Abstract
This study considered whether experiencing cybervictimization is associated with increased recognition of cybervictimization intervention opportunities (i.e., witnessing others’ cybervictimization), as well as greater engagement in self-protective (e.g., changing usernames and privacy settings) and other-protective cybervictimization bystander response behaviors. We collected cross-sectional self-report data from an age-diverse (M = 46.29 years, SD = 19.14, range = 15–93) national sample (n = 3002). We hypothesized that: (1) personal experiences with cybervictimization would be associated with increased reports of witnessing opportunities to intervene when others are cybervictimized, greater self-reported use of active bystander behaviors in witnessed situations, and greater use of self-protective strategies; (2) We also expected that engagement in self-protective behaviors would be positively associated with engagement in other-protective bystander behaviors in response to witnessed cybervictimization. To test our hypotheses, we estimated a structural equation model wherein four latent variables were constructed: cybervictimization experienced, witnessed opportunities to intervene, engagement in self-protective behaviors, and engagement in other-protective cybervictimization bystander behaviors. As hypothesized, cybervictimization was associated with witnessing more opportunities to intervene in other’s cybervictimization, greater self-reported use of active cyber bystander behaviors, and greater engagement in self-protective strategies. However, the strength of two associations was moderated by age, with stronger relationships between cybervictimization and witnessing opportunities to intervene as well as engaging in bystander behavior for older as compared to younger participants. Contrary to hypothesis, there were no significant associations between use of self and other protective behaviors. Furthermore, greater witnessing of cybervictimization was associated with less engagement in bystander behavior in the final model. The implications for existing bystander intervention programs are described. Longitudinal studies of these associations in multiple age groups and among different cultural groups remain necessary.
The Centers for Disease Control and Prevention identifies cybervictimization (e.g., cyberbullying and cyberharassment) as a public health issue among youth (Brochado et al., 2017). However, experiences of cybervictimization occur to people of all ages (Kim et al., 2017; Völlink et al., 2013), although rates may decrease as individuals grow older (Wang et al., 2019). Given the near-ubiquitous use of the internet (Pew Research Center, 2021) and the detrimental effects of cyberbullying victimization for adolescents and children (Anderson, 2018), understanding cybervictimization among diverse-aged individuals is critical.
It is also essential to determine the degree to which witnessing more cybervictimization among others (i.e., they witness potential intervention opportunities) and engaging in more other-directed (bystander) behaviors as well as self-protective behaviors is associated with one’s own cybervictimization, as these three potential behaviors are key to effective prevention and early intervention strategies (Staub & Vollhardt, 2008; Van Cleemput et al., 2014; Woods et al., 2020). Experiencing cybervictimization may function to raise awareness of the need to intervene and advocate for oneself and others, in essence, providing a pathway to changed positionality and even, in some cases, facilitating post-traumatic growth after victimization (Westphal & Bonanno, 2007). Supporting this, a direct relationship between self-reported cybervictimization and greater willingness to intervene in witnessed situations has been reported for U.S. and Chinese male and female early adolescents (Batanova et al., 2014; Cui et al., 2022). To our knowledge, however, these associations have not yet been studied among older adolescents and adults in the U.S. Furthermore, to our knowledge, only one nationally representative dataset has the ability to test these relationships across individuals of diverse ages (adolescent to geriatric). This dataset forms the foundation of the current study.
Cyberbullying has been defined as use “of electronic media with the intention of causing harm, humiliation, suffering, fear, and despair for the individual who is the target of aggression” (Bottino et al., 2015). Operationally, cyberbullying occurs within a relationship with a power differential and occurs repetitively across time. Behaviorally, cyberbullying often manifests as name-calling, posting of harmful photos or personal content without consent, making inflammatory and derogatory comments about the person in public posts, and/or repeated unsolicited and unwanted messaging of negative comments (Gahagan et al., 2016). However, these same behaviors can occur in a variety of relationship contexts and can also be a one-time occurrence; thus, in the current study we use the term cybervictimization. While cybervictimization shares features with other types of victimization experiences, it also has unique characteristics. For example, cyberspace tends to afford perpetrators greater anonymity; correspondingly, it generates less social and legal accountability. Actions in cyberspace can be repetitively viewed, potentially by a wider audience than in-person events and with greater endurance. Thus, a one-time post may have a repetitive victimization impact as it is viewed over time. Cybervictimization responses are often hidden from the perpetrator, and they can occur asynchronously, making it more difficult for the perpetrator to gauge their impact. Yet, efforts to prevent and respond to cybervictimization are essential, as these experiences are associated with numerous negative mental and physical health consequences, including increased depression, substance use, suicide ideation and attempts, and a reduced overall sense of well-being (Bottino et al., 2015; Stevens et al., 2021; Tokunaga, 2010; Völlink et al., 2013).
To prevent deleterious outcomes among cybervictims, increases in the following three behaviors have been the focus of prevention and early intervention efforts, as they have already been shown to be central to prevention and early intervention efforts for numerous other types of aggressive behaviors. They are: (1) increased recognition of opportunities to intervene, (2) greater engagement in active bystander behaviors, and (3) greater use of self-protective strategies in response to one’s own victimization (Veletsianos et al., 2018). However, the associations among these three behaviors have received relatively little attention, especially in adult, rather than child-focused samples, and in cybervictimization, rather than in other types of aggressive, in-person situations. There is also growing recognition that roles within these situations can transform across time, with prevention specialists highlighting the need to focus on the association between one’s own cybervictimization experiences and subsequent engagement in bystander intervention strategies for others (Cui et al., 2022).
Theoretically, it is possible that one’s own traumatic experiences may sensitize one to the problematic experiences faced by others. First, as elucidated in the Turner et al. (1987) Social-Categorization Theory, individuals are more inclined to help others who are perceived as similar to them or who are in their in-group. As such, victims of past crimes may view themselves as belonging to a “victimized” group, making them more inclined to recognize cybervictimization and more likely to intervene with other victims if they are seen as part of their same group (Stevens et al., 2021). It could also be that previous victims recognize others in similar abusive situations because of the shared location in which the victimization is occurring (e.g., by visiting an area or website that has consistent crime or by belonging to a marginalized group whose members are at increased risk for victimization) (Savard, 2018; Woods et al., 2020). Finally, it could be that the experience of victimization serves to change the individual in ways that may make them more inclined to help others, as evidenced by post-traumatic growth theory (Collier, 2016; Jenaro et al., 2018). This later pathway implies that people’s experience of trauma may potentially facilitate greater awareness of and empathy toward others who are facing similar events while also increasing the survivor’s likelihood of intervening in ways that may be other, as well as self, protective as they potentially seek to derive meaning from their own harmful experiences.
Consistent with these potential pathways, previous research has demonstrated a relationship between victimization and engagement in helpful bystander behaviors and use of self-protective strategies across a variety of violence domains (Clay-Warner, 2003; Woods et al., 2016). However, these relationships have only begun to be considered among those experiencing cybervictimization. Thus, the purpose of this paper is to examine the relationship among self-reported experiences of cybervictimization and three prevention/early-intervention behaviors related to cybervictimization: recognition of opportunities to intervene, active bystander behaviors, and self-protective behaviors. Furthermore, much of the cybervictimization research has focused on youth. Here, we utilize cross-sectional self-report data collected from an age-diverse national U.S. sample to consider these relationships and the degree to which they might be moderated by age.
Opportunities to Intervene and Bystander Behavior to Cybervictimization
Given the high reported rates of cybervictimization (Anderson, 2018), it is likely that many people witness this behavior and could intervene to prevent, address, or stop additional cybervictimization from occurring (Holfeld & Mishna, 2018). Accordingly, improving bystander interventions has focused largely on two related but distinct constructs—bystander opportunity and bystander behavior (McMahon et al., 2015). Opportunity reflects whether an individual is exposed to a concerning situation in which risk of violence is present, whereas behavior refers to the action an individual takes when faced with that opportunity (McMahon et al., 2015). Having the opportunity to intervene is a prerequisite for helpful bystander behaviors to be used; however, there are many circumstances in which a person who has an opportunity to intervene does not act. For example, they may not feel comfortable or empowered; they may judge that it is unsafe to intervene; they may lack intervention skills; they may be from a similarly vulnerable group or culture and worry that intervening will make them the focus of future abuse or will reduce their belongingness; or they may judge the severity of the witnessed event as too low to require action (Anderson et al., 2017; Katz et al., 2018). In spite of these realities, some recent research supports our hypothesis that previous cybervictimization will be associated with an increased likelihood that an individual will report engaging in proactive bystander behavior in cyberbullying situations (Panumaporn et al., 2020; Wang & Kim, 2021). Examples of active bystander intervention behavior for cybervictimization include sending messages of advice and support to the cybervictim, directly confronting the cyberperpetrator, and/or asking others to help (Panumaporn et al., 2020). Consistent with theory, Wang and Kim (2021) in their study of 2,888 adult bystanders found that previous cybervictims were three times more likely to be an active bystander in cyberbullying situations as compared to non-cybervictims.
Wang and Kim’s (2021) work exploring the relationship between cybervictimization, and active bystander intervention behavior provides a critical first step. However, this work has two important limitations. First, the Wang and Kim (2021) study used a single-item assessment of bystander behavior. Secondly, the study did not assess the role of self-protective behaviors when considering the relationship between cybervictimization and active bystander behaviors. Self-protective strategies can be actively deployed in response to online harassment and victimization (Veletsianos et al., 2018). They serve to protect cybervictims from further online harassment, and their active nature tends to enhance victim’s sense of agency and promote their overall well-being post-use (Wozencroft et al., 2015). Active self-protective strategies include blocking harmful messages, telling the person to stop harassing behavior, changing one’s username, and telling friends or parents about their cybervictimization (Aricak et al., 2008). The most popular active self-protective strategy appears to be blocking the sender/harasser (Wozencroft et al., 2015), whereas seeking help is the least common (Aricak et al., 2008; Wozencroft et al., 2015).
Additionally, previous research on self-protective behaviors in the face of cybervictimization has focused on adolescents and bullying; we extend this work to an age-diverse adult national sample experiencing cybervictimization. Theoretically, we expect an association between self-protective behaviors and active bystander intervention behavior, under the assumption that those who are more likely to engage in self-protective behaviors may also have a higher perceived self-efficacy for addressing threatening cybervictimization situations happening to others, but there has been very little empirical work testing these hypothesized connections (Bannon & Foubert, 2017; Banyard, 2011).
Finally, although the literature is nascent, there is some suggestion that age may be an important factor. Almenayes (2017) found that the effect of cyberbullying on depression is amplified in older versus younger college students. Relatedly, Barlett and Coyne (2014) report that age-moderated sex differences in rates of cyberbullying perpetration among adolescents.
Overall Purpose and Hypotheses
Therefore, as shown in the conceptual model depicted in Figure 1, we hypothesize that exposure to cybervictimization will be associated with witnessing more opportunities to intervene, greater use of active bystander intervention behaviors, and use of more self-protective strategies in cybervictimization situations. We further predict that reported use of self-protective behaviors in response to one’s own cybervictimization will be positively correlated with the use of active bystander intervention behaviors when witnessing the cybervictimization of others. Finally, given the unique nature of this sample and previous literature, we consider the degree to which participants’ age functions as a moderator in the relationships between cybervictimization and witnessing cybervictimization, engaging in bystander activity, and use of self-protective strategies. However, given that this sample contains a much broader age range of participants than in other related work, moderation analyses were considered exploratory in nature.

Conceptual model of cybervictimization experiences predicting self-protective, witnessed opportunities, and active bystander behavior.
Method
Participants
Participants were between 15 and 93 years of age (M = 46.29 years, SD = 19.14) and were evenly divided among those who were interviewer-identified as female versus male (50.4% female; see Table 1). The majority of the sample self-identified as White (76.1%), and most self-endorsed a heterosexual identity (87.9%).
Demographic Characteristics of the U.S. National Telephone Sample (n = 3,002).
Design
The current study represents a secondary data analysis of the Measuring Cyberabuse Survey, which was conducted from May 17 to July 31, 2016 (see description of full survey and procedures in Lenhart et al., 2016); the study was funded by the Digital Trust Foundation. The protocol was reviewed and approved by the Chesapeake Institutional Review Board (IRB); ethical procedures were followed throughout. The protocol included a waiver of parental permission for youth 15 to 17 years old. It constitutes one of the only known datasets to include a broad range of participants that represent a national adult sample as well as multiple items assessing cybervictimization.
In total, 3,002 internet users aged 15 or older living in the United States participated in this telephone study. Interviews were conducted via landline (NLL = 1,051) and cell phone (NC = 1,951). The survey was conducted by Princeton Survey Research Associates International and interviews were administered in English and Spanish.
Procedure
For the landline sample, interviewers began the encounter by asking to speak with the youngest adult aged 15 or older currently at home. They asked to speak with the youngest male or female based on a random rotation. If the person of the specific sex requested was not available, interviewers asked to speak with the next youngest person, aged 15 or older of the other sex. For the cellular sample, interviews were conducted with the person who answered the cell phone. In all cases, interviewers verified that the person was aged 15 years or older and in a safe place before obtaining informed, verbal consent or assent, and administering the survey. Once a potential respondent was on the phone, interviewers determined that that person used the internet and was thus eligible for the entire survey. A total of 3,834 contacts were made, which resulted in a total sample of 3,002 internet users.
Response Rate
Using the American Association for Public Opinion Research RR3 equation for response rates (American Association for Public Opinion Research, 2011), the response rates for the landline and cellular samples were 8% and 7%, respectively.
Measures
For all measures, response options included yes, no, don’t know, and refused to answer for each. For the current study, don’t know and refused to answer were both re-coded as missing in order to get the most precise estimate of the known reported prevalence of the behaviors (yes, this did occur). Thus, n’s vary slightly across analyses, and reported rates may be underestimated.
Cybervictimization
Participants were asked to indicate whether they had personally experienced eight different types of online cybervictimization behaviors, at least once, by anyone, including a romantic partner, a friend, or even someone they did not know (n = 8, see Table 2).
Rates of Cybervictimization, Witnessing, Bystander, and Self-Protective Behaviors.
Note. Bystander behaviors only asked among those who reported opportunity (n = 1,936).
Denominator is # of people who indicated that they used a bystander behavior (n = 1,215).
Denominator is # of individuals who reported experiencing cybervictimization and were social media users (n = 2,318).
Denominator is # of individuals who reported experiencing cybervictimization and had a cell phone (n = 2,938).
Opportunities to Intervene
Participants were asked whether they had witnessed a number of different types of online cybervictimization behaviors (n = 7 items, see Table 2).
Bystander Behavior
If participants responded yes to witnessing any potential opportunities to intervene (n = 1,988) in a cybervictimization situation, they were presented with three bystander response questions to determine what they did in that situation (n = 3, see Table 2). No way to report or flag was only a response option for “report or flag this behavior through the online platform where it took place.” This response was also re-coded as a missing response in order to clearly identify those who engaged in any of the assessed bystander behaviors (positive prevalence).
Self-Protective Behaviors
Self-protective behaviors were assessed by asking participants whether they had engaged in any of nine types of potentially protective behaviors; rates were considered among those reporting cybervictimization only (n = 9, see Table 2).
Analytical Plan
First, frequencies were calculated. Next, we estimated a structural equation model using Mplus, wherein four latent variables were constructed: cybervictimization experiences, witnessing opportunities to intervene against cybervictimization directed toward others, engagement in cybervictimization bystander behaviors, and use of self-protective behaviors in response to experienced cybervictimization. To account for the dichotomous nature of the data, weighted least squares with mean and variance adjusted estimation was utilized (Liang & Yang, 2014). The model explored the degree to which cybervictimization experiences predicted witnessing more opportunities to intervene, and then engaging in active bystander intervention behaviors among those who perceived an opportunity to intervene. In the same model, self-reported cybervictimization experiences were also used to predict engagement in self-protective behaviors in response to acts of cybervictimization.
Last, three additional exploratory analyses were conducted using the PROCESS macro, Model 1, in SPSS, version 28.0, which operates with bias-corrected 95% confidence intervals (n = 5,000). In these models, significant effects are indicated by the absence of zero within the resulting confidence intervals (CI). These analyses were conducted to test the conditional indirect effect of age as a moderator on the relationship between the predictor (i.e., cybervictimization) and each of the three outcome variables: witnessing opportunities to intervene; engaging in bystander behavior, and doing more types of self-protection.
Results
Descriptive Statistics
One in three (n = 988, 32.9%) participants reported experiencing at least one type of cybervictimization. The most commonly reported cybervictimization experiences were being called offensive names on-line (21.5%), followed by having someone try to embarrass them on-line (19.9%, see Table 2). Individuals were then grouped into four age categories (15–19, n = 290; 19–39, n = 852; 40–69, n = 1387; 70–95, n = 377). As expected, self-reported victimization decreased across age groups, χ2 (6) = 100.69, p < .001; however, some endorsement of cybervictimization occurred in every age group (29%; 27%; 17%; 9%, respectively).
Two in three (n = 1,936, 64.5%) participants endorsed witnessing at least one opportunity to intervene in another’s cybervictimization experiences. The most common witnessed cybervictimization behaviors were witnessing someone trying to embarrass another person online on purpose (56.1%) and seeing someone being called offensive names online (57.8%; Table 2).
A similar number, two in three participants (n = 1,215, 62.7%), who reported witnessing an opportunity to intervene online, indicated that they engaged in at least one active bystander behavior in response to witnessing cybervictimization. The most commonly reported behavior was saying something supportive online to the person who was targeted (72.2%; Table 2).
Finally, among those who reported experiencing cybervictimization, less than one in three participants (n = 303, 30.7%) reported engaging in at least one of the nine named self-protective strategies. The most common behavior reported was changing an email address or telephone number (9.9%; Table 2). Very few cybervictimized respondents indicated that they stopped using their cell phones (1.0%) or got help from a domestic violence center, hotline, or website (1.3%).
Cybervictimization’s Association with Bystander Outcomes
As shown in Table 3, factor loadings for the four latent variables were all above .60. Furthermore, as hypothesized, the fit of the proposed model was acceptable (318, N = 3,002) = 129.10, p < .001, Confidence Fit Index or CFI = .97, root mean square of approximation or RMSEA = .03 (90% CI [.030, .034]), and Standardized Root Mean Square Residual or SRMR = .08. As shown in Figure 2 and as predicted, paths from cybervictimization predicted opportunities to intervene, active bystander behaviors, and self-protective behaviors, such that endorsing more types of cybervictimization was associated with greater recognition of opportunities to intervene, use of more active bystander behaviors, and engagement in more types of self-protective behaviors in response to cybervictimization (Figure 2). Unexpectedly, witnessing more opportunities to intervene in other’s cybervictimization and engagement in active bystander behaviors were negatively related. Participants who indicated that they had witnessed more opportunities to intervene were less likely to engage in more different types of active bystander behaviors in response to these bystander opportunities. Greater witnessing of opportunities to intervene was also not associated with greater engagement in various self-protective behaviors. Finally, contrary to prediction, greater use of diverse self-protective strategies was not significantly associated with use of more types of active bystander behaviors in witnessed cybervictimization experiences.
Factor Loadings for Full Model.
Note. ***p < .001, two-tailed.

Cybervictimization experiences as a predictor of witnessing opportunities to intervene, engagement in bystander intervention behaviors, and self-reported use of self-protective behaviors among those victimized.
The three exploratory multiple regression moderation analyses were conducted using the PROCESS macro, Model 1, in SPSS, version 28.0, which operates with bias-corrected 95% confidence intervals. Each tested the conditional indirect effect of the moderating variable (i.e., age, measured continuously) on the relationship between the predictor variable (i.e., cybervictimization) and each of the three outcome variables as tested separately: witnessing opportunities to intervene, engaging in bystander behavior, and self-protection behaviors.
In the first model, the interaction of age and cybervictimization on witnessing opportunities to intervene was significant, b = 0.0032, Lower Level Confidence Interval or LLCI, Upper Level Confidence Interval or ULCI [0.006, 0.0058], t = 2.41, p = .016. Results showed a significant main effect of experiencing cybervictimization on witnessing others’ cybervictimization, b = .58, LLCI, ULCI [.48, .68], t = 11.34, p < .001, and a significant main effect of age on witnessing as well, b = −.04, LLCI, ULCI [−.04, −.03], t = −18.84, p < .001. The total model accounted for 44% of the variance, F (3, 2602) = 684.24, p < .0001, R2 = .44. Importantly, tests of the moderation effect showed it was statistically significant; however, the amount of variance accounted for by the unconditional interaction was very small, F (1, 2602) = 5.80, p = .0161, R2 = .0012. Inspection of the unconditional effects indicated that the association between experiencing acts of cybervictimization and witnessing cybervictimization was present for all ages, as none of the tested confidence intervals included zero. However, older participants (mean age = 67 years, one SD above the mean) experienced a greater effect of cybervictimization on witnessing opportunities to intervene (standardized slope or β = .80), when compared to average aged (mean age = 47 years; β = .73,) or younger participants (mean age = 27 years, one SD below the mean, β = .65).
Similarly, in the second exploratory model, there was a significant interaction of age and cybervictimization on engaging in bystander behavior, b = 0.0022, LLCI, ULCI [0.0005, 0.0040], t = 2.51, p = .012. Main effects were also noted such that there was an effect of cybervictimization on engaging in bystander behavior, b = .17, LLCI, ULCI [.10, .24], t = 4.90, p < .001. There was also a significant main effect of age on witnessing opportunities to intervene, b = −.01, LLCI, ULCI [−.011, −.005], t = −5.02, p < .001. In this case, the total model accounted for only 18% of the variance, F (3, 1785) = 131.47, p < .0001, R2 = .18. The moderation was statistically significant; however, the amount of variance accounted for by the interaction in the overall model was also quite small, F (1, 1785) = 6.28, p = .0123, R2 = .0029. Inspection of the effects indicated that the relationship between greater cybervictimization and engaging in bystander behavior held at all ages. However, older participants (mean age = 61 years, one SD above the mean) experienced a greater effect of cybervictimization on engaging in bystander behavior (standardized slope or β = .31), when compared to average aged (mean age = 39 years; β = .26) or younger participants (mean age = 21 years, one SD below the mean, β = .22).
In the third exploratory analysis, age was not found to be a significant moderator of the relation between cybervictimization and engagement in self-protective behaviors (F < 1, p = .39), although the overall model was significant and accounted for 19% of the variance in engagement in self-protective behaviors, F (3, 505) = 39.81, p < .0001. There was only one significant main effect in this model. More types of cybervictimization were associated with greater engagement in self-protective behavior, b = 0.28, LLCI, ULCI [0.1015, 0.4614], t = 3.07, p = .0022.
Discussion
As has been summarized previously (Lenhart et al., 2016), nearly a third of participants in this national survey of 3,002 people aged 15 to 93 years reported at least one cybervictimization experience, underscoring the commonality of this behavior. The vast majority of existing cybervictimization research has focused on the experiences of adolescents (Ang, 2015; Brochado et al., 2017) or young adults (Watts et al., 2017). The current study documents that cybervictimization extends later into life. Indeed, one in six adults aged 40 to 69 years and one in 11 adults 70 years of age and older self-reported at least one experience or type of cybervictimization. To date, very little research has focused on older populations or examined an age-diverse sample (Jenaro et al., 2018; Wang & Kim, 2021). As such, the current research fills an important gap by highlighting that cybervictimization, while declining in prevalence with age, continues to be reported across the lifespan. Intervention and response services that are age-inclusive and age-relevant need to be developed and disseminated in order to mitigate the potentially deleterious consequences of these experiences (Völlink et al., 2013). Tracking changes in prevalence rates and in types of cyberperpetration behavior across time will also be essential given the nearly ubiquitous use of cell phones and social media in current society.
As predicted by theory (Stevens et al., 2021), greater exposure to cybervictimization was positively associated with greater concurrent reports of witnessing opportunities to intervene, more reported use of active bystander behaviors when witnessing the cyberbullying victimization of others, and greater engagement in self-protective behaviors among those reporting victimization. The relation between experiencing cybervictimization and recognizing greater opportunities to intervene was the strongest of these three associations. Although we cannot determine temporality, it is possible that experiencing victimization may sensitize people to recognize this type of aggression when it happens to others. Alternatively, they may routinely visit certain sites where problematic behavior is commonplace. It may also be that those who recognize victimization when it occurs for others are more likely to be victimized themselves because they are part of a friend or social group that is engaged in a variety of problematic online behaviors that are impacting the group. It is further possible that this relationship is confounded by the amount of time an individual spends online, as those more frequently online may be at greater risk for being victimized while simultaneously having greater opportunities to witness cyberbullying victimization directed toward others. Disentangling these relationships as they occur among different subgroups will be essential. Understanding how these relationships play out over time may also be critical given suggestion in the current study that these relations are stronger among older than younger participants.
Consistent with current work that suggests negative personal online experiences make individuals more likely to intervene when witnessing the victimization of others (Panumaporn et al., 2020; Wang & Kim, 2021), there is a moderately strong relation in the current study between being the victim of cyberabuse and intervening when others are being harassed online. Prior research in other domains (e.g., child abuse and sexual assault) has also demonstrated a positive association between one’s own victimization and use of active bystander behaviors (Christensen & Harris, 2019; Christy & Voight, 1994; Edgington et al., 2019; Woods et al., 2016, 2020). Thus, the current work adds to the literature by extending this finding to cybervictimization. According to the situational model of bystander intervention (Darley & Latané, 1968), an individual must first notice a situation in order to subsequently intervene. In this case, it is possible that cybervictims intervene at higher rates than non-victims because they are primed to notice more opportunities to intervene. Another plausible pathway might be that experiencing victimization increases a person’s proclivity to intervene because of their heightened levels of empathy, which in turn, may increase their motivation to protect others from harm. Alternatively, it may be that people intervene first and are victimized second; perhaps the instigator turns their aggression on the person intervening as a result of their intervention behavior. Thus, this association also needs to be studied as it unfolds across time. Irrespective of the temporality, it is interesting to note that, in the final model, people who are victims of cybervictimization are more likely to witness but not intervene when they see other people being victimized online. Efforts to increase people’s self-efficacy to move from witnessing to action may be helpful in prevention and intervention programming. A follow-up analysis also indicated a significantly stronger relationship between cybervictimization and bystander action among older as compared to younger participants, suggesting a need to redouble our efforts to promote bystander behavior among younger participants who are experiencing the highest rates of cybervictimization.
Contrary to prediction, the relation between experiencing cybervictimization and engaging in self-protective behaviors was the weakest, suggesting that victims may more be sensitized to notice and give support and advice to other people in need than to try to protect themselves from online aggression. However, it may also be that the self-protective behaviors assessed in the study were seen as more personally disruptive and effortful than a brief online intervention into other people’s experiences, and so were only implemented as a last resort (Olenik-Shemesh et al., 2017). As an example, changing one’s email or phone number has a much bigger impact on one’s life and ability to connect with others than does saying something nice or supportive online to someone else. Future research should examine whether use of any particular self-protective strategies is associated with reduced odds of future victimization so as to inform victims which self-protective behaviors are empirically associated with reduced opportunity for revictimization. It would also be helpful to have a range of effective self-protective strategies available for those experiencing cybervictimization, given that many of the current alternatives would be very disruptive to the victim’s rather than the perpetrator’s life.
Also, contrary to expectation, there was a non-significant relation between engaging in self-protective behaviors and using bystander behaviors when witnessing other’s cybervictimization. While experiences of personal cybervictimization were associated with increased witnessing of opportunities to intervene and also engaging in personal protective behaviors, the proposed direct positive relationship between helping oneself and helping others was not demonstrated. As noted above, it may be that these two constructs are unrelated because they reflect different levels of burden. Additionally, other psychological and social constructs may better explain the lack of relation between these helping behaviors, including lack of empathy, cultural norms against getting involved in other’s difficulties, and lack of self-efficacy post-victimization (DeSmet et al., 2016; Van Cleemput et al., 2014).
Importantly, witnessing opportunities to intervene was significantly and negatively associated with engaging in active bystander behavior in our final model. Perhaps this reflects a learned helplessness such that the more cyberbullying one is exposed to, the less empowered one feels to counter-act it. It also may affect one’s norms: the more cybervictimization one is exposed to, the less likely one may be to identify it as something outside of normative or appropriate online behavior and thus worthy of intervention. It may further be that individuals were bystanders first and, having had a negative experience with intervening, continue to witness harassment without getting involved. If bystanders are experiencing negative consequences to their pro-social actions, it might serve as more of a deterrent than a motivator.
Implications and Future Research
Bystander intervention programs have yet to be applied to violence perpetrated in virtual settings. The current study suggests that existing prevention strategies, which largely focus on face-to-face interactions, could be updated to teach individuals how to recognize abusive online behaviors while increasing participants’ knowledge and self-efficacy about how to effectively intervene in cyberspace. Given that the diffusion of technology has led to a blurring of “online” and “offline” worlds, less siloed intervention programs are likely to result in safer in-person and online environments.
A tailored approach to prevention efforts also may be beneficial. While bystander intervention programs are often implemented universally, these efforts may have differential effects for individuals with differing histories of exposure to violence or who are embedded in different cultures. For example, Mennicke et al. (2021) found that a bystander intervention program was more effective at reducing teen dating violence outcomes (including perpetration and victimization) among high school youth who had previously witnessed parental intimate partner violence as compared to youth not exposed to parental intimate partner violence. Universal programs might be most effective for low-risk individuals or as the opening to a stepped-prevention approach. These broad programs should address both online and offline behavior and should contain more introductory information related to noticing cybervictimization and identifying and practicing realistic and effective ways to intervene once witnessed. While universal, these programs may also benefit from being age-relevant; we may also need to consider delivering these programs in additional venues outside of the school or university environment.
Next, targeted extensions of these universal programs could address family of origin violence and/or previous victimization histories (cyber and other). These latter programs could emphasize risk reduction and use of self-protective behaviors to prevent revictimization. They could also include more advanced bystander intervention content. Finally, given participants’ previous experiences with victimization, these extensions should also focus on identifying and working through barriers to action, given the weak association between exposure to cybervictimization and use of self-protective behavior observed in the current study.
Limitations
There are several limitations to the current study. First, data were collected cross-sectionally. Thus, we are unable to make causal inferences about the relationships among the key variables of interest. Understanding how these behaviors occur across time and in and across in-person and online environments is critical. Second, data were self-reported. We cannot be certain that the reported bystander or self-protective behaviors occurred. However, past research in the domain of cybervictimization (Panumaporn et al., 2020) suggests that self-report may be more predictive of outcomes than others’ assessment of a person’s behavior. We also only included those who endorsed the experience in the current study; those who indicated that they did not know were counted as non-experiencers. Thus, reported prevalence rates may be underestimated. Finally, there are several other factors, including identity (e.g., gender identity; Panumaporn et al., 2020), characteristics of the perpetrator (romantic partner, friend, stranger; Ybarra et al., 2023); psychological indices (e.g., self-efficacy; Olenik-Shemesh et al., 2017), and cultural norms (e.g., norms on joining in on cybervictimization; DeSmet et al., 2016; cultural taboos on interfering in others’ behavior) that were not explored in the current study. Future research should consider how these factors may moderate currently observed relationships.
Conclusion
Bystander research has focused on understanding under what conditions a person will help others (Allison & Bussey, 2016). Much of this previous work has focused on in-person environments that are occurring among adolescents and young adults. The current research highlights the need to target interventions toward middle-aged and older adults, who are also experiencing and witnessing cybervictimization. Findings further suggest that violence reduction programming could better acknowledge the interplay between one’s own victimization experiences and one’s ability to identify opportunities to intervene to help others. Contrary to expectation, moderation analyses indicated that this relationship was weaker among younger than middle-aged and older adults. Our findings are consistent with emerging research focusing on age as a moderator of impact and sex differences in perpetration rates (Almenayes, 2017; Barlett & Coyne, 2014). It is important to note that these previous studies focused on a considerably narrower age range among participants (i.e., school-aged children, and college students). Additional research on this topic is essential to determine if tailoring prevention and intervention efforts to older cohorts will enhance efficacy. Finally, contrary to expectation, people who engaged in self-protective behaviors were no more likely than others to report engaging in bystander behaviors when others were being aggressed upon. These processes may be unrelated. Or it may require more to move from self-protection to focusing on the safety of the larger community.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interests with respect to the authorship and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research and/or authorship of this article: This work supported by a Foundation grant to Amanda Lenhart (Data and Society) & Michele L. Ybarra (Center for Innovative Public Health Research). (September 2015–August 2016). Digital Trust Foundation. Measuring Cyberstalking and Digital Domestic Abuse across the Lifespan.
