Association of Sexting With Sexual Behaviors and Mental Health Among Adolescents

Key Points

Question

Is youth sexting associated with sexual behaviors and mental health?

Findings

A meta-analysis of 23 studies comprising 41 723 participants found that adolescent sexting is significantly associated with sexual activity, multiple sexual partners, lack of contraception use, delinquent behavior, internalizing problems, and substance use. The associations between sexting and multiple sexual partners, drug use, smoking, and internalizing problems were stronger in younger compared with older adolescents.

Meaning

Results of this study suggest that sexting is associated with various sexual behaviors and mental health risk factors; moving forward, education campaigns should focus on providing youth with comprehensive information about sexting and digital citizenship.

This systematic review and meta-analysis of 23 studies examines associations of sexting with behaviors such as multiple sexual partners, drug use, smoking, and internalizing problems in persons younger than 18 years, using sex, age, publication characteristics, and study quality as moderators.

Abstract

Importance

Sexting is the exchange of sexual messages, photographs, or videos via technological devices and is common and increasing among youth. Although various studies have examined the association between sexting, sexual behaviors, and mental health, results are mixed.

Objective

To provide a meta-analytic synthesis of studies examining the associations between sexting, sexual behavior, and mental health using sex, age, publication date, and study methodological quality as moderators.

Data Sources

Electronic searches were conducted in April 2018 in MEDLINE, PsycINFO, Embase, and Web of Science, yielding 1672 nonduplicate records.

Study Selection

Studies were included if participants were younger than 18 years and an association between sexting and sexual behaviors or mental health risk factors was examined.

Data Extraction and Synthesis

All relevant data were extracted by 2 independent reviewers. Random-effects meta-analyses were used to derive odds ratios (ORs).

Main Outcomes and Measures

Sexual behavior (sexual activity, multiple sexual partners, lack of contraception use) and mental health risk factors (anxiety/depression, delinquent behavior, and alcohol, drug use, and smoking).

Results

Participants totaled 41 723 from 23 included studies. The mean (range) age was 14.9 (11.9-16.8) years, and 21 717 (52.1%) were female. Significant associations were observed between sexting and sexual activity (16 studies; OR, 3.66; 95% CI, 2.71-4.92), multiple sexual partners (5 studies; OR, 5.37; 95% CI, 2.72-12.67), lack of contraception use (6 studies; OR, 2.16; 95% CI, 1.08-4.32), delinquent behavior (3 studies; OR, 2.50; 95% CI, 1.29-4.86), anxiety/depression (7 studies; OR, 1.79; 95% CI, 1.41-2.28), alcohol use (8 studies; OR, 3.78; 95% CI, 3.11-4.59), drug use (5 studies; OR, 3.48; 95% CI, 2.24-5.40), and smoking behavior (4 studies; OR, 2.66; 95% CI, 1.88-3.76). Moderator analyses revealed that associations between sexting, sexual behavior, and mental health factors were stronger in younger compared to older adolescents.

Conclusions and Relevance

Results of this meta-analysis suggest that sexting is associated with sexual behavior and mental health difficulties, especially in younger adolescents. Longitudinal research is needed to assess directionality of effects and to analyze the mechanisms by which sexting and its correlates are related. Educational campaigns to raise awareness of digital health, safety, and security are needed to help youth navigate their personal, social, and sexual development in a technological world.

Introduction

It has been 10 years since the advent of sexting research, and, during this time, sexting has moved into mainstream culture. Sexting is the exchange of sexual messages, photos, or videos via technological devices.^1,2 Initially, much of the research on youth sexting was focused on establishing the prevalence of this behavior. A recent synthesis of the prevalence of sexting in 39 studies comprising more than 110 000 youth reveals that 1 in 4 youth are receiving sexts and 1 in 7 are sending sexts.³ The prevalence of youth sexting is also reportedly on an upward trend,³ a finding commensurate with the rapid rise in rates of smartphone ownership, with 95% of teens in 2018 owning mobile phones compared with 71% 10 years ago.^4,5

Although the prevalence of youth sexting is now more definitive, research on the types of risks associated with sexting is not. Various studies have found associations between sexting and increased sexual activity, substance use, delinquency, and internalizing problems such as depression and anxiety^6,7,8,9; however, other studies indicate no associations between sexting and these behaviors.^10,11 A lack of congruent findings may be a barrier to the development and delivery of educational programs that seek to inform youth of sexting and its potential risks. With a decade of research now in hand, enough data exist to resolve discrepancies via a meta-analysis. Although several literature reviews have compiled reports of sexting and its correlates, none has focused exclusively on adolescents,^12,13 nor to our knowledge has a meta-analytic synthesis been conducted.

The primary aim of this meta-analysis is to provide clarity to otherwise incongruent results, by compiling and organizing sexting and associated variables into a synthesized analysis. A secondary aim is to identify potential methodological and sample characteristic moderators that may explain between-study variability. The current analysis will examine sex as a moderator, as studies have shown that female adolescents who sext are more likely to experience adverse health consequences than male adolescents.^14,15 Age will also be examined, because the prevalence of sexting increases with age,^3,7 and, mirroring what is seen in youth who engage in sexual activity at an earlier age,¹⁶ the association between sexting and risk behaviors may be more pronounced in youth who begin sexting earlier vs later in adolescence. Given the rapid increase in technology use across the last decade, publication date is examined as a moderator, and owing to variation in the methodological rigor across studies, method quality will also be assessed to determine if it predicts between-study variability.

Methods

Definitional Concepts

This study followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline. Sexting has emerged as a relatively recent cultural phenomenon; as such, no universal definition of sexting has been used consistently within the literature. Studies range in their definition of the context, content, and format of sexting. Context specificities of sexting include distinctions between sending, receiving, requesting, and forwarding sexts.^{17,18,19,20,21} Contents of sext messages range from sexually suggestive words to more explicit pictures and videos.^20,22,23 Format of sexts includes messages sent over mobile phone (ie, cellular) networks via text (SMS) message, and others include sexual messages that are sent via alternate communication technologies, such as online platforms, accessed by smartphone and/or tablet app (eg, Snapchat, Facebook Messenger) or computer browser.^24,25 To capture both the body of research that has been conducted, and an evolving cultural understanding of what sexting entails, the current meta-analysis defines sexting as any messages containing sexually suggestive words, pictures, or videos that are sent or exchanged over mobile phone and/or cellular network or online, be it app or web-based platforms for smartphone, tablet, or computer. Participants consented to take part in the original studies.

Search Strategy

Searches were conducted by a health sciences librarian in MEDLINE, Embase, PsycINFO, and Web of Science in June 2016, and updated in April 2018. Results were limited to publications between 2000 and 2018 (Figure 1). The concept of sexting was captured by searching numerous text word phrases, using both adjacency operators and truncation to capture variations in spelling and phrasing (eTable 1 in the Supplement). The concept of adolescents was searched using text words, database-specific subject headings, and age limits, when available. Synonymous terms were first combined with the Boolean OR. These 2 concepts were then combined with the Boolean AND.

Study Inclusion and Exclusion Criteria

Studies were included if they met the following inclusion criteria: (1) examined an association between sexting and 1 or more sexual behavior(s) or mental health risk factor(s); (2) assessed participants younger than the mean age of 18 years; (3) provided sufficient information to allow for effect size calculation; and (4) was available in English. Two coders (C.M. and another coder) assessed all studies for inclusion and exclusion criteria, and discrepancies were resolved through consensus.

Data Extraction

A standard data extraction form was used to code study variables and effect sizes. Any of the following statistics were extracted for effect-size data: odds ratios (ORs), correlations, χ² statistics, and P values. In the case that a study stratified results by context (sent or a mix of sent and received), the present analysis prioritized associations based on sent over other types of messages, because (1) only 5 included studies assessed other sexting types in isolation^{9,26,27,28,29} and thus there were insufficient data to conduct moderator analyses based on the type of sexting behaviors; and (2) an individual may receive a sext without wanting one; therefore, sending sexts is seen as more conceptually related to taking a risk than receiving a sext. In terms of the content of sexting, we prioritized images or videos over text message, because picture and video sexts may have more punitive and negative social consequences for adolescents compared with word-only texts. We also extracted child age and sex (percent of the sample who were female), as well as publication date, to determine if these potential moderators explained between-study variability. When there were multiple publications based on 1 data set, each data set was only represented once by selecting the study with the largest sample size and most comprehensive data extraction information.

Methodological Quality

Each study underwent a methods appraisal to ensure quality and validity of included findings. A 9-point assessment scale, developed based on previous meta-analyses,^30,31,32 was used to evaluate studies. Each criterion was given a score of 0 (No) or 1 (Yes), and scores were summed for a total quality score out of a possible 9, with higher scores indicating better quality (eTable 2 in the Supplement). Consistent with previous research, scores of 1 to 2 were considered low quality, 3 to 5 were of moderate quality, and 6 to 9 were high-quality studies.^3,33 Low-quality studies were eliminated from analyses.

Data Analysis

Analyses were conducted using Comprehensive Meta-Analysis software version 3.0 (Biostat).³⁴ All study effect sizes were transformed into ORs with 95% CIs. Operating on the assumption that all studies possess unique population parameters, effect-size calculations were based on random-effects modeling using the DerSimonian and Laird estimator.³⁴ Between-study heterogeneity was analyzed using Q and I² statistics. The Q statistic indicates whether study variability is greater than sampling error; if significant, moderator variables should be examined. The I² statistic describes the proportion of variation across studies owing to heterogeneity, as opposed to chance, with values of 75% or greater indicating high heterogeneity (75%).³⁵ Random-effects meta-regressions were conducted to assess continuous moderators and their potential association with study variability. Egger test³⁶ and review of forest plots were used to assess publication bias.

Results

Selected Studies

The PRISMA flow diagram (Figure 1) outlines the search strategy used to identify articles. Of 1672 nonoverlapping articles, 1431 were excluded through abstract review. Subsequent full-text review of 226 studies identified 26 studies that met inclusion criteria. Three studies^37,38,39were eliminated from analyses owing to low methodological quality; thus 23 studies met full inclusion criteria for the meta-analytic synthesis. Of the included studies, 1 study used a longitudinal design. All other included studies either used a cross-sectional design or analyzed cross-sectional data from an ongoing longitudinal study.^10,24,29,40

Figure 1. PRISMA Flow Diagram Detailing the Search Strategy.

Study Characteristics

The mean age (range) of participants was 14.9 (11.9-16.8) years, and 21 717 (52.1%) were female. Two of the studies (8.7%) were unpublished dissertations.^41,42 All studies used self-report surveys to assess sexual behavior, mental health, or both; 1 study used a face-to-face interview.⁴³ The Table^{7,8,9,10,24,26,27,28,29,40,41,42,43,44,45,46,47,48,49,50,51,52,53} outlines detailed study characteristics.

Table. Characteristics for All Studies Included in the Meta-Analysis on Youth Sexting.

Study	No.a	Participant Age, Mean (SD), y	Female, %a	Sexting Typeb	Risk Factorc	Message Contentd	Countrye
Amoo et al,43 2013	305	15.00 (NA)	49.0	S	Sx	P, V, T	Nigeria
Brinkley et al,40 2017	181	15.50 (NA)	47.0	RP	Sx, MP, CU	T	United States
Dake et al,7 2012	1289	14.58 (NA)	48.3	S	Sx, MP, CU, IP, Alc, Drg, Sm	NA	United States
Frankel et al,44 2018	6021	16.00 (NA)	49.4	RP	Sx, IP, Alc, Sm	P	United States
Houck et al,8 2014	410	12.34 (0.55)	46.6	S	Sx	P, V, T	United States
Lee et al,45 2016	1612	16.00 (NA)	64.3	S	DB	P, V	Korea
MacDonald et al,46 2018	664	16.20 (1.10)	0	S	Sx	P, V, T	United States
Morelli et al,26 2017	610	16.80 (1.63)	63.1	S	Alc	P, V, T	Italy
Patrick et al,27 2015	2114	16.00 (NA)	61.6	S	Sx, Alc, Drg, Sm	P, V	Australia
Rice et al,9 2014	662	11.86 (NA)	48.5	S	CU	P,V, T	United States
Rice et al,28 2018	1208	16.01 (1.33)	51.5	RP	Sx, CU	P, V, T	United States
Romo et al,47 2017	333	16.00 (2.00)	66.4	S	Sx, MP, CU	P, V, T	United States
Schloms-Madlener,41 2013f	420	16.02 (1.49)	49.4	S	Sx	P, V	South Africa
Ševčíková,48 2016	16 137	13.55 (1.68)	50.2	S	Sx, IP, Alc	P, V, T	Multinational (Europe)
Speno,42 2016f	201	16.01 (1.03)	53.7	RP, RQ	Sx	P, V	United States
Temple et al,29 2012	948	15.80 (NA)	56.3	S	Sx, MP	P	United States
Temple et al,10 2014	932	16.05 (NA)	57.0	S	IP, Alc, Drg, Sm	Explicit	United States
Titchen et al,49 2018	256	15.00 (1.10)	100	S	Sx	P	United States
Tomić et al,50 2018	1265	16.20 (0.50)	37.0	S	Sx	P, V	Croatia
Van Ouytsel et al,24 2014	1028	16.68 (0.67)	58.0	S	IP	P	Belgium
West et al,51 2014	975	14.50 (NA)	65.7	RP	DB	NA	Peru
Woodward et al,52 2017	437	15.90 (1.24)	60.0	RP	DB, IP, Alc, Drg	P	United States
Ybarra and Mitchell,53 2014	3715	15.50 (0.07)	56.6	S	Sx, MP, CU, IP, Alc, Drg	P	United States

Abbreviations: Alc, alcohol use; CU, contraception use; DB, delinquent behavior; Drg, drug use; IP, internalizing problems; MP, multiple partners; NA, not available; P, pictures; RP, sending and receiving sexts; RQ, requesting sexts; S, sending messages only; Sm, smoking; Sx, sexual activity; T, text-only messages; V, videos.

^aNumber varies slightly based on outcome type; N = 41 723, with 21 717 (52.1%) female participants.

^bS, study focuses only on sending messages (73.9%); RP, study provides data on both sending and receiving sexts (ie, reciprocal) sexting (26.1%); RQ, study provides data on requesting sexts (4.3%). The percentages represent the number of studies that assess that particular sexting type (eg, 17 studies [73.9%] only looked at sending sexts). The total percentage exceeds 100% because 1 study assesses multiple types of sexting.

^cSx, 69.6%; MP, 21.7%; CU, 26.1%; DB, 13.0%; IP, 30.4%; Alc, 34.8%; Drg, 21.7%; Sm, 17.4%. The total percentage exceeds 100% because some studies assess multiple types of risk factors.

^dExplicit, sexually explicit material. Eight studies assessed pictures, videos, and text (34.8%); 5 studies assessed pictures and videos (21.7%); 6 studies assessed pictures only (26.1%); 1 study assessed text-only messages (4.3%); 1 study assessed explicit messages (4.3%); 2 studies failed to specify (8.7%).

^eOf the 23 studies, 14 (60.9%) took place in the United States; 4 studies (17.4%) took place in Europe; 5 studies (21.7%) took place in Africa, Asia, Australia, and South America.

^fUnpublished dissertation.

Meta-analysis

Results of meta-analyses for each sexual behavior and mental health factor are presented below. The mean pooled effect size is provided, followed by results of publication bias and moderator analyses. Unless otherwise indicated, moderators examined in each meta-analysis were age, sex (percentage female), publication date, and methodological quality score.

Sexual Activity

A total of 16 nonoverlapping studies (35 467 participants) were available to estimate the pooled effect size for the association between youth sexting and engagement in sexual activity. A random-effects meta-analysis produced a significant combined effect size (OR) of 3.66 (95% CI, 2.71-4.92), indicating that youth who sexted were 3.66 times more likely to have engaged in sexual activity. Odds ratios greater than 3.0 are regarded as moderately strong.⁵⁴ Figure 2 provides a forest plot of these data. No publication bias was suggested. Statistically significant heterogeneity between the studies was found (Q = 405.10; P < .001; I² = 96.30). Only study methodological quality emerged as a significant moderator. For every 0.5-unit increase in study quality, the log risk ratio increased by 0.26 (95% CI, 0.01-0.50), suggesting that effect sizes become greater when study methodology is of higher quality.

Figure 2. Effect Sizes for Each Study Included in the Meta-analysis on Sexting and Sexual Behavior.

Contributing studies are sorted in reverse chronological order. Square data markers represent odds ratios, with size of the markers corresponding to 95% CIs, and diamond data markers represent the overall effect size based on included studies.

Multiple Sexual Partners

Five studies (6466 participants) examined the association between sexting and having multiple sexual partners, and the pooled effect size was significant with an OR of 5.37 (95% CI, 2.72-12.67) (Figure 2). There was no indication of publication bias. Statistically significant heterogeneity between the studies was found (Q = 96.21; P < .001; I² = 95.84). Age emerged as a significant moderator of between-study variability: for every 0.2-year unit increase, the log risk ratio decreased by 1.67 (95% CI, –2.92 to –0.41), suggesting that effect sizes diminished as adolescents aged. Being male was also a significant moderator: for every 2.5% unit increase in the percentage of male participants in a sample, the log risk ratio increased by 0.13 (95% CI, 0.04-0.22).

Contraception Use

Six studies (7388 participants) examined the association between sexting and lack of contraception use, and the pooled effect size was significant, with an OR of 2.16 (95% CI, 1.08-4.32) (Figure 2). No publication bias was detected. Because statistically significant heterogeneity between the studies was observed (Q = 51.55; P < .001; I² = 90.30), moderator analyses were conducted. No moderators were detected.

Delinquent Behavior

Three studies (3024 participants) report on sexting and delinquency, with a significant pooled effect size (OR, 2.50; 95% CI, 1.29-4.86) (Figure 3). There was no evidence of publication bias. Moderator analyses were not conducted owing to insufficient sample size.

Figure 3. Effect Sizes for Each Study Included in the Meta-analysis on Sexting and Mental Health.

Contributing studies are sorted in reverse chronological order. Square data markers represent odds ratios, with size of the markers corresponding to 95% CIs, and diamond data markers represent the overall effect size based on included studies.

Internalizing Problems

Seven studies (29 559 participants) examined the association between youth sexting and internalizing problems (anxiety/depression) and the pooled effect size across these studies was significant, with an OR of 1.79 (95% CI, 1.41-2.28) (Figure 3). No publication bias was detected. Statistically significant heterogeneity between the studies was found (Q = 44.5; P < .001; I² = 91.02). Age emerged as a significant moderator: for every 0.5-year unit increase, the log risk ratio decreased by 0.23 (95% CI, –0.34 to –0.12), suggesting that effect sizes weakened as adolescents aged.

Alcohol Use

Eight studies (31 255 participants) examined associations between youth sexting and alcohol use and the pooled effect size was significant, with an OR of 3.78 (95% CI, 3.11-4.59) (Figure 4). No publication bias was detected. Statistically significant heterogeneity between the studies was found (Q = 65.15; P < .001; I² = 89.25), but no significant moderators were detected.

Figure 4. Effect Sizes for Each Study Included in the Meta-analysis on Sexting and Substance Use.

Contributing studies are sorted in reverse chronological order. Square data markers represent odds ratios, with size of the markers corresponding to 95% CIs, and diamond data markers represent the overall effect size based on included studies.

Drug Use

Five studies (8487 participants) examined associations between youth sexting and drug use, and the pooled effect size was significant, with an OR of 3.48 (95% CI, 2.24-5.40) (Figure 4). No publication bias was detected. Statistically significant heterogeneity between the studies was found (Q = 82.37 P < .001; I² = 95.14). Age was a significant moderator: for every 0.2-year unit increase, the log risk ratio decreased by 0.63 (95% CI, –1.20 to –0.06), suggesting that effect sizes weakened as adolescents aged.

Smoking

Four studies (10 356 participants) examined associations between youth sexting and smoking, and the pooled effect size was significant, with an OR of 2.66 (95% CI, 1.88-3.76) (Figure 2). No publication bias was detected. Statistically significant heterogeneity between the studies was found (Q = 19.13; P < .001; I² = 84.32). Age was a significant moderator: for every 0.2-year unit increase, the log risk ratio decreased by 0.61 (95% CI, –0.88 to –0.33), suggesting that effect sizes weakened as adolescents aged.

Discussion

Youth sexting and its potential risks have grabbed the attention of parents, educators, health practitioners, and the public at large.^55,56,57,58 The present meta-analysis provides an aggregated assessment of sexual behaviors and mental health risk factors associated with youth sexting. Results reveal that sexual behaviors and mental health factors are implicated to a greater extent in youth who are sexting. Sexting youth, relative to their nonsexting counterparts, are more likely to use substances, experience anxiety, depression, and delinquency, and to engage in sexual activity, sex with multiple partners, and lack of contraception use. These findings, together with research suggesting that approximately 1 in 4 youth are sexting,³ imply that discussions regarding sexual health and education between parents, teachers, and youth should include digital health and citizenship.

Sexual activity, although associated with risks such as sexually transmitted infections (STIs)⁵⁹ and pregnancy,¹⁶ is not an inherently dangerous or delinquent behavior among youth. Sexual exploration at various points in development is normative and can be indicative of healthy exploration and positive relationship building. However, an important finding to emerge from this study is that the associations between sexting and having multiple partners, experiencing internalizing problems, and engaging in smoking and drug use were stronger in younger compared with older adolescents. Although sexual exploration becomes an increasingly normative part of development as adolescents age^48,60; it is possible that younger adolescents may be more susceptible to risks associated with sexting owing to their relative immaturity compared with older adolescents. A span as short as 3 years within the adolescent period could reflect a difference between the onset of puberty and advanced sexual, physical, and cognitive maturation.⁶¹ Younger adolescents may therefore be less equipped to handle the social and physical implications of engaging in sexual exchanges such as sexting. Furthermore, younger adolescents may be more vulnerable to online dangers such as sextortion (being blackmailed to send nude photos, or money to prevent their pictures being posted online)⁶² and may be at risk of experiencing the negative outcomes that can accompany sexual behavior at an early age, such as teen pregnancy, or sexually transmitted infections, as younger adolescents are less likely than older adolescents to get tested for sexually transmitted infections.^3,16,63 Finally, sexting at a young age, like sex at a young age, may cluster with other risky behaviors.^64,65 However, given that the association between sexting and certain sexual behaviors and internalizing problems remains, although to a lesser extent, in older youth, early education may act as a preventive measure to reduce risk exposure in late adolescence and young adulthood.

Sex moderated the association between sexting and having multiple sexual partners, such that the association was stronger among male than female adolescents. Given that sex differences are not found with respect to actual sexting behavior in youth,³ the weaker association between sexting and having multiple sexual partners seen in female adolescents may be owing to the cultural emphasis placed on the negative sexual, social, and psychological consequences of female sexuality,^14,66 leading female adolescents to engage in, and report, fewer sexual encounters relative to male adolescents. The differing strengths of the association between sexting and having multiple sexual partners may therefore be associated with social mores, which normalize male sexual exploits and lead to male individuals having or reporting a greater number of sexual partners overall.⁶⁷

One possible mediating variable linking sexting to sexual behaviors and mental health risks is intention. Youth may sext with relatively harmless intentions of sexual exploration or intimacy building; such intentions are normative developmental occurrences in adolescence. However, in the case that harmless intentions result in harmful or unwanted outcomes, the likelihood of internalizing problems may increase. For example, youth may sext with the intention of building intimacy; however, when intimacy does not develop, emotional distress or disappointment ensues.^48,68 The link between sexting, sexual behaviors, and mental health difficulties may also be owing to extenuating disinhibitory influences that occur in conjunction with sexting. For example, the environmental disinhibitory effect of alcohol and drugs may contribute to a greater tendency to engage in sexting and other risky behaviors (eg, drunk texting, lack of contraception use).^69,70 Taken altogether, a better understanding of the antecedents and outcomes of sexting, which requires further research to parse out the potential longitudinal and causal mechanisms by which sexting, sexual behaviors, and mental health risks are related, is needed.

A common conclusion in sexting-related research is that education is fundamental to minimizing risk.^44,71 Dake and colleagues⁷ outline ways to educate adolescents about the risks of sexting, such as emphasizing the permanence of messages sent via technology, the reputational damage that may be incurred if pictures were to fall into the wrong hands, and the legal consequences of sexting. Although these are legitimate risks that must be addressed, research shows that youth are already aware that sexting is risky^72,73; therefore, several additional strategies may be needed. Open and frequent conversations with youth about sexual behavior and digital citizenship (ie, how to be safe, legal, and ethical in their online lives) may be essential to safeguarding youth. It is also important to discuss the emotional and legal ramifications of sending or forwarding sexts without consent, which is reportedly occurring in 1 of 8 youth.³

Limitations

Several limitations should be addressed. First, and most importantly, meta-analyses provide a pooled association between 2 variables, and thus, are correlational in nature, and therefore do not imply causation. Second, because small sample sizes may not fully capture population variability and may have less power to detect variable effects, analyses using small sample sizes should be interpreted with caution. This caution also applies to moderator analyses with small sample sizes. Third, as with all research, the methods used to examine topics invariably influence results. In the case of the present meta-analysis, we found some effect sizes were stronger in studies with greater methodological rigor, pointing to the need to uphold consistent and rigorous methodology. Methodological criteria, such as using consistent operational definitions and multi-informant approaches to gather data, have often been sparse within sexting research, likely owing to the recent emergence of sexting in the literature. Ensuring methodological rigor has therefore been identified as a key component to moving the sexting literature forward.^3,74 Fourth, single-rater bias may be present in studies where an individual informant was asked to provide data on multiple types of substance use and sexual behaviors. Fifth, some moderator variables could not be examined systematically owing to insufficient data available in individual studies (eg, percentage single or percentage in romantic relationships). Recent research suggests that the association between sexting and its correlates may be moderated by context: when sexting takes place within the context of a romantic relationship it may not be a marker of risky behaviors, but when it occurs outside of a romantic relationship, participation in risky behavior, such as drug use, is more likely.⁷⁰ Thus, future research should consider relational context when evaluating the sexual behaviors and mental health factors associated with youth sexting behaviors. Overall, the sexting literature would benefit from studies that assess various moderators, such as relationship status, the media through which sexting occurs (smartphones or computers), measures used to assess sexting (self-report vs other-informant), as well as different types of sexting (eg, sending, receiving, forwarding). As such, a recommendation for ongoing and future research is to stratify findings by the content, context, and format of sexting.

Conclusions

Sexting is correlated with sexual behavior and mental health risk factors in youth. However, sexting is also being recognized as a normative occurrence in the current digital era. This shift in perspective is mirrored in US, European, and Canadian laws surrounding sexting, which recognize instances in which sexting can be both harmful and harmless.^75,76 With the goal of mitigating youth risks, more research is needed to clarify the directionality of the associations between sexting, sexual behaviors, and mental health factors. Further research is also needed to discern the mechanisms behind sexting and its association with potential risk factors. Educational initiatives must be prioritized to ensure that adolescents and youth are equipped with the tools they need to navigate their personal and social development in a technological world.

Supplement.

eTable 1. Search Strategy

eTable 2. Study Quality Scoring for Each Study Included in the Meta-analysis

eReferences.