Jump to ContentJump to Main Navigation
Mixed MessagesNorms and Social Control around Teen Sex and Pregnancy$

Stefanie Mollborn

Print publication date: 2017

Print ISBN-13: 9780190633271

Published to Oxford Scholarship Online: March 2017

DOI: 10.1093/acprof:oso/9780190633271.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 17 September 2019

(p.223) Appendix Research Methods

(p.223) Appendix Research Methods

Mixed Messages

Stefanie Mollborn

Oxford University Press

College Student Interviews

This book is primarily based on 57 in-depth qualitative interviews with college students at a large public university in the western United States. Students discussed norms and attitudes about and experiences with sex, contraception, and pregnancy during high school. This retrospective approach had the advantage of recruiting interviewees when they were no longer in their high school settings, allowing them to see the norms and social control they had been immersed in more easily than if our research team had tried to talk to them while they were still immersed. Interviewers asked participants for accounts of the norms in their social contexts and stories about teens they knew, as well as information about their own experiences. The interviews, which received institutional review board approval from the University of Colorado Boulder, were conducted in two phases. The first phase consisted of 43 peer interviews conducted during 2008–2009. Undergraduate students in two senior-level sociology classes received training in qualitative interviewing techniques and were given specific instructions for collecting data for this research project. They then conducted a semi-structured in-depth interview with a college student acquaintance of their choice.1 A minority of students chose a friend from their high school community, and most selected an interviewee with whom they already had rapport. A detailed interview guide specified the main interview questions, as well as suggested probe and follow-up questions. I also encouraged interviewers to ask their own follow-up questions when interesting information or comments arose in the interviews. This resulted in peer interviews that were fairly standardized but also customized to each interviewee’s unique situation. Peer interviewers recorded post-interview field notes describing the interviewee’s appearance and reactions during the interview, the setting, their thoughts on anything of note that happened, and their (p.224) take on what was interesting about the interview. The student interviewers then transcribed the interviews according to specific templates, as well as submitting an audio file of the interview.2

There were many advantages to the peer interviews. The interviewer-based sampling resulted in a group of interviewees that came from all over the United States and had a wide variety of backgrounds, rather than coming from shared social networks as a snowball sampling strategy would have yielded. The interaction between peer interviewer and interviewee also provided different information than did the second-phase interviews in which the interviewers were adult strangers. Peer interviewing techniques have been used successfully for sensitive topics, such as sexual behavior, that may result in a less managed presentation of self and more open disclosure with a familiar peer interviewer than with an older stranger.3 This project certainly seemed to benefit from including peer interviews. For example, two interviewees disclosed previous pregnancies and abortions to the peer interviewers, but none did so to adult interviewers. Interviewees also used more openly negative language when talking about teen parents with peer interviewers than with adult strangers. The minority of interviews that were conducted with a peer who had gone to the same high school yielded interesting data, as the shared experiences of interviewer and interviewee seemed to result in greater detail and richer description.

Alongside their many benefits, the peer interviews had two main drawbacks that prompted my decision to collect additional interview data: A small minority of peer interviewers had done a perfunctory job (asking far fewer probe and follow-up questions than I would have done), and the set interview guide didn’t allow me to change the questions as themes emerged from the data. Thus, to supplement the peer interviews, I conducted a second phase of 14 in-depth, less structured interviews that employed a purposive sampling strategy. I led a female research team that included a graduate student and a senior undergraduate student who had participated earlier as a high-quality peer interviewer. My research team conducted interviews with 14 undergraduates, who were paid $10 each. We worked alone or in pairs, and I conducted more than half of the interviews.

Phase two students were recruited through a campus-wide student email list, which generated a pool of 89 potential respondents. Recruitment emphasized the project’s particular interest in first-generation college students and students who came from a rural or poor community. This focus was intended to balance out the sample because these groups were underrepresented in the university’s student population, but other students were also included.4 The 20 students selected to be interviewed were split evenly between the targeted lower-SES or rural population and the overall sample, and 70 percent of those contacted completed an interview. Although the interviewees and interviewers were strangers, which may have hampered the disclosure of sensitive information, this phase of data collection permitted more detailed probing about important themes that arose inductively in the first round of interviews. This strategy of “abduction” is useful for developing and testing theoretical ideas as they evolve.5 I used an interview guide that went into greater depth and allowed for more customized follow-up than did the interview guide for the first phase. If interviewees addressed a theme that had been identified as emerging from the first round of interviews, the interviewer sometimes articulated that theme and asked them for feedback about whether that made sense to them. This often yielded detailed information to expand on a theme.6 Interviews continued until saturation was reached.7

Both phases of interviews focused on teens’ experiences during high school. The topics covered included interviewees’ perceptions of norms about teenage sex, contraception, and childbearing (p.225) in their peer groups, families, schools, and communities during high school; how these norms were communicated; how they influenced their own and their peers’ sexual behaviors; and what sanctions violators of these norms faced. My research team asked about communication with parents and peers about these issues, as well as for stories about peers’ experiences. I replaced a section from the first phase’s interview guide that had asked interviewees about their current support from family members with a new section asking them to compare their high school’s normative climate around sexuality with the normative climate at their current university. This yielded interesting data that highlighted the consistently negative messages about sex communicated to teenagers in all communities, but not to college students. Questions about norms started out open-ended so teens could identify normative messages and targeted behaviors on their own, and the interviewer followed the open-ended questions with extensive probe questions. Although the bulk of the interview concerned events and attitudes from the past and was therefore probably subject to some degree of recall bias, interviewees’ high school experiences were still fairly recent, and they seemed to have little trouble recalling them.8 About a third of interviewees came from the same city as at least one other interviewee, and sometimes they had even attended the same high schools. Their hometowns fit into the same community type in a clear majority of cases, bolstering my confidence in the reliability of interviewees’ accounts of community norms.

Peer interviewers in phase one, and three graduate students and one undergraduate student in phase two, transcribed all interviews. The research team imported the transcripts into QSR NVivo qualitative analysis software. Transcripts were then coded using three techniques. First, the research team coded responses by interview question, including simple distinctions between answers (e.g., positive versus negative attitudes toward teen sex among the interviewee’s close friends). Second, I read all of the transcripts and identified important themes that emerged from the data. These themes were then coded in other transcripts. Third, I identified key characteristics of interviewees and their communities and compared findings across categories. Community-level SES and religion and teens’ sexual experience emerged through this process as important for understanding the findings. The four predominant community types described in Chapter 6 arose in the second stage of coding and were connected with demographic variables during the third stage. A graduate student member of the research team and I coded cases into these community types independently, arriving at 88 percent initial agreement. We then discussed discrepant cases to arrive at a final coding decision. Fourth, I read all the transcripts one at a time to conduct whole-case analyses that linked themes together. Finally, I searched the body of electronic transcript files for key terms to make sure particular quotes hadn’t been missed in earlier coding. Some of these steps were conducted multiple times at different points in the analysis and writing process.

I compiled demographic information at both the community and individual levels. Table 1.1 in Chapter 1 shows interviewee characteristics on a variety of factors, and Table A.1 displays the age distribution of interviewees.

Ninety-two percent of interviewees identified as heterosexual or straight at the time of the interview. Of the sexual minority interviewees, half identified as gay or bisexual and half as “mostly heterosexual.” This latter group receives less recognition as sexual minorities but experiences a wide range of disadvantages compared to heterosexual individuals.9 It seemed that many or most of the sexual minority respondents had started self-identifying as such sometime after (p.226) finishing high school. For this reason, unfortunately, the interviews didn’t include enough data on the high school experiences of sexual minority students to permit analysis. This is an important topic for future research.

Table A.1 College student interviewees’ ages


% of Interviewees















Table A.2 Interviewees’ sexual experience in high school, by gender


% of Interviewees

Within Gender, % Not Sexually Experienced

Within Gender, % Always Used Contraception

Within Gender, % Didn’t Always Use Contraception











None of the interviewees had become parents, though two told us they had aborted a pregnancy. These proportions are lower than in the general population of young people, likely because the interviewees were all college students—few teen parents end up in four-year colleges.10 A similar proportion of interviewees had heterosexual intercourse in high school compared to national estimates (Table A.2). Forty percent of interviewees (42 percent of women and 37 percent of men) reported remaining sexually abstinent throughout high school. Another 40 percent (again, 42 percent of women and 37 percent of men) were sexually active, but they or a partner always used contraception. Twenty percent of interviewees (16 percent of women and 26 percent of men), or one third of all sexually active interviewees, had sex without contraception at least once during high school.

Table A.3 shows that interviewees’ sexual experience also varied by community type (described in Chapter 6). I do not include information on the sexual experiences of the very small group of interviewees with hybrid community types, although the theoretically important hybrid communities were analyzed in Chapter 6. Interviewees in the generally lower-SES “it could be you” and “it’s wrong, but” communities had higher percentages of inconsistent contraception and lower percentages of sexual abstinence than those in the higher-SES community types. Thus, the starkly different norm sets in these two community types were not reflected in differences in sexual behavior. Interviewees from the generally higher-SES, highly religious “it’s wrong” communities reported the highest levels of sexual abstinence (50 percent of interviewees in this group) and the lowest prevalence of inconsistent contraception (13 percent). (p.227)

Table A.3 Interviewees’ community types, by sexual experience in high school

Community Type

% of Interviewees

% Not Sexually Experienced

% Always Used Contraception

% Didn’t Always Use Contraception

“It could be you”





“Be careful”





“It’s wrong”





“It’s wrong, but”







Teen Parent Interviews

I use interviews with teen mothers and fathers to illustrate normative processes and norm enforcer and norm target strategies among people who have violated teen sexuality norms. By having a child before age 20, these interviewees had all violated societal norms against teen sex, inconsistent contraception, and teen childbearing. My research team conducted in-depth interviews with current and former teen mothers and teen fathers (the latter is a particularly difficult group to reach), combined with limited participant observation, at a school and a clinic in the Denver metropolitan area.11 With teen mothers accounting for 12 percent of all births, Denver is fairly typical of U.S. cities.12 The interviews were conducted in the fall and winter of 2008–2009, with University of Colorado Boulder institutional review board approval. The researchers, who conducted interviews alone or in pairs, consisted of Professor Janet Jacobs and I, one male and one female graduate student, and one undergraduate who had been a teen mother. To be eligible, interviewees had to have had a child before turning 20. Most had babies or toddlers, but a few had older children, allowing us to observe both short- and long-term experiences of teenage childbearing. Fifty-five interviewees had been teen mothers, and 21 had been teen fathers.

When studying a population such as teen mothers and especially teen fathers that is hard to reach because teen parents make up a very small percentage of the overall population, qualitative researchers tend to rely on either samples drawn from sites serving the populations or snowball samples. Our research team chose the former strategy because we could reach a wider variety of interviewees, both mothers and fathers, who were not from the same social networks and neighborhoods. We selected the sites because they permitted access to a large pool of teen fathers and mothers from around the metropolitan area; because they served different populations in terms of race/ethnicity, age, geographic location, and educational aspirations; and because they provided quite different levels of resources to young parents. We conducted 28 interviews at a school for pregnant and parenting teen girls that offered considerable resources to its typically financially needy students, including onsite childcare, basic medical and psychological services, career (p.228) counseling, and a “school store” to exchange attendance credits for diapers and other items. The school also had a satellite program for young fathers. Another 48 interviews were conducted at a hospital-based medical clinic for privately insured and Medicaid patients that provided tandem healthcare to teen mothers and their children. Despite the considerable support provided by the school, many of the school-based interviewees were younger and appeared to have fewer personal and family resources than those from the clinic. Reflecting their substantial social disadvantage, substantial minority of interviewees from both sites talked about personal backgrounds of incarceration, substance abuse, victimization, or mental health problems, and these issues were even more common in their families.

We recruited young parents by distributing flyers at the sites and giving them to interviewees to share with friends (though this latter strategy resulted in very few interviews). Staff at the sites also handed out flyers and described the study to parents they identified as eligible. Interviewees’ ages ranged from 15 to 38, but only four were older than 24. Our interviewees had their first child between ages 14 and 19 with an average of 17. This is younger than the average age for teen births in the United States, tapping into a more marginalized population.13 The average age of their oldest child was two. Two thirds of mothers and half of fathers had just one child, and most had grown up in or around Denver. Ten interviewees identified as multiracial or multiethnic, while others described themselves using a single racial or ethnic label: 36 self-identified as Latina/Hispanic/Mexican American, 21 as African American/Black, 3 as White, 2 as Native American, and 1 as Middle Eastern. Compared to Denver’s 2006 teen births (of which 75 percent were to Latina mothers, 12 percent to White mothers, and 11 percent to Black mothers), our sample likely overrepresents African Americans and underrepresents Latinas and Whites, though our inclusion of multiracial identities makes comparison difficult.14 In analyzing our data to examine the role of race/ethnicity, we found considerable similarity in Denver-area teen parents’ experiences across categories. Forty interviewees were enrolled in school, and 26 were working for pay. Eighteen interviewees (24 percent) were married at the time of the interview; among all Denver teen mothers, 23 percent were married at the time of their child’s birth.15 Twenty interviewees were single, and the others identified themselves as dating, in a relationship, living with someone, or engaged. Nearly all mothers and about half of fathers were living with at least one of their children. More than one third lived with a parent or parent-in-law, about one third with a sibling or sibling-in-law, and about one third with some other adult, such as an aunt or stepparent. Many lived with people from more than one of these categories.

Our research team took an ethnographic approach to the interviews and sites, observing the sites and interactions involving interviewees during our visits to meet with staff and conduct interviews, and taking field notes. Interviews about interviewees’ experiences with teen parenthood lasted about 45 minutes, and they received a $30 gift card. The interviews tended to include the same questions, but they were often asked in a different order depending on the direction the conversation went. We also asked follow-up questions to flesh out interviewees’ accounts, and many of these were unique for each interview. Interviews were digitally recorded and transcribed. Topics covered during the interview included interviewees’ perceptions of messages about teen pregnancy from their peers, families, communities, and broader society; their fertility decisions and behaviors; and people’s reactions to the pregnancy. Their experiences before, during, and after pregnancy were discussed, as well as their ideas of what a good parent is. Interviewees described the resources available to them, who provided them, how these arrangements were negotiated, and how available resources had affected their lives.

(p.229) Transcribed interviews were kept in searchable electronic format and were manually coded in the NVivo qualitative software package. First, the research team coded responses to each question, making simple distinctions between answers (e.g., a certain family member’s positive versus negative reactions to the pregnancy). Second, I read whole transcripts and identified important emergent themes, which were then coded for all transcripts. Third, I searched for key words and phrases to make sure all instances of a theme had been captured by coding.

How the Study Came About

I have studied teen parenthood for more than 15 years, analyzing national longitudinal surveys and in-depth open-ended interviews to examine two topics: the consequences of teen parenthood for women and men and their children, and norms and stigma around teen parenthood. In my first years of research on the topic, through statistical analyses of nationally representative surveys I identified socioeconomic, racial/ethnic, and gender influences on individuals’ perceived norms against teen pregnancy; linked these norms to the withholding of needed resources from teen parents; and found that a lack of resources after the birth largely explains why teen mothers’ and fathers’ educational attainment is compromised.

While doing this quantitative research, I felt strongly that it would be important to talk to young parents about their life experiences to flesh out the processes that had been identified using these large surveys. Qualitative interview data complement quantitative survey data. While the latter can statistically demonstrate significant relationships between two or more researcher-identified factors, such as norms against teen parenthood and the provision of resources to teens, interviews can reveal how people make sense of what is actually happening beneath the numbers, highlighting factors that may have been initially overlooked by the researcher.16 Qualitative interview data can capture a more complete picture of the complex and intertwined factors that figure into a teenager’s experiences. It can also show how young people talk about teen sexuality and parenthood.17 Together with my colleague and mentor Janet Jacobs and a research team, I set out to conduct in-depth interviews with teen mothers, fathers, and couples. These interviews are the supplemental data used throughout the book and are described below. Hearing about the experiences of young parents yielded many important insights beyond those included here.18

But conducting the interviews with teen parents left me convinced that an important piece of the puzzle was still missing. I had heard about normative messages and social control of teens who were sexually active, didn’t use contraception consistently, didn’t get an abortion, and became parents. Both their experiences and their communities—almost all urban and of low SES—were unusual compared to the majority of teens in the United States. I needed a second set of interviews as a counterweight to these. I wanted to talk to teens who had a variety of sexual experiences, who hadn’t become parents, and who came from a more diverse set of communities and social contexts.

This approach is quite different from nearly all the existing research on teen sexuality and parenthood. That research tends to fall into one of two camps: statistical analyses of large-scale survey data that examine relationships between variables that the researchers have identified, or qualitative research (observations and/or interviews) that examines a particular community—almost always one that is considered high risk—in detail. I would characterize the former (p.230) approach (which I have often taken) as “an inch deep and a mile wide,” providing a broad and informative national picture of a specific outcome or relationship. The latter approach is “an inch wide and a mile deep,” yielding a rich understanding of one particular social context without comparison to others. In this approach, “typical” lower-risk groups of teens tend to be ignored.

Strengths and Limitations of the Study

The analytic space between these two approaches interested me for this project. I wanted to gather a moderate amount of qualitative detail about a large and diverse set of communities. The goal is not for the findings to be generalizable to a broader population, nor can they provide anywhere near the depth that single-site ethnographic studies can.19 Yet they do provide breadth for understanding differences and similarities in messages about teen sexuality and social control of teens in the United States today. Because the findings represent individuals and communities from around the country, their reach is quite broad for a qualitative study. I think this strategy can help elucidate complex processes underlying social norms that regulate teen sexual behavior, and these theoretical tools may be of use to researchers working in other settings or studying other phenomena.

Especially because I purposively sampled additional people from low-SES and rural communities (who are underrepresented in the college student population), the interviews with college students better characterize the national diversity of normative climates around teen sexuality than the interviews with teen parents do. For that reason, the college student interviews are the book’s primary data source. But the teen parent interviews are an important supplementary dataset because they can tell us about what happens when teens violate the strongest teen sexuality norms—those that proscribe teen parenthood—thereby illuminating sanctioning processes. I was surprised by the consistency in descriptions of normative climates and social control between the two sets of interviews, since most of the interviewees came from very different walks of life. For example, almost no adults anywhere are giving teens encouragement to be sexually active or to become pregnant in their teenage years. My data reinforce Amy Schalet’s assessment that in the United States, pretty much everyone views teen sex as inherently risky and perceives teen sexual behavior as a struggle of adults against teens and boys against girls.20 Having such a wide variety of communities in the samples allows me to identify these kinds of universal normative messages, as well as many messages that are not universal.

Although my data collection strategy has many advantages, there are also drawbacks. First, relying on people’s own reports of social phenomena, as both in-depth interviews and surveys do, means that the data represent an altered version of social reality that has been “filtered” through the respondent’s own perception and verbal account.21 I make use of that perspective in this book. Using self-reported data from interviews or surveys means that the researcher has to accept people’s accounts of their own and others’ behaviors and motivations, seeing them as important narrative constructions that articulate culture. Interviewees’ representations of their sexual behaviors could be exaggerated or untrue, a problem that also applies to surveys. But these accounts can also be an advantage for studying culture (of which norms are a part) because they capture people’s justifications for their behavior.

(p.231) Second, I chose to do retrospective interviews with both the college students and teen parents, which means that interviewees’ accounts are probably different than they would have been at the time an event occurred. But an advantage of retrospective interviewing is that the interviewees have gained some distance from the events in question. Social norms tend to be invisible to people until they are violated or until people are put in a different normative climate. So it can be helpful to talk to people when they are in a different time and place—and hopefully thereby a different normative climate—and can perceive and articulate the earlier situation’s norms more clearly. The college student interviewees often contrasted their normative climates during high school with those in college, whether prompted to do so or not. This suggests that the contrast helped shed some light on their high school experiences, making their surprisingly articulate accounts of those normative climates easier to talk about.

Third, the standpoints of both the interviewees and my research teams shaped both the data that were collected and my analysis of the data. Even if some of them came from less privileged communities, all of the interviewees in my primary dataset were still college students. This means that they either came from privileged communities or occupied at least a partial “outsider” and presumably more privileged position in their less privileged communities. Because being an outsider—who probably has access to an alternative set of norms or attitudes—can also help make norms more visible, it may have been helpful for sharpening interviewees’ accounts in less privileged communities. But it also probably shaped those narratives, for example because interviewees’ close friends were also more likely to be college-bound and not typical of less privileged communities.

My standpoint also matters. I am a White woman and was in my early thirties at the time of data collection. Like those of the sample, my experiences represent considerable breadth in norms around sexuality. I grew up in a solidly low-SES, highly religious rural community like many of the “it’s wrong, but” communities in Chapter 6. Since then, I have lived in Sweden for four years, attended two private universities in different regions of the United States, and spent a decade working at a public university in a liberal, highly educated city. I have had the advantage of being exposed to radically different normative climates regarding sexuality and adolescence, and these experiences have shaped my reading of the data. I worked to triangulate standpoints in data collection and interpretation by relying on many different peer interviewers and giving the interviewers some leeway in what questions they asked. I approached coding and data analysis from a variety of different angles and worked with collaborators to improve the reliability of some key coding decisions.


The data analyzed in this book are meant to fill several gaps in previous research. I sampled interviewees from a wide variety of communities, included boys as well as girls, and focused primarily on fairly typical teens rather than those from particularly marginalized urban communities who had violated norms against teen parenthood. Combining interviews with college students and those with teen mothers and fathers broadens the reach of the study, engaging demographically distinct kinds of communities and including both people who did and people who did not violate teen sexuality norms. Yet the norm sets described by the college students and the teen parents (p.232) sounded very similar when they came from similar communities. The goal of this book was to document U.S. normative climates around teen sexuality and their links to social control and norm enforcer and norm target strategies. This goal required a broad approach. Next steps for future research include examining particular communities in greater depth to more fully articulate these dynamics of norms, social control, and agency in specific settings, as well as applying the ideas about norms to other phenomena.


(1.) This interview was part of a graded research project for the courses, one of which was taught by a colleague and the other by me. Students could choose an alternative assignment instead of conducting the interview, and they or the interviewee could choose not to release the interview data to this research project. To avoid the possibility that students would feel coerced into releasing the data, this decision was not made known to the course instructor or me until after final grades had been submitted.

(2.) Students were graded on the quality of their interviews and supporting materials, as well as on a paper analyzing their interview and linking it to course materials.

(4.) This strategy of “sampling for range” is useful when a particular subgroup has been identified that is important for understanding the phenomenon at hand (Small 2009).

(6.) Here is an example of this process at work, from the interview with Finn (after he talked about his community taking a practical approach to encouraging condom use): (p.253)

So often in communities where I’m hearing this really practical attitude toward condoms, it’s in places where teachers and parents are fairly resigned to the fact that teens are sexually active. Was that the case here?

Yeah. They accepted it. They never made a conscious effort to stop it because they would be chastised by the teens probably. They had the power. So if the parent was like, “Don’t have sex,” the teen would be like, “You’re so stupid, shut up.” At the same time, the smarter parents would be like, “All right, boys and girls can come over and do stuff, but leave the door open in the room. We know you’re going to go in the room; just leave the door open.” It was like, I guess, sort of common. It was mainly like, we weren’t put in very many situations where it was possible. We don’t have cars. If you do have a car we’re on base and freaked out about the law. Yet to go somewhere, you don’t have a bed anywhere. You have to be in the perfect scenario because every girl has grown up in this romantic paradigm, … it has to be perfect, you know, a beautiful bed. And so all these situations have to be good.

(11.) One pilot interview, recruited through a personal contact, was conducted in the first author’s office.

(13.) Ibid.

(14.) Ibid.

(15.) Ibid.

(17.) For perspectives on ongoing debates about measurement and methodology in cultural and qualitative sociology, see Berezin (2014); Jerolmack and Khan (2014); Lamont and Swidler (2014); and Pugh (2013).