSCRIPT

Del Rey, R., Casas, J. A., Ortega-Ruiz, R., Schultze-Krumbholz, A., Scheithauer, H.,

Smith, P., Thompson, F., Barkoukis, V., Tsorbatzoudis, H., Brighi, A., Guarini,

A., Pyzalski, J., & Plichta, P. (2015). Structural validation and cross-cultural

robustness of the European Cyberbullying Intervention Project Questionnaire.

Computers in Human Behavior, 50, 141–147. doi: 10.1016/j.chb.2015.03.065

Del Rey, R., Elipe, P., & Ortega-Ruiz, R. (2012). Bullying and cyberbullying:

Overlapping and predictive value of the co-occurrence. Psicothema, 24(4), 608-

613.

Drouin, M., Ross, J., & Tobin, E. (2015). Sexting: A new, digital vehicle for intimate

partner aggression? Computers in Human Behavior, 50, 197-204. doi:

10.1016/j.chb.2015.04.001

Fernández-Montalvo, J., Peñalva, A., & Irazabal, I. (2015). Internet Use Habits and

Risk Behaviours in Preadolescence. Comunicar, 44, 113-120. doi: 10.3916/C44-

2015-12

Flores, J. (2015). Sexting: Teens, sex, smartphones and the rise of sextortion and gender

based digital violence. In S. Cortesi & U. Gasser (Eds.), Digitally connected:

Global perspectives on youth and digital media (pp. 22-25). Cambridge, MA:

Harvard University - Berkman Center for Internet & Society.

Gámez-Guadix, M., Orue, I., & Calvete, E. (2013). Evaluation of the cognitive-

behavioral model of generalized and problematic Internet use in Spanish

adolescents. Psicothema, 25(3), 299-306. doi: 10.7334/psicothema2012.274

Gámez-Guadix, M., Villa-George, F., & Calvete, E. (2014). Psychometric properties of

the Cyberbullying Questionnaire (CBQ) among Mexican adolescents. Violence

and Victims, 29(2), 232-247. doi: 10.1891/0886-6708.VV-D-12-00163R1

40
 ACCEPTED MANUSCRIPT

Garaigordobil, M. (2011). Prevalence and consequences of cyberbullying: A review.

International Journal of Psychology and Psychological Therapy, 11(2), 233-

254.

Garaigordobil, M. (2015). Psychometric Properties of the Cyberbullying Test, a

Screening Instrument to Measure Cybervictimization, Cyberaggression, and

Cyberobservation. Journal of Interpersonal Violence, 19, 1-21. doi:

10.1177/0886260515600165

Hinduja, S., & Patchin, J.W. (2015). Bullying Beyond the Schoolyard (2nd Ed.).

Thousand Oaks, CA: Corwin.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure

analysis: Conventional criteria versus new alternatives. Structural Equation

Modeling: A Multidisciplinary Journal, 6(1), 1-55. doi:

10.1080/10705519909540118

Hunt, C., Peters, L., & Rapee, R. M. (2012). Development of a measure of the

experience of being bullied in youth. Psychological Assessment, 24(1), 156-165.

IBM Corp. (2012). SPSS Statistics for Windows, version 21.0. Armonk, NY: IBM Corp.

Izquierdo, I., Olea, J., & Abad, F. J. (2014). Exploratory factor analysis in validation

studies: Uses and recommendations. Psicothema, 26(3), 395-400. doi:

10.7334/psicothema2013.349

Jiménez, A. E., Castillo, V. D., & Cisternas, L. C. (2012). Validación de la Escala de

Agresión entre Pares, y subescala de Agresión Virtual en escolares chilenos

[Validation of the Peer Aggression Scale and the Virtual Aggression subscale in

Chilean students]. Revista Latinoamericana de Ciencias Sociales, Niñez y

Juventud, 10(2), 825-840.

41
 ACCEPTED MANUSCRIPT

Juvonen, J., & Gross, E.F. (2008). Extending the school grounds? - Bullying

experiences in cyberspace. Journal of School Health, 78(9), 496-505. doi:

10.1111/j.1746-1561.2008.00335.x

Klettke, B., Hallford, D. J., & Mellor, D. J. (2014). Sexting prevalence and correlates: A

systematic literature review. Clinical Psychology Review, 34(1), 44–53. doi:

10.1016/j.cpr.2013.10.007

Lam, L. T., & Li, Y. (2013). The validation of the E-Victimisation Scale (E-VS) and the

E-Bullying Scale (E-BS) for adolescents. Computers in Human Behavior, 29, 3-

7. doi: 10.1016/j.chb.2012.06.021

Lloret-Segura, S., Ferreres-Traver, A., Hernández-Baeza, A., & Tomás-Marco, I.

(2014). El análisis factorial exploratorio de los ítems: una guía práctica, revisada

y actualizada [Exploratory Item Factor Analysis: A practical guide revised and

updated]. Anales de Psicología, 30(3), 1151-1169. doi:

/10.6018/analesps.30.3.199361

Menesini, E., Nocentini, A., & Calussi, P. (2011). The measurement of cyberbullying:

Dimensional structure and relative item severity and discrimination.

Cyberpsychology, Behavior, and Social Networking, 14(5), 267-274. doi:

10.1089/cyber.2010.0002

Morelli, M., Bianchi, D., Baiocco, R., Pezzuti, L., & Chirumbolo, A. (2016). Not-

allowed sharing of sexts and dating violence from the perpetrator's perspective:

The moderation role of sexism. Computers in Human Behavior, 56, 163-169.

doi: 10.1016/j.chb.2015.11.047

Muñiz, J., Elosua, P., & Hambleton, R. K. (2013). International Test Commission

Guidelines for test translation and adaptation: Second edition. Psicothema,

25(2), 151-157. doi: 10.7334/psicothema2013.24

42
 ACCEPTED MANUSCRIPT

Navarro-González, D., Lorenzo-Seva, U., & Vigil-Colet, A. (2016). How response bias

affects the factorial structure of personality self-reports. Psicothema, 28(4), 465-

470. doi: 10.7334/psicothema2016.113

Nocentini, A., Calmaestra, J., Schultze-Krumbholz, A., Scheithauer, H., Ortega, R., &

Menesini, E. (2010). Cyberbullying: Labels, behaviours and definition in three

European countries. Australian Journal of Guidance & Counselling, 20(2), 129-

142.

Ortega-Ruiz, R., & Núñez, J. C. (2012). Bullying and cyberbullying: Research and

intervention at school and social contexts. Psicothema, 24(4), 603-607.

Palasinski, M. (2013). Turning assault into a “harmless prank”—teenage perspectives

on happy slapping. Journal of Interpersonal Violence, 28(9), 1909-1923. doi:

10.1177/0886260512469107

Palladino, B. E., Nocentini, A., & Menesini, E. (2015). Psychometric properties of the

Florence CyberBullying-CyberVictimization Scales. Cyberpsychology,

Behavior, and Social Networking, 18(2), 112-119. doi: 10.1089/cyber.2014.0366

Rios, J., & Wells, C. (2014). Validity evidence based on internal structure. Psicothema,

26(1), 108-116. doi: 10.7334/psicothema2013.260

Rose, Ch. A., & Tynes, B. M. (2015). Longitudinal associations between

cybervictimization and mental health among U.S. adolescents. Journal of

Adolescent Health, 57(3), 305-312. doi: 10.1016/j.jadohealth.2015.05.002

Ruiz, M. A., Pardo, A., & San Martín, R. (2010). Modelos de ecuaciones estructurales

[Structural equation models]. Papeles del Psicólogo, 31(1), 34-45.

Sijtsma, K. (2009). On the use, misuse, and the very limited usefulness of Cronbach’s

Alpha. Psychometrika, 74(1), 107-120. doi: 10.1007/s11336-008-9101-0

43
 ACCEPTED MANUSCRIPT

Stewart, R. W., Drescher, Ch. F., Maack, D. J., Ebesutani, Ch., & Young, J. (2014). The

development and psychometric investigation of the Cyberbullying Scale.

Journal of Interpersonal Violence, 29(12), 2218-2238. doi:

10.1177/0886260513517552

Sumter, S. R., Valkenburg, P. M., Baumgartner, S. E., Peter, J., & Van der Hof, S.

(2015). Development and validation of the Multidimensional Offline and Online

Peer Victimization Scale. Computers in Human Behavior, 46, 114–122. doi:

10.1016/j.chb.2014.12.042

Tokunaga, R. S. (2010). Following you home from school: A critical review and

synthesis of research on cyberbullying victimization. Computers in Human

Behavior, 26, 277-287. doi: 10.1016/j.chb.2009.11.014

Topcu, Ç., & Erdur-Baker, Ö. (2010). The Revised Cyber Bullying Inventory (RCBI):

Validity and reliability studies. Procedia Social and Behavioral Sciences, 5,

660-664. doi: 10.1016/j.sbspro.2010.07.161

Tsitsika, A., Janikian, M., Wójcik, S., Makaruk, K., Tzavela, E., Tzavara, C.,

Greydanus, D., Merrick, J., & Richardson, C. (2015). Cyberbullying

victimization prevalence and associations with internalizing and externalizing

problems among adolescents in six European countries. Computers in Human

Behavior, 51, 1-7. doi: 10.1016/j.chb.2015.04.048

Tynes, B. M., Rose, Ch. A., & Williams, D. R. (2010). The development and validation

of the Online Victimization Scale for Adolescents. Cyberpsychology: Journal of

Psychosocial Research on Cyberspace, 4(2), article 1.

http://cyberpsychology.eu/view.php?cisloclanku=2010112901&article=1

Van Geel, M., Vedder, P., & Tanilon, J. (2014). Relationship between peer

victimization, cyberbullying, and suicide in children and adolescents. A meta-

44
 ACCEPTED MANUSCRIPT

analysis. JAMA Pediatrics, 168(5), 435-442. doi:

10.1001/jamapediatrics.2013.4143

Wolak, J., & Finkelhor, D. (2011). Sexting: A typology. Durham, NC: Crimes Against

Children Research Center.

Wright, M., & Li, Y. (2013). The association between cyber victimization and

subsequent cyber aggression: The moderating effect of peer rejection. Journal of

Youth and Adolescence, 42, 662–674. doi: 10.1007/s10964-012-9903-3

45
 ACCEPTED MANUSCRIPT

Appendix A

Indicate how often you've been victim of the following situations, in the last 3 months:

1 2 3 4

1. Se han hecho pasar por mí en Internet publicando comentarios a mi

nombre, como si fuese yo. [Someone has impersonated me on the

Internet, posting comments under my name, as if they were me.]

2. Me han hecho fotos o grabaciones de vídeo sin mi consentimiento con

un contenido sexual o sugerente (por ej., en la playa, en un vestuario,..) y

las han difundido a través del móvil o Internet. [Someone has taken

pictures or video recordings of me with a sexual or suggestive content

(e.g., on the beach, in a locker room,..) without my consent and they have

disseminated them over the mobile phone or the Internet.]

3. Han colgado en Internet fotos mías trucadas (modificadas), para

hacerme daño o reírse de mí. [Someone has hung doctored (modified)

photos of me on the Internet to harm me or laugh at me.]

4. Me han echado o no me han aceptado en la lista de contactos de algún

chat, red social (por ej., Tuenti) o programa de mensajería instantánea (por

ej., Messenger, WhatsApp), sin haber hecho nada, sólo por ser yo. [I was

kicked out or not accepted on some chat list, social network contact list

(e.g., Tuenti) or instant messaging program (e.g., Messenger, Whatsapp),

without having done anything, just for being me.]

5. He recibido llamadas a mi móvil, que no contestan, supongo que para

46 that are not answered, I

fastidiar. [I have received calls on my mobile
 ACCEPTED MANUSCRIPT

guess to annoy me.]

6. Han colgado en Internet fotos o vídeos reales comprometidos, sin mi

permiso, para hacerme daño o reírse de mí. [Someone has hung real

compromising photos or videos of me on the Internet without my

permission, to harm me or make fun of me.]

7. He recibido llamadas para insultarme o burlarse de mí. [I have received

calls insulting me or making fun of me.]

8. Se han burlado de mí con comentarios ofensivos o insultantes en las

redes sociales. [Someone has made fun of me with offensive or insulting

comments on social networks.]

9. Han difundido sin mi permiso a través del teléfono móvil o Internet

imágenes o vídeos míos comprometidos (de tipo sexual, sugerente o

insinuante), que yo mismo/a había hecho. [Someone has disseminated,

without my permission, via mobile phone or Internet, compromising

images or videos of me (of a sexual, suggestive, or insinuating nature) that

I had taken.]

10. Me han pegado, lo han grabado y luego lo han difundido. [I have been

beaten, and others have recorded it and then disseminated it.]

11. He recibido insultos a través de mensajes cortos de texto (sms) o

programas de mensajería instantánea (por ej., WhatsApp). [I have received

insults through short text messages (sms) or instant messaging programs

(e.g., WhatsApp).]

12. Se han hecho pasar por mí en Twitter, Tuenti,.., creando un falso perfil

de usuario (foto, datos personales,..) con el que se me ha insultado o

47
 ACCEPTED MANUSCRIPT

ridiculizado. [I have been impersonated in Twitter, Tuenti, ... creating a

false user profile (photo, personal details,..) with which I was insulted or

ridiculed.]

13. Se han hecho quejas falsas sobre mí en algún foro, red social o juego

on-line, que han hecho que me expulsasen. [Someone has made false

complaints about me in some forum, social network, or online game,

which have caused me to be expelled.]

14. Me han presionado para hacer cosas que no quería (haya accedido

finalmente a hacerlas o no), amenazándome con difundir conversaciones o

imágenes íntimas mías. [I have been pressured to do things that I didn't

want to (whether or not I finally agreed to do them), threatening me with

disseminating my intimate conversations or images.]

15. Me han obligado a hacer algo humillante, lo han grabado y luego lo

han difundido para burlarse de mí. [They have forced me to do something

humiliating, they have recorded it, and then disseminated it to ridicule

me.]

16. Se ponen de acuerdo para hacerme el vacío (ignorarme) en las redes

sociales. [They agree to ignore me on the social networks.]

17. He recibido llamadas anónimas, para amenazarme o atemorizarme. [I

have received anonymous phone calls, to threaten me or intimidate me.]

18. Alguien que ha conseguido mi contraseña ha enviado mensajes

molestos a algún conocido, como si hubieses sido yo, para meterme en

líos. [Someone who has gotten my password has sent annoying messages

to someone I know, as if it were me, to get me into trouble.]

48
 ACCEPTED MANUSCRIPT

19. Se han publicado rumores falsos sobre mí en alguna red social. [There

have been false rumors about me on some social network.]

1 = Never; 2 = Rarely; 3 = Often; 4 = Always.

49
 ACCEPTED MANUSCRIPT

Acknowledgements

This work was financed by the Consejería de Economía y Empleo del

Principado de Asturias [Council of Economy and Employment of the Princedom of

Asturias] (Spain) (Ref. FC-15-GRUPIN14-053).