Discussion

Sexting and its potential links to sexual behavior have garnered significant scholarly scrutiny in recent years, but this literature is marked by divergent perspectives and disparate findings (Doring, 2014). The deviance perspective that dominates this literature assumes that sexting is a risky practice and, therefore, should be associated with other risky practices, such as having multiple sex partners or unprotected sex; however, an emerging normalcy discourse challenges these assumptions (Cooper et al., 2016). To assess these claims, we conducted a meta-analytic and critical review of published research on sexting and its relationship to sexual behavior. Our review revealed several trends in the sexting literature, including the types of study designs and samples that were most common. In addition, we uncovered considerable variability in how sexting was measured across the 15 studies including in our analysis. Despite this variability, our meta-analysis was able to parse out significant relationships between sexting and three aspects of sexual behavior: general sexual activity, history of unprotected sex, and number of sexual partners. The following sections describe possible explanations for and implications of each of these findings.

First, our meta-analytic review of the 10 studies and 23 effects that dealt with sexting and general sexual activity revealed a substantive relationship between two, which undermines the catharsis argument. Although this finding might confirm parental concerns about sexting, these results must be interpreted with caution because the vast majority of studies reviewed were cross-sectional in nature. As such, all we know is that sexting and sexual behavior co-occur. Sexting might act as a gateway to other sexual activity, but it is just as likely that sexting is simply a part of an already sexually active person's sexual repertoire. The lack of attention to relational context in the sexting measures employed also makes this finding difficult to interpret. Including questions or question wording regarding the relationship between the sexters in future studies would provide valuable data. As Davis and colleagues (2016) noted, this information is needed to “differentiate between healthy sexual expression within a steady partnership versus a strategy to engage new and casual partnerships that may increase HIV risk” (pp. 139–140). Contextualizing these practices within particular relationships should invite questions regarding the relational (not just the sexual) implications of sexting and bring an important relational lens to bear. For example, studies linking sexting to sexual activity in committed relationships have limited utility; research on the possible links between sexting and other aspects of established relationships, such as relational stability and/or satisfaction, would be of greater use. Two recent publications (i.e., McDaniel Drouin, 2015; Parker, Blackburn, Perry, Hawks, 2013) have examined these links, but the use of small convenience samples in these studies make additional research warranted. Although communication scholars and theory are particularly well suited to these kinds of pursuits, none of the articles included in this meta-analysis were published in communication journals or were framed by communication theory. Studying sexting as a way of communicating and relating should encourage more communication research on this decidedly discursive activity and help bring to light previously unexamined aspects of sexting.

Our meta-analytic review also explored the relationship between sexting and two indicators of sexual risk—sex with multiple partners or without protection. Our analysis of the nine studies that examined unprotected sex and the seven studies that attended to the number of sex partners reported by participants uncovered relatively weak overall associations between engaging in sexting and risky sexual practices. These findings suggest that, although sexting might be an indicator of risky sexual practices, it is not a particularly good one. Experimental or longitudinal research on sexting and risky sexual behavior would help establish causality or time-order, but the weak association observed in this meta-analytic review indicates that the money, time, and energy spent on collecting experimental or longitudinal data might be better invested elsewhere. Although this finding provides support (albeit weak) for the deviance perspective and problem behavior theory, future work informed by either or both of these perspectives would be wise to move beyond discussions of sexual risk to focus on other risky practices that might have a stronger tie to sexting. Researchers have linked sexting practices to other problem behaviors, such as intimate partner violence (e.g., Drouin, Ross, Tobin, 2015) and school truancy or suspensions (e.g., Perkins, Becker, Tehee, Mackelprang, 2014), but, in a review of the literature on sexting and adolescent risk behavior, Van Ouytsel and colleagues (2015) noted that the number of studies on sexting and sexual risk greatly exceeded the number of articles that dealt with other problem behaviors. Identifying problem behaviors with stronger ties to sexting would bolster the arguments advanced by proponents of the deviance perspective and provide more fertile ground for testing problem behavior theory.

When coupled with the findings from our critical review, these results suggest additional avenues for future research. For example, potentially useful demographic information, namely the sexual identity or orientation of participants, is missing from the majority (60%) of the articles we reviewed. Furthermore, only four studies made mention of transgender participants, with one that erroneously characterized transgender as a sexual rather than gender identity (i.e., Rice et al., 2012) and three that excluded data from gender nonconforming individuals (i.e., Jonsson et al., 2015; Rice et al., 2014; Ybarra Mitchell, 2014). The lack of attention to sexual and gender identity in the extant sexting literature is important for a few reasons. First, members of sexual minority groups and/or the transgender community are especially likely to use communication technologies to find potential sexual partners (Kosenko, Bond, Hurley, 2016) and several studies (e.g., Beymer et al., 2014) have linked locating sex partners online or with the help of apps to risky sexual practices. It might be that sexting is also associated with risky sexual practices among sexual minorities and/or transgender individuals, but without information regarding the sexual or gender identity of participants, researchers cannot make this determination. This represents a missed opportunity to explore a potential risk vector in communities disproportionately affected by sexually transmitted infections, including HIV (Centers for Disease Control and Prevention, 2016). Including questions about sexual identity, allowing participants to select “transgender” or “gender non-conforming” when asked about their sex, and/or targeting members of sexual minority groups or the transgender community for future sexting studies would help give us greater insight into the specific practices of these subgroups.

Like others who have reviewed the sexting literature (e.g., Cooper et al., 2016; Klettke et al., 2014), we noted inconsistencies in how sexting is measured across studies. Most (85.7%) of the articles we examined included questions regarding sending sexts; only half asked about receiving sexts. Although sending sexts is more of an active, agentic activity than receiving sexts (Temple Choi, 2014), passive media consumption, such as exposure to sexual content in television and film, has been linked to early sexual debut (O'Hara, Gibbons, Gerrard, Li, Sargent, 2012) and teen pregnancy (Chandra et al., 2008). Moreover, receiving a sext might mean being in possession of child pornography and, thus, being subject to prosecution (Lorang, McNiel, Binder, 2016). Given the potential legal and behavioral consequences of passive sexting, researchers should consider including questions or question wording about receiving sexts in future sexting studies. Future studies should also involve a careful consideration of the content and form of sexts. The majority of the studies we reviewed defined sexting content as sexually suggestive or explicit but did not explain what messages or images counted as such. For instance, do breastfeeding images, which have been flagged on social media sites as sexual and obscene (Ibrahim, 2012), qualify as sexually suggestive or explicit? Relying on such nebulous terminology has proven problematic in research and legislation on pornography and could create similar problems with respect to sexting (Suarez, 2008). Some researchers attempted to clarify what was meant by sexually suggestive or explicit by specifying that sexts involved nude or seminude photos, which leaves written messages out of the definition. Although focusing on sexual photos might make sense given that the legal cases involving sexting also center on images (Strassberg et al., 2014), researchers and policymakers will need to determine if sexting involves more than just pictures. If researchers limit sexting to photographic content, then sexting measures will need to clearly establish the subject of the photographs. Ten of the studies we reviewed specified this information in their sexting measures, but one-third of our sample did not, which could be a source of measurement error. For example, participants may count sending sexual images of known or unknown others, such as celebrities, as sexting. Given the concerns about sexting resulting in cyberbullying (e.g., D'Antona, Kevorkian, Russom, 2010), defining sexting content in terms of images of oneself seems unnecessarily limiting, but broadening these definitions to include images of others creates its own set of problems. Future research in this area must consider which activities AND what types of content are constitutive of sexts.

Our critical review of the studies included in this meta-analysis revealed additional trends in sexting research that warrant further discussion. References to theory were all but absent from the articles that we reviewed, which made it more difficult to interpret study findings. Although these articles establish an excellent base for theory building, using extant theory to inform future sexting research would provide a stronger basis for study predictions and make findings easier to explain and interpret. For example, we noticed that problem behavior theory (Jessor Jessor, 1977) offered an explanatory framework for studies rooted in the deviance perspective and that catharsis theory (Feshbach, 1955), although not supported by our findings, shared some key assumptions with the normalcy perspective. In future studies, sexting researchers could turn to models of health behavior, such as the health belief model (Rosenstock, 1974), and/or theoretical perspectives rooted in the communication discipline, including the uses and gratifications perspective (Rosengren, 1974). Each of these theories would lead researchers to examine different aspects of sexting and sexual behavior and would inevitably improve our understanding of the relationship between the two. For example, researchers who believe that sexting causes risky sexual behavior might choose to view this relationship through the lens of the health belief model and to argue that sexting is a cue to action that prompts subsequent unhealthy sexual behaviors. Whereas, a uses and gratifications perspective would draw a researcher's attention to the gratifications sought and obtained from mediated sexual encounters versus face-to-face ones. Using extant theory to inform sexting research would improve the quality of these studies and broaden the scope of the theories applied.

Our critical and meta-analytic review is not without limitations. First, this meta-analysis was constrained in its scope. Because of discrepancies in how variables were measured and reported, not all studies on sexting and sexual behavior could be included or compared using meta-analysis. We attempted to circumvent this issue by requesting original datasets from authors, but our attempts to gather these data from all authors were unsuccessful. Reasons for data being unavailable ranged from general unwillingness to share data (including data sets collected using U.S. federal funds), authors not storing data after publication, corresponding author nonresponse, and the unfortunate death of one of the data sets' gatekeepers. In addition, inconsistencies in how sexual behavior was measured made comparisons across studies difficult. For example, assessing the relationships between sexting and the number of sexual partners was complicated by inconsistent time frames measured in each study. Some studies asked participants about their lifetime number of sexual partners while others asked about the number of sexual partners one had had in the last 3 months. In our analyses, we had to assume that this variability between studies was random and did not introduce systematic error into our overall effect estimates, but future analysis should test this assumption when more published work is available. We also selected Pearson's r assuming that sexting and sexual behaviors represent constructs that are normally distributed on a continuous measurement scale. However, very few studies measured sexting or sexual behavior in this way. Most commonly, these constructs were operationalized with dichotomous measures (e.g., sexted or not, had sex or not, had multiple sexual partners or not), which provides data that offer straightforward interpretation but weaker measurement validity. We were also unable to follow recent recommendations to use standardized regression coefficients (Ferguson, 2015), which may provide a more accurate estimate of statistical relationships by removing “noise” from the bivariate relationship, because we could not obtain original data and because nine of 15 articles did not use multivariate models with sexting as the independent variable and sexual behavior as a dependent variable.

Overall, our research presents an overview of the state of the growing research on this communicative activity and its correlates in real-life sexual behaviors, as well as some important routes to follow in future research on these topics. Most importantly, additional attention is needed in study design and variable measurement. Moreover, researchers should put effort into identifying key demographic or background variables that shape this behavior, ideally from probability samples of the general population. Pursuing these avenues would allow for greater consistency in comparisons across studies and, ultimately, greater understanding of how sexual communication using new technologies is reshaping the contours of human sexual experience.