Battling the Bullies: A Text Analysis of Student Interventions at University

Thomas Mueller

McKenzi Wallin

Appalachian State University

 Battling the Bullies: A Text Analysis of Student Interventions at University

Abstract

An unfortunate, yet prevalent peril at University is the role of peer association and bullying. Research indicates that often, bullying occurs within a social context (O’Connell, Pepler, & Craig, 1999). This study used linguistic inquiry and word count (LIWC) text analysis to explore university students’ perception of bullying intervention through essay writing. Student subjects lived in the same residence learning community (RLC) and were enrolled in the same first year seminar (FYS). It was hypothesized problem solving and bullying resolution would significantly differ before and after the course was completed. Pretest, students exhibited traits identified in the variables “Clout” and “Tone.” In posttest, students exhibited dominant traits “Authentic” and “Tone.” It is suggested social identity among student residents is a constant. Students initially intervened with bravado and confidence. Knowledge gained throughout the FYS course, and within the RLC, led to increased strategic thinking, then transparency, in better understanding the core dysfunction in the bullying persona.

Keywords: Text analysis, Residence Learning Community, First Year Seminar, “Luke,” LIWC, student leadership, linguistic inquiry and word count.

Introduction

An unfortunate, yet prevalent peril at University is the role of peer association and bullying. Research indicates that often, bullying occurs within a social context (O’Connell et al., 1999). As the reach of social interaction increases, there is potential for a systematic process that leads to bullying episodes. The university setting has become endangered and for many students, deemed “unsafe” in both emotional and physical terms. It has become imperative for university administrators and counselors to better understand the dynamics of bullying and the social protocol that Drives it.

The U.S. Department of Education and the Centers for Disease Control released a research based definition of bullying. It summarized “bullying is any unwanted aggressive behavior(s) by another youth or group of youths who are not siblings or current dating partners that involves an observed or perceived power imbalance and is repeated multiple times or is highly likely to be repeated. Bullying may inflict harm or distress on the targeted youth including physical, psychological, social, or educational harm (Gladden, Vivolo-Cantor, Hamburger, & Lumpkin, 2014).”

Farrington (1993) identified bullying as an act between perpetrators and recipients, one which is Driven by a power differential and imposed negative actions. Bullying may be verbal or physical, in some instances repeated over time. Craig (1998) supported that posit through research that indicates male bullying is predominantly physical at the 5th grade level, but transitions to verbal aggression by the 8th grade. For female respondents, it was documented differences in aggression did not occur until the latter years. When there is peer association, bullying can assume an indirect form. Bullying can include acts of exclusion for targeted individuals, or social framing that includes perpetuating rumors or propagating gossip (Salmivalli, Lagerspetz, Björkqvist, Österman, & Kaukiainen, 1998).

Although it is documented bullying occurs within many social settings, it is difficult to control (Kyriacou, Mylonakou-Keke, & Stephens, 2016). Among students, bullying is a major problem in schools, yet not well managed by administration officials. It is a complex issue. The study relates that although there is consensus bullying is prevalent, there are alternative views on how to best manage and alleviate this deviant behavior.

Garland, Policastro, Richards, and Miller (2017) stated that while most students are proactive in their defense against bullying, there was a faction that employed victim blaming. This group consisted of males, heterosexuals, and a grade school and high school history of bullying. Students who used illegal drugs or frequent alcohol use were also prone to minimize bullying. Bullying is an issue without boundaries, where counseling and courses specific to the topic are now being developed to support abused students (Webber, 2017).

Now, in the digital age, cyber-bullying has become a frequent form of libelous and slanderous attack. In a study with 226 Greek students, almost one third indicated a bullying assault each month. Male students indicated a greater level of bully events, related to females in the study (Kokkinos, Baltzidis, & Xynogala, 2016).

If bullies exist within educational environments, who are the “bully stoppers?” Who are the protectors who bring interventions to disengage the perpetrators from their actions? In a university setting, future leaders have been developed through First Year Seminars (FYS), which are known as foundational in students’ successful college experience (Gardner, 1986). These specific academic curriculums are most commonly part of general education requirements and tend to be interdisciplinary. FYS focus on honing academic skills such as critical thinking and expository writing (Barefoot & Fidler, 1992). They focus on a variety of topics, or can be associated in content for a specified discipline or profession.  Some institutions require a basic study skills FYS, which is offered to prepare students for the rigors they will face throughout their college careers. Topics tend to include grammar, how to take notes, efficient text reading, and proper research techniques (Barefoot & Fidler, 1992). Common learning outcomes for FYS courses include identity within a peer group, student and faculty bonding within the seminar course, and education on specific skills associated with success in college (Murphy, 1989).

First Year Seminars increase meaningful connections with faculty and staff, prompt higher engagement with campus services, increase involvement in activities, and lead to greater satisfaction (Barefoot, Warnock, Dickinson, Richardson, & Roberts, 1998). Prompting cognitive process regarding the construct of bullying can build future leaders who can campaign against bullies and break the recurrent bullying process. Students in this study were housed in a Residential Learning Community (RLC) associated with the course. RLC’s house a predominant population of first year students, brought together through either academic or lifestyle associations (Kuh & Vesper, 1997). Advantages to students in RLC communities include planned programming, academic themes, and focus on student-specific needs. Research indicates students who are housed in residential communities are exposed to productive educational experiences. In many cases, RLC students track through university with a notably higher GPA. Pike (1999) concluded involvement, interaction, integration and intellectual development were significantly higher (over a control group) for students housed in residential communities.

        An emerging advantage to the RLC lifestyle is a diverse community (Aber et al., 2010). Engaging others from diverse ethnic backgrounds has become an enriching experience within these dedicated housing communities. Students and faculty within RLCs believed critically analyzing and addressing racial inequality would extend beyond residence life and graduation. Ellett and Schmidt (2011) found that faculty identified the need for equality and sustained interactions in community building. Dialog that holds a meaningful context, coupled with shared experiences, was part of successful relationships. This move towards social justice may be related to a sense of identity within the RLC (Spanierman et al., 2013).

This study examines bullying intervention, as perceived by university students enrolled in a First Year Seminar (FYS) and residing in a RLC. Students were prompted to provide a problem-solving response to a specific bullying scenario at the beginning of the course (pretest) and once again as the course concluded (posttest). Each essay was converted into quantifiable data using Linguistic Inquiry and Word Count (LIWC) software.

Based on our knowledge, LIWC has not been applied to the construct of bullying. Published academic literature on bullying is sometimes quantitative, conducted through surveys and data collection. Other published studies are qualitative in nature, using interviews or focus groups to discern general themes. This study presents a “hybrid” that collected qualitative input to capture expression, then transitioned to quantitative analysis through LIWC. Statistical testing was used to identify the motivators of bully-stopping, testing for dissimilarities as university Freshman mature through their first semester.

It was important to consider the nature of leadership related to bullying resolution, as well as characteristics exhibited by which those with leadership qualities. Scholars have defined leadership as the ability to impart interactive influence, within a set context, where individuals accept someone as leader, to reach common goals (Silva, 2016). Specific to this study, Kouzes and Posner (1993) defined student leaders as those who challenge the current process; empower others to act; set an example by modeling behaviors; develop a shared vision; and support the affective response in others through encouraging psychological well-being as part of the process.

History and relevance of LIWC

LIWC (referred to as “Luke”) was developed in the mid 1980’s, originally intended as a simple word search and categorizing software. It has steadily evolved over decades of psychological research and technological advancement, and now serves as a fluid and ever-changing tool used for the reliable prediction of linguistic inquiry (Tausczik & Pennebaker, 2010). The concept of linguistic inquiry as an insight as to one’s beliefs, fears, thinking patterns, social relationships and personalities have long been established.  Therefore, LIWC serves as a modern technological platform for which to efficiently analyze a person’s psychological or behavioral status: priorities, intentions, and thoughts, degree, valence, and expression of emotion, status of social relationships, status, dominance, and social hierarchy, social coordination and group processes, honesty and deception, thinking styles, and individual differences (Pennebaker, Boyd, Jordan, & Blackburn, 2015). “Luke” tabulates frequency of word use and word associations, then categorizes responses into psychological and social variables. Prior studies have documented the importance of using LIWC to assess affective outcomes and emotion (Kahn, Tobin, Massey, & Anderson, 2007).

Students in a university setting have been featured in several prominent studies. Research found that testing for students who wrote about thoughts and feelings, use of positive emotion words and words associated with the development of insight were associated with an improvement in physical health (Kahn et al., 2007). Another university study found that those who wrote about deeply personal topics were subsequently healthier (both physically and psychologically) than those writing about controlled topics (Pennebaker, 1993).

LIWC has two central features, which by working together produce a process for which to sort and analyze text files. Processing, the first feature, allows LIWC to open text files and sort them word by word. The second process is the program’s “dictionaries.” Each word in a text file is grouped into a type of word based on the program’s pretest-established dictionaries. Each of the program’s 80 dictionaries are collections of words that define a category and include such groupings as impersonal pronouns, auxiliary verbs, past tense verbs, positive emotions, and negative emotions (Tausczik & Pennebaker, 2010). While some of the categories are straightforward language dimensions (such as articles and pronouns), the subjective categories (such as positive and negative emotion) requires the input of human judges.  A three-step process was established to judge which subjective words fall into which category. Over the years, some word categories and dictionaries have been deleted while others have been added. Overall, the program evolves and changes to provide an accurate and modern reflection of the role it serves within linguistic inquiry research.

LIWC calculates two categories of variables. Most variables are assessed by the proportion of the whole, based on frequency for which those words are presented in the essay. As an example, LIWC might discover that 3.74% of all the words in a poem were impersonal pronouns, while 2.8% of the words were negative emotions. This is the “transparent” measure, set to a percent-of-the-whole 0-100% scale. Four “summary language variables” are scored on a 0-100 standardized scale measure. They are analytical thinking, clout, authenticity, and emotional tone. These measures are derived from a compilation of research from previously published findings in large comparison files (Pennebaker et al., 2015).

The results from LIWC’s capabilities have been numerous and widespread (Newman, Pennebaker, Berry, & Richards, 2003). Overall, these outcomes have established a strong foundation for exploring health, psychological, and clinical impact through LIWC software. The ever-evolving nature of LIWC and the outcomes it measures are the foundation of an ever growing and influential field of research.

Methods

The research study was approved by the university Institutional Review Board (IRB). Student participants submitted demographic information including gender, age and ethnicity. Two sections of First Year Seminar classes were included in the study, 59% were female and 41% were male. An almost equal proportion of students originated from urban (49%) and rural (51%) settings. Most were 18 years of age (84%) while 14% were 19 or 20 years of age. The sample set was predominantly Non-Hispanic White or Euro-American (84%). Five percent were Black, Afro-Caribbean or African American. Three percent were Latino or Hispanic American. Three percent reported East Asian or Asian American.

        Most students (57%) stated “family funds my university expenses.” Another faction stated that finances were self-funded, with family help (24%). Eight percent were financed through scholarships, while 11% used a combination of financial aid, scholarships and family. GPAs among the test group were high. Forty one percent held a 3.5-4.0 GPA. The largest proportion stated they came to university with above 4.0 (54%). Some of the Freshman students (32%) carried 1-10 advanced credits from high school to university, while 16% carried in 11-20 credits. Half (50%) did not bring advanced high school credits to university.

        Psychologically, the group paired almost even with 51% identifying as introverted and 49% identifying as extroverted. Most (62%) viewed themselves as leaders, while 38% would choose to follow within a group setting. Approximately half (49%) invest one to four hours a week in out of classroom university sponsored activities.

Participants were presented with a complex bullying storyline that incorporated group dynamics; students exhibiting restricted physical capabilities; and those with restricted ability being marginalized and mocked by an external group. The following was posited in the form of an essay question assignment as an introduction to the FYS. The same essay question was presented as a concluding assignment in the course:

You are at a recreational sporting event with a group for the (redacted) RLC. A hiking event is included and several students in your group are taking the hike at a slower pace and are back from the lead group. Another group is also on the hike. They begin to ridicule the slower hikers, telling them they aren’t athletic enough, that the slower hikers should have stayed home. What would you say to the slower hikers from your group? What would you say to the external group that is ridiculing your slow hikers? How did you make decisions on what to say and do? How would you take a leadership role in this situation?

Essay responses were exported from a university learning management system into data management software, then imported into LIWC to achieve text conversion to variable responses. Analysis for this study focused on Pennemaker’s (2015) summary language variables Analytic, Clout, Authentic and Tone. These psychological and social composite variables were relatable to leadership qualities. Summary language variables were examined for mean score comparison pretest FYS, posttest FYS and related to LIWC composite average scores. Essays indicating the highest score for each summary variable, pretest and posttest, were examined for indicative narratives.

The composite variable “Drive” was incorporated as a dependent variable. Drive captures the dimension of group affiliation; levels of achievement; perception of power; and levels of reward based on actions. Data sets for pre and post essay narrative were tested for multiple regression, using the summary variables as independent, and the Drive variable as dependent. Correlations among the five variables were tested pre and post for significant associations.

Though descriptive variables are presented to define the respondent audience, the N of the subjects do not allow for statistically robust testing related to Analysis of Variance (ANOVA) measures.

Results

A mean score comparison between pretest and posttest FYS essays was developed. The developers of LIWC provide a benchmark mean score, compiled from their national and international testing dictionaries. Analytic, Clout and Tone variables saw a decrease in mean score, from pretest to posttest. The variable Authentic indicated a sharp increase, from pretest to posttest (Figure 1).

Excerpts for each variable were compiled, to illustrate narratives representative of summary variables Analytic, Clout, Authentic and Tone (Table 1).

Correlation testing in pretest essay data presented a strong negative correlation between Authentic and Clout (r = -.65, p <.001) and a strong positive correlation between Drive and Tone (r = .69, p <.001). In posttest essay data Authentic and Clout was the only associated pair, holding a negative relationship (r = -.36, p = .02).

LIWC summary variables, with other variables selected for their leadership behavior outcomes, were tested in a network analysis. Network analysis determined the linear association of variables within pretest FYS essays (Figure 2) and posttest FYS essays (Figure 3).

Multiple regression analysis was performed to test if personality traits Analytic, Clout, Authentic and Tone significantly predicted the variable Drive. The results of the pretest essays indicated Tone was the unique significant predictor of Drive (R2 = .65, F (4,32)=14.88, p<.001). The regression for the posttest essays was not significant.

Discussion

This study quantified perception of bullying and intended intervention tactics for Freshman university students, with a First Year Seminar serving as the treatment/intervention between pretest and posttest text analysis. Prior to the FYS experience, students addressed the bullying scenario with pretest dominant Clout and Tone. High Clout measures indicate the author holds a perspective of high expertise and confidence. High Tone indicates an approach framed through a positive, upbeat point of view. It should be noted student responses indicate a much higher mean response than is represented in the standard mean provided by LIWC across its cumulative text analysis.

Following one semester in FYS and its related Residence Learning Community, students indicated a slight drop in their Analytic approach and maintained a positive Tone. However, there was a drastic reduction in perception of lout – shifting to a tentative, humble and possibly anxious demeanor, which was more representative of the LIWC standardized mean for Clout. Students transitioned from Clout, to Authentic in resolution style. This indicates students believed themselves to be honest, personal and open in their approach to alleviating the bullying scenario. This partially supports the (Kouzes & Posner, 1993) study, in which student leaders were defined as having compassion for the emotional response in others, promoting psychological wellbeing. A transparent sense of self invites the affective response in others.

“We must be stronger…through reproof of inappropriate actions and encouragement of appropriate actions of members in our group,” wrote the respondent most highly ranked for Clout in the pretest. This shifts in perception in the posttest, note the position taken by the high ranked Clout respondent: “If I want people to respect me as a leader, I have to respect them…No person is of any greater value than another.  Everyone is equal and accomplish greater things when they work together….to help everyone fulfill their goals and keep a positive mindset until the end.”

Association of variables through network analysis provides a deeper explanation of leadership narrative. Drives – defined as an overarching dimension that captures needs and motives – became the variable “hub” in pretest essays, anchored by Clout, Achievement and Affiliation. The network also reveals that Affiliation holds a negative relationship with Power. The Drive of students in this pretest had an association to Power, but in a lesser extent. After mediation through FYS, students again set Drives as the foundational variable, this time supported by Analytic, Affiliation and Focus Present. What proves interesting is that Analytic, Affiliation and Focus Present hold negative associations, mediated through the variable Power.

This model supports the definition of bullying as aggressive behavior that is prompted by a perceived power imbalance (Gladden et al., 2014). Students who held a high-power view in pre testing, have learned to reidentify without power traits in posttests. When having Drive(s) one should stay in the present and be Analytical in assessment; when the issue is Power, focus on present is not as relevant.

“Making these decisions (to support the bullied group) extend from my previous experience of stepping up and being a leader…taking a stand…making my presence known in a positive manner,” stated the author of the highest-ranking posttest Analytic response. Here, the residence life community might have presented a positive treatment. This supports the study by (Pike, 1999), where intellectual development was significantly higher for students housed in residential communities.

This study indicates that students in the pretest may have been less Driven by knowledge, but rather by passion. Clout and Achieve were associated with achieving Drive, as noted in the following pretest excerpts:

“The power comes from the comparison of the bully to the bullied, which is often shallow and relative.”

“…do it because it’s never right to watch someone get bullied and realize you had the power and ability to stop them.”

“(We have) the power and should stand up for the ones getting bullied.”

A more thoughtful, reflective student formed a solution in posttest. Still anchored by Drive, processing was now filtered through an analytical approach, staying in the moment to ascertain needed information for the best decisions as demonstrated in the following posttest excerpts:

“I do not appreciate conflict between people for silly reasons such as bullying.”

“It is the actions that occur after the harmful words that truly determine who is the real winner. Bullies are lonely people who are desperately seeking for attention at any cost.”

“It is clear the third group of the hikers has taken the former approach to athleticism. Typically, bullies are the ones who are self-conscious and are unable to have high levels of self-esteem.”

“At first I struggled with staying diplomatic and I gave the bullies the reactions they wanted. I eventually learned to stay calm and collected and learned that saying nothing at all was better.”

Through this process, students became less assertive and more transparent in attaining positive outcomes. As the students began and concluded the semester, affiliation remained a constant. Resident Learning Communities attract students with similar interests and passions. Social identity within the group was essential in alleviating bullying brought by external parties, against their affiliated group.

Limitations

Since LIWC is a rather new science in social and psychological research, there is a broad interpretation in conceptual meaning. Dictionaries and algorithms designed in LIWC were developed over a broad range of topics and writings (Humphreys & Wang, 2017). Some of the interpretations might be actionable, while other associations might prove spurious in nature. Text analysis now provides substantial data for statistical computing. However, analysis should be part of a methodology that references prior study and theories. The rigor of quantitative testing remains in need of the proper operationalization and framing of constructs. Repeated testing with other students from other population groups is required before results can be generalized across a larger population.

References

Aber, M. S., Dutta, U., Neville, H. A., Spanierman, L. B., De La Rosa, B., & Communities, L. L. (2010). Fostering Diversity, Dialogue and Democracy in the Intersections Living Learning Community at the University of Illinois. Implementing Diversity, 187.

Barefoot, B. O., & Fidler, P. P. (1992). National Survey of Freshman Seminar Programming, 1991. Helping First Year College Students Climb the Academic Ladder. The Freshman Year Experience: Monograph Series Number 10. Retrieved from https://eric.ed.gov/?id=ED354842

Barefoot, B. O., Warnock, C. L., Dickinson, M. P., Richardson, S. E., & Roberts, M. R. (1998). Exploring the Evidence: Reporting Outcomes of First-Year Seminars. The First-Year Experience. Volume II. Monograph Series, Number 25. National Resource Center for the First-Year Experience and Students in Transition, 1629 Pendleton St. Retrieved from https://eric.ed.gov/?id=ED433742

Craig, W. M. (1998). The relationship among bullying, victimization, depression, anxiety, and aggression in elementary school children. Personality and Individual Differences, 24(1), 123–130.

Ellett, T., & Schmidt, A. (2011). Faculty Perspectives on Creating Community in Residence Halls. Journal of College & University Student Housing, 38(1), 26–39.

Farrington, D. P. (1993). Understanding and Preventing Bullying. Crime and Justice, 17, 381–458.

Gardner, J. N. (1986). The Freshman Year Experience. College and University, 61(4), 261–274.

Garland, T. S., Policastro, C., Richards, T. N., & Miller, K. S. (2017). Blaming the Victim: University Student Attitudes Toward Bullying. Journal of Aggression, Maltreatment & Trauma, 26(1), 69–87.

Gladden, G., Vivolo-Cantor, Hamburger, & Lumpkin. (2014). Bullying Surveillance Among Youths: Uniform Definitions for Public Health and Recommended Data Elements. Retrieved from https://calio.dspacedirect.org/handle/11212/1303

Humphreys, A., & Wang, R. J.-H. (2017). Automated Text Analysis for Consumer Research. Journal of Consumer Research. https://doi.org/10.1093/jcr/ucx104

Kahn, J. H., Tobin, R. M., Massey, A. E., & Anderson, J. A. (2007). Measuring Emotional Expression with the Linguistic Inquiry and Word Count. The American Journal of Psychology, 120(2), 263–286. https://doi.org/10.2307/20445398

Kokkinos, C. M., Baltzidis, E., & Xynogala, D. (2016). Prevalence and Personality Correlates of Facebook Bullying Among University Undergraduates. Comput. Hum. Behav., 55(PB), 840–850. https://doi.org/10.1016/j.chb.2015.10.017

Kouzes, J. M., & Posner, B. Z. (1993). Credibility: How leaders gain and lose it, why people demand It. San Francisco, CA: Jossey-Bass.

Kuh, G. D., & Vesper, N. (1997). A comparison of student experiences with good practices in undergraduate education between 1990 and 1994. The Review of Higher Education, 21(1), 43–61.

Kyriacou, C., Mylonakou-Keke, I., & Stephens, P. (2016). Social pedagogy and bullying in schools: the views of university students in England, Greece and Norway. British Educational Research Journal, 42(4), 631–645. https://doi.org/10.1002/berj.3225

Murphy, R. (1989). Freshman Year Enhancement in American Higher Education. Journal of The First-Year Experience & Students in Transition, 1(2), 91–101.

Newman, M., Pennebaker, J. W., Berry, D., & Richards, J. (2003). Lying Words: Predicting Deception from Linguistic Styles. Personality and Social Psychology Bulletin, 29(5), 665–675. https://doi.org/10.1177/0146167203029005010

O’Connell, P., Pepler, D., & Craig, W. (1999). Peer involvement in bullying: insights and challenges for intervention. Journal of Adolescence, 22(4), 437–452. https://doi.org/10.1006/jado.1999.0238

Pennebaker, J. W. (1993). Putting stress into words: health, linguistic, and therapeutic implications. Behaviour Research and Therapy, 31(6), 539–548.

Pennebaker, J. W., Boyd, R. L., Jordan, K., & Blackburn, K. (2015). The Development and Psychometric Properties of LIWC2015. Retrieved from https://repositories.lib.utexas.edu/handle/2152/31333

Pike, G. R. (1999). The effects of residential learning communities and traditional residential living arrangements on educational gains during the first year of college. Journal of College Student Development. Retrieved from http://psycnet.apa.org/psycinfo/1999-05202-006

Salmivalli, C., Lagerspetz, K., Björkqvist, K., Österman, K., & Kaukiainen, A. (1998). Bullying as a Group Process: Participant Roles and Their Relations to Social Status Within the Group. Aggressive Behavior, 22, 1–15. https://doi.org/10.1002/(SICI)1098-2337(1996)22:1<1::AID-AB1>3.0.CO;2-T

Silva, A. (2016). Can Leaders of Public Companies Make Them Innovative? Journal of Leadership Studies, 10(1), 64–65. https://doi.org/10.1002/jls.21449

Spanierman, L. B., Soble, J. R., Mayfield, J. B., Neville, H. A., Aber, M., Khuri, L., & De, L. R. B. (2013). Living Learning Communities and Students’ Sense of Community and Belonging. Journal of Student Affairs Research and Practice, 50(3), 308–325. https://doi.org/10.1515/jsarp-2013-0022

Tausczik, & Pennebaker. (2010). The Psychological Meaning of Words: LIWC and Computerized Text Analysis Methods. Journal of Language and Social Psychology, 29(1), 24–54. https://doi.org/10.1177/0261927X09351676

Webber, M. A. (2017). Bullying: University Students Bring a Moral Perspective to Middle School Students. Journal of Education and Learning, 6(3), 157.

Figure 1. Mean scores psychological and social summary variables

Table 1. Highest rank excerpts LIWC psychological and social summary variables (0 to 100 point scale)