I. EDUCATOR INTERVIEWS
In 2008, we conducted interviews with forty long-standing classroom teachers (twenty-four men, sixteen women) to cull their observations about how current students may be different from the students they taught in the predigital era. Educators and researchers affiliated with Harvard Project Zero recommended these educators to us based on their years of experience and teaching excellence. Participants averaged 23.5 years of teaching experience, and all but two had been teaching since 1992.
Participants were drawn from eighteen schools in the greater Boston area and one in central New Hampshire. All served students from affluent families. In total, we interviewed educators from two middle schools, two colleges, and fifteen high schools.
The educators in our sample represent a broad range of intellectual disciplines, including history (6), general social studies (1), English and/or English literature (6), foreign language (2), art (5), theater arts (7), music (5), biology (3), chemistry (1), physics (2), athletics (2), and general education (1). Several also coached a sport or, in the case of boarding school teachers, served as housemasters, and could therefore comment on their students’ lives outside the classroom.
Two researchers conducted the interviews, which followed a semistructured interview protocol and lasted approximately 90 to 120 minutes. Participants were first asked general questions about the changes they’ve noticed in various aspects of students’ lives, including academic engagement and performance, peer relationships, and extracurricular activities and interests. In an effort to avoid biased responses, we deliberately did not raise the topic of digital media during this part of the interview. However, participants typically introduced the topic on their own, and we were ready with a series of follow-up questions.
All but one of the interviews was audio-recorded. Both interviewers took detailed notes during each interview, which they later synthesized into a single record. Attached to each record was a briefer topsheet that summarized key points of interest. Halfway through the interview process, project staff constructed a matrix grid to organize salient data from each master record and topsheet. Categories were determined by the strength and frequency of a finding and later amended as more data were captured.
II. FOCUS GROUPS
Members of our research team conducted seven focus groups between May 2009 and March 2011. Participants were fifty-eight veteran professionals who each had over twenty years of experience working with young people (roughly ages twelve to twenty-two) in a variety of settings. The professionals included psychoanalysts; psychologists and other mental health workers; camp directors and longtime counselors; religious leaders; arts educators; and high school teachers and after-school educators who worked primarily with students living in low-income neighborhoods.
The focus group facilitator asked participants to reflect on changes they have observed in youth over the last twenty years and to offer their thoughts on the causes of these changes. Each participant was given 5 to 10 minutes to share his or her initial reflections, with the facilitator asking clarifying questions and summarizing themes as appropriate.
Members of the research team followed up with questions that encouraged participants to elaborate on their answers and invited them to respond to comments made by participants in earlier focus groups. The majority of these follow-up questions related to the “three Is” (identity, intimacy, imagination) identified as dominant themes from earlier, one-on-one interviews with veteran educators. To guard against leading questions, digital media were not introduced as a topic of conversation until a participant explicitly made reference to them.
Following each focus group, researchers compiled their individual field notes into a single memo summarizing the major themes discussed. One researcher then synthesized these themes into a series of formal reports, one for each group of professionals.
III. TEENS’ CREATIVE PRODUCTIONS
Between February 2011 and August 2012, we conducted three related studies of artwork and fiction writing produced by teens between 1990 and 2011. We analyzed 354 works of high school students’ visual art, 50 short stories written by high school students, and 44 short stories written by middle school students.
Our visual art sample included a random selection of 354 artworks published in a teen literary and art magazine based in Massachusetts. The same husband-and-wife editorial team has published the magazine, called Teen Ink, since its inception in 1989. Half of the pieces in our sample (n = 177) appeared in issues published in 1990–1995, and half of the pieces appeared in annual issues published in 2006–2011. Though most submissions are two-dimensional, a small number of pieces are photographs of three-dimensional artworks, such as sculptures and installation work.
We selected artwork to include in our sample from the “Art Gallery” page that appears in every issue of Teen Ink. We used a random number generator to select three pieces from each issue, with the exception of the February 1991 issue, which was not available in the Teen Ink archives. To balance out the number of issues included in the early and late periods, we additionally omitted the December 2011 issue.
The low print quality of the early issues limited the amount of visible detail. Therefore, we collected original pieces from Teen Ink’s physical archives. Occasionally, an original artwork was missing. In such a case, we randomly selected substitute pieces from the available originals in the archive. In total, approximately 20 percent of the early sample was chosen from this reselection process. Due to the high printing quality of the magazine issues published after September 1999, we were able to code artworks from the later period directly from the printed Teen Ink issues.
Two research assistants with formal arts training devised the coding scheme used to analyze the visual art. They drew on the themes that emerged from the educator interviews and focus groups, as well as formal elements of technique and interpretation in the visual arts, such as background, composition, and medium. The final coding scheme included eighteen codes in total.
To ensure consistency and accuracy in interpretation, the coding process followed a coder/shadow coder approach. Each researcher served as the primary coder for half of the pieces, while the other researcher, as the shadow coder, reviewed the decisions made by the primary coder. The two coders discussed any divergent interpretations (which were few in number) and updated the code classification after achieving consensus.
The coding was entered into the qualitative software program NVivo 9, which enabled researchers to identify trends across pieces from the early (1990–1995) and late (2006–2011) periods. These trends were documented in a series of coding reports, one for each of the eighteen codes in the coding scheme.
High School Fiction
Our high school creative sample included fifty fiction pieces written by high school students attending a school with a strong creative writing program in New Orleans. Half of the stories in our sample (n = 25) were written in 1990–1995, and half of the stories were written in 2006–2011. All stories were published in annual issues of the school’s literary magazine, which selects stories for publication through a peer-review process.
Drawing stories from the same school ensured that the population of students remained relatively constant throughout the twenty-year period that comprised the focus of our investigation. Our sample initially included all of the short stories published during the two periods of interest. However, in order to avoid having a sample composed heavily of repeat authors, we ultimately excluded a number of stories written by the same student.
Two research assistants with training in English literature and composition developed the coding scheme used to analyze the high school fiction writing. As with the visual art analysis, the researchers drew on the themes that emerged from our educator interviews and focus groups, as well as technical elements in fiction writing, such as plot, setting, characters, and narrative structure. The final coding scheme included twenty-two codes in total.
Like the visual art analysis, the coding process for the high school fiction employed a coder/shadow coder approach. Discrepancies in coding were rare (<1 per story) and resolved through discussion between the coders. NVivo 9 was used to explore trends across stories from the early (1990–1995) and late (2006–2011) periods.
Middle School Fiction
Our middle school creative writing sample comprised forty-four stories written by seventh- and eighth-grade students attending an independent K–8 school in Maine. Half of the stories in our sample (n = 22) were written in 1995–1998, and half of the stories were written in 2007–2009. As with our high school writing sample, all stories were published in issues of the school’s literary magazine. The stories published in the magazine were not juried; the published stories are, therefore, a genuine representation of the range of work that students submitted.
The sample initially included all of the stories published during our time periods of interest: 1995–1998 and 2007–2009. If a student had multiple pieces published, we included the latest work in our sample and discounted the others. Since some issues included a greater number of stories than others, after we excluded pieces with duplicate authors, we assigned a number to each story and used a random number generator to select an approximately equal number of stories from each period.
The data analytic approach for the middle school fiction stories mirrored directly the process used for coding and analyzing the high school fiction pieces.
IV. THE GOODPLAY PROJECT
In 2008–2010, we conducted in-depth, qualitative interviews with 103 youth aged ten to twenty-five living in the greater Boston area. Funded by the MacArthur Foundation’s Digital Media and Learning Initiative, this investigation focused on the ethical dimensions of young people’s digital media activities, including youth’s experiences with and thoughts about the “thorny” situations they’ve encountered online.
The middle and high school participants were recruited from one suburban and two urban public school districts, while the college-age participants were recruited from two- and four-year private and public colleges. Our research team recruited postcollege participants using Craigslist and by posting flyers in the same geographic areas as the student participants. These recruitment efforts yielded a socioeconomically and racially diverse sample.
In the interviews, participants were asked a series of questions about their experiences with digital media and how they have responded to challenging situations online, such as witnessing hate speech or deciding how much personal information to share. We also posed to them a series of hypothetical dilemmas involving ethically charged issues that are salient online, such as privacy, ownership and authorship, and community participation.
All interviews were audio recorded and transcribed verbatim. The research team devised a coding scheme that contained etic codes, identified a priori from our research questions and review of prior scholarship, and emic codes, which reflected themes that emerged directly from our line-by-line reading of the transcripts.
To ensure that researchers applied the coding scheme consistently and accurately, each researcher coded a subset of transcripts separately and then met to discuss areas of disagreement. In these meetings, the research team clarified code definitions and resolved areas of disagreement. This process was repeated until satisfactory levels of inter-coder reliability were reached for each code. Afterward, researchers divided up the codes evenly among them and applied their group of codes to the remaining transcripts. Periodic coding meetings were held throughout the coding process to ensure that researchers continued to apply their codes reliably.
After coding was complete, researchers produced a series of coding reports that summarized the patterns across participants. Also included in the reports were representative quotes that illuminated key themes.
V. TEEN BLOGGER STUDY
In 2007, Katie interviewed twenty girls who had each been blogging throughout their middle and high school years in a popular online journaling community called LiveJournal. Participants were between the ages of seventeen and twenty-one and represented all school years between tenth grade of high school and senior year of college. The majority were White (12), with a sizable minority identifying as Asian (5). The remaining girls identified themselves as Hispanic (1), Pacific Islander (1), and a mix of Native American, Black, and White (1). All participants were either residents of or attending college in the greater Boston area during the time of the study.
Each participant took part in a face-to-face interview that lasted approximately 60 minutes and followed a semistructured protocol. The questions focused on how the participants used their blog to express themselves, explore their personal interests, and connect to other people. Participants were also asked about how their blog had changed over the years, as well as the relationship between their blogging and other digital media activities like social networking, texting, and instant messaging. These questions elicited responses that touched on the themes of identity, intimacy, and imagination that we explore in this book.
Following the analytic approach used by the GoodPlay team, the interviews were audio-recorded and transcribed verbatim. The transcripts were then coded using a coding scheme that included both etic and emic codes. To establish inter-coder agreement, two members of the GoodPlay team used the coding scheme to code one of the transcripts. They discussed discrepancies in their code application, and Katie used this discussion to guide her subsequent coding. To identify themes within and across participants, Katie produced analytic memos for each participant and conducted analyses using the qualitative software package NVivo 8.
VI. BERMUDA STUDY
In 2010, Katie conducted a mixed-method study in Bermuda’s secondary schools (grades 8–12). The first phase of the study involved a large-scale survey of 2,079 students (57 percent female) between the ages of eleven and nineteen years (M = 15.4 years) attending public and private schools in Bermuda. With approximately 2,600 students attending secondary school in Bermuda, the survey sample accounted for roughly 80 percent of all secondary students on the island. The second phase of the study (whose findings we draw on throughout this book) comprised in-depth interviews with 32 students who had participated in the survey. The interview participants spanned the same schools and grade levels as the survey participants.
The interviews took place during school hours and followed a semistructured protocol that allowed the interviewer to explore unanticipated responses and topics that were of particular interest to participants. All participants were asked questions about their technology ownership and digital media activities, including how often they engage in these activities and their motivations for doing so. They were asked about a wide range of digital media pursuits, including social networking, texting, instant messaging, and gaming. The interview also included questions regarding the quality of participants’ relationships with their friends, parents, and teachers; their experiences in school; and their self-understanding.
The analytic approach for the interviews mirrored the approaches taken in the GoodPlay Project and blogger study, described above. Each interview was audio-recorded and transcribed verbatim. Katie created a coding scheme that contained both etic and emic codes. In an effort to ensure that she applied the codes consistently and accurately, she enlisted a graduate student experienced in qualitative data analysis to code a subset of the transcripts. They each coded the same group of transcripts and then met to discuss areas of discrepancy. Once satisfactory levels of agreement had been reached, Katie coded the remainder of the transcripts.
During the coding process, Katie produced a coding memo for each participant that summarized all comments they made in relation to the codes in the coding scheme. After completing this process, she used a qualitative software program (NVivo 9) to aid in the identification of salient patterns across participants.