The Influence of Academic Tracking on Adolescent Social Networks

Abstract

This study examined adolescents’ social capital, through social network analyses (i.e., ego network analyses), in two high schools where students were placed into academic tracks adopted by the schools and shaped by disability status (i.e., general education, co-taught, segregated special education classrooms). The impact of academic tracks, as well as the impact of personal and environmental characteristics, on ego networks was examined. Differences in ego network structural characteristics (i.e., network size, density, effective size, and efficiency) across academics tracks, differences in information communication technology (ICT) use, and participation in informal social activity for students in the segregated special education track are highlighted. Implications for research and practice, including supporting the use of ICT and ego network mapping during instruction, are provided.

Keywords

social networks academic tracking social capital adolescents disability

Social capital includes the social connections and structures available to human beings that produce trust and reciprocity (Putnam, 2000). It has been defined as a tool that provides direct benefits to a person or community as he or she utilizes these connections and structures, often referred to as social networks, to achieve desired outcomes and bring benefits to the community (Coleman, 1988; Lin, 2001; Portes, 1998). Researchers have suggested that people with higher social capital have (a) greater access to education (Calvo-Armengol, Patacchini, & Zenou, 2009; Maroulis & Gomez, 2008) and employment opportunities (Briggs, 1998), (b) better physical (Cornwell, 2009; Poortinga, 2006) and mental health outcomes (Drukker, Kaplan, Feron, & van Os, 2003), and (c) higher well-being (Cattell, 2001). As a result, governments have recognized the importance of social capital by calling for its measurement when examining the economic and social health and outcomes of economies (Stiglitz, Sen, & Fitoussi, 2009).

For educational researchers, understanding the role of social capital within and across contexts (e.g., classrooms, after-school and community activities, etc.) is critical. This is particularly important for adolescents as they begin to establish their own social connections outside the family, with peers exerting greater influence (see Brown, Bakken, Ameringer, & Mahon, 2008). Adolescents also begin to spend a substantial amount of their time communicating with friends through technology (Lenhart, 2012) and engaging in leisure activities with friends (see Bartko & Eccles, 2003). All of these activities shape social networks and short- and long-term social, educational, and life outcomes.

Examining the outcomes of adolescents with disabilities through a social capital framework highlights the relational (i.e., supports-based) rather than the individualistic (deficits-based) nature of supports for successful short- and long-term outcomes. Specifically, understanding the importance of social connections and social network membership embodies the shift from simply promoting physical access to inclusive environments to a social-ecological approach where promoting quality of life by identifying support needs is the focus (World Health Organization [WHO], 2002)

Social Capital Through Network Analysis

To analyze the factors that shape social capital in adolescents with and without disabilities, systematic measures must be used. Measuring social capital can be challenging (Carolan, 2014; Scrivens & Smith, 2013) but social network analysis (SNA) has emerged as a recommended methodology (Frank, Mueller, & Mueller, 2013). In SNA, one identifies (a) individuals in a network and (b) the directed and undirected connections those individuals have. Key features of empirical SNA include (a) analyses of relationships between individuals using empirical data, (b) graphic representation of relations, and (c) mathematical computations that model social life (see Carolan, 2014; Wasserman & Faust, 1994). SNA has been used in sociology and education to examine the relationship between social capital and outcomes such as social support (Briggs, 1998; Tracy & Whittaker, 1990), community membership (Poortinga, 2006), and participation in extracurricular activities (Schaefer, Simpkins, Vest, & Price, 2011).

Social networks can be represented using whole network analysis (WHA) or ego network analysis (ENA). In WNA, researchers map connections across the social structure of an identified group with specified boundaries such as a specific classroom. Using WNA, researchers have suggested that disability status influences social network membership with students with disabilities experiencing some isolation from peer social networks or peripheral social network membership in specific contexts (e.g., a classroom; Calvo-Armengol et al., 2009; Farmer et al., 2011; Farmer, Van Acker, Pearl, & Rodkin, 1999). In ENA, researchers map the network’s structure where connections occur without bounding the network by a particular setting. This provides particularly useful information when studying adolescents, as adolescents, typically, experience many different environments that vary in the nature of social opportunities and relationships available. Furthermore, it allows for the examination of influencing factors across contexts that may not be ascertained with whole networks where boundaries are determined by researchers. Although ENA has been used less frequently, researchers have found when looking across contexts, people with disabilities, particularly those with intellectual and developmental disabilities, have limited ego networks (van Asselt-Goverts, Embregts, & Hendriks, 2013). Furthermore, variations in specific ego network characteristics, such as network size, crowd membership, and network effectiveness across contexts, are present.

Specific vocabulary is used in ENA to refer to the actors or the people included in the analysis: (a) the ego is the person whose network is examined, (b) the alters are the individuals the ego indicates as network members, and (c) the ties are the connections the alters have with the other listed alters. Structural network characteristics, in addition to descriptive information such as age or gender, are also examined. These include size and density, and the connectedness of network members, all of which indicate network breadth and constraints or available support (i.e., bonding social capital; Marsden, 2002). Measures of effective size and efficiency examine the ego’s impact on network members and are based on Burt’s (1992) structural holes theory that suggests that structural holes, places in the network where alters are not tied to other alters except through the ego, provide the ego with novel information. This is referred to as bridging social capital or the ego’s ability to transfer new information from one network to another.

Factors Affecting Social Capital

Social capital theory suggests that personal and environmental characteristics influence the development of social capital. For adolescents, who are becoming increasingly aware of their peer relationships and the need to create social status through affiliations with similar peer networks (Brown et al., 2008), how peer networks are defined and supported or not supported in schools and communities exerts a substantial influence on social capital. Personal characteristics that affect peer relationships and networks include factors like gender, disability status, and personal interests. Environmental characteristics can also influence networks and include social interaction and network development opportunities available in the environment, as well as access to technology to communicate and establish social connections. For example, when students learn in the same classroom and have access to the same academic content and experiences, this shapes their peer networks and how groups are defined. This can also shape the degree to which adolescents are included in formal (e.g., sports) and informal (e.g., watching movies) social activities with peer networks.

Social interactions with peers in inclusive general education environments, then, have the potential to affect adolescents’ social network membership, formal and informal social activity (Schaefer et al., 2011), and, potentially, access to social capital. However, students with disabilities, particularly those with the most intensive support needs, continue to be served in more restrictive and frequently less academically challenging settings (Hallinan, 1994; Harris, 2011). This limits the degree to which they are perceived as members of peer groups. In the high-school setting specifically, where academic tracking or grouping students based on common academic skills (Wouters, De Fraine, Colpin, Van Damme, & Verschueren, 2012), in some form, is a common practice, students in more restrictive tracks experience not only limited access to a challenging and rigorous academic curriculum (Hallinan, 1994; Harris, 2011) but are also less likely to be perceived as part of the networks that are created in less restrictive classes. The practice of academic tracking, then, may provide a useful explanatory variable to understand social network development above and beyond disability status, as tracking into inclusive or segregated classrooms may function to exaggerate differences between students through social comparison (Byrne, 1988).

Aside from access to inclusive environments, having access to modern means of communication is also a key environmental factor that affects how adolescents interact with peers and build social networks and social capital (Lenhart, Ling, Campbell, & Purcell, 2010). Particularly, in the last 5 years, adolescents have shifted to primarily using email, texting, cell phones, and social media (Reich, Subrahmanyam, & Espinoza, 2012). They use this technology to communicate socially and also to plan informal and formal social activities (Ling, 2010). However, youth with disabilities, particularly those with most intensive support needs, have less access and usage (Tanis et al., 2012). This is a concern as there is a growing body of work suggesting effective information communication technology (ICT) use can promote social activity and perhaps expand social networks (Carrasco & Miller, 2006; Carrasco, Miller, & Wellman, 2008; see Raine & Wellman, 2012),

Purpose

Little research has explored how adolescents with disabilities access social capital in the school context. Given the body of research that suggests adolescents with disabilities experience isolation from peer relationships and social networks (and, in turn, social capital) and how academic tracking further segregates these students, further research is needed to examine the personal and environmental factors that influence social capital development. To explore these issues, we implemented a cross-sectional survey design study to examine (a) how personal and environmental factors, beyond disability label, affect adolescent social capital (measured through ego networks) and (b) how social networks affect social activities with peers. Accordingly, our research questions were as follows:

Research Question 1 (RQ1): How do high-school students’ ego network structural characteristics (i.e., size, density, effective size, efficiency) differ across academic tracks defined by the school system?

Research Question 2 (RQ2): What is the relationship between student characteristics (i.e., gender, grade, race/ethnicity, academic track) and students’ ego network structural characteristics (i.e., size, density, effective size, efficiency)?

Research Question 3 (RQ3): What is the relationship between (a) academic track and ICT usage and (b) academic track and informal and formal social activity?

Research Question 4 (RQ4): What factors (i.e., student characteristics, social network size, and ICT use) predict frequency of informal and formal social activities?

Method

Participants

Adolescents

We used stratified sampling procedures (see “Procedures” section) to recruit 350 students in two high schools across three academic tracks (described subsequently). A total of 297 adolescents in two high schools across the following academic tracks (see “School characteristics” section) completed the survey: general education (GENED), n = 141; co-taught (CO-T), n = 116; special education (SPED), n = 40 (57% boys, 43% girls). Fifty-three students across the following academic tracks did not assent to participate: GENED = 13; CO-T = 27; SPED = 11 (response rate = 84%). Two students in the special education track were unable to complete the survey due to survey demands not matching their support needs. The distribution of students across racial and ethnic groups and academic tracks closely resembled the demographics of the two schools, the representative districts, and the midwestern state where both schools reside (see Table 1). Fifty-three percent of the sample were freshman, with the rest divided fairly equally between sophomores, juniors, and seniors. The representativeness of the sample assists in extrapolating the findings to the greater school population.

Table 1.

Study, School, and State Demographic Information.

School	Race/ethnicity (%)						SES (%)	Academics (%)
School	Asian	AI	Black	Latino	M	White	LI	IEP	ELL	ME
Study	4	1.0	23	14	11	48	NA	14	NA	NA
Hinman	15	0.3	14	13	3	54	26	13	4	72
Judson	2	0.3	6	12	2	78	31	13	4	59
State	4	0.3	18	24	3	51	50	14	10	53

Note. We retrieved the demographic data from the State Board of Education report card. The name of the state has been redacted to maintain confidentiality. AI = American Indian; M = Multiracial; LI = low-income; IEP = Individualized Education Program; ELL = English language learner; ME = meet/exceeds standards; NA = not available.

School characteristics

The participating high schools were located in two school districts in the suburbs of a large midwestern city. Each district had specific, nearly identical, academic tracking policies to place students with and without disabilities into core academic classes (Redacted, 2012a, 2012b). The academic tracks, labeled as Levels 0 through 9, were defined both by the curriculum content (i.e., functional curriculum, intensive instruction linked to the general education curriculum, general education core content) and the classrooms within which students were placed. Levels 0 (i.e., functional) and 1 (i.e., instructional) were segregated SPED classes. Level 0 courses focused on building and strengthening basic skills. Level 1 courses provided students the entire curriculum with a focus on improving student skills. General education core content was delivered either in co-taught general education class (Level 7; CO-T) or general education classes with special education consultation (Levels 2–6; GENED). Levels 8 and 9 were accelerated and advanced placement courses, respectively. For the purposes of this article, we have named Levels 0 through 1 as SPED, Level 7 as CO-T, and all other tracks as GENED. CO-T and GENED classes were considered college-bound. Approximately 30% of students in CO-T classes had Individualized Education Programs (IEPs), and approximately 5% of students in GENED classes had IEPs. According to staff, students with disabilities in GENED typically did not need instructional support, and students without disabilities were randomly placed in CO-T or GENED. SPED classes were only for students with IEPs. Students with less intensive support needs (i.e., learning disability, other health impairments) tended to receive instruction in Level 1 classes whereas students perceived as having more intensive support needs (i.e., intellectual, multiple disabilities, autism spectrum disorder) tended to receive instruction in Level 0 classes.

Procedures

Recruitment

Directors of Special Education and/or Directors of Research in 15 school districts in a midwestern state were contacted and provided information about the study. We targeted recruitment to metropolitan public school districts that had clearly defined academic tracking policies for students with and without disabilities. Of the 15 districts initially contacted, two did not respond, nine declined, and four agreed to participate. Two high schools in two districts participated in a pilot study, and two high schools in the remaining districts contributed data to this project. In agreement with district/school staff and the university’s Institutional Review Board (IRB), we used a waiver of informed consent and student assent. The waiver letters were sent home through previously established home/school communications (e.g., paper copies, electronic communications, parent listserv).

Sampling procedures

To obtain a representative sample of students with and without disabilities at each high school, we stratified our sample across academic tracks. Academic tracks were defined based on their level of inclusiveness. Segregated SPED classes were grouped together in sampling. CO-T and GENED were both considered inclusive but because students with disabilities were overrepresented in CO-T classes, unique samples were generated. We sampled classes in core content area classes (i.e., English, math, or science) because all students across different grade levels were required to take these classes.

Sample size

Using GPower3.1 (Faul, Erdfelder, Buchner, & Lang, 2009), we conducted a priori power analyses to determine the sample size needed to answer each question (required n = 159). We planned for imbalance in the design, as there are fewer students served in the SPED track. Based on these results we sampled a total of 350 students across three academic tracks and two high schools.

Measures

Survey

We merged existing research tools and protocols to create the Student Survey of Social Networks, Information Communication Technology Use and Social Activities (SNITA). Specifically, the survey incorporated existing questions and tools (described below) to measure (a) personal and environmental factors, (b) ego networks, and (c) social activity.

Personal and environmental factors

Student’s academic tracks were provided, through the sampling process, by school staff who assisted in randomly sampling from classrooms in each track; all other information was collected directly from students. Students provided information on (a) gender, (b) age, (c) race/ethnicity, (d) grade, and (e) crowd membership. For crowd membership, students were provided a list of crowds identified by previous researchers (Ryan, 2000) and asked to mark all of the crowds they belonged to (i.e., popular, jock, brain, normal, tough, outcast, none, or other). Finally, students were asked the frequency and number of friends with whom they used the following ICT modes (Lenhart et al., 2010): (a) call by landline, (b) call by cell phone, (c) email, and (d) text.

Ego networks

Ego network data were collected on the SNITA through name generators, name interpreter, and alter–alter questions (Carolan, 2014; Wasserman & Faust, 1994). These questions came from two established tools, The Social Network Map (SNM) and The Social Network Grid (Tracy & Whittaker, 1990). These tools have been used in the ego network literature focused on youth with and without disabilities (Blakeslee, 2012; Hulbert-Williams, Hastings, Crowe, & Pemberton, 2011; Lau-Barraco & Collins, 2011).

The SNM provides a structured methodology to elicit ego–alter relationships. Students are instructed to freely list the peers “with whom they hang together a lot” or who “have been important to” them by providing emotional or academic support. Then, for each alter listed, students are asked a series of questions that elicits network structural and functional characteristics. Only data on the structural characteristics (i.e., alter–alter ties) are reported in this study. Students indicated (a) alters (name generator), (b) characteristics of those alters (name interpreter), and (c) to which alters the named alter is connected (alter–alter relationships). Kogovšek and Ferligoj (2005) indicate that eliciting network data on larger networks and having respondents answer name interpreter (i.e., characteristics) questions by alter rather than referring back to each alter independently after all alters are named provide more reliable (Cronbach’s α = .835) and valid (Cronbach’s α = .951) means of gathering these data.

Social activity

After providing demographic information and egocentric network data, students were asked to indicate their participation in informal and formal social activities. We modified the Children’s Assessment of Participation and Enjoyment (CAPE; King et al., 2004), an assessment that examines both informal and formal social activity for young children and adolescents with and without disabilities. For informal social activities, participants indicated the frequency per week and number of friends with whom they hung out at home or at their friend’s house. For formal social activities, students indicated the frequency per week and the number of friends with whom they participated in school-sponsored activities (e.g., clubs/organizations).

Survey implementation

On the SNITA, students were provided with a landscape table to indicate the structural and functional characteristics of their ego networks. We created and implemented two modified versions of the SNITA (described subsequently) to address the support needs of students in the instructional SPED track (SNITA Modified) and in the functional SPED track (SNITA Modified–2). The SNITA Modified differed only in the layout of the ego network questions. This modification was created after instructional class teachers (who provided feedback on the measure prior to implementation) suggested the existing layout might be confusing for the students. For both the SNITA and the SNITA Modified, the classroom teacher administered the survey. Students in functional classes completed the SNITA Modified–2, a computer-assisted survey, delivered through Intellitools Classroom Suite©. Although the questions in the SNITA Modified–2 were similar to questions in both the SNITA and the SNITA Modified, we followed computer-assisted survey procedures for ego network data collection outlined by Gerich and Lehner (2006). Specifically, the SNITA Modified–2 was administered using a computerized and text-to-speech delivery of questions, computerized response opportunities, and automatic sequential questions. The first author individually administered the SNITA Modified–2 to students by providing the students with the computer and answering any questions that came up. Interaction was minimal as the students primarily interacted with the computer.

Analytic Plan

As ego network structural characteristics was a primary variable of interest across research questions, only students who completed the name generator, name interpreter, and alter–alter ties were included in the analyses. There were missing data for these variables in five cases. Prior to analysis, network size, density, effective size, and efficiency were calculated, using standard procedures (Marsden, 2002). Size was calculated by summing the total number of alters. Density was calculated by dividing the number of ties among alters by the total number of ties available. The density formula used was d_i = (n_i (n_i − 1) / 2) × a_i (Marsden), where n_i is the number of possible alter ties in a given ego’s network and a_i is the total number of ties present among indicated alters. Effective size was determined by subtracting the total number of redundant ties (i.e., alters who were tied together without the presence of the ego) from the total number of alters. Finally, efficiency was calculated by dividing effective size by the network size.

Network characteristics by academic track (RQ1)

Four separate one-way ANOVAs were used to examine differences in means of the four ego network characteristics (network size, density, effective size, and efficiency). Academic track (SPED, CO-T, GENED) was the independent variable. If there was a significant main effect of academic track for any of the analyses, post hoc pairwise comparisons were used to determine the pattern of differences.

Student and network characteristics (RQ2)

Four separate regression models for each of the four dependent variables (network size, density, effective size, and efficiency) were run with gender (male/female), grade (9th, 10th, 11th, 12th), race/ethnicity (Black, Hispanic, White, Other, Multiracial), academic track (GENED, CO-T, SPED), and crowd membership (popular, jock, brain, normal, tough, outcast, none, brain/normal, jock/normal) as predictors. We entered data using forward selection to determine whether each predictor was correlated with the dependent variable and whether the addition of subsequent predictors explained significant additional levels of variability in the model (Cohen’s f = 0.15, α = .05, Power = .80). Non-significant variables were excluded. Each predictor was entered as dummy coded variables.

ICT use and social activity by academic track (RQ3)

Chi-square tests were used to examine differences between students in the three academic tracks in (a) use of ICT (landline, cell phone, email, texting) and (b) participation in informal and formal social activity.

Factors associated with social activity (RQ4)

We used fixed-effects regression models to assess the degree to which personal factors (gender, grade, race/ethnicity, and crowd membership), academic track, ego network size, and ICT use predicted the variability in frequency of (a) informal social activity and (b) formal social activity. Predictors were entered using forward selection.

Results

Ego Network Structural Characteristics by Academic Track (RQ1)

Overall, students in the SPED track reported smaller network sizes, less dense or less connected networks, smaller effective size or more redundancy in their networks, and less efficient networks where information traveling through the network takes more time. Less differentiation was found between the CO-T and GENED groups (see Table 2).

Table 2.

Means and Standard Deviations of Ego Network Characteristics, Overall and by Academic Track.

Ego network characteristic	Overall		Academic track
	Overall		SPED		Co-taught		GENED
	M	SD	M	SD	M	SD	M	SD
Number of alters	4.00	1.34	2.22	1.51	4.30	1.00	4.29	1.20
Density	0.42	0.37	0.27	0.39	0.46	0.39	0.43	0.34
Effective size	3.58	1.34	1.96	1.47	3.79	1.13	3.86	1.12
Efficiency	0.87	0.18	0.77	0.34	0.88	0.13	0.89	0.14

Note. SPED = special education; GENED = general education.

Tests of the degree to which the differences in network characteristics between tracks were significant (see Table 3) revealed that, indeed, all four network characteristics were significantly different (p < .01). To decompose the differences related to academic track, post hoc Tukey’s Honestly Significant Difference tests were run for each dependent variable. Results indicated that the means of all four ego network structural characteristics of students in the SPED track were significantly different than students in both the CO-T and GENED tracks. The CO-T and GENED tracks did not differ from each other. Post hoc analysis indicated a medium effect size (1 − β = 0.98, df₁ = 2 df₁ = 295; Cohen, 1988).

Table 3.

Analysis of Variance of Ego Network Structural Characteristics by Academic Track.

Characteristic	Source	Sum of squares	df	Mean square	F	Significance*
Network size	Between groups	145.194	2	72.60	54.76	.001
	Within groups	389.80	294	1.33
	Total	535.00	296
	Welch		2/100.39		33.79	.001
Density	Between groups	1.19	2	0.60	4.41	.01
	Within groups	39.67	294	0.14
	Total	40.86	296
	Welch		2/105.40		3.85	.02
Efficiency	Between groups	0.46	2	0.23	7.19	.001
	Within groups	9.48	294	0.03
	Total	9.94	296
	Welch		2/101.43		29.61	.001
Effective size	Between groups	121.25	2	60.62	43.77	.001
	Within groups	407.17	294	1.39
	Total	528.42	296
	Welch		2/93.28		93.28	.11

p < .01.

Predictors of Ego Network Structural Characteristics (RQ2)

Table 4 shows the significant predictors for each model. We describe the specific pattern of results for each dependent variable below.

Table 4.

Model: Student-Level Factors on Ego Network Structural Characteristics.

Predictors	Source	B	SE	β	R ²	ΔR²
1. Network size	Constant	4.27	0.07
	SPED**	−1.72	0.21	−.44	.27	.27
	Ninth grade**	0.37	0.14	.14	.29	.02
	12th grade*	0.45	0.20	.15	.29	.01
	Asian*	−0.68	−0.34	−.10	.30	.01
2. Density	Constant	0.35	0.31
2. Density	Ninth grade**	0.24	0.04	.19	.04	.04
3. Effective size	Constant	3.83	0.07
	SPED**	−1.71	0.22	−.44	.23	.23
	10th grade*	−0.53	0.26	−.11	.024	.01
4. Efficiency	Constant	0.88	0.01
4. Efficiency	SPED**	−0.12	0.03	−.22	.05	.05

Note. Post hoc analysis indicated effect size f² = 0.15, Power (1 − β = 0.999), df = 10.

p < .01. **p < .001.

Network size

The SPED track, ninth grade, 12th grade, and Other race/ethnicity were kept in the model as significant predictors. Specifically, being in the SPED track was a strong predictor of network size, with students in this track having approximately 25% fewer network members. Being in ninth grade was associated with a slight increase (.14) in the number of network members as was being in Grade 12 (.15 increase). Finally, being identified as in the Other racial/ethnic category was associated with .10 fewer network members. The model had good fit with approximately 30% of the variance in network size explained by the predictors.

Density

The only significant predictor was being in ninth grade, which improved network density by .19. Although ninth grade was significantly associated with network density, the model only explained about 3% of the variance in network density (R² = .03).

Effective size

The strongest predictor of effective size was placement in the SPED track, which predicted a .44 smaller effective size (p < .01, R² = .23) than placement in the CO-T or GENED tracks. Being in 10th grade was associated with an additional .11 smaller effective size (p < .01, R² = .02).

Efficiency

Being a member of the SPED track, the only significant variable in the model, decreased the individual’s network efficiency by 22% (p < .01, R² = .05).

ICT Use and Social Activity by Academic Track (RQ3)

Table 5 provides descriptive statistics across academic tracks. Although most students indicated using cell phones and texting for communication and that they hung out with their friends at home or at a friend’s house (88%) and at school (75%), we found significant differences by academic track. Students in the SPED track communicated with friends less frequently through cell phone use (χ² = 54.082, df = 2, p < .01) and texting (χ² = 77.078, df = 2, p < .01). Similarly, there was a significant association between a respondent’s academic track and whether or not they hung out with friends at home or at a friend’s house (χ²= 148.14, df = 2, p < .01) and whether they hung out with friends at school (χ² = 18.29, df = 4, p < .05), with students in the SPED less likely to engage in either.

Table 5.

ICT Use and Social Activity by Academic Track.

Activity	n
Activity	GENED	CT	SPED
ICT use to communicate with friends
Calls friend by landline	32	34	14
Calls friend by cell phone*	121	104	16
Emails friends	38	25	7
Texts friends**	134	107	18
Social activity
Hangs out with friends at home or at a friend’s house***	135	107	19
Hangs out with friends at school-sponsored activities/events****	116	83	26

Note. All analyses have an effect size w = 0.3, Power (1 − β) =0.99, and df = 2. ICT = information communication technology; CT = co-taught.

χ² = 54.082, df = 2, p < .01. **χ² = 77.078, df = 2, p < .001. ***χ² = 148.136, df = 2, p < .01. ****χ² = 6.893, df = 2, p < .05.

Predictors of Informal and Formal Social Activity (RQ4)

Informal and formal social activity

Because the analyses in RQ3 indicated, overall, that students did not use landline phone or email to communicate, these forms of ICT were not included. As shown in Table 6, using a cell phone to communicate with friends was associated with hanging out with friends at home or at a friend’s house .25 more times per week. Being White was associated with an increase of .16 times per week. However, being in ninth grade was associated with hanging out with friends at home or at a friend’s house 0.25 fewer times per week. Using a cell phone had the largest predictive value as it explained 46% of the model. Other student-level factors, particularly academic track and network characteristics, were excluded from the model as they were not significant. When examining predictors of formal social activities at school-sponsored events, this model (see Model 2; Table 6) had a much poorer fit (R² = .01) with only one predictor, female, staying in the model. Although this model has a strong effect size and power, being female explained so little of the variance that there are likely other unmeasured factors that influence this outcome.

Table 6.

Modeling Social Activity.

Model	Predictors	B	SE	β	R ²	ΔR²
1. Informal social activity frequency	Constant	1.49	0.32
	Cell phone**	1.48	0.34	.25	.46	.05
	Ninth grade**	−1.17	0.30	−.25	.08	.03
	White***	0.74	0.30	.16	.09	.02
2. Formal social activity frequency	Constant	3.52	0.36
2. Formal social activity frequency	Female***	−1.10	0.55	−.12	.01	.01
3. Informal social activity frequency	Constant	0.98	0.84
	Density***	0.93	0.47	−.10
	Efficiency***	2.02	1.05	.16	.16
4. Formal social activity frequency	Constant	1.07	1.67
4. Formal social activity frequency	Network size*	0.70	0.26	.20	.19

Note. Models 1 and 2 used student-level factors, network size, and ICT use (cell phone and texting) as predictors. Models 3 and 4 used ego network characteristics only. Post hoc analysis indicated effect size f² = 0.15, Power (1 − β = 0.999), df = 10.

p < .05. **p < .01. ***p < .001.

Ego network characteristics and social activity

Given that the results of previous analyses suggesting the strength and utilization of an individual’s ego network can influence one’s activity level (RQ3), two additional models were run with informal and formal social activity as dependent variables and structural network characteristics as the only predictor. This was performed because in RQ1, structural characteristics significantly differed across academic tracks and it was important to explore whether focusing on ego network structural characteristics would produce a better model fit. The model for informal social activity (Model 3, Table 6) showed a significant effect of network density and efficiency. Specifically, an increase in network density, where redundant ties are included, was associated with a .10 times per week decrease in hanging out with friends. An increase in efficiency, where information flows efficiently through the network, was associated with a .16 increase in times per week hanging out with friends. The model for formal social activities also showed significant results (see Model 4, Table 6) where network size was associated with a .20 increase in times per week a student hung out with friends at school.

Discussion

The purpose of this study was to examine how personal and environmental factors affect access to social capital through the analysis of ego network characteristics. We found students served in segregated environments reported less adaptive social network characteristics, ICT use, and participation in social activities. These findings are congruent with previous research that suggests that disability status affects social connections; however, specifically examining the impact of the placement of students with and without disabilities, an environmental factor under the influence of educational systems, extends this work. Although it is not surprising that those in segregated settings had the worst outcomes, it is important to note that students with and without disabilities in CO-T and GENED classrooms reported similar social network characteristics, despite students with disabilities being overrepresented in CO-T classrooms. This suggests that students with disabilities in inclusive placements may have more adaptive outcomes. However future research is needed to control for disability status and the impact of support needs, as we found that disability status and support needs interacted with both placement and social network characteristics. Overall, this preliminary research provides important information on how inclusive school environments affect ego networks and social network membership.

Specifically, as found in RQ1, students in more segregated environments (i.e., SPED track) experienced smaller, less dense, and more redundant networks (i.e., smaller effective size), where information took longer to pass through the network (i.e., less network efficiency). Given this, students who were in segregated environments had fewer opportunities to draw on support compared with students in more inclusive settings. Although less dense networks can be an indicator of bridging social capital (as described earlier), the participants in our study experienced more redundant network ties, meaning information passed through the network less efficiently. So, opportunities to benefit from bridging social capital were restricted. Given that participation in inclusive programming (i.e., CO-T or GENED) was a significant predictor of more positive outcomes, it is therefore possible that inclusive opportunities may improve network size, decrease redundancy, and improve the flow of information in an individual’s network. The findings suggest this might be best done at the ninth grade when all students are new to the school—although this may be unique to these schools or schools of this size where students come from many different middle schools across different towns and have less opportunity to know each other. Overall, an individual who has different networks of friends across environments (e.g., drama club, English class, or homeroom) with stronger effective size may have increased opportunities to learn about potential social activity, receive academic assistance (peer-mediated assistance), and even other information or support regarding community involvement or employment (e.g., part-time jobs, volunteer work). Furthermore, the individual may be perceived as having more value in their network, independent of disability label. Further research is needed in examining the relationship between ego network characteristics and outcomes.

We also found differences in the way students communicated (i.e., ICT use) and interacted (i.e. informal and formal social activity) with their peers across placements. In fact, in contrast with students in the other tracks, when asked about communication and social activities, several students (n = 7) in functional and instructional classes reported that they did not have access to a cell phone. In addition, only students in the SPED track indicated that they use a landline, a form of ICT rarely used by adolescents. This is a concern given that recent research has indicated that 78% of teens have a cell phone (Madden, Lenhart, Duggan, Cortesi, & Gasser, 2013) and 54% of teens use texting as the primary means of communicating with their friends (Lenhart et al., 2010). Discrepancy in ICT access may reduce these students’ ability to (a) interact and build peer networks and (b) identify and participate in social activities, particularly in informal social activities as texting is the primary means of communicating to schedule these types of activities (Ling, 2010). This suggests the need to, in transition plans or IEPs, make mobile technology use a priority through specific instruction and support (as opposed to using a landline or desktop computer to communicate). Furthermore, the design and implementation of accessible mobile technology may also increase ICT use. Although we did not examine why students in the special education track reported using a cell phone and texting friends less, other research on adults with disabilities suggests barriers related to access (physical and cognitive demands) and parental or caregiver beliefs and skill levels (Bryen, Carey, & Friedman, 2007; Li-Tsang, Yeung, Chan, & Hui-Chan, 2005; Tanis et al., 2012). Further research may examine the role of accessibility, caregiver perceptions, and access.

Limitations

Our findings provide important guidance for future research and practice, but there are inherent limitations that must be considered in interpreting the findings. Even though WNA has been used extensively in education research, ENA has not (Carolan, 2014). Given the various contexts experienced by adolescents, this methodology provides meaningful information on social capital; however, more research is needed, particularly as we did not use objective confirmation of the tie data (e.g., whether adolescents are accurate reporters of their ego networks across contexts; Feld & Carter, 2002). In addition, we only focused on structural network characteristics and not functional network characteristics or what adolescents gets out of the connection; further research is needed in addressing the actual support individuals derive from their social networks (Putnam, 2000). Also, as our data are cross-sectional, the results may not reflect the change in networks over time, an identified area of research by social network researchers (Wellman, 2007).

We used academic placement based on school tracking policies as a proxy for access to inclusive opportunities. It is possible the sampled schools have unique structures that affect network characteristics. We also did not analyze the impact of differing disability labels and differences between students with and without disabilities in more inclusive tracks. Thus, it is possible that there are differences within the SPED and CO-T tracks based on student support needs (Byrne, 1988). Future research should examine the interaction of placement, disability label, and support need. And, due to the sample size and school specific policies, the findings may not generalize to schools with different policies and characteristics.

Finally, future research should explore the relationship between social networks and social capital (Scrivens & Smith, 2013). For instance, do highly dense networks provide social support or bonding social capital (Putnam, 2000) at the expense of novel information and opportunities obtained through unique ties in less dense networks (a form of bridging social capital; Burt, 1992)? Understanding the interaction of these mechanisms is critical to inclusive policies across school contexts.

Implications for Research and Practice

Results suggest access to inclusive classrooms has a significant impact on students’ ego networks with restricted social networks influenced not only by disability label but also by access to inclusive classrooms. In practice, however, this suggests the need to consider the impact of tracking policies and placement decisions not only academically but also socially. Researchers have suggested that social connections affect not only social outcomes but also academic and life outcomes (Calvo-Armengol et al., 2009; Drukker et al., 2003; Maroulis & Gomez, 2008). Systematically considering how to promote access to social and academic opportunities in inclusive settings for all students may lead to more adaptive outcomes in multiple domains related to education and transition.

The findings also suggest the importance of considering social networks across contexts. Network boundaries should not, necessarily, be artificially defined by researchers or only considered by practitioners within the settings they see adolescents. Instead, considering the characteristics of social networks across contexts can provide a meaningful way to develop interventions so students can be supported to leverage resources across settings (e.g., use effective connections from informal social activities to leverage support in the classroom). It can also be used to evaluate the impact of inclusive programming and associated supports (e.g., funding, staff support, peer-mediated strategies).

In transition planning, ego network data provide a tool for adolescents with disabilities and their support networks to visualize network ties and the associated support each tie provides across contexts (e.g., job/volunteer opportunity, social activity, transportation). Transition teams can build upon existing network ties and look to create ties where holes exist (e.g., ties needed for employment). This has the potential to directly involve adolescents in understanding and leveraging their social networks. For instance, interventions could be developed that enable teachers to teach students to map their connections across contexts, via paper-pencil or electronic means, and use those maps to identify opportunities for (a) gaining social support, (b) obtaining new information, and (c) participating in social activities. Finally, additional research is needed on the relationship between social capital and long-term outcomes (e.g., independent living, employment, transportation). In doing so, researchers can better understand the impact of supporting social networks and how inclusive policies affect network characteristics and access to social capital. Further research is needed on how environmental factors can be manipulated within educational contexts to build social networks and social capital to enhance long-term outcomes for adolescents with and without disabilities.

Although the findings are preliminary and replication is needed, our findings provide important information on the structural characteristics of networks, the support those networks provide to adolescents with disabilities, and how school policies can affect opportunities to build networks. Examining access to and the utilization of social capital by adolescents is essential. In making and sustaining social connections, the individual recognizes that each relationship provides a benefit (e.g., emotional, occupational) either immediately or in the future. If it is understood and leveraged, social capital, the information and support provided and exchanged through social ties, can lead to improvements across many life domains.

Footnotes

Acknowledgements

The authors thank Drs. James Halle, Michaelene Ostrosky, and Michael Wehmeyer who served on the first author’s dissertation committee and provided welcomed feedback in planning and finalizing the project. The authors also acknowledge the late Dr. Philip Rodkin who provided the first author with important and critical input in the planning process of this study as an original member of the dissertation committee.

Authors’ Note

The first author completed this study in partial fulfillment of a Doctor of Philosophy degree in Special Education in the Department of Special Education at the University of Illinois. Two references are listed as redacted. These are both Hinman’s and Judson’s curriculum guides, respectively (Hinman and Judson are pseudonyms). The information on the specific district names are redacted to maintain confidentiality.

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.

References

Bartko

W. T.

Eccles

J. S.

(2003). Adolescent participation in structured and unstructured activities: A person-oriented analysis. Journal of Youth and Adolescence, 32, 233–241. doi:10.1023/A:1023056425648

Blakeslee

(2012). Expanding the scope of research with transition-age foster youth: Applications of the social network perspective. Child & Family Social Work, 17, 326–336. doi:10.1111/j.1365-2206.2011.00787.x

Briggs

D. S. X.

(1998). Brown kids in white suburbs: Housing mobility and the many faces of social capital. Housing Policy Debate, 9, 177–221. doi:10.1080/10511482.1998.9521290

Brown

B. B.

Bakken

J. P.

Ameringer

S. W.

Mahon

S. D.

(2008). A comprehensive conceptualization of the peer influence process in adolescence. In Prinstein

M. J.

Dodge

K. A.

(Eds.), Understanding peer influence in children and adolescents (pp. 17–44). New York, NY: Guilford Press.

Bryen

D. N.

Carey

Friedman

(2007). Cell phone use by adults with intellectual disabilities. Intellectual & Development Disabilities, 45, 1–9.

Burt

R. S.

(1992). Structural holes: The social structure of competition. Cambridge, MA: Harvard University Press.

Byrne

B. M.

(1988). Adolescent self-concept, ability grouping, and social comparisons: Reexamining academic track differences in high school. Youth & Society, 20, 46–67. doi:10.1177/0044118X88020001003

Calvo-Armengol

Patacchini

Zenou

(2009). Peer effects and social networks in education. The Review of Economic Studies, 76, 1239–1267. doi:10.1111/j.1467-937X.2009.00550.x

Carolan

B. V.

(2014). Social network analysis and education: Theory, methods, & applications. Thousand Oaks, CA: SAGE.

10.

Carrasco

J. A.

Miller

E. J.

(2006). Exploring the propensity to perform social activities: A social network approach. Transportation, 33, 463–480. doi:10.1007/s11116-006-8074-z

11.

Carrasco

J. A.

Miller

E. J.

Wellman

(2008). How far and with whom do people socialize? Empirical evidence about distance between social network members. Transportation Research Record: Journal of the Transportation Research Board, 2076, 114–122. doi:10.3141/2076-13

12.

Cattell

(2001). Poor people, poor places, and poor health: The mediating role of social networks and social capital. Social Science & Medicine, 52, 1501–1516. doi:10.1016/S0277-9536(00)00259-8

13.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.

14.

Coleman

J. S.

(1988). Social capital in the creation of human capital. American Journal of Sociology, 94, S95–S120. doi:10.1080/09687599.2012.741517

15.

Cornwell

(2009). Good health and the bridging of structural holes. Social Networks, 31, 92–103.

16.

Drukker

Kaplan

Feron

van Os

(2003). Children’s health-related quality of life, neighborhood socio-economic deprivation and social capital: A contextual analysis. Social Science & Medicine, 57, 825–841.

17.

Farmer

T. W.

Leung

M.-C.

Weiss

M. P.

Irvin

M. J.

Meece

J. L.

Hutchins

B. C.

(2011). Social network placement of rural secondary students with disabilities: Affiliation and centrality. Exceptional Children, 78, 24–38.

18.

Farmer

T. W.

Van Acker

R. M.

Pearl

Rodkin

P. C.

(1999). Social networks and peer-assessed problem behavior in elementary classrooms: Students with and without disabilities. Remedial and Special Education, 20, 244–256. Retrieved from http://dx.doi.org/10.1177/074193259902000408

19.

Faul

Erdfelder

Buchner

Lang

A. G.

(2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41, 1149–1160. doi:10.3758/BRM.41.4.1149

20.

Feld

S. L.

Carter

W. C.

(2002). Detecting measurement bias in respondent reports of personal networks. Social Networks, 24, 365–383. doi:10.1016/S0378-8733(02)00013-8

21.

Frank

K. A.

Muller

Mueller

A. S.

(2013). The embeddedness of adolescent friendship nominations: The formation of social capital in emergent network structures. American Journal of Sociology, 119, 216–253. doi:10.1086/672081

22.

Gerich

Lehner

(2006). Video computer-assisted self-administered interviews for deaf respondents. Field Methods, 18, 267–283. doi:10.1177/1525822X06287535

23.

Hallinan

M. T.

(1994). Tracking: From theory to practice. Sociology of Education, 67, 79–84.

24.

Harris

D. M.

(2011). Curriculum differentiation and comprehensive school reform: Challenges in providing educational opportunity. Educational Policy, 25, 844–884. doi:10.1177/0895904810386600

25.

Hulbert-Williams

Hastings

R. P.

Crowe

Pemberton

(2011). Self-reported life events, social support and psychological problems in adults with intellectual disabilities. Journal of Applied Research in Intellectual Disabilities, 24, 427–436. doi:10.1111/j.1468-3148.2011.00624.x

26.

King

Law

King

Hurley

Hanna

Kertoy

. . .Young

(2004). Children’s Assessment of Participation and Enjoyment (CAPE) and Preferences for Activities of Children (PAC). San Antonio, TX: Harcourt Assessment.

27.

Kogovšek

Ferligoj

(2005). Effects on reliability and validity of egocentered network measurements. Social Networks, 27, 205–229. doi:10.1016/j.socnet.2005.01.001

28.

Lau-Barraco

Collins

R. L.

(2011). Social networks and alcohol use among nonstudent emerging adults: A preliminary study. Addictive Behaviors, 36, 47–54. doi:10.1016/j.addbeh.2010.08.017

29.

Lenhart

(2012). What teens do with their phones. Washington, DC: Pew Research Center.

30.

Lenhart

Ling

Campbell

Purcell

(2010). Teens and mobile phones. Washington, DC: Pew Research Center.

31.

Lin

(2001). Social capital: A theory of social structure and action. New York, NY: Cambridge University Press.

32.

Ling

(2010). Teens, texting, and social isolation. Washington, DC: Pew Research Center.

33.

Li-Tsang

Yeung

Chan

Hui-Chan

(2005). Factors affecting people with intellectual disabilities in learning to use computer technology. International Journal of Rehabilitation Research, 28, 127–133. doi:10.1097/00004356-200506000-00005

34.

Madden

Lenhart

Duggan

Cortesi

Gasser

(2013). Teens and technology 2013. Washington, DC: Pew Research Center.

35.

Maroulis

Gomez

L. M.

(2008). Does “connectedness” matter? Evidence from a social network analysis within a small-school reform. Teachers College Record, 110, 1901–1929.

36.

Marsden

P. V.

(2002). Egocentric and sociocentric measures of network centrality. Social Networks, 24, 407–422. doi:10.1016/S0378-8733(02)00016-3

37.

Poortinga

(2006). Social relations or social capital? Individual and community health effects of bonding social capital. Social Science & Medicine, 63, 255–270. doi:10.1016/j.socscimed.2005.11.039

38.

Portes

(1998). Social capital: Its origins and applications in modern sociology. Annual Review of Sociology, 24, 1–24. doi:10.1146/annurev.soc.24.1.1

39.

Putnam

R. D.

(2000). Bowling alone: The collapse and revival of American community. New York, NY: Simon & Schuster.

40.

Raine

Wellman

(2012). Networked. Cambridge, MA: MIT Press.

41.

Redacted. (2012a). Curriculum guide. Author.

42.

Redacted. (2012b). Imagine the possibilities: A curriculum guide. Author.

43.

Redacted State Board of Education. (2013). Redacted report card. Author.

44.

Reich

S. M.

Subrahmanyam

Espinoza

(2012). Friending, IMing, and hanging out face-to-face: Overlap in adolescents’ online and offline social networks. Developmental Psychology, 48, 356–368. doi:10.1037/a0026980

45.

Ryan

A. M.

(2000). Peer groups as a context for the socialization of adolescents’ motivation, engagement, and achievement in school. Educational Psychologist, 35, 101–111. doi:10.1207/S15326985EP3502_4

46.

Schaefer

D. R.

Simpkins

S. D.

Vest

A. E.

Price

C. D.

(2011). The contribution of extracurricular activities to adolescent friendships: New insights through social network analysis. Developmental Psychology, 47, 1141–1152. doi:10.1037/a0024091

47.

Scrivens

Smith

(2013). Four interpretations of social capital: An agenda for measurement (OECD Statistics Working Papers, 2013/06). OECD Publishing. Retrieved from http://dx.doi.org/10.1787/5jzbcx010wmt-en

48.

Stiglitz

J. E.

Sen

Fitoussi

J.-P.

(2009). Report by the Commission on the Measurement of Economic Performance and Social Progress. Retrieved from http://www.stiglitz-sen-fitoussi.fr/en/index.htm

49.

Tanis

E. S.

Palmer

Wehmeyer

Davies

D. K.

Stock

S. E.

Lobb

Bishop

(2012). Self-report computer-based survey of technology use by people with intellectual and developmental disabilities. Intellectual and Developmental Disabilities, 50, 53–68. doi:10.1352/1934-9556-50.1.53

50.

Tracy

E. M.

Whittaker

J. K.

(1990). The social network map: Assessing social support in clinical practice. Families in Society, 71, 461–470.

51.

van Asselt-Goverts

A. E.

Embregts

P. J. C. M.

Hendriks

A. H. C.

(2013). Structural and functional characteristics of the social networks of people with mild intellectual disabilities. Research in Developmental Disabilities, 34, 1280–1288. doi:10.1016/j.ridd.2013.01.012

52.

Wasserman

Faust

(1994). Social network analysis: Methods and applications. New York, NY: Cambridge University Press.

53.

Wellman

(2007). The network is personal: Introduction to a special issue of social networks. Social Networks, 29, 349–356. Retrieved from http://dx.doi.org/10.1016/j.socnet.2007.01.006

54.

World Health Organization. (2002). Towards a common language for functioning, disability and health ICF. International Classification (Vol. 1149). Retrieved from http://www.who.int/classifications/icf/training/icfbeginnersguide.pdf

55.

Wouters

De Fraine

Colpin

Van Damme

Verschueren

(2012). The effect of track changes on the development of academic self-concept in high school: A dynamic test of the big-fish–little-pond effect. Journal of Educational Psychology, 104, 793–805.