Qualitative Analyses of Couples’ Text Messages: Methods Reflection

Abstract

Qualitative data analysis of text messages presents a unique methodological challenge for researchers and is largely absent specific guidelines to determine best practices in analysis as well as team management. This is especially true when the text messages are sexual and intimate in nature as researchers have new insight into previously private partnered communication. This research note explores the challenges and considerations in qualitative text message, specifically sexting, research. Although most current text message research is quantitative in nature rather than using qualitative analyses, the ubiquity of text messaging as a form of communication necessitates other approaches. Specifically, we explore how emotional this type of study can be for researchers and provide recommendations for how to manage researchers’ interactions with the data. We also discuss how reading and coding sexts and text messages between new couples can force researchers to uncomfortably reflect on their own relationships and how best to manage these challenges. We propose that this work is “messy” and teams interested in analysis of text messages should be structurally prepared for the emotional labor required.

Introduction

Sexting is one form of communication young adults use to explore their sexual identities allowing them to express their sexual interests (Valkenburg & Peter, 2011), establish intimacy (Walrave et al., 2015), or negotiate condom use (Broaddus & Dickson-Gomez, 2013). Sexting is defined as “the interpersonal exchange of self-produced sexualized texts and above all images (photos, videos) via cell phone or the internet” (Döring, 2014, p. 1). Rarely have prior investigations explored how sexting patterns are developed and maintained in the early stages of a relationship. Furthermore, the work that has been done has been more quantitative in nature (Holtgraves, 2011; Underwood et al., 2012) and has allowed for researchers to emotionally distance themselves from the content of the sexts. The current paper offers a methodological reflection on the emotionally draining nature of qualitative work focused on sexting in early couple relationships.

Early on in relationships, couples discuss and develop norms surrounding beliefs, risk behavior, and health behaviors, which may then be maintained over time. For example, young adult dating couples tend to discontinue condom use early on in relationships (e.g., a sharp decline after nine coital events or 21 days; Fortenberry et al., 2002; He et al., 2016). However, we know very little about the conversations that may or may not occur surrounding the decision to discontinue condom use. Though romantic relationships can and do begin at any age, young adulthood is a particularly important time for the development of relationships as well as engagement in sexual behaviors that can have both short- and long-term repercussions for health and well-being. Therefore, expanding our understanding of how sexting and overall communication occurs among partners early in the dating relationship is critical to designing interventions that promote healthy behaviors in young romantic couples. It is also important to explore the content of sexts to understand the nuance and evolution of this particular form of communication—a program of research ideally approached from a qualitative perspective. This allows researchers to understand people’s natural communication instead of changing the approach to test a hypothesis (Gough & Lyons, 2016; Tracy, 2020).

The widespread use of mobile phones technology has rapidly changed the way young people communicate, flirt, date, and even participate in sexual activity (O'Sullivan, 2014; Pew Resarch Center, 2019; Walrave et al., 2015). Research to date suggests that sexting is prevalent among older adolescents and young adults (Davis et al., 2016; Klettke et al., 2014; Madigan et al., 2018), and has increased since the beginning of the COVID-19 global pandemic (Döring, 2020; Lehmiller et al., 2020). In the past, many were concerned about the relationship between sexting and negative consequences, specifically risky sexual behavior (Klettke et al., 2014). However, more recent work focusing on sexuality changes due to the pandemic highlights a continuing trend of technology-mediated sexuality, specifically that text messaging allows for a novel way to develop digital intimacy (Banerjee & Rao, 2020; Döring, 2020; Lehmiller et al., 2020; Lopes et al., 2020).

Many sexting studies rely on self-report assessments that are prone to reporting biases (Benotsch et al., 2013; Davis et al., 2016). This has likely resulted in inconclusive evidence regarding how sexting as a form of intimacy develops (Burkett, 2015; Van Ouytsel et al., 2018), is related to sexual or relationship satisfaction (Drouin et al., 2017; McDaniel & Drouin, 2015), and even how sexting relates to sexual health outcomes (Benotsch et al., 2013; Gordon-Messer et al., 2013). There are, however, a minority of studies that have analyzed the content of text messages between romantic partners and/or peers (Brinkley et al., 2017; Ehrenreich et al., 2014; Holtgraves, 2011; Kemp & Grace, 2017; Underwood et al., 2012). These studies have provided naturalistic assessments of the digital communication patterns of participants (Rizzo et al., 2019). Still, the few studies that have examined text message content have some limitations. Much of this research is limited to analyzing the text messages between romantic partners and peers for a brief period of time (e.g. Brinkley et al., 2017; Ehrenreich et al., 2014; Kemp & Grace, 2017), focused on adolescent relationships (Brinkley et al., 2017; Ehrenreich et al., 2014; Holtgraves, 2011; Kemp & Grace, 2017; Underwood et al., 2012), and/or utilized computational approaches like Linguistic Inquiry and Word Count (LIWC, Holtgraves, 2011; Pennebaker et al., 2001; Underwood et al., 2012). Our study sought to expand on this foundational work by (1) analyzing the content of text messages over a longer period of time (i.e., 6 months), (2) focusing on those in a relationship for 6 months or less, young adult couples, and (3) conducting manual coding to identify sexting themes both within and across couples. The nascent field of sexuality-related texting research calls for exploratory approaches. As such, we prioritized a qualitative approach rather than the quantization that can result from using computational approaches (such as LIWC).

The increasing use of text messages as a form of data for social and behavioral researchers has created new ethical and methodological considerations, and this article aims to discuss the emotional work and impact of this from the researcher’s perspective (Hanna, 2019). Qualitative work is often considered “messy” with less prescriptive epistemologies (Lester et al., 2020), norms, and procedures than quantitative research (Denzin & Lincoln, 2011). As Hanna (2019) describes in their work analyzing online posts, analysis may entail large amounts of emotional labor from the researchers. Hanna describes feeling “wrung out” (Hanna, 2019, p. 352) in a way that may be more typical for emotional face-to-face data collection methods. Nelson (2020) also explored the idea of researcher emotions and how they might influence interpretation of the data as well as force researchers to develop an emotional management system for themselves. We argue, along with several qualitative researchers (Dickson-Swift et al., 2006, 2007; Hanna, 2019; Nelson, 2020), that this emotional labor researchers experience should not be ignored; rather, it should be incorporated into a general framework of the research process.

The current paper focuses on qualitative text message methodology rather than findings from the study. Specifically, we discuss the difficulties in collecting this type of data (i.e., text messages between intimate partners), as well as the “dynamic and contextual complexity” of the researchers’ interactions with the data during data analysis (Irvine & Gaffikin, 2006, p. 116). We asked the research team to reflect on a number of questions using a shared written document and two synchronous online conversations embedded within our project meetings. Team members were encouraged to reflect on all the questions, but not required to respond to all prompts. The faculty members selected responses from the team to identify main themes as well as ensure representation of the authors as much as possible. Selected responses are reported verbatim. After our themes were identified, we organized them based in part Hanna’s (2019) themes, such as the phrasing of “What happened next?“.

Background of the Study

The larger study sought to understand sex communication practices among a sample of young adult couple in new relationships (0–6 months) using naturally occurring communication (i.e. text messages).

Study Procedures

Participants were eligible for the study if: (1) they were a part of an different gender couple, (2) one person in the couple identified as a cisgender woman aged 18–25 years old; or a cisgender man at least 18 years old, (3) in a relationship for 6 months or less, (4) engaged in sexual activity with their partner in the past 6 months, (5) were able to speak English, and (6) one member of the couple owned an iPhone. We focused on different gender couples due to the anticipated nuances in same-gender dynamics of sexual health communication, such as perceptions of risk or contraception needs (e.g., Dalessandro, 2020; Haas et al., 2020). Participant age eligibility requirements considered potential age differences in emerging adult relationships. Participants were recruited via a large Southeastern university, although participation was not limited to university students.

Participants were recruited using various methods. These included the undergraduate psychology participant pool (SONA), university-wide email blasts, and announcements in undergraduate courses. Couples who met all eligibility criteria were invited into the lab space to complete an informed consent form. Once consent was received, all text messages sent and received by the couple in the preceding 6 months were downloaded using the program: Ecamm Phoneview. Ecamm Phoneview was compatible with iPhones making it necessary for one person in the couple to own an iPhone. Additionally, at the time of data collection (2018–2021) the program did not offer other communication platforms for download (e.g., WhatsApp); thus, only text messages were collected. Text messages included any use of emojis but did not include pictures or videos, and any messages that participants did not want to be included were able to be deleted prior to downloading. This was a requirement arranged with the university institutional review board.

Due to the impacts of the global Covid-19 pandemic in 2020, in-person lab procedures were halted, and procedures changed to be entirely remote. Remote procedures involved participants meeting with undergraduate research assistants virtually to complete the informed consent and text message downloads.

Research Team

Our team consists of two faculty members and five undergraduate students, from psychology and public health sciences. Of the seven of us, all are under the age of 40 and identify as (cisgender) women, two of us identify as people of color and five identify as white, three identify as bisexual and four are heterosexual, and two of us were single and dating while the others were in committed relationships during this project. We used a team of a total of 10 coders, which was necessary given the amount of data to be coded.

What happened before?

Institutional Challenges

Erika: The original conceptualization of the study attempted to track text messages over a 6-month time period shared with all sexual partners (e.g., casual and steady) since sexting and digital intimacy can occur within all relationship types. Given the sheer number of potential interactions with various people during that time, we requested a waiver of third-party consent from the Institutional Review Board (IRB). At the time (2017), the federal and state telephone privacy laws (18 U.S.C. Sec. 2511(2) (d)), [redacted for peer review]) stated that only one party on a telephone communication is required to provide consent to monitoring communication. However, federal and state laws are lagging behind technological developments and the IRB and the University legal office had questions about how these laws may or may not have been applicable to text messages. This was a difficult process to navigate for a number of reasons. All communication about the project had to go through the IRB; I, as the principal investigator on the project, was advised against communicating directly with the research development legal administrators at the university. Second, [redacted for peer review] is designated as a R2 institution and is working to be more research active. As the university builds its capacities, its processes and policies related to “riskier” projects, such as the current study, will be more established and streamlined. In the end, however, members of the IRB and I were able to compromise, which included obtaining consent from both members of new couples (i.e., dating for 6 months or less), and move the study forward.

Data Collection

Recruitment

Sydney: The process of recruiting participants is daunting even when only dealing with contacting a single person and scheduling them to come into the lab. However, trying to contact two people and correspond their schedules so that we could go through the screening process and then schedule them to come into the lab together was extremely challenging.

Erika: We had anticipated recruitment to continue being difficult once we moved the study completely online due to the COVID-19 pandemic. Actually, our protocol became even more complicated since we could not bring couples into the lab, so research assistants used video conferencing with couples and instructed them how to download and send their texts to the research team. However, prior to the pandemic, we had to turn interested participants away because we needed them to come into our lab to complete the study. We no longer had those restrictions, which means we’ve been able to successfully recruit more long-distance couples and those not at the university. Also, during our first massive listserv recruitment during the pandemic we had greater interest in the study than ever before. In some very surprising ways, the pandemic has made this study easier to recruit for by being more accessible to more couples and perhaps folx having more time to participate in studies.

Screening

Sydney: When partners were together during my scheduling call, it saved a lot of time for me but also the participants since they could look at their schedules together. There were a few more couples that were together when I called during the COVID pandemic than when I would call pre-COVID. COVID made it simpler to reach individuals because they could answer the phone during other online activities. It was much easier to schedule them to participate too because they could be in different places as long as they were on the call together.

Erika: We attempted to design a screening process that was quite thorough--both members of the couple were asked the same questions and their answers had to align for them to be enrolled. However, once we began coding the content of the text messages it became clear that a minority of couples were (1) fabricating some portion of the screener, and (2) there was some level of coercion occurring within the partnership to participate in the study. For example, one couple reported to us during the eligibility screening that they had been together for less than 6 months. We discovered while coding that we were capturing the couples’ final texts before enrolling in the study and this particular couple had an entire conversation about getting their story straight about how long they had been dating. Given that this project focused on how new couples develop digital intimacy, I am currently grappling with how we can prevent this in the future and if/when to exclude those couples from data analysis. Other couples negotiated being in the study at all and usually one partner was way more excited about the idea of the study than the other. Some, but not all, couples had one member use coercion (e.g., “you owe me this.“) to get the other to agree. Bearing witness to couples’ negotiations and coercion about being in the study was deeply uncomfortable for me. It feels like an ethically gray area to have signed consents from all parties involved, but also data to indicate that maybe some were not participating in the study completely voluntarily.

What Happened During?

Coding

In order to explore the content and purpose of texting beyond self-report, we used a qualitative inductive approach (Thomas, 2006). We first conducted open coding to identify meaningful excerpts from the texts and then axial coding to group common ideas together for categories. These categories were then refined and combined to create themes. When put into Word from the spreadsheet output, the text messages between participants ranged from 17 to 714 pages. We used Dedoose qualitative software and had challenges when pulling in emojis that would not convert for coding. As a response to these challenges, we decided to redact the emojis by inserting Apple emoji descriptions which increased the data cleaning and redaction time.

How Did Your Identity Influence Your Coding?

Jessamyn: As a White person who has been thinking about racism in light of the continued social uprisings and movement for Black lives, I noticed that I focused on cues for Black identities in text messages and wondered about how that was influencing the dynamics. Yet I didn’t wonder the same thing about how Whiteness was showing up. This may have affected how I interpreted comments that may have been intended as humorous (such as one male partner saying to a female partner, “Don’t be a fag, just come over”) as violence (including homonegativity).

Stuti: As someone who identifies as bisexual, I have been a part of hetero and homosexual relationships, but with data from strictly heterosexual couples I found myself comparing my coding experiences to solely my heterosexual past relationships which I thought was interesting. I also found myself thinking more heteronormative things such as presuming the female participant is needy if she initiated conversations more and realized I had expectations for how each gender’s texting behaviors should vary.

Erika: I found my Whiteness top of mind when reading through the text messages and tended to focus on cues of racism and microaggressions. I noticed this happening particularly when I coded our bi-racial couples. While I am not in an interracial relationship now, I have been in the past and I found myself wondering if I had ever communicated with my partners so thoughtlessly.

What was Frustrating About this Method?

Jessamyn: The lack of videos and photos sent by the participants through text messages was inhibiting since it seemed like such a rich piece of the conversation.

Diana: I agree with Jessamyn in that not having photos was a drawback since we couldn’t see more content. Furthermore, if they were talking about multiple friends we were not able to see who they were talking about specifically later on.

Stuti: The platform switching (e.g. Snapchat) and gaps in conversation were really frustrating. Also it was frustrating when couples would be talking about something important and agree to finish the conversation on a call or in person. I feel as though we missed many important parts and experienced some disconnect and jumps in their growth as a couple.

Sarai: It’s confusing when couples make references to things they talked about in real life. It’s also frustrating when partners say they have a lot to tell the other and decide to talk about it over the phone because we suspect this is where much of the emotional support and disclosures occurred.

Erika: It never ended - it was really hard to manage my time. I wanted to feel like when I was coding the couples that if I put in an hour of time, then I would see changes but often couples had thousands of lines of text messages. Inside jokes were hard to place. Defining flirting as a code required long conversation about it that necessitated a self-examination of my position in relation to romantic relationships. Am I interpreting this as flirting because that’s what I do? Or, on the other hand, am I interpreting this as annoying because I don’t like saccharine communication and I don’t communicate like that? It forced me to think about my relationship to these themes, which I didn’t always want to do. After our meeting with students [research assistants] on re-calibrating our flirting definition, in which we as faculty members described what we thought was flirting based in part on how we flirt, we were unsure if we’d done the right thing in revealing those details.

What was New About this Method?

Jessamyn: As a qualitative researcher who is more accustomed to interview and focus group discussion data, not being able to further inquire about specific things that were said in the texts felt challenging. However, this was tempered by the chance to analyze data that is naturally occurring and might be more reflective of how sexual health is discussed between partners.

Erika: The sheer magnitude of data that we had to manage was new. The cleaning and coding of it. The emotional labor was new too - managing that within the team was not something I’ve had to do. I’ve had to do it with quantitative data; I’m comfortable guiding students through frustrations like dealing with missing data and seeing ceiling effects. However, reading through couples’ text messages and re-examining personal relationships is not something I’m used to talking about with my research assistants but it was obviously something everyone needed for this study. This is the first dataset where when I’ve run into challenges where I talked to mentors and consulted guides in a variety of data analytic techniques that I haven’t gotten answers or guidance from experience; there is no best practice for this that I can find.

What Happened Next?

How did this Impact your Views on Relationships or Your Own relationship?

Jessamyn: I found myself comparing my relationship to the couples I was coding, wondering if I supported my partner’s goals enough or how that would show up in my texts.

Diana: While I was coding I realized how common/frequent unhealthy relationships are. Specifically with controlling behavior such as dictating who either the male or female could talk to (one couple would log onto each other’s social media accounts and delete pictures of the significant other with other people/ex-boyfriends/friends). In my own relationship with my boyfriend, I found that our texting behaviors can be boring from a [sexuality] researcher’s perspective since it is mostly scheduling and short conversations.

Stuti: My views on relationships, specifically my own, were changed as I began evaluating things I did not read into much before. For example, when I am coding couples who do not match in the attention they give each other, I reflect back on whether or not I give too much. I also reflect more on how healthy my relationship and dating habits would look like if someone were to read through my texts. Starting to date someone while coding was really interesting as I started to root for couples when they were being cutesy with one another, and feeling a little cringe at myself for recognizing unhealthy relationship norms in other couples that I also exhibit.

Sarai: I started to remember how my relationship was when we first started (we’ve been together for 4 years) and began to compare it with new couples in the honeymoon phase. I wanted to be more affectionate with my partner and I started to send loving text messages.

Jennifer: My current partner and I have been off and on, so I found myself thinking about how we handled more serious discussions the first time around and how we handled similar discussions again more recently. I’m not sure how interesting our text messages would be to someone analyzing them because I tend to wait until we FaceTime to bring up more serious topics since we are currently long distance.

Erika: When this study began, I had just begun dating my spouse and identified with the getting-to-know-you text exchanges that occur at the beginning of a relationship. Now, we have been married over a year and my connection to the text exchanges has changed. I am both relieved when reading text exchanges when it is clear one member of the couple does not feel secure in the relationship and also miss the excitement of a new relationship.

How did This Impact Your Thoughts on Text Messages?

Jessamyn: As someone who texts often, I sometimes send texts without thinking too much about it or editing my language; this raises questions for me about the weight to put on these data.

Diana: I think text messages should not be used when having important conversations since it’s easy to misinterpret messaging without seeing body language and facial expressions.

Stuti: I used to think that my texts were comparable with how I talk to my friends normally, but after dissecting many different couples’ texts I have now realized how different texting can be from normal conversation and how I use it to hide behind a screen and say what I am not able to convey in person. Also, I started a new relationship while coding and began mentally coding some of my texts as due to COVID-19 a lot of our conversations were through text.

Sarai: I never really use my text messages to have conversations, only make plans to meet in real life. Until this study I forgot that texts can actually be used to have real conversation and connect.

Jennifer: Overall, I now think text messages might be the most difficult type of communication. I found myself constantly wondering if the participants were joking and being playful or being serious, especially when it involved harsher language.

Erika: At the beginning, when I came across couples who were sexting I considered sexting more in my own relationship. That didn’t really work in my own situation. In the past, it’s always occurred at the beginning of relationships for me and I noticed a longing for that excitement and noticed that we’ve moved past that initial phase.

How did this Affect you Emotionally?

Stuti: When couples were not matching in how emotionally invested they were in the relationship or practiced other unhealthy relationship behaviors that I have personally experienced, it made me feel sad and a little hurt for my past for not being able to see my relationships from the outside until it was too late. Basically, it made me realize that I was missing a lot of red flags and patterns of behaviors which made me feel down.

Sarai: I was mostly shocked at how much new couples are willing to disclose and forgive during the beginning stages of their relationship. I also began to get frustrated with couples in which one was treating the other with less respect than they should be. Some couples were very sweet to each other and it made me hopeful that they would stay together.

Erika: This entire study has been emotionally draining--from the difficulties getting the study off the ground to the coding of the content of the texts. The text coding has been particularly difficult especially when the couple often fights. I found myself needing to decompress after those coding sessions. It’s also hard to “move-on” to a new couple after coding all of one couple’s messages. We discuss in our lab meetings about how these texts often read like a story and we’re left without an ending!

Conclusions

We have highlighted the challenges and considerations in conducting qualitative text message research, with notes about changes due to the COVID-19 pandemic. Given the relative paucity of data related to this methodology, we hope this will provide guidance for researchers moving forward with lessons learned.

Hanna (2019) extends Campbell’s (2002) argument for researchers to be “emotionally engaged researchers” with data that addresses sensitive topics, even when the data collection is not face-to-face. We find evidence that this may be needed for couples’ text message research as well. Our research team became emotionally invested in the success of couples they liked and concerned for couples who fought frequently via text. This emotional investment is like face-to-face emotionally draining research. However, the magnitude of text messages our researchers had to read through, and the time committed to each couple’s text exchanges may differ from face-to-face emotionally engaging research. Face-to-face research is often time limited even if it is emotional. Conversely, text message research can take several dozen hours to complete, and researchers spend much more time with the couples than if they had been interviewing them face-to-face. We found it vitally important to allow our researchers the space to explore these feelings in lab meetings. This support system follows the recommendations of Dickson-Swift et al. (2007) when researchers are doing sensitive research. Following the recommendations of Hanna (2019), we additionally created the opportunity for student researchers in the lab to keep diaries and reach out to faculty members for mentor support.

The paradigm that researchers should be detached from the data is pervasive and our team found that impossible. We spent hours with a couple’s data, getting to know and understanding their unique communication pattern, hoping for success and happiness for the partners. Our focus on text messages did not allow for participants’ subjective experiences or meaning-making in this study. As reflected in researchers’ frustration in this lack of follow-up, future research may pair naturalistic data (e.g. text messages) with participant experiences. This expansion may allow for more ethnographic approaches, in which researchers gather rich data regarding participants’ culture (Jorgenson, 2002). Text message research may lend itself toward ethnographic approaches, given the availability of “insider” culture between the texters (see for example, McSweeney, 2018; Rew & Hosterman, 2018; Spilioti, 2011; Quinn, 2020).

We argue that future guidelines for conducting qualitative analysis of text messages should include a focus on researchers’ well-being. This includes planned breaks, debriefing with formal and informal support systems, and guidance on when and how to disclose personal information (e.g., relationship experiences) as researchers may be experiencing similar things participants describe in the text messages (Dickson-Swift et al., 2007; Nelson, 2020). To date, researchers’ well-being is generally discussed in the context of emotionally engaging face-to-face research and we argue that this same consideration should be given to text message research as it can be equally emotionally taxing and sometimes more so.

Research on text methods is a low-cost and naturalistic approach to understanding how new young adult couples communicate and establish relationship norms. Similarly to Rizzo et al. (2019), we argue that this is not a low-burden data collection approach. Data management and analysis are incredibly time consuming and emotionally taxing for researchers. However, we hope with the help of the current paper, we have offered valuable insight into how best to prepare and manage similar projects. Collecting text messages between new young adult couples allows researchers the unique opportunity to explore the full range of communication and experiences within relationships.