Change-Point Analysis of Cyberbullying-Related Twitter Discussions During COVID-19
CChange-Point Analysis of Cyberbullying-RelatedTwitter Discussions During COVID-19
Sanchari Das , , Andrew Kim , and Sayar Karmakar Indiana University University of Denver University of Florida Abstract.
Due to the outbreak of COVID-19, users are increasinglyturning to online services. An increase in social media usage has alsobeen observed, leading to the suspicion that this has also raised cyber-bullying. In this initial work, we explore the possibility of an increase incyberbullying incidents due to the pandemic and high social media usage.To evaluate this trend, we collected 454 ,
046 cyberbullying-related pub-lic tweets posted between January 1 st , 2020 – June 7 th , 2020. We sum-marize the tweets containing multiple keywords into their daily counts.Our analysis showed the existence of at most one statistically significantchangepoint for most of these keywords, which were primarily locatedaround the end of March. Almost all these changepoint time-locationscan be attributed to COVID-19, which substantiates our initial hypoth-esis of an increase in cyberbullying through analysis of discussions overTwitter. Cyberbullying has become more prevalent, as targeted victimization has movedfrom in-person to digital platforms, reaching users regardless of geographic con-straints [52,44]. Victims of cyberbullying can be targeted through various sources,including mobile phones, video cameras, emails, and web pages [53]. Targets ofcyberbullying– particularly adolescents– are more likely to show signs of depres-sion, anxiety, and, in some cases, suicidal behavior [28,42,51]. Online harassmentcan carry into adulthood, with bullied victims being more likely to show men-tal health problems later on [3,9]. Such online harassment can negatively impactmental health, with 32% of victims reporting symptoms of stress and 38% of vic-tims experiencing emotional distress, even after the online abuse stopped [1,54].Thus, it is critical to detail cyberbullying and understands the victims’ perspec-tives.Social media privacy and security have been a concern for many researchersand industry practitioners [15,17,11]. Researchers have often noted that usersexperience several privacy-focused issues in these social media platforms, whichcan also lead them to leave such platforms [35]. Privacy policy and recommendedchanges to the same addressed some of the users’ concerns [12,16], but prior a r X i v : . [ c s . S I] A ug Das et al. studies have shown that social media usage has increased the extent of cyber-bullying [54]. On social networking sites and applications, cyberbullying is par-ticularly common, with 66% of all incidents on these platforms [6]. Platformssuch as Twitter allow people to sometimes interact with strangers (includingcelebrities) [13]; however, this also leads others to imitate and forge identitiesonline and trick users [47].Furthermore, with the current COVID-19 pandemic, people have increasedtheir social media usage to seek information and stay connected with otherswhile social distancing [49]. Social media can be used to support others dur-ing crises [38]. However, there have also been reports of incivility through suchplatforms [27]. A sudden rise in social media usage– combined with childrenand adolescents regularly using such platforms– could create a spike in cyber-bullying [41]. Thus, we specifically wanted to see whether that is the case andanswer the following research question:
How do crises, such as a global pandemic(COVID-19), impact cyberbullying trends over social media?
To understand users’ perspectives and the impact of COVID-19 on cyber-bullying, we collected 454 ,
046 of publicly available tweets about cyberbullyingto understand user experiences online. As hypothesized, we noticed an increasein cyberbullying incidents and discussions about it during the pandemic. Afterdiscussing the impact of cyberbullying and some related works in section 2, weprovide a detailed methodology, analysis, and findings in section 3. We brieflydiscuss the data collection, pre-processing the data in meaningful categories, andgive an overview of the change-point analysis.
Cyberbullying is a major concern for digital communication that can lead tocritical consequences. Cyberbullying has increased, given the advent of socialmedia and billions of users being online everyday [44]. Additionally, becausetimes of crisis can increase users’ online presence and, as a result, cyberbullying,it is important to consider human factors to protect users during such situations–especially with the current pandemic situation [14].
Mason defined cyberbullying as “. An individual or a group willfully using in-formation and communication involving electronic technologies to facilitate de-liberate and repeated harassment or threat to another individual or group bysending or posting cruel text and/or graphics using technological means” [30].The source of the attack can vary from mobile phones to personal computers toother digital mediums. While studying the various sources of cyberbullying, itis critical to study the behavior and reaction of the attackers and their victims.Nocentini et al. studied the behavior of attackers for different types of cyber-bullying, including an imbalance of power, intention, repetition, anonymity, andpublicity [34]. itle Suppressed Due to Excessive Length 3
Previous works have explored the effects of cyberbullying on targets, especiallyteenagers; sometimes, such abuse can impact both the cyberaggressors and cy-bervictims. Bonanno and Hymel found that both victims and perpetrators ofcyberbullying were more likely to develop depression and suicidal thoughts thanthose involved in other types of bullying [5]. Dredge et al. noted the detrimentaleffects of cyberbullying on targets’ social and emotional lives, with the sever-ity of the impact of the harassment depending on different factors, includingthe anonymity of the perpetrators and bystanders’ presence [18]. Similarly, Wis-niewski et al. noted that lower online risk could help in the teens’ develop-mental stages while developing and enhancing crucial interpersonal skills, suchas boundary setting, conflict resolution, and empathy [50]. In addition to themental impact, ˇSl´eglov´a and Cerna found that cyberbullying led to behavioralchanges, with victims displaying more cautious browsing habits and avoidancestrategies [43]. McHugh et al. noted the negative emotions victims of cyberbully-ing experience, though they also found that the impact may be more short-termthan previously thought, emphasizing the importance of resilience [31].
Cyberbullying can occur across a range of different online platforms, includingsocial networks, chat rooms, and mobile messaging applications, regardless ofgeographic proximity; such bullying can last as little as a week or go on formuch longer [45]. Because social networking platforms are often used as a meansof self-comparison, they are a prime source of self-esteem issues [48]. Severalhigh-profile incidents of cyberbullying have taken place over major social mediaplatforms. In May 2020, a Japanese reality TV star took her own life after beingsubject to abuse on social media [4]. Similar incidents across the world have ledlawmakers to pass legislation that would make cyberbullying criminal [20].As a mitigating measure, some prior work has focused on improving socialmedia policies to prevent perpetrators from abusing their victims. Milosevicexamined the responsibilities of social media companies’ in addressing cyber-bullying among children [33]. They mention concerns on the transparency andaccountability of these platforms in addressing and mitigating such issues.
Studies that analyze trends in cyberbullying help understand how events canimpact digital users. Schneider et al. conducted four surveys across 17 highschools and found that the overall rate of cyberbullying increased from 2006to 2012 [26]. Through survey-based analysis, Snell and Englander found thatfemales are more likely to be involved in cyberbullying as both victims and asperpetrators, indicating the importance of gender as a factor in mitigating onlinebullying [46]. Mangaonkar et al. used a distributed design for analyzing tweetsand detecting cyberbullying in real-time [29].
Das et al.
Twitter allows users to express themselves in 280 character ‘tweets;’ priorstudies have analyzed these messages for cyberbullying [2,36]. Cortis and Hand-schuh analyzed bullying tweets in the context of two trending events (the Ebolaoutbreak and the shooting of Michael Brown in Ferguson, Missouri). They identi-fied commonly used hashtags and named entities in bullying tweets [10]. Whetheror not such crises increased, bullying tweets were not studied. Due to an increasein individuals’ online digital presence, assumptions have been made that the pan-demic situation from COVID-19 can increase cyberbullying incidents. Thus, ourgoal is to understand the trend and find evidence to support or contradict thishypothesis.
Twitter is a social networking site where users can post real-time messages.With over 300 million active daily users, it is an ideal data source [8]. To assessthe impact of COVID-19 on cyberbullying, we collected 454 ,
046 public tweetson Twitter, all of which mentioned cyberbullying. We outline our process forcollecting and analyzing the relevant tweets below.
We scraped Twitter for user-posted, publicly available tweets related to the top-ics of cyberbullying, social media bullying, online harassment, etc. More specif-ically, we used the following key terms when conducting our search:
Internetbullying, Internet bully, Internet bullies, online abuse, online harassment, onlineshaming, online stalking, cyberbullying, social media bullying, stop cyberbullying,cyber bully, cyber bullies, FB bullying, FB cyberbullying, FB harassment, FBvictim, Facebook bullying, Facebook cyberbullying, Facebook victim, Facebook ha-rassment, Twitter bullying, Twitter cyberbullying, Twitter harassment, Twittervictim, Insta bullying, Insta cyberbullying, Insta harassment, Insta victim .The data was collected using the Get Old Tweets API [19], which allowedus to access tweets older than one week. This API was used in the web crawler,written in Python, and the data was stored with MongoDB. The data collectionspanned from January 1 st , 2020–June 7 th , 2020. This timeline was mainly se-lected to note the impact of COVID-19 on online users and determine whetherthe crisis led to an increase in online abuse. We specifically only collected directtweets and removed any retweets or duplicate tweets. After completing the data collection, we performed a trend analysis to evaluatethe impact of the crisis. Using the timestamp of the post, we obtained the dailycount of the tweets, which including at least one of these keywords. Figure 1shows the daily count for the 159 days from 01 st January, 2020 to 07 th June,2020. itle Suppressed Due to Excessive Length 5
Fig. 1.
Daily count of total tweets related to cyberbullying
There were 28 different keywords (mentioned above). Some of the keywordshad fewer tweets with a negligible impact on the analysis. Thus, we broadlydivide them into three sub-classes: keywords containing ”cyber” (CY) for thekeywords – cyberbullying, cyber bully, cyber bullies, stop cyberbullying, FB cy-berbullying, Facebook cyberbullying, and Insta cyberbullying; ”online/internet”(ON) for the keywords – internet bullying, Internet bully, Internet bullies, onlineabuse, online harassment, online shaming, online stalking, and ”twitter” (TW)for keywords – Twitter bullying, Twitter cyberbullying, Twitter harassment,and Twitter victim. The total daily counts for these sub-classes are shown inFigure 2. We also tabulate them later in the changepoint analysis in Section 3.3(Table 1).One can see a pattern prevalent to all the counts and the sub-classes thatwe present here. Overall, except for the sub-class ‘ON’ , there does not seemto be a considerable change in mean except it went slightly upwards since mid-March and in all categories, including the total. We notice a more considerablespike in the cyberbullying related tweets in the second half of May. The suddenrise in the frequency of tweets in the second half of May can be due to thesuicide of the Japanese TV star [32]. Moreover, for the class ‘ON’ , one can seea spike in the second half of February, and the overall mean also had an upwardtrend. This may or may not be due to the pandemic. Since these are cumulativegraphs of the prevalence of such words, we wanted to introspect in each of the18 keywords in the subclasses (CY=seven, ON=seven, TW=four keywords).Out of them, we observed three keywords: “cyberbullying, cyber bully, cyber
Das et al.
Fig. 2.
Daily count of total tweets for the three sub-classes (CY, ON, and TW) bullies” having a significant impact which we summarize in Figure 3. We providesome mathematical details in the next subsection about how to formally testchangepoint.
Assume we observe X , . . . , X T over T time-points, and we are suspecting atmost ONE changepoint (AMOC) location τ in mean i.e. E ( X i ) = µ if i ≤ τ, and E ( X i ) = µ + δ if i > τ where δ (cid:54) = 0 . In layman’s term this mean that the realized counts as random variables have adifferent mean before and after the change-point location τ . Statistically speak-ing this difference needs to be significant to be able to be detected from theobserved data. We adopt a CUSUM technique for estimating the changepointlocation. We define it as:ˆ τ = argmax ≤ s ≤ T s ( n − s ) n (cid:32) s s (cid:88) i =1 X i − n − s T (cid:88) i = s +1 X i (cid:33) (1)Intuitively, the above equation is the location that maximizes the differenceof normalized cumulative sum before and after this point. There have been pre-vious literature on offline change-point detection [37,21,39]. Here, for simplicity, itle Suppressed Due to Excessive Length 7 Fig. 3.
Impact of the Cyberbullying Incidents dependent on Three Major Keywords(cyberbullying, cyber bully, and cyberbullies) we assume independence over time-horizon. In statistics literature, consistencyresults typically assume the observations to be Gaussian; however, since this isa count data, it can be questionable. In light of the weak law of large num-bers, however, one can assume normality as the counts are large. To detect thechangepoints, we employ the changepoint package in R and observe the follow-ing changepoints in the three sub-classes and the three significant individualkeywords. These are tabulated in Tables 1 and 2. All these changepoints weresignificant at the type-1 error level α = 0 . Name Contains
Table 1.
Changepoint analysis of 3 sub-classes
We note that, for all the series and sub-classes we observe a changepoint. Ex-cept for the subclass ‘ON,’ all of the changepoints can be possibly be attributedto COVID. However, the total count did not show any changepoint, and wethink this can be due to multiple reasons. First, we are adding a lot of keywords.
Das et al. Keyword
Table 2.
Changepoint analysis of 3 specific keywords
Thus, the effects might get confounded. A more important reason could be thesimplicity of the assumption of independence. We show in Figure 4 that thetotal count of the individual keywords and all three sub-classes show significantcorrelation over time. Once the dependence is taken care of, it is possible thateven the total count data will show changepoints somewhere around the end ofMarch. We wish to explore this as a future work.
Fig. 4.
Auto correlations for the total count and sub-classes
In this work in progress, we wish to explore a comprehensive time-trend analysisof the impact of COVID-19 on cyberbullying as suspected by many experts.We found that certain class of keywords show a change in cyberbullying relatedtweets from the end of February or March when the pandemic fear primarily itle Suppressed Due to Excessive Length 9 started. As a future extension of this work, we would like to comprehensivelyaddress this using a change-point analysis for a time-series of count data. One canalso implement possible changes in variance since we observed some fluctuationsin the tweet counts. An interesting finding from this initial analysis is that thechange points for different sub-class and tweets are not necessarily close. This canlead us to employ methods from prior work Karmakar et al. [23] to statisticallyvalidate the hypothesis of synchronization of changepoints, as the authors thereinallowed for non-linear non-Gaussian time-series.We are also working on an alternative formulation of the same problem usinga Bayesian time-varying paradigm [24]. We assume the parameters of the modelsdo not change abruptly if there is a change-point but instead shows a moregradual change. We wish to explore time-varying models in a frequentist senseas done in [25,22] or Bayesian methods from a relatively recent work by Roy andKarmakar [40] in the regime of count autoregressive series. This would allow usto incorporate dependence in the analysis and give a clearer picture of how themean or the dependence coefficient changed over time (and thus if COVID-19had a telling effect on the increase). A time-series formulation often asks forprediction of the future, and such a work has not yet been done in the field ofcyberbullying trend analysis. Instead of a single k -step ahead forecast, we wouldlike to predict the trend for an entire month or two. We wish to explore statisticalmethods developed by Zhouwu et al. and Chudy et al. [55,7] to this non-Gaussiancount time series and build statistically valid prediction intervals. This can helpcreate a mitigating strategy in case we can predict a rise of cyberbullying forthe next one or two months. We would like to thank Umang Mehta for his help with the data collection.We would also like to acknowledge the research institutions and labs of the re-searchers involved with this project- Secure and Privacy Research in New-AgeTechnology (SPRINT) Lab, University of Denver; and Human and TechnicalSecurity (HATS) Lab, Indiana University, and the University of Florida. Anyopinions, findings, and conclusions or recommendations expressed in this mate-rial are solely those of the author(s).
References Online Victimization: A Report on the Nation’s Youth . ERIC, 2000.2. Sophia Alim. Analysis of tweets related to cyberbullying: exploring informationdiffusion and advice available for cyberbullying victims.
International Journal ofCyber Behavior, Psychology and Learning (IJCBPL) , 5(4):31–52, 2015.3. Rienke Bannink, Suzanne Broeren, Petra M van de Looij-Jansen, Frouwkje Gde Waart, and Hein Raat. Cyber and traditional bullying victimization as a riskfactor for mental health problems and suicidal ideation in adolescents.
PloS One ,9(4), 2014.0 Das et al.4. BBC. Hana kimura: Netflix star and japanese wrestler dies at 22, 2020 (accessedJuly 09, 2020).5. Rina A Bonanno and Shelley Hymel. Cyber bullying and internalizing difficulties:Above and beyond the impact of traditional forms of bullying.
Journal of Youthand Adolescence , 42(5):685–697, 2013.6. Pew Research Center. Part 2: The online environment. ,2014 (accessed July 09, 2020).7. M Chud`y, Sayar Karmakar, and Wei Biao Wu. Long-term prediction intervals ofeconomic time series.
Empirical Economics , 58(1):191–222, 2020.8. J. Clement. Twitter: number of monthly active users 2010-2019, 2019 (accessedJuly 09, 2020).9. William E Copeland, Dieter Wolke, Adrian Angold, and E Jane Costello. Adultpsychiatric outcomes of bullying and being bullied by peers in childhood and ado-lescence.
JAMA Psychiatry , 70(4):419–426, 2013.10. Keith Cortis and Siegfried Handschuh. Analysis of cyberbullying tweets in trendingworld events. In
Proceedings of the 15th International Conference on KnowledgeTechnologies and Data-driven Business , pages 1–8, 2015.11. Sanchari Das, Jayati Dev, and L Jean Camp. Privacy preserving policy modelframework.
Available at SSRN 3427634 , 2019.12. Sanchari Das, Jayati Dev, and Kaushik Srinivasan. Modularity is the key a newapproach to social media privacy policies. In
Proceedings of the 7th Mexican Con-ference on Human-Computer Interaction , pages 1–4, 2018.13. Sanchari Das, Javon Goard, and Dakota Murray. How celebrities feed tweeples withpersonal and promotional tweets: Celebrity twitter use and audience engagement.In
Proceedings of the 8th International Conference on Social Media & Society ,pages 1–5, 2017.14. Anneliese Depoux, Sam Martin, Emilie Karafillakis, Raman Preet, AnneliesWilder-Smith, and Heidi Larson. The pandemic of social media panic travels fasterthan the covid-19 outbreak, 2020.15. Jayati Dev, Sanchari Das, and L Jean Camp. Understanding privacy concerns ofwhatsapp users in india: poster. In
Proceedings of the 5th Annual Symposium andBootcamp on Hot Topics in the Science of Security , pages 1–1, 2018.16. Jayati Dev, Sanchari Das, and Linda Jean Camp. Privacy practices, preferences,and compunctions: Whatsapp users in india. In
HAISA , pages 135–146, 2018.17. Jayati Dev, Sanchari Das, Yasmeen Rashidi, and L Jean Camp. Personalizedwhatsapp privacy: Demographic and cultural influences on indian and saudi users.
Available at SSRN 3391021 , 2019.18. Rebecca Dredge, John Gleeson, and Xochitl De la Piedad Garcia. Cyberbullying insocial networking sites: An adolescent victims perspective.
Computers in HumanBehavior , 36:13–20, 2014.19. Jefferson Henrique. Getoldtweets-python, 2016 (accessed July 09, 2020).20. Sameer Hinduja and Justin W Patchin. Cyberbullying legislation and case law.
Implications for School Policy and Practice , 5:2016, 2015.21. D. V. Hinkley. Inference about the change-point from cumulative sum tests.
Biometrika , 58:509–523, 1971.22. Sayar Karmakar. Asymptotic theory for simultaneous inference under dependence.Technical report, University of Chicago, 2018.23. Sayar Karmakar, Maggie Cheng, and Wei Biao Wu. Test-ing synchronization of change-point in multiple time-series. https://sayarkarmakar.github.io/publications/sayar5.pdf , 2020+.itle Suppressed Due to Excessive Length 1124. Sayar Karmakar and Sanchari Das. Evaluating the impact of covid-19 on cyber-bullying through bayesian trend analysis.
The European Interdisciplinary Cyber-security Conference (EICC) , 2020.25. Sayar Karmakar, Stefan Richter, and Wei Biao Wu. Simultaneous inferencefor time-varying models.
In revision, https: // sayarkarmakar. github. io/publications/ sayar1. pdf , 2020+.26. Shari Kessel Schneider, Lydia O’Donnell, and Erin Smith. Trends in cyberbullyingand school bullying victimization in a regional census of high school students,2006-2012.
Journal of School Health , 85(9):611–620, 2015.27. Bumsoo Kim. Effects of social grooming on incivility in covid-19.
Cyberpsychology,Behavior, and Social Networking , 2020.28. Robin M Kowalski and Susan P Limber. Psychological, physical, and academiccorrelates of cyberbullying and traditional bullying.
Journal of Adolescent Health ,53(1):S13–S20, 2013.29. Amrita Mangaonkar, Allenoush Hayrapetian, and Rajeev Raje. Collaborative de-tection of cyberbullying behavior in twitter data. In , pages 611–616. IEEE, 2015.30. Kimberly L Mason. Cyberbullying: A preliminary assessment for school personnel.
Psychology in the Schools , 45(4):323–348, 2008.31. Bridget Christine McHugh, Pamela J Wisniewski, Mary Beth Rosson, Heng Xu,and John M Carroll. Most teens bounce back: Using diary methods to examinehow quickly teens recover from episodic online risk exposure.
Proceedings of theACM on Human-Computer Interaction , 1:1–19, 2017.32. Patrick Michel. Public begins to wake up to sheer volume of harassment online,2020 (accessed July 09, 2020).33. Tijana Milosevic. Social media companies’ cyberbullying policies.
InternationalJournal of Communication , 10:22, 2016.34. Annalaura Nocentini, Juan Calmaestra, Anja Schultze-Krumbholz, Herbert Schei-thauer, Rosario Ortega, and Ersilia Menesini. Cyberbullying: Labels, behavioursand definition in three european countries.
Journal of Psychologists and Counsel-lors in Schools , 20(2):129–142, 2010.35. Abu Saleh Md Noman, Sanchari Das, and Sameer Patil. Techies against facebook:Understanding negative sentiment toward facebook via user generated content. In
Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems ,pages 1–15, 2019.36. Hani Nurrahmi and Dade Nurjanah. Indonesian twitter cyberbullying detectionusing text classification and user credibility. In , pages 543–548. IEEE,2018.37. E. S. Page. A test for a change in a parameter occurring at an unknown point.
Biometrika , 42:523–527, 1955.38. Leysia Palen. Online social media in crisis events.
Educause Quarterly , 31(3):76–78,2008.39. A. N. Pettitt. A simple cumulative sum type statistic for the change-point problemwith zero-one observations.
Biometrika , 67(1):79–84, 1980.40. Arkaprava Roy and Sayar Karmakar. Bayesian semiparametric time varyingmodel for count data to study the spread of the covid-19 cases. arXiv preprintarXiv:2004.02281 , 2020.41. Mythili Sampathkumar and Maya Shwayder. Cyberbullying increases amid coro-navirus pandemic. heres what parents can do, 2020 (accessed July 09, 2020).2 Das et al.42. Allison M Schenk and William J Fremouw. Prevalence, psychological impact, andcoping of cyberbully victims among college students.
Journal of School Violence ,11(1):21–37, 2012.43. Veronika ˇSl´eglov´a and Alena Cerna. Cyberbullying in adolescent victims: Percep-tion and coping.
Cyberpsychology: Journal of Psychosocial Research on Cyberspace ,5(2), 2011.44. Robert Slonje and Peter K Smith. Cyberbullying: Another main type of bullying?
Scandinavian Journal of Psychology , 49(2):147–154, 2008.45. Peter K Smith, Jess Mahdavi, Manuel Carvalho, and Neil Tippett. An investigationinto cyberbullying, its forms, awareness and impact, and the relationship betweenage and gender in cyberbullying.
Research Brief No. RBX03-06. London: DfES ,2006.46. Patricia A Snell and Elizabeth Englander. Cyberbullying victimization and behav-iors among girls: Applying research findings in the field.
Journal of Social Sciences ,2010.47. Alexander Tsoutsanis. Tackling twitter and facebook fakes: Id theft in social media.
World Data Protection Report , 12(4):1–3, 2012.48. Erin A Vogel, Jason P Rose, Lindsay R Roberts, and Katheryn Eckles. Socialcomparison, social media, and self-esteem.
Psychology of Popular Media Culture ,3(4):206, 2014.49. Brenda K Wiederhold. Social media use during social distancing, 2020.50. Pamela Wisniewski, Heng Xu, Mary Beth Rosson, Daniel F Perkins, and John MCarroll. Dear diary: Teens reflect on their weekly online risk experiences. In
Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems ,pages 3919–3930, 2016.51. Michele L Ybarra. Linkages between depressive symptomatology and internetharassment among young regular internet users.
CyberPsychology & Behavior ,7(2):247–257, 2004.52. Michele L Ybarra, Marie Diener-West, and Philip J Leaf. Examining the overlapin internet harassment and school bullying: Implications for school intervention.
Journal of Adolescent Health , 41(6):S42–S50, 2007.53. Michele L Ybarra and Kimberly J Mitchell. Online aggressor/targets, aggressors,and targets: A comparison of associated youth characteristics.
Journal of childPsychology and Psychiatry , 45(7):1308–1316, 2004.54. Michele L Ybarra, Kimberly J Mitchell, Janis Wolak, and David Finkelhor. Exam-ining characteristics and associated distress related to internet harassment: Find-ings from the second youth internet safety survey.
Pediatrics , 118(4):e1169–e1177,2006.55. Zhou Zhou, Zhiwei Xu, and Wei Biao Wu. Long-term prediction intervals of timeseries.