The researchers in our lab focus on the significance of the social context to adolescents’ peer experiences. In past and current work, lab members attend to multiple influential levels of the social context. These include different features of school and social media contexts (e.g., ethnic composition and behavioral norms), and more proximal levels of the social context such as students’ friendship groups and episode characteristics (i.e., features of specific real-life interactions).
Daily Experiences with Diversity: Academic and Social Adjustment in High School (IES, Nishina, Witkow, & Bellmore)
Recent nationally visible events reflecting interethnic conflict have renewed concerns about race relations in the United States. At the same time, many geographic areas are seeing increased ethnic diversity, particularly within youth populations. The purpose of the proposed study is to examine malleable factors that may allow schools and districts to parlay this ethnic diversity into positive academic outcomes for adolescents. Specifically, the study follows high school students longitudinally across grades 10-12 with 3 main aims, which are to (1)predict social-cognitive flexibility from ethnic diversity proximity (i.e., opportunities) and experiences (i.e., actual interactions with peers from ethnically diverse backgrounds), (2) assess social-cognitive flexibility as a mediator in the association between diversity exposure and academic outcomes (grades, attendance), and (3) explore key factors—ethnicity, prior diversity experiences in middle school, and interethnic school climate—that may moderate these processes. The study relies on daily reports of group-level diversity exposure, a methodology expected to capture a reliable and valid assessment of students’ actual social experiences on high school campuses.
Advancing the Scientific Understanding of Bullying Through the Lens of Social Media (NSF, Zhu & Bellmore)
Stemming from the clear evidence that bullying is a serious national health issue, this project was designed to develop machine learning models that can analyze social media data to understand bullying. Two types of data were collected: over 1.5 million public Twitter posts that used keywords related to bullying between the years 2012-2018, and public Twitter posts along with daily report surveys of social media use and school experiences from over 500 high school students in the years 2014-2018. With these data we built a machine learning system to filter the relevant bullying-related tweets, and we demonstrated that bullying can be investigated with social media data through Natural Language Processing techniques. Using these techniques and the public Twitter posts, we conducted sentiment analysis of bullying tweets, we investigated factors associated with the deletion of tweets about bullying, we identified cross-cultural differences in posts about bullying on Twitter and Weibo (used in China), we identified how different categories of hashtags are associated with influential bullying posts, we investigated who is most likely to post about bullying (including celebrities who use social media in mostly positive ways to serve as advocates and confidants), and we identified the temporal and geographic representation of bullying tweets. Capitalizing on the dynamic nature of the data, we designed a mathematical model for properly modeling spatio-temporal distribution of online events such as bullying. With the high school sample, we were able to demonstrate how victims and non-victims of bullying use social media and how their social media content differs from one another. In these ways, we provided significant new scientific data toward understanding and intervening bullying on social media. The broader impact of the work beyond the scientific outcomes yielded involved dissemination of findings to the public through popular media and scientific training for over 25 undergraduate and graduate students in interdisciplinary work between computer science and social science.
Day-to-Day Coping with Peer Victimization: Diversity of Friends, Bystanders, and Responses (NSF, Bellmore, Nishina, Witkow)
The current study was designed to uncover the coping responses sixth grade students choose when they are victimized, whether their social environments (including the ethnic diversity of their friendship groups and the presence of peer bystanders) impact these choices, and which coping responses work best for victimized youth. Our overall focus was on the significance of school social contexts to students’ peer victimization experiences so that we can lay the groundwork for uncovering solutions that can be adopted by schools to reduce the harm associated with peer victimization. We hypothesized that the responses chosen by the youth and the effectiveness of these responses would be dependent on the immediate social environments in which they were used. Good fit between a coping response and the social environment was expected to predict the best adjustment. Inconsistent fit between coping responses and environment—for example, using social support coping when no friends are present—was expected to be associated with poor adjustment. When taken together, we expected our results to help explain why the existing research literature suggests that there is no one-size-fits-all approach to coping with peer victimization where certain responses are always effective and others are always ineffective.
A daily-report methodology where students completed “real time” reports of their daily victimization experiences, coping, and well-being as well as global, one-time measures of these same phenomena were used to test these hypotheses with a sample of 1053 students (47% boys, 53% girls) across 78 classrooms and 3 states (Wisconsin, California, and Oregon). As planned, the overall sample was ethnically diverse: 31% White, 30% Latino/a, 14% African American, 11% Asian, 2% Native American, and 12% multiethnic, and reflected each schools’ ethnic composition. With this large sample that spanned different geographic regions of the U.S., we found that daily victimization at school predicted increased daily feelings of humiliation, anxiety, and for boys, physical symptoms, and that the coping choices in response to peer victimization were tied to adjustment. In line with our hypotheses, results also indicated that there was considerable variation in the contextual features of adolescents’ daily victimization experiences and that these features were predictive of the type of coping responses that victimized youth used. Adolescents’ coping goals such as whether they wanted to avoid a fight or stand up for themselves were also found to predict their coping responses. We also found evidence that the contextual features of the events were related to the behavior and adjustment of witnesses to peer victimization. The number of witnesses at a victimization event is associated with the likelihood of the victim receiving help. When more than one witness was present the victim was more likely to receive help. End-of-day anger increased on days in which students witnessed peer victimization, and for those students who had higher levels of empathic pity, they became increasingly angry with repeated witnessing over time.
There is evidence from this study that school contextual factors also impact adolescents’ adjustment more broadly. With the daily report measures, we found that frequently eating lunch with a cross-ethnic peer is associated with higher GPA (in core academic courses), and higher teacher expectations of academic attainment. This finding holds after controlling for gender and ethnic main effects, and largely holds across ethnic groups, suggesting that cross-ethnic interactions may offer benefits for all adolescents. Other work examined adolescents’ peers at different levels. For example, being above the average body mass index (BMI) of same-sex friends (i.e., friendship group level) was associated with greater depressive symptoms and being above the BMI average for same-sex grademates (broader peer group) was associated with more peer victimization. In another study, we found peer-reported acceptance and popularity were associated with increased odds of receiving a crush nomination (i.e., being perceived romantically desirable). For those high in peer acceptance, self-reported empathy also predicted greater romantic desirability, whereas for those high in perceived popularity, aggression diminished it. Combined, these findings highlight the role individual by school social context interactions play in students’ adjustment. It is not just the student’s own characteristics or social context factors that matter, but the individual combined with the school social environment that matters for individual outcomes.
The broader impact of the work beyond the scientific outcomes involves dissemination of findings and scientific training for undergraduate and graduate students. We issued a study newsletter for participating schools. We presented feedback in person for schools/parent workshops and have worked with other local schools, and state and national organizations providing feedback to address bullying. Across the 3 study sites, NSF funding facilitated the recruitment of a highly diverse research team. Overall, the project provided research opportunities to 35 undergraduates (many from underrepresented backgrounds based on ethnicity or education) and 12 graduate students (6 from underrepresented backgrounds).
Who is Involved?
For details of our lab director, graduate students, undergraduate students, and lab alumni, please visit the People page.
Publications and Presentations
Please visit the Collaborators page to see the information and links of each of our lab’s collaborators.