Introduction
New technologies have occupied a privileged space in people's lives (Castells & Cardoso, 2005), including relationships, although they can also be used to exert violence. In this sense, the empirical tradition on abuse and technological aggressions has developed around four main lines of research: (a) cyber violence in adolescence (Aboujaoude, Savage, Starcevic & Salame, 2015; Rodríguez-Domínguez, Martínez-Pecino & Durán, 2015), (b) sexually-motivated cyber violence perpetrated by adults towards minors (Górriz, 2016), (c) gender-based cyber violence (e.g., Donoso, Rubio & Vilà, 2017; 2018), also including cyber violence towards women by their romantic partners (e.g., Torres, Robles & De Marco, 2013); and (d) cyber violence produced within underage and/or adult relationships, whether exercised by men or women in same-sex or different-sex relationships (e.g., Borrajo, Gámez-Guadix & Calvete, 2015a; Dank, Lachman, Zweig & Yahner, 2013; Gámez-Guadix; Borrajo & Calvete, 2018; Martinez-Pecino & Durán, 2016; Zweig, Lachman, Yahner & Dank, 2014).
As far as this last line of research is concerned, certain methodological shortcomings are identified.
Firstly, the diversity of contexts in which violent cyber behaviors are reproduced is compounded by the use of different terms to describe this problem (Gámez-Guadix et al., 2018).
Secondly, it is difficult to establish a definitive classification of the different manifestations of this type of violence, as new behaviors emerge as technologies and our use of them evolve (Rodríguez-Domínguez, 2015; Torres et al., 2013). An example of this would be the recent emergence of new sexual practices related to cyber violence, such as sexting (Ibarra, 2014; Quesada et al, 2018; Rodríguez-Domínguez, Moreno & Durán, 2017) where the contents exchanged, at first, with consent (McLaughlin, 2010; Pérez et al., 2011; Wolak & Finkelhor, 2011), are subsequently used criminally to extort or harm the victim (Rodríguez-Domínguez et al., 2017).
Finally, another problem in cyber dating violence research was pointed out by Borrajo et al. (2015b), who maintain that the instruments used to study this construct do not show adequate psychometric properties, either because they underrepresent the construct by focusing on measuring specific types of cyber dating violence, or because they do not provide evidence of validity. The analysis of the evidences of validity and reliability would be a fundamental requirement in scientific research on human behavior (Prieto & Delgado, 2010). Thus, in the measurement of psychological variables it would be essential for studies to report on the reliability or stability of the measure and also on the factorial structure of the instruments as an approximation to the validity of the construct, in addition to generating additional evidence of empirical validity referring both to the construct and to the content or criterion (Fernández, 2008).
This disparity of study perspectives, together with the lack of evidence of validity and reliability of the instruments used, hinders a proper understanding of cyber dating violence (hereinafter CDV). As a public health problem (David-Ferdon & Feldman, 2007) with social and psychological repercussions (Donoso-Vázquez, 2018; Jabaloyas, 2015), it becomes necessary to solve both problems.
The objective of this review work is to analyze the methodology used to date in the evaluation of the CDV, identifying the dimensions of the phenomenon, as well as the main measurement instruments and the evidence of validity and reliability obtained through their use.
Method
Search
The search for scientific articles was conducted in April 2018 in the databases Psycinfo, Scielo, Pubmed and Web of Science. The following keywords were used in English: "cyber stalking", "cyber harassment", "online violence", “cyber dating violence”, "cyber dating abuse", "cyber victimization", "cyber aggression", "electronic dating violence", "sextortion" and "revenge porn", and in Spanish: “ciberviolencia”, “ciberacoso”, “violencia online”, “violencia internet”, “cibervictimización”, “ciberagresión”, “sextorsión” and “pornografía de venganza”. The search was not limited to a temporary period, nor was the country of origin of the study limited.
Refining and systematization of information
The selection of articles was made according to the following inclusion criteria: (a) the research must provide findings in relation to CDV regardless of the methodology used; (b) the participants may be minors and/or adults (early and intermediate adulthood); (c) the work must include the CDV instrument used or provide information to access it; (d) the work must be published in English or Spanish. Theoretical articles were excluded. Figure 1 shows the diagram of search and selection of the sources analyzed, finally obtaining a total of 59 scientific articles on CDV.
Results
Psychometric properties of measuring instruments
From the sample of 59 valued items, thirty CDV measuring instruments were identified. Table 1 presents the main characteristics of these instruments and the corresponding research. Only the first published study that uses the instrument or the one that provides the analysis of its psychometric properties is included.
The range of years in which this research has been published is from 2007 to 2018. All instruments examined were self-reports, with 50% using an online format as opposed to the traditional paper format. The items were predominantly scalar response (n = 23; 76.6%), in a range of 4 to 7 options depending on the frequency or timing of the CDV indicators. Twenty different terms were identified to designate the construct to be measured for each instrument, most of them Anglo-Saxon terminology. In 13.3% (n = 4) of the cases no reference was made to the psychometric properties of the instrument. On the other hand, in 50% (n = 15) of the cases data were provided on evidence of reliability, in 33.3% (n = 10) on reliability and validity, and only in one case (3.3%) on validity exclusively.
As for reliability evaluations, the study of internal consistency using Cronbach's alpha prevailed, with coefficients higher than .70 in most cases, except in three papers (Sánchez, Muñoz-Fernández, Lucio & Ortega-Ruíz, 2017; Smith-Darden, Kernsmith, Victor & Lathrop, 2017; Temple et al., 2016). In terms of evidence of validity, the internal structure was studied to a greater extent by means of exploratory and/or confirmatory analysis.
In three cases, sexual CDV dimensions were specifically addressed (Dick et al., 2014; Smith-Darden et al., 2017; Zweig, Dank, Yahner & Lachman, 2013), and a specific instrument within this typology was found (Drouin et al., 2015). According to the five Spanish instruments, only the one published by Borrajo and collaborators (2015b) contained a single item of sexual CDV (sending and/or uploading photos and/or videos with intimate or sexual content of a partner) without permission.
Research characteristics and main findings
As Table 2 shows, the instruments come mainly from the United States (n = 22; 73.3%) and to a lesser extent from Spain (n = 5; 16.6%) and from other countries such as the Czech Republic, Australia and Canada (n = 3; 9.9%). The phenomenon was especially studied in university populations (n = 16; 53.3%) and non-probability samples were taken (n = 21; 70%); 43.3% of the studies (n = 17) specified the existence of temporal restrictions in the measure of the CDV, most commonly assessing these behaviors in the last year elapsed (n = 11; 36.6%) and also in the current relationship (n = 22; 73.3%).
Reported prevalence rates of CDV perpetration in minors ranged from 8.8% (Muñiz, 2017; Muñiz & Monreal, 2017) to 38% (Smith-Darden et al., 2017), and in victimization from 22.3% (Temple et al., 2016) to 41% (Dick et al., 2014). In adults, perpetration figures ranged from 16.3% (Ramos et al., 2017) to 71% (Melander, 2010), and victimization from 20% (Drouin et al., 2015; Strawhun et al., 2013) to 75% (Melander, 2010), 76.5% in women and 77.1% in men (Bennett et al., 2011), considering only global indices, although values above these limits were observed in specific items. As for the frequency of CDV, the mean scores found were generally at the lower end of the scale, in adults (Bennett et al., 2011; Reed et al., 2016), and in minors (Sánchez et al., 2017; Temple et al, 2016), although there were results as disparate as those obtained in the two papers by Borrajo and his team, on the one hand, reporting an average victimization of 23 times in the last six months (Borrajo et al., 2015a), and on the other hand, a much lower average in the last year of the relationship, according to the type of CDV (Borrajo et al., 2015b).
Discrepancies were also observed regarding the participation of men and women in CDVs. In adults, some studies did not detect gender differences in the frequency of perpetration (Ramos et al., 2017; Reed et al., 2016; Tokunaga, 2011), in the frequency of victimization (Drouin et al., 2015; Wolford-Clevenger et al., 2016), or in the chronicity of CDVs (Bennett et al., 2011; Borrajo et al., 2015a; Strawhun et al., 2013; Wolford-Clevenger et al., 2016). This was similar in studies with minors (e.g., Muñiz, 2017; Muñiz & Monreal, 2017; Sánchez et al., 2017; Reed et al., 2016; Smith et al., 2018). Conversely, other studies reported a higher degree of perpetration in older men, in prevalence (Martinez-Pecino & Durán, 2016) and frequency (Leisring & Giumetti, 2014), while others indicated that women exercised these aggressions more often (Burke et al., 2011; Smoker & March, 2017; Strawhun et al., 2013) and men suffered them more often (Cutbush et al., 2018; Reed et al., 2016). These discrepancies were also observed according to the type of CDV assessed (Burke et al., 2011; Dick et al., 2014; Leisring & Giumetti, 2014; Melander, 2010; Zweig et al., 2013).
Discussion
The objective of this review study was to analyze the methodology used throughout these years in the evaluation of CDV. In the 59 scientific works analyzed, thirty instruments of CDV measurement were detected.
Of the results obtained, the scarce number of instruments that provide evidence of their psychometric properties stands out, as Borrajo et al., argue (2015b). Some of the published works report on the reliability of the instruments, but there are currently no specific studies on evidence of validity to support their statistical conclusions. These studies would be necessary to ensure that the conclusions obtained are scientifically justified (Prieto & Delgado, 2010). On the other hand, the evidence of validity displayed is mostly from the internal structure, preferably through exploratory factor analyses. In addition to other techniques, such as confirmatory factor analyses, which could reveal specific aspects of measurement models, there are also other sources of evidence of validity (AERA, APA & NCME, 2014) for which scarce information is available.
The instruments of CDV have been used mainly in United States research, which could limit the understanding of this phenomenon in the Spanish population. In addition, it highlights the variety of terms used to refer to the construct, as other works have also pointed out (Gámez-Guadix et al., 2018). The data indicate that there are few international or national studies that include electronic sexual assault or the fraudulent use of this type of content. As previously pointed out (e.g., IEIG, 2017; Rodríguez-Domínguez et al., 2017), other violent practices carried out through technologies would be underestimated, which also has a negative impact on the understanding of this phenomenon.
This systematic review also provides relevant information on the approach to CDV in research. The methodological differences identified (attention to the prevalence or chronicity of the phenomenon, temporality of the aggressions studied, type of relationship analyzed, etc.) make it difficult to compare the studies with each other and are possibly responsible for the plurality of evidence found in this area. It is essential to clarify questions of this type in the studies in order to better interpret the results obtained and better understand the phenomenon.
The findings reviewed do not allow us to conclude with confidence the existence of gender differences in this phenomenon. In addition to the plurality of the methodological approach, another reason could lie in the motives and the type of CDV exercised, as happens in offline violence. In this type of violence, some hypotheses suggest that young women use violence as a form of relief and in a reactive manner, as a response to unacceptable acts by their partners. Young men, on the other hand, use violence as a way of dominating and controlling women (Foshee, Bauman, Linder, Rice & Wilcher, 2007). Similarly, young women express verbal aggressions towards their partners to a greater extent, while young men emphasize physical and sexual aggressions (Harned, 2001; Muñoz-Rivas, Graña, O'Leary & González, 2007). Something similar seems to happen in CDV. An example of this would be the study by Zweig et al., (2013) in which more boys who are sexually abusive and more girls who are victims of general and sexual CDV are identified. Also the recent findings of Smith et al., (2018), which, although similar in terms of prevalence between boys and girls, observed higher levels of psychological distress in female victims, which could be a better indicator of the severity of the abuse, rather than the number of times it occurs.
Among the limitations of this study are those referring to the process of searching for scientific studies and the inclusion criteria that have been considered. Future studies could analyze possible differences according to the degree of commitment in the relationship or the moment in which CDV is exercised.
In conclusion, the evaluation of CDV shows a great methodological diversity, as well as a scarce attention to manifestations of sexual CDV. Although some studies have estimates of the reliability of the measure, a considerable number of them have not been approached from the perspective of validity, which would limit the construction of scientific knowledge in this field. The information obtained in this systematic review is fundamental to understand the possible reasons for the divergence of results about the CDV, since it shows how the methodological approach conditions the interpretation of the evidence found, issues that have to be addressed from the scientific rigor and adequately detailed in the research.