Abstract

Keywords: bullying; cyberbullying; adolescence

Introduction

Despite increased awareness of cyberbullying– or the “willful and repeated harm inflicted through the use of computers, cell phones, and other electronic devices” (Patchin & Hinduja, 2006, p.152) - cases of youth committing suicide as a result of online abuse continue to occur (Dean, 2012; Salazar, 2010). These cases illustrate that bullying behaviors (e.g., harassment, insults, public shaming) are no longer isolated to school, but instead follow young people home and are visible to potentially hundreds of witnesses through the use of social technology. Although some cyberbullied youth are unfazed by such occurrences, the youth who are affected may experience emotions like anger, depression, embarrassment, and sadness, which, in turn, can lead to other negative consequences (Aoyama, Saxon, & Fearon, 2007; Hinduja & Patchin, 2007; Hinduja & Patchin, 2011; Li, 2010; Sabella, Patchin, & Hinduja, 2013). As a result of these consequences, scholars must continue to study various risk factors associated with bullying – both cyberbullying and offline bullying. Accordingly, this study examined various risk factors associated with experiencing cyberbullying that have largely been uninvestigated (e.g., lying about one’s age, sexting, sociability, etc.). Additionally, we explore whether these same risk factors, used to assess risk of experiencing cyberbullying, could also inform parents and caregivers of the risk of experiencing offline bullying. As such, the findings from this study have important implications for both prevention and intervention programs combatting these problems.

Although a substantial amount of research has examined various aspects of cyberbullying (Ackers, 2012; Beran & Li, 2005; Bossler, Holt, & May, 2012; Marcum, Higgins, & Ricketts, 2010; Navarro & Jasinski, 2012, 2013; Patchin & Hinduja, 2006; Sengupta & Chaudhuri, 2011; Staksrud, Ólafsson, & Livingstone, 2013), research investigating risk factors of experiencing cyberbullying is still largely in a nascent stage. Therefore, this study aims to fill in some important gaps in previous research. First, we investigated several risk factors that broadly addressed the question of which online and offline behaviors predicted cybervictimization. While previous studies have greatly increased our collective knowledge of which online behaviors increase the risk of victimization (e.g., blogging, chat room use, etc.; Marcum et al., 2010; Navarro & Jasinski, 2012), the rapid advancement of technology necessitates continued exploration of other factors not as widely investigated (e.g., sexting). Second, this study investigates whether parental supervision successfully mitigated the risk of experiencing victimization. Unfortunately, given the nature of the Internet, monitoring online activities by system administrators is impossible. Thus, it is paramount – especially when considering the victimization of youth – that other forms of guardianship are investigated. Finally, some contention among scholars regarding the degree of overlap between cyberbullying and offline bullying exists. While most studies have found a high degree of overlap (e.g., Hinduja & Patchin, 2008; Kowalski & Limber, 2013; Olweus, 2012; Raskauskas & Stoltz, 2007; Vandebosch & Van Cleemput, 2009) the extent of the crossover varies due to methodological challenges still prevalent in the cyberbullying field (see Patchin & Hinduja, 2012, for more extensive discussion). Accordingly, this study revisits the crossover between the two types of victimization and whether there are similarities and differences in risk factors between the two groups (i.e., victims of cyberbullying and victims of offline bullying).

Literature Review

Bullying among youth, particularly in online fora, continues to inspire international concern - especially given the rapid advancement of technology and the subsequent rise in new types of social media. In fact, findings from a recent study on teens’ social media use indicate that over 80 percent of teens use Facebook and an increasing number of teens use services such as Twitter, SnapChat, and Instagram for interacting with friends (Madden, Lenhart, Duggan, Cortesi, & Gasser, 2013). Moreover, young online users are also sharing more information about themselves through social media compared to years past (Madden et al., 2013). While the utilization of social media can have several benefits for young people (e.g., greater connectivity with friends; increased online sociability; increased access to information) (Cummings, Sproull, & Kiesler, 2002; Neustadl & Robinson, 2002; Van Cleemput, 2010), this reliance on technology can also increase one’s risk of experiencing cyberbullying or lead to other serious problems (e.g., Internet addiction) (Christakis, 2010; Griffiths, 1999; Widyanto & Griffiths, 2006; Young, 1998). Indeed, previous studies document prevalence rates of cyberbullying ranging from less than 10 percent to over 50 percent among youth (Ackers, 2012; Gomez-Garibello, Shariff, McConnell, & Talwar, 2012; Marcum, 2008; Navarro & Jasinski, 2012, 2013; Patchin & Hinduja, 2012; Tomşa, Jenaro, Campbell, & Neacşu, 2013). Although these rates stem from several methodological variations across studies (Patchin & Hinduja, 2012), they underscore the point that the risk of experiencing cyberbullying is a reality for many youth.

As the blending of online and offline social life continues to occur with the rapid advancement of technology (Gergen, 2002; Turkle, 2008), research has called attention to the overlap between traditional (offline) bullying and cyberbullying (Olweus, 2012). However, cyberbullying sets itself apart from traditional bullying in several important ways: (1) the inability of victims to escape cyberbullies, (2) the increased exposure of the cybervictimization to potentially hundreds of bystanders, (3) and the anonymity provided to cyberbullies by the Internet that is not present offline (Slonje & Smith, 2008). Yet, despite these differences, a number of studies suggest a high degree of overlap between cyberbullying and offline bullying likely exists (Hinduja & Patchin, 2008; Kowalski & Limber, 2013; Raskauskas & Stoltz, 2007; Sabella et al., 2013; Vandebosch & Van Cleemput, 2009). In fact, Olweus (2012) found that over 85 percent of youth exposed to cyberbullying had also been exposed to some form of offline bullying. Accordingly, we hypothesize (H1) that children who were bullied offline were more likely to be cyberbullied than children who had not been bullied offline.

Impact of Social Technology Use, Sociability, and Emotions on Cyberbullying

Previous studies of various forms of cyberbullying found that youth react to experiencing victimization in various ways. While some cyberbullied youth report being unfazed by their experiences, others experienced a range of emotions: anger, anxiety, depression, and low self-esteem (Campbell, Spears, Slee, Butler, & Kift, 2012; Li, 2010). Research also indicates that, while both offline and online forms of interpersonal aggression are problematic, cyberbullied youth reported greater social difficulties and psychological ramifications compared to victims of offline bullying (Campbell et al., 2012). Perhaps most alarming, cyberbullying has been associated with suicide ideation and attempts (Bauman, Toomey, & Walker, 2013; Hinduja & Patchin, 2010a). As such, it is not surprising that being cyberbullied is a reason some youth abstain from using SNS entirely (Baker & White, 2011). In fact, findings from several studies suggest SNS have become a venue for cyberbullying and other problematic Internet behaviors (Ackers, 2012; Bossler et al., 2012; Dake, Price, Maziarz, & Ward, 2012; Navarro & Jasinski, 2012; Sengupta & Chaudhuri, 2011; Staksrud et al., 2013). Therefore, we hypothesize (H2) that children who used SNS daily were more likely to be cyberbullied than children who did not use SNS daily.

Previous research also suggests that children with low levels of sociability are at greater risk of victimization than children with high levels of sociability (Boulton & Smith, 1994; Boulton & Underwood, 1992; Boulton, Trueman, Chau, Whitehand, & Amatya, 1999; Gest, Graham‐Bermann, & Hartup, 2001; Hodges & Perry 1996). Boulton and Underwood's (1992) study revealed that bullying victims often did not have close friends and, thus, tended to spend more time alone at school. Yet, children who had close friends tended to exhibit an emotional disposition (e.g., prosocial behaviors, higher self-esteem) and have the social resources (e.g., the ability to call upon friends for help) that helped deter bullies (Gest et al., 2001; Kochenderfer & Ladd, 1997). With this in mind, we hypothesize (H3) that children who spent more time with friends outside of school were less likely to be victims of cyberbullying.

Taking the aforementioned studies into account, we also sought to examine a different aspect of “risk” than what has been previously discussed – that being whether youths’ emotional attachment to interactions within SNS contributed to victimization, which has not been examined extensively to date (for exception, see:(Vandebosch & Van Cleemput, 2009). Thomas, MacInnis, and Park (2005, 78) conceptualized emotional attachment as an “emotion-laden target-specific bond between a person and a specific object.” Specifically, we hypothesize (H4) that children who were emotionally attached to their interactions within their SNS were more likely to be cyberbullied than children who were not emotionally invested in their online interactions within their SNS. We argue that children who are emotionally invested in their online interactions were more likely to identify and recall events as cyberbullying, because these events had a greater impact on their feelings compared to children less invested in online social interactions. Aside from these variables, other less explored risk factors were included in this study (e.g., sexting, lying, and cyberbullying others) to extend the current literature.

Risky Online Behaviors and Risk of Experiencing Cyberbullying

As social life becomes increasingly intertwined with technology, more youth are accessing mobile devices at an early age. Indeed, children as young as 11 years old are accessing the Internet via cell phones, while some are “logging on” as early as eight years old, which has resulted in unintended ramifications (D'Antona, Kevorkian, & Russom, 2010). Sexting – or the transmission of nude or partially nude photos or videos via some technology – has gained increasing recognition as a social problem (D'Antona et al., 2010; Dake et al., 2012; Hinduja & Patchin, 2010b). In four nationally representative studies, eight to 19 percent of youth reported sending “sexts,” and 13 to 31 percent of youth received such images/videos (Hinduja & Patchin, 2010b). The aforementioned figures are especially troubling given that several high profile teen suicides that stemmed from unrelenting cyberbullying ultimately began with sexting (e.g., Jesse Logan, Amanda Todd, Hope Witsell; Dean, 2012; Hinduja & Patchin, 2010b). Unfortunately, although previous research has also shown that cyberbullying by photos/videos (in general) is more hurtful than other forms of aggression (Dake et al., 2012; Hinduja & Patchin, 2010b; Slonje & Smith, 2008; Smith, Mahdavi, Carvalho, Fisher, Russell, & Tippett, 2008), few studies to date have specifically investigated sexting as a risk factor for experiencing cyberbullying. Of the few studies of sexting and cyberbullying, Dake et al. (2012) provide evidence that not only is sexting associated with cyberbullying and indirect bullying, but also sending explicit content results in attempted/completed suicide, physical abuse, sexual victimization, and feelings of sadness and hopelessness. This led us to hypothesize (H5) that children who sent risqué photos or videos were more likely to be cyberbullied than children who abstained from such activities.

Emerging cyberbullying research has also begun to focus on the overlap between victim and offender as a risk factor for experiencing cyberbullying. Many studies provide evidence that youth are more likely to cyberbully others if they have been victims of cyberbullying themselves (Bauman, 2010; Marcum, Higgins, Freiburger, & Ricketts, 2013; Mishna, Khoury-Kassabri, Gadalla, & Daciuk, 2012; Navarro & Jasinski, 2013; Vandebosch & Van Cleemput, 2009). Such studies support a growing body of research suggesting that engaging in reckless behavior, in general, increase children’s chances of becoming victims (Reyns, Burek, Henson, & Fisher, 2013; Schreck, 1999). In conjunction with previous literature, we hypothesize (H6) youth who cyberbullied others were more likely to be bullied than to those who did not cyberbully others. Finally, this study examines a risk factor that has not been considered in any other cyberbullying study (to the best of our knowledge): lying about one’s age online. As with other risk factors, we hypothesize (H7) that children who lied about their age online were more likely to be cyberbullied than children who did not engage in this behavior.

Impact of Parental Guardianship on Cyberbullying

Unfortunately, while the aforementioned research has advanced our understanding of risk factors associated with experiencing cyberbullying, less is known about what factors effectively mitigate risk – particularly concerning parental guardianship. Unlike offline bullying, cyberbullying presents unique challenges to parents – and to users themselves – in guarding against victimization. One of these unique challenges is the ability of cyberbullies to remain anonymous (Navarro & Jasinski, 2012). Moreover, while certain online activities are officially monitored (e.g., chat rooms), typically e-mails and instant messages are not (Patchin & Hinduja, 2006). Although many employ the use of filtering software to protect against various forms of online victimization, various studies have largely found these programs to be ineffective at reducing the chances of experiencing cybervictimization (Bossler & Holt, 2009; Bossler et al., 2012; Marcum et al., 2010; Navarro & Jasinski, 2013).

Due to previous findings regarding the largely ineffective role of filters and software programs to prevent cybervictimization, scholars are beginning to explore other forms of parental guardianship. For example, Lwin, Stanaland, and Miyazaki (2008) found that parental monitoring of children’s Internet behavior minimized the risk of them engaging in risky online activities. However, other studies found that despite establishing parental rules and mandates, caretakers remained largely unaware of their children’s activities (Dehue, Bolman, & Völlink, 2008). Moreover, other studies found that, as parents employed increased guardianship over youth, the risk of cyberbullying actually increased (Marcum, 2008). Although previous studies were limited in their ability to establish directionality in these relationships (e.g., whether increased guardianship occurred before cybervictimization or as a result of it), further exploration of this relationship is necessary. Accordingly, we hypothesize (H8) that the children of parents who talked to them about online safety were less likely to experience cyberbullying than children whose parents did not talk to them about appropriate online behaviors.

Methods

Data

The data for this study came from the 2011 Teens and Digital Citizenship Survey conducted by the Pew Research Center’s Internet and American Life Project (Pew Research Center, 2011). The survey consisted of telephone interviews with a sample of parents and their children – ages 12 to 17 – in the U.S. (n = 799); however, to ensure that these data were nationally representative of the U.S. population, weights were included in the original dataset to correct for demographic discrepancies, disproportionate sample designs, and patterns of nonresponse (n = 4257).¹ The current study focused primarily on the children’s responses to the surveys, in which they were asked about their online behaviors as well as a variety of experiences both on and offline. The margin of sampling error for the survey was about +/-5 percentage points, which provides a reliable sample of the families with children between the ages of 12 and 17 in the U.S. population.

Measures

The dependent variables utilized in this study were “bullied online” and “bullied offline,” which were based upon the following survey questions (both coded “No” = 0/“Yes” = 1): “In the past 12 months, have you been bullied online?” and "In the past 12 months, have you been bullied in person?”

The independent variables analyzed were associated with the use of SNS and Internet via cell phones, sociability, emotional attachment to interactions within SNS, parental involvement in their children’s online behavior, and risky online behaviors (See Table #1 for descriptive statistics). Specifically, the use of SNS was operationalized as the daily use of these sites, which was created by conducting a recode of the following survey question in which the values were reversed: “About how often do you visit SNS.” The response options were (1) “Several times a day,” (2) “About once a week,” (3) “3 to 5 times a week,” (4) “1 to 2 days a week,” (5) “Every few weeks,” and (6) “Less often.” However, for “daily social networking site (SNS) use,” options (1) and (2) were recoded as (1) “Daily,” and options (3) through (6) were recoded as (0) “Not daily.”

In relatively few studies have researchers taken issue with the notion of “being online.” Conventionally, “being online” commonly refers to using the Internet, which, if only implicitly, assumes sitting in front of a computer. Yet, being “on the cell” is becoming indistinguishable from “being online” (Turkle, 2008). SNS undoubtedly constitutes online behavior; however, children can be connected to the Internet, generally, and SNS, specifically, through their cell phones. It is necessary, then, to move beyond the binary notion of children either being online or being offline. To account for this, Internet use via cell phone was also measured. Specifically, the use of Internet via cell phones was operationalized as responses to the question, “In the last 30 days, have you used the Internet of a cell phone,” which was originally coded as (1) “Yes,” (2) “No,” and (3) “I do not have this device/ it does not apply to me.” This variable was recoded to (0) “No” and (1) “Yes.”

Conceptually, sociability is simply the extent to which a person is social, i.e., how often they spend time with friends. It was operationalized as the response to the following survey question: “About how often do you spend time with people in person, doing social activities outside of school?” The response options ranged from (0) “Never” to (4) “Everyday.” However, the variable was recoded as (0) “Never,” (1) “Once to a few times per week,” and (2) “Daily.”

Emotional attachment to interactions on SNS is operationalized as the extent to which asynchronous interactions via SNS (e.g., via wall posts, comments, likes, tags, tweets) have an emotional impact on a person. Due to secondary data constraints, the measure used in this study did not focus on the bond between the person and the SNS, as objects, but rather attachment to the interactions SNS provide. Because the survey did not include a question asking children specifically about their emotional reactions to such asynchronous interaction on SNS, the following tacit measure of an “emotion-laden” response to social interactions on SNS was utilized instead (coded “No” = 0/“Yes” = 1): “Have you, personally, ever had an experience on a social networking site that made you feel good about yourself?”

Parental intervention in online behavior was measured by the following survey question (coded “No” = 0/“Yes” = 1): “Have your parents ever talked with you about what kinds of things should and should not be shared online or on a cell phone?”

Finally, engagement in risky online behaviors was operationalized into three separate variables: (1) lying about one’s age online, (2) sending risqué photos or videos, and (3) online harassment of others within SNS. First, children were asked, “Have you ever said you were older than you are so you could get onto a web site or sign up for an online account, such as for email or a social networking site?” (coded “No” = 0/“Yes” = 1). Second, children were asked the following: “Have you ever sent a sexually suggestive nude or nearly nude photo or video of yourself to someone else?” (coded “No” = 0/“Yes” = 1). Third, following a series of questions about how the respondent (i.e., the child) reacts upon observing other people being mean or cruel to someone on an SNS, the following questions were asked: “And how about you? How often have you joined in the harassment?” The response options were (1) “Frequently,” (2) “Sometimes,” (3) “Once in a while,” and (4) “Never;” however, since the focus is on simply whether children have joined in harassment or not, (1) “Frequently,” (2) “Sometimes,” and (3) “Once in a while” were recoded as (1)“Yes,” and (4) “Never” was recoded as (0) “No.”

Gender, race, and age were control variables in the analyses. Gender was coded as (0) “Male” and (1) “Female.” Due to the lack of a race variable specific to the respondent, we utilized parent’s race as a proxy variable for the child: (0) “White” and (1) “Non-White.” Age was a continuous variable ranging from 12 to 17 years old.

Table 1. Descriptive Statistics of Variables, for Unweighted and Weighted Data.

Analytical Strategy

Using SPSS 22, a 2x2 cross-tabulation of the weighted data that included a chi-square test and risk estimate and three multivariate logistic regression analyses were conducted to examine the impact of the independent variables on the likelihood of being cyberbullied as well as being bullied offline.² In addition to reporting comparative odds (“odds ratio”), probabilities (“prob.”) of being bullied – both online and offline – under certain conditions were calculated in some cases. In such cases, the following equation was used to extrapolate probabilities from the log-odds provided in the SPSS output:

The goals of reporting probabilities were simply to clarify the nature of and further develop the findings from the odd ratios.

Results

In this study, approximately 8 percent of children were victims of cyberbullying, while over 12 percent of children were victims of traditional (offline) bullying. These figures support previous research indicating the average across-time prevalence for experiencing cyberbullying is about five percent, whereas the average across-time prevalence for being bullied offline is nearly 18 percent (Olweus, 2012). These percentages reveal that bullying is not necessarily shifting from traditional offline bullying to cyberbullying or becoming an epidemic (Sabella et al., 2013).

A simple 2x2 cross tabulation using weighted data revealed that nearly 55 percent (54.5%) of children who had been cyberbullied had also been bullied offline compared to children who had solely experienced cyberbullying (45.5%); in fact, only about nine percent of children who had not been cyberbullied (8.8%) had been bullied offline (χ² = 585.53; p <.001). A post-hoc risk estimate indicated that children who were bullied offline were over 12 times more likely to be cyberbullied than children who were not bullied offline (odds ratio = 12.38; p <.001), which addresses our first hypothesis (H1). Our first multivariate logistic regression (see Table #2) in which we also controlled for gender, race, and age further underscores the strength of this relationship (Nagelkerke R² = .29; Model χ² = 516.89; p <.001). When controlling for these demographic factors, children who were bullied offline were 14 times more likely to be cyberbullied than children who were not bullied offline (odds ratio = 14.03; p <.001). Although causal directionality cannot be deduced from these analyses, these statistics do confirm our first hypothesis (H1) that children who have been cyberbullied are more likely to be bullied offline than children who have not been cyberbullied, and, thus, suggest a fluid connection between online and offline bullying.

Table 2. Logistic Regression Analysis of the Likelihood of Being Bullied Online
(weighted n = 4257).

The second and third multivariate logistic regression analyses, in Table #3, model youths’ online and offline behaviors to predict the likelihood of being both cyberbullied (Model 1 χ² = 286.60, p <.001) and bullied offline (Model 2 χ² = 235.10, p <.001). These analyses suggest that 32 percent of the variance in cyberbullying and 31 percent of the variance in offline bullying is explained by the independent variables (Model 1 Nagelkerke R² = .32; Model 2 Nagelkerke R² = .31).

In Model 1, we predicted the likelihood of experiencing cyberbullying in terms of children’s daily SNS use, Internet use via cell phones, emotional attachment to interactions within SNS, parental intervention, and risky online behaviors while controlling for gender, race, and age. First, general behavioral factors – i.e., daily use of SNS, use of Internet via cell phones, and sociability – had a statistically significant impact on the likelihood of being cyberbullied. Consistent with hypothesis two (H2), children who used SNS daily as well as children who used the Internet on cell phones were more likely to be cyberbullied than children who did not use SNS daily or use the Internet on cell phones (odds ratio = 3.48; p <.001; odds ratio = 2.45; p < .001, respectively). In fact, 17 out of every 50 children who use SNS daily and who use the Internet on their cell phones are likely to be cyberbullied (prob. = .34). Furthermore, consistent with hypothesis three (H3), the more often children spent time with friends in person outside of school, the less likely they were to be cyberbullied. That is, children who spent time daily with friends outside of school were less likely to be cyberbullied than children who did not spend any time with friends outside of school (odds ratio = .32; p <.001). Therefore, only around one out of every 25 children who spend time daily with friends outside of school are likely to be cyberbullied (prob. = 04), whereas three out of every 25 children who do not spend time in person with friends outside of school are cyberbullied (prob. = .06). However, seven out of every 50 children who spend time with friends in person daily outside of school, use SNS daily, and use the Internet on their cell phone are likely to be cyberbullied (prob. = .14).

Similarly, children who were not emotionally attached to interactions within SNS were less likely to be cyberbullied (H4). In other words, children who expressed an emotional attachment to interactions within SNS were three times more likely to experience cyberbullying than children who did not express emotional attachment to interactions within their SNS (odds ratio = 3.06; p < .001). In fact, the probability of experiencing cyberbullying among children who use SNS daily, use the Internet on their cell phones, and are emotionally attached to interactions within SNS is quite high (prob. = .61).

Generally, engaging in risky online behaviors does increase the likelihood of being cyberbullied (H5, H7).Children who sent risqué photos or videos to others were five times more likely to be cyberbullied than children who had not sent risqué photos or videos to others (odds ratio = 5.88; p < .001; H5). Additionally, children who lied about their age online were more likely to be cyberbullied than children who did not lie about their age online (odds ratio = 1.79; p < .01; H7). However, contrary to previous research, children who harassed others online were not more likely to be cyberbullied than children who had not harassed others online, which is counter to our sixth hypothesis (H6).

Parental involvement, as predicted, acts as a protective factor to minimize risk of victimization (H8). That is, children whose parents talked with them about appropriate information to post online or send via cell phones were less likely to be cyberbullied than children whose parents did not talk with them about appropriate information to post online or send via cell phones (odds ratio = .38; p <.001). Parental involvement dropped the probability of being cyberbullied among children who use SNS daily and Internet on their cell phones by nearly half. Recall that the probability of children who use SNS daily and use the Internet on their cell phones experiencing cyberbullying was .34. However, the probability of experiencing cyberbullying is lower for children whose parents have talked with them about appropriate online behavior (prob. = .16).

Having firmly established how the aforementioned factors impact the likelihood of being cyberbullied, Model 2 incorporates the same independent variables previously discussed in order to explore the impact of online behaviors on the likelihood of being bullied offline (see Table #3).

In addition to confirming previous research suggesting that children who spend more time with friends are less likely to be bullied, the findings provide evidence that online behaviors impact the likelihood of being bullied offline. That is, consistent with previous research (e.g., Boulton & Smith, 1994; Boulton & Underwood, 1992; Boulton et al., 1999; Gest et al., 2001), children who spent time daily with friends outside of school were less likely to be bullied offline than children who did not spend time with friends outside of school (odds ratio = .35; p < .01). Additionally, the findings provide evidence that online behaviors impact the likelihood of being bullied offline. First, children who used SNS daily as well as children who used the Internet on their cell phones were more likely to be bullied offline compared to children who did not do these things (odds ratio = 2.97, p < .001; odds ratio = 2.10, p <.01). Second, children’s attachment to interactions in SNS impacted the likelihood of being bullied offline; that is, children who were emotionally attached to interactions in SNS were more likely to report being bullied offline than children who were not emotionally attached to interactions in SNS (odds ratio = 2.70; p <.001). Third, parental involvement in children’s online activities also mitigated the risk of offline bullying, as children whose parents talked with them about the information they post online or send via cell phones were less likely of being bullied offline than children whose parents did not talk with them about these things (odds ratio = .55; p <.05).

The impact of certain risky online behaviors on the likelihood of being bullied offline is also quite interesting. First, although sending risqué photos was not statistically significant, children who lied about their ages online were more likely to be bullied offline compared to others who did not lie about their age (odds ratio = 1.80; p <.01). At the very least, the aforementioned finding requires additional exploration. Furthermore, children who joined in the harassment of others online were more likely to be bullied offline than children who did not join in the harassment others online (odds ratio = 1.83, p <.05), which is consistent with previous research on the overlap between victims and offenders (Bauman, 2010; Mishna et al., 2012; Navarro & Jasinski, 2013; Vandebosch & Van Cleemput, 2009).

It should also be noted that gender, race, and age influenced the likelihood of being cyberbullied and being bullied offline in a similar manner. According to Model 1, girls were more likely to be cyberbullied than boys (odds ratio = 2.54; p <.001); white children were more likely to be cyberbullied than non-white children (odds ratio = .22; p <.001); and older children were less likely to be cyberbullied than younger children (odds ratio = .56; p <.001). Similarly, as presented in Model 2, girls were more likely than boys to be bullied offline (odds ratio = 1.94; p <.01); non-whites were less likely to be bullied offline than whites (odds ratio = .26; p <.001); and older children were less likely to be bullied offline than younger children (odds ratio = .68; p <.001).

Table 3. Logistic Regression of the Likelihood of Children (1) Being Bullied Online and (2) Being
Bullied Offline (weighted n = 4257).

Discussion

This study sought to investigate the impact of several relatively unexplored risk factors on the likelihood of experiencing cyberbullying and offline bullying using a nationally representative sample of the U.S. population. Ultimately, these results will add to the growing body of literature on risk factors for experiencing cyberbullying and bullying in general, which can assist various stakeholders (e.g., educators, parents, policy makers) in creating effective prevention and intervention programs to combat these problems. Results from our first logistic regression support previous research calling attention to the overlap between offline bullying and cyberbullying (Hinduja & Patchin, 2008; Kowalski & Limber, 2013; Olweus, 2012; Raskauskas & Stoltz, 2007; Vandebosch & Van Cleemput, 2009). Thus, adults and other guardians should remain cognizant of the fact that offline bullying may very well be continuing in victims’ homes (i.e., cyberbullying). And vice-versa, parents who suspect youth are experiencing cyberbullying may want to stay alert to whether offline bullying is also occurring.

The results of this study suggest that both daily SNS use as well as Internet use via cell phones are particularly problematic for children in terms of risk of experiencing cyberbullying and offline bullying, which is also in line with previous research (Ackers, 2012; Bossler, Holt, & May, 2012; Navarro & Jasinski, 2012; Sengupta & Chaudhuri, 2011; Staksrud et al., 2013). Indeed, children who used SNS daily and children who used the Internet on their cell phones were at greater risk of being cyberbullied as well as being bullied offline than those who did not use SNS daily or use the Internet on their cell phones. However, to be clear, this finding should not be interpreted as support for prohibiting SNS use or a barring from the Internet in general. Indeed, previous research has found that, while youth still engage in face-to-face communication for the majority of their time with friends, electronically mediated communication supplements face-to-face communication in a very meaningful way (Van Cleemput, 2010). To prohibit the utilization of SNS or Internet access, in general, would cut off a significant method of communication among youth (Van Cleemput, 2010) and would be a completely unrealistic approach to combatting the problem (Sabella et al., 2013). This finding, then, highlights the reality that children currently live within a technologically mediated world and that harassment is no longer isolated to the schoolyard or the school day. Accordingly, parents and caregivers must take a balanced approach to combatting the problem rather than simply cutting off access to these forms of technology; they should discuss the risk factors with youth and empower them to engage in self-care while using technology and to intervene when victimization occurs.

Similarly, consistent with previous research suggesting that children who have stronger social relationships are less likely to be bullied (DeSmet, Veldeman, Poels, Bastiaensens, Van Cleemput, Vandebosch, & De Bourdeaudhuij, 2014; Olweus, 1994; Perren & Alsaker, 2006), we found that increased sociability with other children outside of school decreased the likelihood of being bullied both offline and online. This finding is likely related to recent research investigating bystander behavior in cyberbullying incidents, which contends that bystanders are more likely to intervene on a cyberbullying victim’s behalf if the victim is an in-group member (DeSmet et al., 2014). On the one hand, this finding highlights the benefits of having established social relationships in a virtual environment; on the other hand, it also suggests that children without strong social ties might be especially vulnerable to online victimization. This finding, too, should not be interpreted as evidence of a need to bar access, but rather to inform parents and caregivers of the importance of paying particularly close attention to the online experiences of children who lack strong social ties, as they are particularly vulnerable to online victimization.

The findings here also provide tacit evidence that children who have an emotional attachment to their social experiences on SNS are more likely to report being cyberbullied as well as being bullied offline. Substantively, these findings imply that children might not compartmentalize online social life and offline social life; rather, they might experience both as normal social life. Accordingly, parents and caregivers should recognize this potential parallel in children’s lives and, thus, treat negative experiences youth have online with the same level of importance as offline experiences, as online experiences are likely to be just as emotionally significant as events experienced offline.

General Internet use alone does not sufficiently explain bullying; rather, the ways in which children use the Internet clarifies the impact of the Internet on bullying. Our analyses suggest that children who engage in risky online activities were at greater risk of being bullied both online and offline than children who did not engage in such activities. First, lying about one’s age to gain access to age-restricted websites increased the odds of experiencing both types of bullying. Given that the aforementioned has largely not been investigated to date, scholars in the field should continue to explore the role of lying, or misrepresenting oneself, as a risk factor for experiencing victimization. Secondly, online harassers were not more likely to be cyberbullied than children who had not harassed others online; however, online harassers were indeed more likely to be bullied offline, which does align with previous literature on the victim-offender overlap (e.g., Bauman, 2010; Mishna et al., 2012; Navarro & Jasinski, 2013; Vandebosch & Van Cleemput, 2009). However, the lack of a relationship between online harassment and experiencing cyberbullying is inconsistent with previous research. One possible explanation of this inconsistency is that there is a qualitative difference between children who join in the harassment of others versus those who initiate the harassment. The aforementioned supports recent research that has identified that not all youth who engage in cyberbullying are motivated in the same way (Sabella et al., 2013). For example, many cyberbullies engage in abuse due to their own anger or emotional instability, while others engage in such activities as retaliation against someone who has hurt them, and yet other cyberbullies engage in the abuse of others as a method of entertainment or tomfoolery (Sabella et al., 2013).

Finally, sexting increased the likelihood of being cyberbullied but did not have a statistically significant impact on the likelihood be being bullied offline. This relationship underscores the need for continued research on the relationship between sexting and experiencing cyberbullying as relatively few studies have explored this factor to date (Dake et al., 2012); however, the lack of an impact on offline bullying is a curious contrast as recent suicides have highlighted the role of risqué photos/videos in the offline harassment of victims (e.g., Jesse Logan, Amanda Todd, Hope Witsell; Dean, 2012; Hinduja & Patchin, 2010b; Navarro & Jasinski, 2012; Salazar, 2010). Such current research and mass media reports provide sufficient precedent for parents and caregivers to strongly discourage such behaviors; however, future research is necessary before firm conclusions can be drawn.

Not only would reducing children’s risky online behaviors decrease the likelihood of being both cyberbullied and being bullied offline, but also our findings suggest that parental involvement in children’s online behaviors decreases the likelihood of such bullying. In short, the role of parents and guardians in teaching children appropriate online behaviors as a tactic to guard against potential victimization cannot be overstated. Although previous research suggests that relatively few forms of effective guardianship have been found to mitigate the risk of experiencing victimization (Bossler & Holt, 2009; Bossler et al., 2012; Marcum et al., 2010; Navarro & Jasinski, 2013), our findings clearly indicate that parental involvement in children’s online lives, generally, can decrease the likelihood of being bullied both online and offline.

Limitations

Although the dataset used in this study is statistically representative of the U.S. population, it has a number of substantive and operative limitations. First, the majority of the survey questions had only dichotomous responses, which limited our ability to investigate any context surrounding certain statements. For example, while we accounted for whether or not children had been bullied offline and cyberbullied, we had no ability to account for how often children were victimized in these ways or how they felt following the incident. Relatedly, due to our utilization of secondary data, we were limited to the variables we had in the original data file. As a result, our conceptualizations of certain variables were limited. For example, the measure of emotional attachment to interactions on SNS lacks specificity, as it does not account for the substantive dynamics – or consequences – of social interactions on SNS. That is, it does not address how often experiences on SNS make children feel good about themselves or the nature of the experiences on SNS that make them feel good about themselves. Moreover, it does not account for the degree to which children are emotionally attached to these interactions or for the types of interactions that produce emotional reactions among children. Therefore, this is a measure that we believe deserves particular attention among researchers, as children certainly have varying degrees of emotional attachment to such interactions.

Secondly, the dependent variables used in this study, being bullied offline and being cyberbullied, are missing important context information regarding how the victimization occurred or what it included. Moreover, the participants were not provided with a conceptual definition of “bullying” prior to answering the questions. Consequently, these measures only account for whether children felt like they were being bullied. Previous research suggests that people conceptualize bullying in different ways – even among researchers active in the field (Patchin & Hinduja, 2012; Sabella et al., 2013). Although the cyberbullying measures lacked a definition of bullying commonly used in cyberbullying research, our employed method can provide first indications that risk factors for experiencing cyberbullying are also relevant to consider in offline bullying, which can be further investigated through replication in future research utilizing more robust indicators of cyberbvictimization replicated in future research with more robust indicators of cybervictimization.

Finally, given the cross-sectional nature of these data, it is important to remind the reader that these risk factors do not cause cyberbullying or offline bullying. Rather, our results point out important relationships that might exist upon replication of this study through the efforts of scholars in the field. Moreover, given the statistical challenges that arose when trying to separate “pure offline victims” from “pure online victims,” this study should be replicated by specifically investigating these risk factors across those three groups: pure cyberbullying victims (who were not bullied offline), pure offline victims (who were not cyberbullied), and victims of both bullying types. Unfortunately, our data did not allow for this examination, as the size of each group was too small to make reliable predictions through logistic regression analyses.

Conclusion

The social landscape of childhood has dramatically changed over the past 30 years with the introduction and mobilization of the Internet. Accordingly, the normal trials and tribulations as well as the particularly disturbing experiences of childhood have taken qualitatively different form, as they have become digitally-physically hybridized. Online social life for children is only qualitatively different than offline social life. Yet, our findings suggest that children do not experience offline social life and online social life separately. Rather, offline and online social life blend together to constitute “normal” social life for children growing up “connected.” This study suggests that online bullying and offline bullying, too, are inherently inter-connected. Not only do children’s offline sociability and general use of social technology impact their likelihood of being bullied both online and offline, but also their behaviors online dramatically impact the likelihood of being bullied virtually and physically. Moreover, consistent with most research of the childhood experience, parental involvement is crucial. Parents must be diligent in teaching their children about appropriate online behaviors. Although our research only scratched the surface of the complex nature of bullying in a postmodern social world, it does, at least, present a strong and loud call for future research into the qualitative intersections of online and offline social life on cyberbullying and offline bullying.

Notes

1. For details regarding the calculation of the weights, please see Pew Research Center, 2011. Although the analyses in this study utilize the weighted sample (n = 4257) in order to present findings that are nationally representative of the U.S. population, analyses without weights are available upon request to the corresponding author.

2. All three multivariate logistic regression analyses passed tests for multi-collinearity and goodness of fit.

References

Acknowledgement