The findings from 11 reviews were synthesised to provide an overview and assessment of the current evidence in relation to young people’s use of pornography and involvement in sexting, and their beliefs, attitudes, behaviour and wellbeing. Studies on both pornography and sexting have often been framed within a ‘negative effects’ paradigm, which assumes specific sexual behaviours represent inherent risks or harms [
33]. In this paradigm, exposure to sexually explicit media is considered a potential stimulus to engagement in ‘harmful’ behaviours [
33,
34].
Evidence of an association between sexual behaviours and pornography use, in particular, was often inconsistent across reviews and across studies within reviews. Inconsistent findings were also reported on the relationship between both pornography and sexting and mental health, as well as between pornography use and gender-stereotypical sexual beliefs. The relationship between pornography use and sexual violence and aggression appears complex with some studies suggesting an association only with certain sources of pornography, specific pornographic content or for young men who are prone to aggressive behaviour.
Methodological issues
Review quality varied and most had some key limitations, but all eleven were considered to be of an adequate standard. Notably, reviews by Horvath et al. [
21] and Cooper et al. [
25] potentially included evidence from an unknown number of non-empirical publications. Given the uncertainty regarding the sources of evidence presented in these two reviews, their findings should be treated with caution.
Other key methodological issues were identified with reviews and the primary studies included in them. Importantly, most of the evidence on pornography and sexting is derived from observational studies using a cross-sectional design. This means it is not possible to draw conclusions about whether reported associations are a consequence or a cause of viewing pornography or engaging in sexting. For example, it could be the case that sexting encourages young people to engage in sexual activity. However, as Kosenko et al. [
29] pointed out, it is equally likely that sexting is simply an activity performed by individuals who are already sexually active, and the same also holds true with regard to the viewing of pornography. Similarly, individuals who already have stronger permissive attitudes and gender-stereotypical beliefs may be more drawn to pornography.
Review authors cited the cross-sectional nature of the evidence as a significant limitation, and more prospective longitudinal research was suggested to improve understanding of the temporal relationship between pornography or sexting and a range of outcomes. Peter and Valkenburg [
22] emphasised the need to include a range of potentially significant control variables in statistical analyses of longitudinal data to reduce the likelihood of confounding and obtaining spurious associations. Importantly, these authors also highlighted the fact that whilst longitudinal studies generally have greater methodological rigour than cross-sectional designs, they are still correlational in nature and do not demonstrate causality.
Given the potential for spurious associations due to confounding, findings from existing studies should be treated with caution. Peter and Valkenburg [
22] highlighted wide variation in the extent to which researchers had attempted to adjust for confounding in existing studies, with some only controlling for a limited number of variables such as individual demographics. It is likely that recognised predictors of behaviour and other potentially important confounding variables may not have been controlled for during analyses, which limits the degree of confidence that can be placed in findings.
Evidence suggests that insufficient attention has been given to contextual factors in quantitative studies on sexting and young people. For example, none of the studies reviewed by Van Ouytsel et al. [
24] had distinguished between the different contexts in which sexting may occur, and this was recognised to be a potential limitation. Sexting-related outcomes could be influenced by a number of different contextual factors including the relationship status of the individuals involved and their motives for sexting. Van Ouytsel et al. suggested that some of the reported associations between sexting and behaviour may not hold true after controlling for the context in which sexting occurred.
Similar studies reported inconsistent findings on the relationship between pornography and sexting and multiple outcomes of interest. Inconsistency is likely to be related, at least in part, to heterogeneity in how previous research has been operationalised. In particular, there was marked variation in the conceptualisation and definition of both sexting and pornography. For example, multiple sexting reviews [
28‐
31] reported that studies varied in whether the focus was on messages being sent, received or both. Differences were also noted in the types of messages studied, (such as image only, text and images or video), and in the terminology used to describe message content, with terms being open to individual interpretation. For example, terms included ‘sexy’, ‘sexual’ ‘sexually explicit’, ‘suggestive’, ‘provocative’, ‘erotic’ ‘nearly nude’ or ‘semi-nude’. Similarly, differing definitions and terminology have been used in pornography studies, for example ‘X-rated material’; ‘sexually explicit media’; and ‘sexualised media’ [
23]. Such differences were seen to reflect variation between studies in the conceptualisation of pornography and specific content of interest. Review authors highlighted a failure in some studies to provide a definition or explanation of key terms. Variability was also found in other important factors such as age range, specific outcomes studied, outcome measurement and recall periods for behaviour (e.g. ever, within the last year or last 30 days). Together, these factors make comparisons between study findings, and assessing the overall evidence base, extremely difficult.
The problem of heterogeneity was highlighted in the three reviews using meta-analysis. Watchirs Smith et al. [
31] stated that a pooled estimate was not calculated for the association between pornography use and sexting and several forms of sexual activity due to high statistical heterogeneity. In addition, both Kosenko et al. [
29] and Handschuh et al. [
30] reported substantial levels of heterogeneity in their pooled analyses. Handschuh et al. [
30] reported multiple meta-analyses related to sexting and sexual activity: findings were reported for all adolescents combined, and then for males and females separately. Analyses revealed heterogeneity to be greater than expected by chance alone, with
I2 estimated at 65% for all adolescents. Values for
I2 of 50% and 75% are considered to represent moderate and high heterogeneity respectively [
36]. When analysed by sex, very high levels of heterogeneity were found:
I2 = 86.4% for males and
I2 = 95.8% for females. Subgroup analyses were conducted, but could not explain the heterogeneity. Kosenko et al. [
29] also reported analyses for various types of sexual activity and sexting in which heterogeneity was calculated to be
I2 = 98.5% (general sexual activity);
I2 = 87.5% (unprotected sex) and
I2 = 42.7% (number of sex partners). Given the high levels of heterogeneity found, findings should be treated with caution.
It was not possible to assess the extent of study overlap in reviews for all reported outcomes. However, as expected, we found that for some outcomes there was considerable overlap in the studies included across reviews and in meta-analyses. This included overlap in studies reporting on the association between pornography use and sexual beliefs, attitudes and activity and between sexual activity and engaging in sexting. The inclusion of the same study or studies in multiple reviews may offer some reassurance that individual reviews have been conducted in a consistent manner and their results reflect the available literature. However, the presence of overlapping primary studies in reviews is recognised to be a potential issue for RoRs [
16,
18]. For example, study overlap can be a potential source of bias, when specific studies, particularly those that are small or of poorer quality, become over-represented through their inclusion in multiple reviews [
16]. It may also lead to an overestimation of the size and strength of the evidence base.
Key evidence gaps and future research
The term pornography covers an array of different material and the type of content watched may be important in terms of potential harms, as indicated by the findings on the relationship between violence and pornography (i.e. a link with aggression was identified only when violent pornography was viewed). Whilst some research has focused on specific sources of material, such as online pornography, studies with young people appear to have largely treated pornography as a homogenous entity in terms of content. As some authors have identified, there is a need for more research that investigates separately, or disaggregates the effects of, different types of pornographic content [
23].
Whilst there is concern that many young people are accessing highly stylised, degrading or violent pornography, there also exists a general lack of knowledge and understanding about what pornographic material young people are actually viewing [
21,
22]. Current discourse is based largely on opinion or speculation about what young people are accessing [
21]. More research is required to investigate the type of pornographic content that young people are viewing rather than relying on speculation.
Evidence was identified to suggest that young people are not uncritically accepting of what they see in pornographic material. For example, Peter and Valkenburg [
22] indicated that on average young people did not view pornography as a realistic source of sexual information. Similarly, Horvath et al. [
21] reported evidence that many young people recognised that pornography may portray distorted messages about sexual activity, relationships, power and body ideals. Such findings are consistent with other media research, which indicated that young people are not simply passive ‘dupes’ or ‘victims’ of media messages. Instead, young people were found to adopt a critical and active role in interpreting various media [
37‐
40].
Various authors including Attwood [
34] and Horvath et al. [
21] have highlighted the value of conducting more research focused on the ways in which young people actually view, understand and engage with various forms of explicit media. Further qualitative research that explores the factors that influence young people’s perceptions of pornography, and their reactions to it, may be particularly informative.
Non-consensual forwarding of sexts was identified as a significant concern. Potential negative consequences for the sender were reported if sexts were made public, which included reputational damage, harassment and cyberbullying. However, it is important to recognise that such consequences are not a direct or inevitable outcome of sending a sext. Rather they result from a betrayal of trust as well as from victim blaming and gendered cultural norms related to what is acceptable sexual behaviour and self-representation, particularly for girls [
14,
41]. Qualitative studies suggest that the non-consensual sharing of sexts most commonly affects girls, but this is not supported by existing quantitative data. A meta-analysis conducted by Madigan et al. [
42] found no association between sex/gender and the prevalence of either having a sext forwarded without consent or perpetrating non-consensual sexting. The authors cautioned that the meta-analyses on the non-consensual sharing of sexts were based on small sample sizes and recommended additional research to examine prevalence. In addition to further quantitative studies, the non-consensual forwarding of sexts by young people warrants a specific and more in-depth examination using qualitative methods. Research aimed at informing strategies to prevent non-consensual sharing of sexts could be particularly valuable.
Multiple review authors identified a lack of research on the influence of social identities such as ethnicity, sexual orientation or disability on outcomes. This is a potentially important gap in knowledge, especially as the reported prevalence data suggest that involvement with sexting and/or pornography may be higher in LGBT individuals and those from ethnic minority groups [
22,
25,
28,
43]. Notably, some studies have indicated that LBGT young people use pornography as a key source of information about sex, as well as to explore their sexual identity and to determine their readiness to engage in sexual activity [
21,
22,
33,
44]. Research that adopts an intersectionality perspective would be beneficial for understanding the combined influence of social identities on outcomes of interest.
The current evidence base lacks geographical diversity, with the majority of findings originating from studies conducted in a small number of countries only. The extent to which findings are generalisable across countries is unclear. One review identified the extent to which a country has a liberal culture as a factor determining the existence, or extent, of gender differences in pornography use [
22]. Culture as well as other country-specific factors are also likely to influence the relationship between pornography use and sexting and individual beliefs, attitudes, behaviour and wellbeing. For example, access to comprehensive, relevant and high-quality sex and relationship education.
Whilst some positive aspects to pornography and engaging in sexting were identified, the predominant focus of the studies reported across reviews, was on potential negative outcomes, or outcomes that were framed by review authors as negative. The need for more quantitative studies to adopt a wider perspective and examine the potential positives associated with pornography use for young people was highlighted in reviews by Peter and Valkenburg [
22] and Koletić [
23].
Limitations
We conducted this RoR using methods that were consistent with the key principles outlined in published guidance, for example Pollock et al. 2016 [
45] and 2020 [
46]. This RoR is limited by the specific focus adopted in individual reviews, and the quality of reporting on primary studies and their findings by review authors. Some findings may have been omitted, selectively reported or reported inaccurately. Both pornography use and sexting are potentially sensitive issues and consequently reporting of behaviours may have been influenced by social desirability bias. Almost all of the reviews only included studies published in peer-reviewed journals and written in English, which may also have been a source of bias.
The age group of interest for this RoR was children and young people up to early adulthood, but multiple reviews included studies that had an upper age limit over nineteen years old. In addition, the reviews by both Kosenko et al. [
29] and Watchirs Smith et al. [
31] included at least three studies with individuals aged 18 years and older only. The wide age range of the included studies in some reviews, and the fact that data in a number of studies were derived from individuals aged 18 years and over only, are therefore potential limitations in the context of examining the experiences of children and younger adults.
We identified reviews published up to early autumn 2018, but inevitably findings were based on data obtained from earlier primary studies. Review authors did not search beyond 2017 for primary studies on sexting and 2015 for ones on pornography. Thus, data published in the last three to five years are not represented in this RoR. There may also have been reviews published since 2018 on pornography use and sexting amongst young people. However, it is extremely unlikely that any relevant reviews published in that short period of time would have significantly altered our findings and assessment of the evidence base.
We used modified DARE criteria to critically appraise included reviews and this is acknowledged as a potential limitation. The DARE criteria were not originally designed as a tool for quality assessment and have not been validated for the task. Whilst the criteria focus on a relatively small number of characteristics, reviewers were able to supplement the criteria when conducting the appraisal by recording any key observations regarding potential methodological issues or sources of bias. We incorporated these observations into the findings of the appraisal process.