Identifying elementary school student opinion leaders for engagement in evidence-based program adaptation and implementation

Abstract

Evidence-based programs, such as bullying prevention, often demonstrate disappointing outcomes when widely disseminated. Engaging opinion leaders – those individuals whom others emulate and go to for advice – in the process of adaptation and implementation may improve outcomes. However, opinion leaders have the most influence on individuals who are similar to them, making social groups an important consideration in opinion leader recruitment. In a series of studies we examined the social groups of childhood and whether teacher nominations can be used to identify opinion leaders within these groups. In Study 1, students (N = 35) and school staff members (N = 23) reported on social groups at their school. Results suggest four predominant social groups (Elites, Athletes, Academic, and Deviants), and that students and adults are equally likely to identify these groups. In Study 2, students (N = 29) and school staff members (N = 10) identified opinion leaders from each of the four predominant social groups. Student and school staff members agreed on the primary opinion leader within the Elites, Athletes, and Academics groups, but identified different opinion leaders from the Deviants group. We conclude with relevance of these findings to schools and suggestions for further research.

Keywords

bullying crowds elementary school students evidence-based opinion leaders social diffusion social groups sociometric analysis

Many children face problems in schools, both academic and non-academic. A variety of school-based interventions have demonstrated the ability to positively affect these problems in controlled efficacy studies; however, wide-scale dissemination often fails to replicate outcomes (Ringeisen, Henderson, & Hoagwood, 2003). This disparity between the promise of research and the real world is known as the research to practice gap.

One promising approach to mitigating the research to practice gap involves engaging student ‘opinion leaders’ in the process of adapting and implementing evidence-based interventions. Opinion leaders are individuals from whom others take advice and whose behavior others emulate (Hansen & Hallum-Hansen, 2005). However, the influence of opinion leaders is generally limited to individuals who are similar to them in salient characteristics (Fisher & Misovich, 1990). Consequently, understanding what types of social groups exist within a particular type of network – in this case, elementary schools – requires consideration. Failing to engage opinion leaders associated with existing predominant social groups will likely limit the effectiveness of their participation. Engagement of opinion leaders has been shown to improve the adoption of healthy behaviors among adults and adolescents (Valente & Pumpuang, 2007). Engagement of child opinion leaders may be important for school-based programs targeting childhood problems.

Sparse research exists to guide the identification of opinion leaders among young children. To advance the potential of opinion leader strategies in the adaptation and implementation of evidence-based interventions in schools, this research was conducted to (1) better understand the social groups of elementary school children, and (2) examine teacher nomination as a strategy for identifying elementary school student opinion leaders among those social groups. Bullying prevention is used to demonstrate one potential application of opinion leaders.

Bullying and bullying prevention

Bullying is one of the most significant problems children face. Bullying is a widespread form of aggression that peaks between late childhood and early adolescence (Wang, Iannotti, & Nansel, 2009). Approximately 21% of American sixth grade students report being physically bullied, 45% report being verbally bullied, and 50% report being relationally bullied in the past two months (Wang et al., 2009). In a recent meta-analysis by Modecki, Minchin, Harbaugh, Guerra, and Runions (2014) that examined 80 studies of bullying perpetration and victimization, they found that 34.5% of students were reporting perpetration of bullying and 36% were reporting victimization. Similar rates have been reported in diverse countries and cultures, making this an international problem (Carney & Merrell, 2001). Students who are bullied have higher rates of academic, social, and psychological problems than their peers (Glew, Fan, Katon, & Rivara, 2008; Hawker & Boulton, 2000).

Bullying prevention is one area in which evidence-based interventions have demonstrated disappointing results when widely disseminated. Overall, bullying prevention programs and approaches have produced modest results, with many studies of real world implementation finding small or no positive effect (Merrell, Gueldner, Ross, & Isava, 2008). For example, the Olweus Bullying Prevention Program (OBPP), arguably the most popular evidence-based bullying prevention program, was associated with large reductions in bullying in Norway where it was developed, but has failed to demonstrate similar results in the US and other Western countries (Bauer, Loranzo, & Rivera, 2007; Olweus & Limber, 2010). While evidence suggests that bullying has decreased in several countries over the past few decades, it remains a significant problem for many children around the world (Rigby & Smith, 2011).

Engaging children in interventions

The diminished effectiveness of bullying prevention programs when widely disseminated may have to do with acceptability of the programs in new settings. Programs are most relevant to the cultures and communities in which they were developed; adaptation and endorsement by local leaders may influence acceptability in other communities (Castro, Barrera, & Martinez, 2004; Flay, 1986). Partnering with those individuals who were once thought of solely as the recipients of programming has shown promise as a strategy for supporting implementation of successful programs in novel settings (Fixsen, Naoom, Blase, Friedman, & Wallace, 2005). Incorporating consumer perspectives leads to programs that are more socially and culturally relevant, which in turn leads to greater sustainability and effectiveness (Alcantara, Harper, & Keys, 2015; Durlak, Weissberg, Quintana, & Perez, 2004; Gosin, Dustman, Drapeau, & Harthun, 2003).

Increasingly, children are being viewed as important stakeholders in the development and implementation of community-based programs (Gallagher, 2004). Youth may better understand the issues, have more enthusiasm for the issues, and be more creative in their approach to addressing the issues than adults (Pain, Francis, Fuller, O’Brien, & Williams, 2002). Furthermore, interventions designed and implemented by youth may be more appealing and acceptable to other youth than those developed without youth input (Flicker et al., 2010; Harper & Carver, 1999). Students can serve as designers, adapters, program champions, data collectors – all in addition to their roles as program recipients.

While some evidence-based bullying prevention programs engage educators in customizing the program to the school, students are rarely engaged in this process (Gibson, Flaspohler, & Watts, 2015). Gibson and colleagues (2015) found that a program that engaged a small group of fifth grade students from three schools in planning, implementing, and evaluating bullying prevention efforts was associated with improvements in students’ perception that peers and adults at school will intervene when bullying occurs. Paluck, Shepherd, and Aronow (2016) randomly assigned 54 middle schools to a control group or to receive an intervention in which a small group of students planned and intervened to prevent peer conflict. Results suggest reduced conflict-related disciplinary reports and increased peer discussion about how to reduce conflict in intervention schools compared to control schools.

Whereas Gibson et al. (2015) had teachers identify opinion leaders from different peer groups to participate in the intervention, Paluck, and colleagues (2016) randomly selected students to participate. Using social network analysis to identify individuals with many social connections (‘seed students’), Paluck et al. (2016) found that the higher the proportion of ‘seed’ students who participated in planning and implementing the intervention, the greater the decline in peer conflict. The seed students were more effective than less connected students in changing social norms. Engaging students in bullying prevention planning and implementation may be useful and is enhanced through participation of influential students. The literature on opinion leaders further supports this idea.

Opinion leaders

The spread of innovation (e.g., a product, an idea, a behavior, or an attitude) occurs within a social network (Rogers, 1983). Opinion leaders may act as ‘gatekeepers’, blocking or promoting adoption of innovation, depending on whether they participate in the innovation or not (Valente, 2010). Other social network members seek advice from and emulate the behaviors of opinion leaders, so they play a critical role in communicating and demonstrating the value of an innovation (Hansen & Hallum-Hansen, 2005). However, the influence of opinion leaders is constrained within social groups; opinion leaders yield greater influence when they share similar characteristics with other members, such as age and lifestyle (Fisher & Misovich, 1990).

The use of opinion leaders has strengthened health promotion efforts among adolescents and adults, including community HIV risk reduction, tobacco prevention in schools, mammography screenings, and other areas (Jaganath, Gill, Cohen, & Young, 2012; Valente & Pumpuang, 2007). A study by Kelly et al. (1991) looked at key opinion leaders and their influence on HIV prevention in adults. The study found that when opinion leaders acted as behavioral endorsers, they promoted positive behavior change among their peers, which led to a decreased risk for HIV infection. Other studies have found similar positive behavior change through the use of key opinion leaders in HIV prevention efforts (e.g., Amirkhanian et al., 2003; Laumann and Youm, 1999). Furthermore, Perry et al. (2003) found that a program that utilized peer opinion leaders had a greater effect on preventing tobacco use among seventh grade students than the same program without peer opinion leaders.

Identifying Opinion Leaders

No evidence exists to suggest that programs can create key opinion leaders by helping individuals develop necessary characteristics (Smith, 2005). Opinion leaders must be identified and recruited using one of several strategies. Valente and Pumpuang (2007) reviewed close to 200 studies and identified ten methods for identifying opinion leaders and the advantages and disadvantages of each (See Supplemental Materials for table summarizing study results). The most precise methods for identifying key opinion leaders – those that provide a mapping of an entire social network (i.e., sociometric methods) – are labor intensive and time consuming. Identification methods that are easier to employ rely on only a few individuals during the identification process. These individuals must be knowledgeable about the network, its subgroups, and the opinion leaders associated with each group (Valente & Pumpuang, 2007).

Despite evidence suggesting a correlation between the influence of opinion leaders and the similarities leaders share with other members of their social network (Fisher & Misovich, 1990), studies commonly neglect investigations of existing network subgroups – groups that share similar characteristics and differentiate themselves from other groups. This study fills a gap in the literature by examining the network subgroups of children, and how to identify opinion leaders from these groups.

Childhood peer groups

Most of the empirical research conducted on youth identification of peer groups has been conducted with adolescent, not child, participants, and has used the word ‘crowd‘ to refer to these groups. Sussman, Pokhrel, Ashmore, and Brown (2007) conducted a meta-analysis of 44 peer-reviewed studies on child and adolescent crowds and found only four studies that included pre-adolescents. Of these, only one targeted elementary school children (Dubow & Cappas, 1988), and this study focused on the relationship between status (e.g., popular, rejected) and adjustment, not how children define and describe existing crowds.

Five predominant crowd categories differentiated by shared values and beliefs, musical and clothing preferences, and a mutual interest in specific activities were found consistently across reviewed studies (Sussman et al., 2007). ‘Elites’ enjoy high peer status, are socially involved, and are somewhat engaged academically. ‘Athletes’ are also high in peer status, are socially involved, but are only slightly involved in academics. ‘Deviants’ enjoy some peer status, are somewhat socially involved, and rebel against school norms and expectations; they are not typically engaged academically. ‘Academics’ are highly academically engaged, enjoy some peer status, but are relatively socially uninvolved. ‘Others’ tend to be relatively low in peer status, are socially uninvolved, and are academically disengaged. These crowds are not found in every school, though their presence is the norm (Urberg, Değirmencioğlu, Tolson, & Halliday-Scher, 1995), and they do not always have the same labels across schools (Sussman et al., 2007).

Crowd affiliation has been used to improve public health interventions. Berger and Rand (2008) found that college students were less likely to select junk food items from a menu when they were informed that the food was commonly selected by members of a disaffiliated crowd. Similarly, peer group targeted anti-smoking advertisements are more efficacious than non-peer group targeted advertisements at producing anti-smoking beliefs (Moran & Sussman, 2014). Relevant to bullying, Pokhrel, Brown, Moran, and Sussman (2010) found identification with high-risk crowds (e.g., ‘Druggies’ or ‘Goths’) predicted higher relational and physical aggression, and identification with high-status crowds (e.g., ‘Jocks’ and ‘Populars’) predicted higher relational aggression compared to identification with ‘Average’ crowds.

The current studies

Engaging child opinion leaders in planning and implementing interventions, such as bullying prevention efforts, may help to improve uptake of a program or intervention through increasing acceptability among students. Social groups are an important consideration when recruiting opinion leaders because their influence is generally restricted to similar individuals (Fisher & Misovich, 1990). We conducted two studies to provide preliminary information to guide elementary schools in identifying children who are opinion leaders.

In Study 1, we examined the crowds identified by elementary school students and school staff members working with them. We wanted to know: Do elementary school students and elementary school staff members identify the same crowd categories in their schools as Sussman and colleagues (2007) identified among adolescents (Sussman et al., 2007)? If so, do students and school staff members identify these crowds with similar frequency across schools, or are these crowds more common in some schools than others? Finally, do students and school staff members identify these crowds with similar frequency to each other, or does one group identify these crowds more readily than the other?

In Study 2 we examined elementary school students and staff members’ identification of opinion leaders from the four predominant crowds (Sussman et al., 2007). We wanted to know: Do elementary school students and school staff members identify the same opinion leaders from each of the predominant social groups as each other, or do they see different students as opinion leaders? This question is important because if teachers are able to identify the same opinion leaders as their students, then resource-intensive methods (i.e., sociometric methods) could be substituted with more feasible methods (i.e., teacher nominations) of identifying elementary school student opinion leaders for engagement in evidence-based program adaptation and implementation.

Study 1

Method

Participants

Participants were from three elementary schools in a rural, Midwestern school district. Schools were chosen due to the fact that they shared pre-existing relationships with several of the researchers. Within this district, 17.5% of the students are economically disadvantaged and approximately 94% are Caucasian (Ohio Department of Education, 2007). Fifth grade students were targeted because the schools had student councils comprised solely of fifth grade student representatives, providing some indication that the schools valued the input of their oldest cohort of students on matters related to events, policies and/or practices. All of the fifth grade students, as well as 25 school staff members working with fifth grade students (i.e., lunch/recess monitors, fifth grade teachers, art, music, and physical education teachers, and principals), were eligible to participate. A total of 23 school staff members (100% female) provided consent, and 35 students (54% female) received parental consent and assented to participate in the study.

Materials and procedures

The materials and procedures used with the students and school staff members were adapted from the social-type ratings (STR) methodology developed to study crowds (Brown, 1990; Cross & Fletcher, 2009). Because STR is typically used with adolescents, the instructions were reworded for use with younger participants (Flesch-Kincaid Grade Level Readability score of 5.6, indicating that instructions were comprehendible to the average fifth grade student in the US). To obtain participant perspectives on existing crowds, STR instructions (oral and written) began by introducing the idea that there are many different types of peer groups. Then, participants were given a stack of index cards and asked to write the names of groups of fifth grade students at their school on separate cards. Finally, participants were prompted to write characteristics of each group on the back of the associated index card. Fifth grade participants completed the study in small groups (five to ten students). School staff members within each school met as a group for one data-collection session. In all sessions, participants had sufficient space for privacy. Participants were informed that there were no right or wrong answers and were permitted to discontinue participation at any time. Each session lasted approximately 15 minutes. A University Institutional Review Board reviewed and approved these procedures and methods.

Data analysis

The cards generated by students and adults were analysed via a card sorting process to determine whether predominant social groups (Elites, Athletes, Deviants, and Academics) were identified by participants. In the initial stage of data analysis, the lead researcher examined all index cards to determine which ones clearly reflected one of the four primary crowd categories identified by Sussman et al. (2007; see Table 1). If the fundamental theme of a particular card was related to high sociability or high social status (e.g., ‘Popular’ and ‘Social’), it was determined to belong to the Elite category. If the fundamental theme was related to academic success, studiousness, exceeding expectations, or high intelligence (e.g., ‘Nerd’, ‘Teacher’s Pet’, and ‘Bookworm’), it was determined to belong to the Academic category. If the fundamental theme was related to involvement in particular sports or athletic ability (e.g., ‘Sports’, ‘Football’, ‘Jock’, and ‘Athlete’), it was determined to belong to the Athlete category. If the fundamental theme was related to rule-breaking behavior, disruptive behavior, or noncompliance (e.g., ‘Troublemaker’, ‘Rule Breaker’, and ‘Bully’), it was determined to belong to the Deviant category. When the name on a card was considered to be synonymous with one of the expected categories, but the description contained information that might also be associated with another category, the card was placed in the category that matched the name.

Table 1.

Frequency distribution for variables in Study 1.

	Frequency (N = 204)	Percentage
Label
Elite	44	21.6
Academic	46	22.5
Athlete	50	24.5
Deviant	64	31.4
School
School 1	68	33.3
School 2	66	32.4
School 3	70	34.3
Participant
Student	119	58.3
Adult	85	41.7

When the fundamental theme of a particular card was unclear or reflected multiple categories, it was subjected to further analysis. Ten research assistants used the online software program Websort (Information Architecture, 2011) to assign the remaining cards to Elite, Athlete, Academic, Deviant, or an Other category. Cards were sorted as Other if they a) belonged to a category other than Elite, Athlete, Academic, or Deviant, or b) were unable to be categorized (e.g., reflected more than one category). Each card was sorted by five of the ten research assistants. Cards sorted with at least 80% inter-rater agreement (i.e., four of the five research assistants agreed) as Elite, Athlete, Academic, or Deviant were determined to belong to that category. Cards sorted with at least 80% inter-rater agreement as Other were further examined by the lead researcher to determine if there were any unexpected predominant social groups (groups other than Elite, Athlete, Academic, or Deviant that constituted at least 10% of the cards created by students and/or school staff members). Cards that were not categorized with at least 80% inter-rater agreement were eliminated from further analyses.

As a quality inspection measure, 13 cards (10%) that were initially sorted by the lead researcher were randomly selected and sorted again by five research assistants using the Websort inter-rater reliability analysis. Ten of these cards were categorized consistently with their initial sort by 100% of research assistants, while the remaining three cards were categorized consistently with their initial sort by 80% of research assistants. This suggests that the initial sorting decisions made by the lead researcher were comparable to those made by the ten research assistants, supporting the use of this initial screen in the data categorization process.

After sorting was complete, non-parametric tests (chi square tests of independence) were conducted to determine whether students and school staff members identified predominant social groups with the same frequency across schools and to determine whether students and school staff members identified the predominant social group categories with the same frequency.

Results and discussion

Study participants generated 321 crowd group cards. Over half (N = 175, 58%) were generated by students, and the remaining (N = 146, 41%) were generated by adults. In the initial stage of data analysis, 134 cards were sorted as belonging to the Elite, Academic, Athlete, or Deviant crowd categories. Of the remaining 187 cards, 70 were sorted as belonging to Elite, Academic, Athlete, or Deviant social group categories after using the Websort program to establish interrater-reliability (see Supplemental Materials, Table 1). In total, 204 (64%) cards were determined to correspond to one of the four expected crowd categories. Of these, 119 (58%) were generated by students and 85 (42%) were generated by adults. Descriptive statistics are presented in Table 2.

Table 2.

Chi square test results for groups by school, Study 1.

	School
	1	2	3	χ²
Elite	10 (−1.93)	12 (0.07)	22 (1.79)	22.32***
Academic	11 (2.01)	9 (1.82)	1 (−3.59)
Athlete	12 (1.35)	1 (−3.09)	14 (1.50)
Deviant	7 (−0.96)	10 (1.35)	10 (−0.30)

Note: Adjusted standardized residuals appear in parentheses below crowd group frequencies.

***

p = 0.001.

A total of 117 cards (36%) failed to correspond to one of the four expected crowd categories with sufficient inter-rater agreement. Of these, 28 cards (24%) were placed into the Other category with at least 80% inter-rater agreement (e.g., ‘The Pokemon Group: Boys, talk about Pokemon and their games, dress regularly and usually swing and talk at recess’). Some of these cards seemed to group together well; however, each cluster was too few in number to constitute an additional predominant crowd. Thus, it was determined that study participants did not identify any unexpected, predominant crowds.

Chi square analysis revealed that the groups identified by the student were not independent of the school the student attended, χ²(6) = 22.32, p = .001 (see Table 2) indicating that the students identified the four different crowds with different frequencies across schools. The adjusted standardized residuals for the chi square test reveal that students from School 1 generated more Academic cards, students from School 2 generated fewer Athlete cards, and students attending School 3 identified fewer Academic cards than expected.

Chi square analysis revealed that the crowds identified by the adults in the study were independent of school (χ² p > 0.05), indicating that adults identified the four different crowds with comparable frequencies across schools. Additionally, the crowds identified were independent of participant (student versus adult; χ² p > 0.05). This indicates that adults and students identified the four different crowds with comparable frequencies.

Study 2

Method

Participants

Participants were from a grade school in New England (K-8) where 12% of the students are economically disadvantaged and approximately 90% are Caucasian (Massachusetts Department of Primary and Secondary Education, 2010). The school was chosen due to the fact that it shared a pre-existing relationship with the primary researcher. All of the school’s 34 fifth grade students, and 13 school staff members working with fifth grade students (i.e., lunch/recess monitors, fifth grade teachers, art, music, and physical education teachers, and principals) were eligible to participate. Ten school staff members (80% female) and 29 fifth grade students (55% female) participated.

Materials and Procedures

Materials and procedures were adapted from the sociometric key opinion leader identification method (Valente & Pumpuang, 2007; Zakriski, Seifer, Sheldrick, Prinstein, & Dickstein, 1999). Each participant received a stack of index cards with each fifth grade student’s name printed on a card. The bottom of the card contained an identification number for the named student, as well as a place for participants to indicate if that student was ranked as a group opinion leader. Participants were introduced to the idea that there are many different peer groups. Six sheets of paper were placed on each participant’s desk. These were labeled: ‘Popular’, ‘Smart’, ‘Sport’, ‘Troublemaker’, ‘Other’, and ‘I don’t know’. The adjectives, ‘popular’, ‘smart’, and ‘sport’ were the most common adjectives used by child participants to describe the Elite, Academic, and Athlete groups, respectively, in Study 1. While the adjective ‘bad’ was most commonly used to describe the Deviant group, ‘troublemaker’ was chosen because it was thought to have less negative connotations and was the second most frequently used adjective. Descriptions did not accompany names to limit bias in responses.

The participants were instructed to sort each index card onto the sheet of paper that corresponded to the named child’s peer group. Students categorized all fifth grade students, as the school had only two fifth grade classrooms and the two classrooms spent significant time together. After sorting the cards, participants were asked to rank the three most influential students in each of the four predominant groups. This was based on research that suggests that providing nominators with the opportunity to identify multiple individuals yields more reliable and valid sociometric data (Terry, 2000). To assure anonymity of the named students, participants then cut off the part of the index cards with the students’ names. A University Institutional Review Board reviewed and approved these procedures and methods.

To complete this task, student participants met in small groups ranging from two to seven students. School staff participants met collectively with the lead researcher. All instructions were read aloud and also provided in writing. All data collection sessions allowed participants enough space so that their answers would remain private. Each session lasted approximately 15 minutes.

Data Analysis

Approximately one-third of student participants and at least two adults reported difficulty ranking key opinion leaders beyond the first, most influential individual. For this reason, analyses focused exclusively on the highest ranked key opinion leaders. We summed the number of crowd-specific key opinion leader nominations that each student received from students and staff members. Then, two sample z-tests for differences between proportions were conducted to examine if similar proportions of students and adults nominated the same crowd-specific key opinion leaders.

Results and discussion

Each of the 29 student participants assigned 33 grade-level peers to one of the four predominant groups or categorized them as ‘Other’ or ‘I Don’t Know’. Of the 957 group assignments made by students, 181 (18%) were categorized as ‘I Don’t Know’. Descriptive statistics for the remaining 776 (81%) assignments are in Table 3. The results of the group assignment task in Study 2 lend further support of the findings of Study 1 because 603 (77%) of the 776 group assignments were categorized as belonging to one of the four predominant crowd groups, and each of the predominant crowd groups received more than 10% of nominations. While a substantial number of students (22.3%) were assigned to the ‘Other’ category, results of Study 1 suggest that this category likely consists of many, less prominent social groupings.

Table 3.

Descriptive statistics for student participants’ use of crowd groups, Study 2.

Crowd Group	N	Range	Mean	Percentage
Popular Kids	177	0–13	6.10	22.8
Sports Kids	102	0–7	3.52	13.1
Smart Kids	175	2–18	6.03	22.6
Troublemakers	149	1–9	5.14	19.2
Other	173	1–16	5.97	22.3

Analyses using two sample z-tests for differences between proportions revealed that the top ranked student in each of the following crowd groups received a similar proportion of both student and adult nominations: Popular Kids crowd (z = −0.34, p = 0.48, two tailed), Sports Kids crowd (z = −0.34, p = 0.48, two-tailed), and Smart Kids crowd (z = −0.28, p = 0.42, two-tailed). Different members of the Troublemakers crowd received the majority of adult and student nominations. Student participants nominated Trouble Maker 1 (TM1) as the most influential member of the Troublemakers crowd with the greatest frequency (16 of 29, 55%). Adult participants, however, nominated Trouble Maker 2 (TM2) as the most influential member of the Troublemakers crowd with the greatest frequency (5 of 10, 50%); TM1 received the second greatest number of adult nominations (2 of 10, 20%). A z-test for two proportions was used to determine if the proportion of student nominations of TM1 was significantly different from the proportion of adult nominations of TM1. The result was significant (z = 1.56, p = 0.014, two-tailed). None of the students nominated as most influential of one crowd were nominated as the most influential of another crowd.

Discussion

We examined a method of identifying child opinion leaders for engagement in the planning and implementation of school-based programs. We were interested in the existence of four predominant crowd groups among fifth grade students, and examining whether fifth grade students and adults would identity the same opinion leaders from each crowd. If these crowds exist and adults can accurately identify the opinion leaders from them, then schools can use adult nomination of opinion leaders from each crowd to select influential students for engagement in school-based programs, such as bullying prevention, in order to improve acceptance and relevance of the program among students.

In Study 1 we asked: Do elementary school students and elementary school staff members identify the same crowd categories in their schools as Sussman and colleagues (2007) identified among adolescents (Sussman et al., 2007)? If so, do students and school staff members identify these crowds with similar frequency across schools, or are these crowds more common in some schools than others? Finally, do students and school staff members identify these crowds with similar frequency to each other, or does one group identify these crowds more readily than the other?

With regard to the first and third questions, the results of Study 1 suggest that the names/descriptions of crowd groups generated by fifth grade students correspond to those frequently cited in the literature for adolescents (Sussman et al., 2007), and that the names/descriptions of crowds generated by adults resembled those generated by their students. The majority of crowds identified by students (68%) and adults (58%) corresponded to the Elite, Athlete, Academic, or Deviant groups. Neither students nor adults identified predominant crowds other than those the four frequently cited in the literature on adolescents. These findings suggest that crowds previously thought to emerge in adolescence are present as early as fifth grade and are visible to both children and adults at school.

With regard to the second question, Study 1 revealed some variability in the prevalence of crowd groups across schools. Students attending School 3 used fewer Academic names/descriptions, students from School 2 used fewer Athlete names/descriptions, and students from School 1 used more Academic names/descriptions than expected. This is consistent with previous research suggesting that adolescent crowd groups are not present in every school (Urberg et al., 1995), and when present, they may not have the same names (Sussman et al., 2007).

In Study 2, we wanted to know if elementary school students and school staff members would identify the same opinion leaders from each of the predominant crowd groups as each other, or if they see different students as opinion leaders. Results suggest that adults were able to correctly identify the same crowd-specific key opinion leaders as their students in three of the four crowds: Elites, Athletes, and Academics. This is consistent with research that suggests that elementary school teachers exhibit reliable agreement with their students in the identification of reciprocal friendships (Gest, 2006), and supports the idea that teachers and their students similarly understand childhood social networks.

It is worth noting that a relatively large number of both students and adults in Study 2 identified the same Academic as the most influential (82.76% and 80%, respectively), but students and adults identified the same Elite and Athlete with relatively less consensus (approximately 40% of each participant group). The indicators of being an Academic (e.g., high grades and test scores, class assignments/presentations that exceed expectations) may be more indisputable and more likely to be observed by all students and adults.

While the adult nomination strategy resulted in accurate identification of most of the crowd-specific key opinion leaders (when compared to student identification), adults failed to accurately identify the same opinion leader of the Troublemakers crowd as their students. One possible explanation is that students and adults conceptualized Troublemakers differently. Students may have identified the student who lacks social skills, and, as a result, commonly disrupts social interactions during unstructured times of the day (e.g., lunch, recess). Adults, on the other hand, may have identified the student who was the most disruptive to classroom instruction and created the greatest conflict with respect to their job responsibilities.

A potential way to improve the correspondence between adult and student nominations might be to encourage adult participants to take the perspective of their students. For example, the adult version of the protocol used in the current project could be amended to say, ‘The object is not necessarily to identify who is the most trouble for you, or disrupts your class the most, but who other students in the Troublemakers group emulate’.

In addition, it is important to note that not all students neatly fit into one of Sussman’s four predominant crowds. In Study 1, 15% of the crowd names/descriptions generated by participants did not correspond to the Elite, Athlete, Academic, or Deviant groups. While some of the alternative crowd names/descriptions seemed to cluster, each cluster was too few in number to constitute a predominant crowd. In Study 2, 22.29% of student names were sorted into the Other group rather than the Elite, Athlete, Academic, or Deviant groups. Together, these findings suggest that a substantial number of fifth grade students are associated with alternative social groups, or perhaps no group. As a result, recruiting only opinion leaders from the Elite, Athlete, Academic, or Deviant groups may mean that a significant proportion of students are without an opinion leader representative engaged in the project. Further research needs to be conducted to determine if engagement of influential members of predominant crowds influences students who are associated with less predominant crowds or no crowds.

Overall, Study 1 suggests that children may naturally form standard crowds at younger ages than previously thought, while Study 2 suggests that teachers can identify the opinion leaders from these crowds. Schools can harness the potential of these crowds and their opinion leaders by having teachers nominate opinions leaders from each crowd for participation in planning and implementing programs aimed at student issues, such as bullying prevention. Our study suggests that adult nominations are a reasonable substitute methodology for peer nominations, which are labor-intensive and commonly elicits apprehension among school personnel and students’ families (Bell-Dolan, Foster, & Sikora, 1989). Adult nominations are likely to be particularly effective for projects in which the aim is to engage students from the Elite, Athlete, and Academic crowd groups.

Two previous studies by Gibson et al. (2015) and Paluck et al. (2016) explored student engagement in planning and implementing bullying prevention programs. The students in these studies worked with adults to assess specific needs at their school and design and implement prevention efforts to address those needs. Paluck et al. (2016) suggest that engaging influential peers is more effective than engaging randomly selected students, and the teacher nomination method evaluated in the current study could be used to identify these students.

Another way to use the nomination strategy described in this paper is to increase the domain specificity of the groups from which opinion leaders are identified. In the case of bullying prevention, adults may consider specifically seeking out opinion leaders that represent the various participant roles students play in bullying: Bully, victim, bully assistant, reinforcer, defender, and outsider (Salmivalli, Lagerspetz, Björkqvist, Österman, & Kaukiainen1996). Salmivalli, Huttunen, and Lagerspetz (1997) found that children who behave similarly in bullying situations tend to belong to the same peer groups. Thus, it may be possible to also identify the opinion leader for each role in bullying and have those students come together to work with adults on planning and implementing prevention efforts that are geared towards other students who take the same participant role in bullying.

As a final note, we caution that in addition to considering which students are most influential, researchers and practitioners may also need to consider which students have the requisite skills to contribute to the program planning and engagement process. Organizational, interpersonal, communication, and problem-solving skills may be necessary and are not assessed using the adult nomination process described in this article. If the influential students struggle with the project’s demands, adults may use wish to identify and recruit additional students who demonstrate the desired skills and could work with the opinion leaders to strengthen the project.

Limitations and future directions

It is important to acknowledge methodological limitations when considering the project’s findings. First, it is possible that the instructions in Study 1 may have influenced participant responses and encouraged them to identify descriptors of interest to the researchers. Instructions of this kind were used with seventh grade participants in similar studies (England & Petro, 1998) and were thought to be necessary because the use of open-ended formats has proven to be difficult, even among college students, when attempting to elicit ideas about between-crowd-group differences (England & Hyland, 1986, as cited in England & Petro, 1998).

Another potential limitation of Study 1 pertains to the validity of the crowd names and descriptions provided by participants. Brown (1989) argues that the crowds identified using the methodology described may not reveal actual groups, but instead social categories used to cognitively organize social information. While this may be true, participants’ ability to identify opinion leaders from the same crowds in Study 2 may provide some evidence that participants suspected the crowds actually exist.

In Study 2, students were often identifying individuals who they felt were the opinion leaders in crowds of which they were not members. A crowd’s actual members may be more attuned to their group dynamic subtleties and be better able to identify their influential members. In addition, we did not look at gender-based groups in this study by asking girls and boys to identity the groups and opinions leaders that exist among children of their same gender. Research suggests that until about age 11, children tend to prefer and spend most of their time with same gendered peers (Maccoby, 1990). Interestingly, gender did not always arise as a key characteristic in describing the groups that child identified, perhaps indicating that gender segregation was less salient in our sample than in previous research.

The examination of the correspondence between teachers’ nominations of particular students as members of particular crowds, and students’ nomination of those same students to the same crowd could have been used as a valid indicator of crowd groupings. Unfortunately, due to limitations of the methodology used in Study 2, the data did not afford the researchers with the opportunity to conduct such an analysis. Future research should rely on methods that will provide insight into the correspondence between teacher and student assignments of students to particular crowds.

A large, representative group of student raters is necessary for the most accurate crowd identification and assignment (Brown, 1990). Replication of our studies with larger and diverse samples would strengthen the findings. In particular, it is noteworthy that both Study 1 and Study 2 were conducted in relatively small schools with whom the researchers had a pre-existing relationship, and it may be that the same method for identifying social groups and opinion leaders would be less successful in a large school. Furthermore, these schools were both in the US, and international replication is suggested. Only three of the 44 studies Sussman and colleagues (2007) reviewed were conducted outside of the US, further indicating the need for this topic to be examined in diverse countries and cultures. Finally, engaging child opinion leaders in bullying prevention and other programs is supported by Gibson et al. (2015) and Paluck et al. (2016), but warrants further investigation.

Conclusion

Schools are complicated systems in which it can be difficult to introduce, implement, and sustain evidence-based programs. Often these programs demonstrate disappointing outcomes. To address this, schools may consider identifying and inviting crowd-specific student opinion leaders to be a part of adapting and implementing evidence-based programs. This study suggests that four predominant crowds exist among students as young as fifth grade and that adults can reliably determine the opinion leaders from three of these crowds. Thus, adult nomination of key opinion leaders may be used to identify and recruit student collaborators from these crowds. While engagement of childhood opinion leaders as collaborators in evidence-based programming is by no means a panacea to childhood problems in schools, it may strengthen relevance and acceptability of programs among students. Further research on this is necessary, as is replication of our studies with larger and more diverse samples.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author biographies

Christopher J. Reiger, PhD is a Program Development and Evaluation Specialist at Wediko Children’s Services, in Boston, Massachusetts, USA. Christopher is dedicated to assisting schools in the development of effective prevention and intervention programming aimed at promoting and supporting the healthy development of all children. His research interests include the involvement and empowerment of children in the development and continuous improvement of a variety of services in their school communities.

Jennifer E. Gibson, PhD is an Associate Professor and the Associate Director of Clinical Training for the School of Psychology at Xavier University, in Cincinnati, Ohio, USA. Her research is primarily focused on promotion and prevention programs in the school setting. Her interest in bullying prevention has recently led to studying bystander intervention on social media sites.

Rebecca E. Passarelli, MA is a graduate student in the Clinical Psychology program at Miami University in Oxford, Ohio, USA. She works with Dr. Paul Flaspohler in the School and Community Research Action Lab. Her research interests focus on ways to promote and evaluate positive social-emotional development in children.

Paul D. Flaspohler, PhD is an Associate Professor of Clinical Psychology at Miami University and the Director of Research and Evaluation for the Center for School-Based Mental Health Programs. In addition to applied research in community development and program evaluation, He works with community organizations and schools to identify needs and develop solutions for community problems.

References

Alcantara

Harper

G. W.

Keys

C. B.

(2015) ‘There’s gotta be some give and take’: Community partner perspectives on benefits and contributions associated with community partnerships for youth. Youth and Society 47(4): 462–485.

Amirkhanian, Y. A., Kelly, J. A., Kabakchieva, E., McAuliffe, T. L., & Vassileva, S. (2003). Evaluation of a social network HIV prevention intervention program for young men who have sex with men in Russia and Bulgaria. AIDS Education and Prevention, 15(3), 205–220.

Bauer

N. S.

Lozano

Rivara

F. P.

(2007) The effectiveness of the Olweus Bullying Prevention Program in public middle schools: A controlled trial. Journal of Adolescent Health 40(3): 266–274. doi:10.1016/j.jadohealth.2006.10.005.

Bell-Dolan

D. J.

Foster

S. L.

Sikora

D. M.

(1989) Effects of sociometric testing on children’s behavior and loneliness in school. Developmental Psychology 25: 306–311. doi:10.1037/0012-1649.25.2.306.

Berger

Rand

(2008) Shifting signals to help health: Using identity signaling to reduce risky health behaviors. Journal of Consumer Research 35: 509–518. doi:10.1086/587632.

Brown, B. (1989). Social type ratings interview manual. Unpublished manuscript. Madison, WI: University of Wisconsin, Department of Educational Psychology.

Brown

B. B.

(1990) At the threshold: The developing adolescent, Cambridge, MA: Harvard University Pressdoi:10.1016/0140-1971(92)90048-A.

Carney

A. G.

Merrell

K. M.

(2001) Bullying in schools: Perspectives on understanding and preventing an international problem. School Psychology International 22: 364–382. doi:10.1177/0143034301223011.

Castro, F. G., Barrera M., Jr, & Martinez C. R. Jr, (2004). The cultural adaptation of prevention interventions: Resolving tensions between fidelity and fit. Prevention Science, 5(1), 41–45. doi:10.1023/B:PREV.0000013980.12412.cd.

10.

Cross

J. R.

Fletcher

K. L.

(2009) The challenge of adolescent crowd research: Defining the crowd. Journal of Youth and Adolescence 38: 747–764. doi:10.1007/s10964-008-9307-6.

11.

Dubow

E. F.

Cappas

C. L.

(1988) Peer social status and reports of children’s adjustment by their teachers, by their peers, and by their self-ratings. Journal of School Psychology 26(1): 69–75. doi:10.1016/0022-4405(88)90033-7.

12.

Durlak

J. A.

Weissberg

R. P.

Quintana

Perez

(2004) Primary prevention: Involving schools and communities in youth health promotion. In: Jason

L. A.

(ed.) Participatory community research: Theories and methods in action, Washington, DC: American Psychological Association, pp. 73–86. doi:10.1037/10726-000.

13.

England, E. M., & Petro, K. D. (1998). Middle school students' perceptions of peer groups: Relative judgments about characteristics. The Journal of Early Adolescence, 18(4), 349–373.

14.

Fisher

J. D.

Misovich

S. J.

(1990) Social influence and AIDS-preventative behavior. In: Edwards

R. S.

Tindale

Heath

Posavac

E. J.

(eds) Applying social influence processes in preventing social problems, New York, NY: Plenum Press, pp. 39–70. doi: 10.1007/978-1-4899-2094-2_3.

15.

Fixsen

D. L.

Naoom

S. F.

Blase

K. A.

Friedman

R. M.

Wallace

(2005) Implementation Research: A Synthesis of the Literature. Tampa, FL: University of South Florida, Louis de la Parte Florida Mental Health Institute. The National Implementation Research Network. (FMHI Publication #231).

16.

Flay

B. R.

(1986) Efficacy and effectiveness trials (and other phases of research) in the development of health promotion programs. Preventive Medicine 15: 451–474. doi:10.1016/0091-7435(86)90024-1.

17.

Flicker

Guta

Larkin

Flynn

Fridkin

Travers

Layne

(2010) Survey design from the ground up: Collaboratively creating the Toronto Teen Survey. Health Promotion Practice 11(1): 112–122.

18.

Gallagher

C. B.

(2004) ‘Our Town’: Children as advocates for change in the city. Childhood 11(2): 251–262. doi:10.1177/0907568204043060.

19.

Gest

S. D.

(2006) Teacher reports of children’s friendships and social groups: Agreement with peer reports and implications for studying peer similarity. Social Development 15(2): 248–259. doi:10.1111/j.1467-9507.2006.00339.x.

20.

Gibson

J. E.

Flaspohler

P. D.

Watts

(2015) Engaging youth in bullying prevention through community-based participatory research. Family and Community Health 38(1): 120–130. doi:10.1097/FCH.0000000000000048.

21.

Glew

G. M.

Fan

M. Y.

Katon

Rivara

F. P.

(2008) Bullying and school safety. The Journal of Pediatrics 152(1): 123–128. doi:10.1016/j.jpeds.2007.05.045.

22.

Gosin

M. N.

Dustman

P. A.

Drapeau

A. E.

Harthun

M. L.

(2003) Participatory action research: Creating an effective prevention curriculum for adolescents in the Southwestern US. Health Education Research 18: 363–379. doi:10.1093/her/cyf026.

23.

Hansen

Hallum-Hansen

M. H.

(2005) Children as innovators and opinion leaders. Young Consumers 1: 44–59. doi:10.1108/17473610510701106.

24.

Harper

G. W.

Carver

L. J.

(1999) ‘Out-of-the-mainstream’ youth as partners in collaborative research: exploring the benefits and challenges. Health Education and Behavior 26(2): 250–265. doi:10.1177/109019819902600208.

25.

Hawker

D. S.

Boulton

M. J.

(2000) Twenty years’ research on peer victimization and psychosocial maladjustment: a meta-analytic review of cross-sectional studies. Journal of Child Psychology and Psychiatry 41(4): 441–455. doi:10.1111/1469-7610.00629.

26.

Information Architecture (2011). Websort [computer software]. Retrieved from: https://dirtarchitecture.wordpress.com/websort/.

27.

Kelly, J. A., St Lawrence, J. S., Diaz, Y. E., Stevenson, L. Y., Hauth, A. C., Brasfield, T. L., … Andrew, M. E. (1991). HIV risk behavior reduction following intervention with key opinion leaders of population: an experimental analysis. American Journal of Public Health, 81(2), 168–171.

28.

Laumann, E. O., & Youm, Y. (1999). Racial/ethnic group differences in the prevalence of sexually transmitted diseases in the United States: A network explanation. Sexually Transmitted Diseases, 26(5), 250–261.

29.

Jaganath

Gill

H. K.

Cohen

A. C.

Young

S. D.

(2012) Harnessing online peer education (HOPE): Integrating C-POL and social media to train peer leaders in HIV prevention. AIDS Care 24(5): 593–600.

30.

Maccoby, E., E. (1990). Gender and relationship: A developmental account. American Psychologist, 45, 513–520. doi:10.1037/0003-066X.45.4.513.

31.

Massachusetts Department of Primary and Secondary Education. (2010). School and District Profiles. Retrieved from http://profiles.doe.mass.edu/profiles/student.aspx?orgcode=02960005&orgtypecode=6&&fycode=2010.

32.

Merrell

K. W.

Gueldner

B. A.

Ross

S. W.

Isava

D. M.

(2008) How effective are school bullying intervention programs? A meta-analysis of intervention research. School Psychology Quarterly 23(1): 26, doi:10.1037/1045-3830.23.1.26.

33.

Modecki

K. L.

Minchin

Harbaugh

A. G.

Guerra

N. G.

Runions

K. C.

(2014) Bullying prevalence across contexts: A meta-analysis measuring cyber and traditional bullying. Journal of Adolescent Health 55(5): 602–611.

34.

Moran

M. B.

Sussman

(2014) Translating the link between social identity and health behavior into effective health communication strategies: An experimental application using antismoking advertisements. Health Communication 29: 1057–1066. doi:10.1080/10410236.2013.832830.

35.

Ohio Department of Education. (2007). Ohio School Report Cards. Retrieved from http://reportcard.education.ohio.gov/Pages/District-Report.aspx?DistrictIRN=046151.

36.

Olweus

Limber

S. P.

(2010) Bullying in school: Evaluation and dissemination of the Olweus Bullying Prevention Program. American Journal of Orthopsychiatry 80(1): 124–134. doi:10.1111/j.1939-0025.2010.01015.x.

37.

Pain

Francis

Fuller

O’Brien

Williams

(2002) Hard-to-reach young people and community safety: A model for participatory research and consultation, London: Home Office, Policing and Reducing Crime Unit.

38.

Paluck

E. L.

Shepherd

Aronow

P. M.

(2016) Changing climates of conflict: A social network experiment in 56 schools. PNAS 113: 566–571. doi:10.7910/DVN/29199.

39.

Perry, C. L., Komro, K. A., Veblen-Mortenson, S., Bosma, L. M., Farbakhsh, K., Munson, K. A., et al. (2003). A randomized controlled trial of the middle and junior high school DARE and DARE Plus programs.Archives of Pediatrics and Adolescent Medicine,157(2), 178–184. doi:10.1001/archpedi.157.2.178.

40.

Pokhrel

Brown

B. B.

Moran

M. B.

Sussman

(2010) Comments on adolescent peer crowd affiliation: a response to Cross and Fletcher (2009). Journal of Youth and Adolescence 39(2): 213–216. doi:10.1007/s10964-009-9454-4.

41.

Rigby

Smith

P. K.

(2011) Is school bullying really on the rise? Social Psychology of Education 14: 441–455. doi: 10.1007/s11218-011-9158-y.

42.

Ringeisen

Henderson

Hoagwood

(2003) Context Matters: Schools and the ‘Research to Practice Gap’ in Children’s Mental Health. School Psychology Review 32(2): 153–168.

43.

Rogers

E. M.

(1983) Diffusion of innovations, 3rd ed. New York, NY: The Free Press.

44.

Salmivalli

Huttunen

Lagerspetz

K. M.

(1997) Peer networks and bullying in schools. Scandinavian Journal of Psychology 38: 305–312. doi:10.1111/1467-9450.00040.

45.

Salmivalli

Lagerspetz

Björkqvist

Österman

Kaukiainen

(1996) Bullying as a group process: Participant roles and their relations to social status within the group. Aggressive behavior 22(1): 1–15. doi:10.1002/(SICI)1098-2337(1996)22:1<1::AID-AB1>3.0.CO;2-T.

46.

Smith

P. A. C.

(2005) Knowledge sharing and strategic capital: The importance and identification of opinion leaders. The Learning Organization 12(6): 563–574. doi: 10.1108/09696470510626766.

47.

Sussman

Pokhrel

Ashmore

R. D.

Brown

B. B.

(2007) Adolescent peer group identification and characteristics: A review of the literature. Addictive Behaviors 32: 1602–1627. doi:10.1016/j.addbeh.2006.11.018.

48.

Terry

(2000) Recent advances in measurement theory and the use of sociometric techniques. In: Cillessen

A. H. N.

Bukowski

W. M.

(eds) Recent advances in the measurement of acceptance and rejection in the peer system, San Francisco, CA: Jossey-Bassdoi:10.1002/cd.23220008805.

49.

Urberg

K. A.

Değirmencioğlu

S. M.

Tolson

J. M.

Halliday-Scher

(1995) The structure of adolescent peer networks. Developmental Psychology 31: 540–547. doi:10.1037/0012-1649.31.4.540.

50.

Valente

T. W.

(2010) Social networks and health: Models, methods, and applications, Oxford: Oxford University Pressdoi:10.1093/acprof:oso/9780195301014.001.0001.

51.

Valente

T. W.

Pumpuang

(2007) Identifying opinion leaders to promote behavior change. Health Education and Behavior 34: 881–896. doi:10.1177/1090198106297855.

52.

Wang

Iannotti

R. J.

Nansel

T. R.

(2009) School bullying among adolescents in the United States: Physical, verbal, relational, and cyber. Journal of Adolescent Health 45: 368–375. doi:10.1016/j.jadohealth.2009.03.021.

53.

Zakriski

A. L.

Seifer

Sheldrick

R. C.

Prinstein

M. J.

Dickstein

Sameroff

A. J.

(1999) Child-focused versus school-focused sociometrics: A challenge for the applied researcher. Journal of Applied Developmental Psychology 20: 481–499. doi:10.1016/S0193-3973(99)00021-0.