Abstract
Adolescent dating and relationship violence (DRV) is widespread and associated with increased risk of subsequent poor mental health outcomes and partner violence. Shifting social norms (i.e., descriptive norms of perceived behavior and injunctive norms of acceptable behavior among a reference group of important others) may be important for reducing DRV. However, few DRV studies assess norms, measurement varies, and evidence on measure quality is diffuse. We aimed to map and assess how studies examining DRV measured social norms concerning DRV and gender. We conducted a systematic review of DRV literature reporting on the use and validity of such measures among participants aged 10–18 years. Searches included English peer-reviewed and grey literature identified via nine databases; Google Scholar; organization websites; reference checking; known studies; and expert requests. We identified 24 eligible studies from the Americas (N = 15), Africa (N = 4), and Europe (N = 5) using 40 eligible measures of DRV norms (descriptive: N = 19; injunctive: N = 14) and gender norms (descriptive: N = 1; injunctive: N = 6). No measure was shared across studies. Most measures were significantly associated with DRV outcomes and most had a defined reference group. Other evidence of quality was mixed. DRV norms measures sometimes specified heterosexual relationships but rarely separated norms governing DRV perpetrated by girls and boys. None specified sexual-minority relationships. Gender norms measures tended to focus on violence, but missed broader gendered expectations underpinning DRV. Future research should develop valid, reliable DRV norms and gender norms measures, and assess whether interventions’ impact on norms mediates impact on DRV.
Keywords
Introduction
Dating and relationship violence (DRV) refers to intimate partner violence (IPV) involving a young person (Young et al., 2017), defined here as aged 12–18 years. It comprises physical, psychological, and/or sexual abuse perpetrated or experienced by a current or former intimate partner (Barter & Stanley, 2016; Breiding et al., 2015; Young et al., 2017). DRV is widespread among girls and boys (Leen et al., 2013; World Health Organization, 2021): in systematic reviews, psychological DRV victimization rates range from 47% to 88% (Exner-Cortens et al., 2016a) and meta-analyses suggest prevalence of 21% for physical and 14% (among girls) and 8% (among boys) for sexual DRV (Wincentak et al., 2017). While specific prevalence rates vary widely by measurement and sampling (Exner-Cortens et al., 2016a; Leen et al., 2013; Wincentak et al., 2017), patterns tend to be consistent: psychological DRV is the most common, followed by physical and then sexual DRV, often with multiple types co-occurring (Leen et al., 2013). Experiencing DRV can lead to injuries (Foshee, 1996) and is associated with increased risk of subsequent depression (Roberts et al., 2003), substance use, antisocial behavior (Exner-Cortens et al., 2013; Roberts et al., 2003), suicidal ideation (Exner-Cortens et al., 2013), and suicide attempts (Castellví et al., 2017) among girls and boys. In addition, it is a leading risk factor for morbidity and mortality among girls aged 15–19 years (Mokdad et al., 2016), with girls who experience DRV reporting harms additional to those reported by boys including fear (Barter et al., 2009), increased substance use (Exner-Cortens et al., 2013), and more injuries (Foshee, 1996). DRV victimization is a longitudinal risk factor for IPV victimization (Exner-Cortens et al., 2017; Herrenkohl & Jung, 2016) and perpetration (Manchikanti Gómez, 2011) in adulthood, highlighting the influence of adolescent relationships on future development (Exner-Cortens et al., 2017) and the importance of early intervention.
Systematic reviews report that interventions have been successful in increasing DRV knowledge (De La Rue et al., 2014; Fellmeth et al., 2013) and changing personal attitudes (De La Rue et al., 2014), but demonstrate little impact on DRV perpetration or victimization (De La Rue et al., 2014; Fellmeth et al., 2013). While little is known about effective DRV prevention, social norms theory posits that harmful social norms can hinder behavior change despite changes in knowledge or attitudes (Alexander-Scott et al., 2016), while protective norms can support behavior change (Cislaghi & Heise, 2018). Social norms comprise perceptions of typical behaviors (descriptive norms) and acceptable behaviors (injunctive norms) among a reference group of important others, with social sanctions playing an important role in holding norms in place (Alexander-Scott et al., 2016; Cislaghi & Heise, 2018).
Empirical research finds that DRV norms are associated with DRV victimization and perpetration. Considering descriptive norms, young people who believe that their friends experience or perpetrate DRV are more likely to report perpetrating DRV themselves (Kinsfogel & Grych, 2004; Reed et al., 2011), including in longitudinal studies (Foshee et al., 2001; Vagi et al., 2013), even when controlling for their own attitudes toward DRV (Foshee et al., 2001). Girls who report having friends involved in violent relationships are at increased risk for subsequent victimization (Arriaga & Foshee, 2004). Although injunctive norms are less explored in the literature, data also suggest that injunctive pro-DRV norms in secondary school are associated with sexual violence perpetration prior to university (Salazar et al., 2018).
Gender norms, “collective beliefs about what behaviors are appropriate for women and men and the relations between them” (The Social Norms Learning Collaborative, 2021, p. 8), can be thought of as a particularly powerful type of social norms (Lokot et al., 2020) and play an important role in DRV risk. Qualitative research finds myriad ways in which inequitable gender norms operate to underpin male DRV in heterosexual relationships, including by forming a basis for the social acceptability of sexual coercion (Barter et al., 2009) and by grounding girls’ status in having a boyfriend (Marston & King, 2006), which could present a barrier to ending abusive relationships (Barter, 2006; Barter et al., 2009). In interviews with young people, norms supporting the legitimacy of male dominance in relationships emerge as drivers of both physical violence and controlling behaviors (Barter et al., 2009; Wood et al., 2011). Although less explored in quantitative research, evidence also suggests that inequitable injunctive norms relating to household gender roles (Shakya et al., 2022) and female sexual availability (Wesche & Dickson-Gomez, 2019) are associated with an increased risk of DRV.
This evidence suggests that interventions may need to shift social norms concerning DRV and gender that support DRV. Social norms theory, and the approach of fostering protective DRV and gender norms among peers, has long informed DRV interventions (Offenhauer & Buchalter, 2011; Stanley et al., 2015; Wolfe & Jaffe, 1999). This is evident in the popularity of bystander interventions, which encourage young people to intervene in DRV (Stanley et al., 2015), and of gender-transformative approaches (Stanley et al., 2015; Whitaker et al., 2006), which aim to reshape gender roles and promote “more gender-equitable relationships” (Gupta, 2000, p. 10). Evaluations suggest that norms-based interventions can be effective in reducing intra-marital and domestic violence (Fulu et al., 2014), and there is emerging evidence that interventions with young people (Plourde et al., 2016) or their parents (Ehrensaft et al., 2018) can shift DRV-specific social norms among adolescents. However, evaluations of DRV interventions rarely assess impact on social norms (Coker et al., 2017; Foshee et al., 2005; Miller et al., 2014; Taylor et al., 2011), and to our knowledge none have assessed social norms as a potential mediator of intervention effects, limiting what is known about intervention mechanisms.
This may be due in part to the lack of consensus on how to measure social norms and to limited evidence as to the reliability and validity of existing measures (Ashburn et al., 2016). Valid measures assess the construct in question (DeVellis, 2017): in this case, social norms that are important to DRV outcomes. Reliable measures do so consistently (DeVellis, 2017). Measures of social norms used with adults are unlikely to be suitable for adolescents due to likely differences between these populations in reference groups, behaviors, and cognitive ability to distinguish between personal attitudes and the views of others (Moreau, 2018; Moreau et al., 2021). We therefore reviewed existing DRV literature to explore (1) what measures exist of adolescent descriptive and injunctive social norms concerning DRV and gender and (2) the validity and reliability of these measures.
Methods
This review was guided by a study protocol registered on the Open Science Framework (Meiksin, 2020) and is exempt from ethical review.
Eligibility, Search Strategy, and Screening
Eligible reports were studies published in English since 1997. We selected this timeframe because cultural shifts might render older measures meaningless or inappropriate for young people today (Reyes et al., 2016), and because 1997 marks the advent of social media (History Cooperative, n.d.), which plays an important role in the initiation and formation of relationships among adolescents (McGeeney & Hanson, 2017). Reports were required to the assess the construct validity of one or more quantitative measure of norms relating to DRV and/or gender (including bystander norms in these domains) by testing these against DRV behavioral outcomes (i.e., by exploring their association with DRV victimization, perpetration, and/or bystander behavior). Measures were assessed among participants aged 10–18 years and comprised one or more survey items, with at least 50% of items assessing one of four domains: descriptive DRV norms, injunctive DRV norms, descriptive gender norms, or injunctive gender norms. Where eligible measures comprised subscales, subscales were also included as unique measures if they independently met eligibility criteria.
Measures of DRV and gender norms overlap where those relating to perceptions of the typicality or social acceptability of DRV are “gender specific,” by which we mean they assess norms governing girls and boys separately (e.g., a measure assessing the social acceptability of a boy hitting his girlfriend). We categorized all measures of DRV norms as DRV norms whether or not they were gender specific. Broader gender norms measures, that is, those that did not focus on violent behaviors in the context of adolescent relationships or dating, were categorized as gender norms. Descriptive norms were operationalized as perceptions of the typicality or frequency of (1) DRV and (2) gendered behaviors, excluding DRV behaviors. Injunctive norms were operationalized as perceptions of (1) DRV’s social acceptability and (2) social expectations based on gender, excluding social acceptability of DRV.
Our search strategy used free-text and controlled vocabulary terms linked by the Boolean connector “OR” for three concepts: (1) social norms concerning DRV and/or gender; (2) DRV; and (3) adolescents. The search terms used within each concept were linked by the Boolean connector “AND” (see Supplemental Appendix A for Medline search strategy). The search strategy was peer reviewed based on the Peer Review for Electronic Search Strategies guidance (McGowan et al., 2016; Shamseer et al., 2015). After piloting the strategy in Medline, in June 2019 we searched nine databases containing reports relevant to our topic: IBSS; Popline; Medline; PsychINFO; PsychEXTRA; EMBASE; Web of Science; Global Health; and Scopus. We conducted additional searches via Google Scholar (July 2019; limited to the first 100 results), websites of relevant organizations (June 2020) (Care Evaluations, n.d.; Explore Our Resources, n.d.; Find a Report, n.d.; Girl Effect, n.d.; Global Early Adolescent Study, n.d.; Publications, n.d.; Resources, n.d.), two online databases of relevant measures (June 2020) (EMERGE, n.d.a; EMERGE, n.d.b; Gender and Power Metrics, n.d.), contacting subject experts (February–March 2020) (Advancing Learning and Innovation on Gender Norms (ALIGN), n.d.; Gender Violence and Health Centre (GVHC), n.d.; Learning Collaborative to Advance Normative Change—IRH, n.d.; Sexual Violence Research Initiative, n.d.), our study team’s internal database of DRV literature, and reference checking. We also screened for eligibility all evaluations included in eight reviews of DRV intervention studies (Cornelius & Resseguie, 2007; De Koker et al., 2014; De La Rue et al., 2014; Fellmeth et al., 2013; Leen et al., 2013; Lundgren & Amin, 2015; Stanley et al., 2015; Whitaker et al., 2006).
Search results were imported into EndNote X9 (The EndNote Team, 2013), de-duplicated (“Removing Duplicates from an EndNote Library,” 2018), and dual-screened on title and abstract by the first author (RM) and another reviewer in batches of 50 until reaching 85% agreement. These reviewers discussed records of uncertain eligibility to reach a consensus. RM then single-screened remaining records on title and abstract and screened all retained records on full text, discussing records of uncertain eligibility with AK and CB.
The database search was updated in March 2022, excluding IBSS (due to lack of institutional access) and Popline (retired in September 2019) (USAID, n.d.).
Data Extraction
From all included reports, RM extracted study information and the following data for each eligible measure: method of development; content; mode of data collection; evidence of reliability, construct validity, content validity, and convergent validity; and statistical properties. A second reviewer (AB) checked all extracted data, flagging areas of disagreement which were then resolved through discussion. We requested missing information on social norms measures and analysis results from study authors.
Analysis and Synthesis
Informed by previous reviews of measures (Costenbader et al., 2017; Hennegan et al., 2020), we report on the quality of included norms measures rather on than the overall quality of included studies. This quality assessment is the key focus of this review. Drawing on existing methods for assessing survey measure quality (Doherty et al., 2016; Lewis et al., 2015, 2018; Pocock et al., 2021; Prinsen et al., 2018; Terwee et al., 2007), we assessed each measure against seven criteria: participatory development; defined reference group; reliability (internal consistency, test–retest reliability, or split-half reliability); content validity (assessed as 75% or more items aligning with a relevant domain); construct validity (significant association with DRV behavior); other evidence of validity (factor analysis; or significant association with theoretically related constructs: DRV/gender attitudes, DRV intentions or perceived behavioral control over DRV); and statistically desirable properties (responsiveness, lack of floor/ceiling effects, or data available on measures of central tendency and distribution of total score for the full measure—or, where absent, for all measure subscales) (Lewis et al., 2015; Terwee et al., 2007). Significance of associations for construct validity criteria was assessed using the threshold of p ≤ .05, or a lower p value where lower values were used by the authors of included reports. Reliability was scored to reflect poor reliability (Cronbach’s alpha or correlation of <0.70) (−1), no evidence (0) or good reliability (Cronbach’s alpha or correlation of ≥0.70) (+1). Construct validity was scored to reflect an inverse relationship between pro-DRV/inequitable gender norms and DRV (−1), no evidence of a significant relationship (0), or pro-DRV/inequitable gender norms associated with increased risk of DRV (+1). All other criteria were scored as evidence absent (0) or present (1). Supplemental Appendix B further details our methods assessing for assessing measure quality.
Within each of the four social norms domains considered in this review, we inductively created categories of constructs assessed by included measures (Hennegan et al., 2020). We then created tables summarizing features of included measures and evidence on their quality; and summarizing characteristics of the measures and of the samples in which their reliability and construct validity were assessed. Drawing on these tables and other extracted data, we summarized the following: features of included studies; features of included measures; and evidence for measures’ validity and reliability, identifying strengths and limitations of existing measures.
Results
Literature Search
Database and Google Scholar searches identified 7,347 unique records (Figure 1), of which 477 were retained to screen on full text and 21 were eligible for inclusion (Aizpitarte et al., 2017; Antônio et al., 2012; Enosh, 2007; Flisher et al., 2007; Foshee et al., 2001; Gagné et al., 2005; Gonzalez-Mendez et al., 2019; Hébert et al., 2019; Helland, 1998; Hopper, 2011; Kernsmith & Tolman, 2011; Kinsfogel & Grych, 2004; Peskin et al., 2017; Pöllänen et al., 2021; Price, 2002; Reed et al., 2011; Reyes et al., 2016; Shamu et al., 2016; Shorey et al., 2018; Van Ouytsel et al., 2020; Wesche & Dickson-Gomez, 2019), reporting on 21 unique studies. Two reports presented analyses of different social norms measures from the same randomized controlled trial (Foshee et al., 2001; Reyes et al., 2016) and were therefore treated as two unique studies. Our updated search identified one new eligible report (Hunt et al., 2022), and two additional reports were identified by screening known studies (Nardi-Rodríguez et al., 2022; Shakya et al., 2022), resulting in the inclusion of 24 eligible reports of 24 unique studies.

Flow diagram of search and study selection.
Included Studies
Of the included studies, 11 were conducted in the United States (Foshee et al., 2001; Helland, 1998; Hopper, 2011; Hunt et al., 2022; Kernsmith & Tolman, 2011; Kinsfogel & Grych, 2004; Peskin et al., 2017; Reed et al., 2011; Reyes et al., 2016; Shorey et al., 2018; Wesche & Dickson-Gomez, 2019), three in Canada (Gagné et al., 2005; Hébert et al., 2019; Price, 2002), three in South Africa (Flisher et al., 2007; Pöllänen et al., 2021; Shamu et al., 2016), three in Spain (Aizpitarte et al., 2017; Gonzalez-Mendez et al., 2019; Nardi-Rodríguez et al., 2022), one in Belgium (Van Ouytsel et al., 2020), one in Brazil (Antônio et al., 2012), one in Israel (Enosh, 2007), and one in Niger (Shakya et al., 2022). All studies were observational, and seven (Enosh, 2007; Foshee et al., 2001; Peskin et al., 2017; Pöllänen et al., 2021; Reyes et al., 2016; Shakya et al., 2022; Shamu et al., 2016) analyzed data collected as part of an evaluation. One report presented only longitudinal associations between social norms measures and DRV (Nardi-Rodríguez et al., 2022), all other reports presented cross-sectional analyses, and three presented both (Foshee et al., 2001; Reyes et al., 2016; Shorey et al., 2018). Considering participants, 17 studies sampled girls and boys, four included only girls and three included only boys (Supplemental Appendices C and D provide further details of study and sample characteristics, respectively). All studies assessed relationships between social norms measures and DRV victimization and/or perpetration; none assessed relationships with bystander behaviors.
Included Measures
Most studies included a single eligible social norms measure assessing a single domain of interest (N = 15) (Aizpitarte et al., 2017; Antônio et al., 2012; Enosh, 2007; Foshee et al., 2001; Hébert et al., 2019; Helland, 1998; Hunt et al., 2022; Kernsmith & Tolman, 2011; Peskin et al., 2017; Price, 2002; Reed et al., 2011; Reyes et al., 2016; Shorey et al., 2018; Van Ouytsel et al., 2020; Wesche & Dickson-Gomez, 2019). Six studies included two eligible measures (Flisher et al., 2007; Gonzalez-Mendez et al., 2019; Hopper, 2011; Kinsfogel & Grych, 2004; Pöllänen et al., 2021; Shakya et al., 2022) and one study included three (Shamu et al., 2016). In addition, one study included a single measure for which half the items assessed one social norms domain and half assessed another (Gagné et al., 2005), and a second study included four such measures (Nardi-Rodríguez et al., 2022). Since half of a measure’s items must assess a domain of interest for inclusion as a measure of that domain, each of these five measures was eligible for inclusion in two separate domains and therefore counted twice for this review. The review therefore includes 35 unique measures, assessed as 40 measures across four domains: 19 measuring descriptive DRV norms, 14 measuring injunctive DRV norms, one measuring descriptive gender norms, and six measuring injunctive gender norms.
Characteristics of Measures
Measures were generally quite short, comprising a median of six items (range = 1–28, mean = 7). For most measures (58%), all items assessed the domain of interest. Where information was provided on measure development, reports suggested that six measures were adapted from measures of DRV outcomes (Aizpitarte et al., 2017; Antônio et al., 2012; Hopper, 2011; Kinsfogel & Grych, 2004; Van Ouytsel et al., 2020); two were adapted from a measure of personal attitudes (Shakya et al., 2022); two were tools used in previous studies (Hébert et al., 2019; Kernsmith & Tolman, 2011); one was adapted from a previous study to ask about physically rather than sexually aggressive behaviors (Helland, 1998); and six were newly developed (Flisher et al., 2007; Peskin et al., 2017; Pöllänen et al., 2021; Wesche & Dickson-Gomez, 2019). For detailed information on each included measure please, see Supplemental Appendices D (measure wording, variable calculation) and E (initial development, reference group, content validity, reliability, and construct validity).
Descriptive DRV norms
The review identified 19 eligible measures of descriptive DRV norms from 14 included reports (Table 1) (Aizpitarte et al., 2017; Antônio et al., 2012; Foshee et al., 2001; Gagné et al., 2005; Gonzalez-Mendez et al., 2019; Hébert et al., 2019; Helland, 1998; Kinsfogel & Grych, 2004; Nardi-Rodríguez et al., 2022; Peskin et al., 2017; Price, 2002; Reed et al., 2011; Reyes et al., 2016; Shorey et al., 2018). Measures ranged from 1 to 26 items (mean = 6, median = 4). Most specified reference groups of friends or peers. Only two referenced social rewards or consequences for adhering to/violating a norm (Flisher et al., 2007; Pöllänen et al., 2021). Questions were typically framed to ask for perceptions of the number or proportion of reference-group members who had experienced or perpetrated DRV, or for perceptions of whether “most” reference-group members had done so (Foshee et al., 2001, p. 133; Nardi-Rodríguez et al., 2022, pp. 12–13; Reyes et al., 2016, p. 353). Most measures referred to specific DRV behaviors among the reference group (e.g., hitting, yelling, threatening, forcing sex).
Eligible Constructs Assessed by Included Measures and Evidence of Measure Quality, by Domain and Inductive Grouping.
Note. DRV = dating and relationship violence; F = female; gen. = general; GN = gender neutral; M = male; participatory dev. = participatory development; perp. = perpetration; psych. = psychological; phys. = physical; sex. = sexual; vict. = victimization.
≥75% items assessing social norms domain.
Relationship to DRV behavioral outcome.
Most measures were gender neutral, that is, they did not specify gender or they included items about girls and boys within the same measure. Most asked about perceptions of DRV perpetration alone. We identified three inductive categories of constructs measured (Table 1). In all, 10 “gender/sexuality-neutral DRV” measures did not specify heterosexual or sexual-minority relationships (Aizpitarte et al., 2017; Antônio et al., 2012; Gonzalez-Mendez et al., 2019; Hébert et al., 2019; Helland, 1998; Kinsfogel & Grych, 2004; Peskin et al., 2017; Shorey et al., 2018). One “mixed DRV” measure assessed perceptions of gender-neutral perpetration and female victimization within heterosexual partnerships (Gagné et al., 2005), and eight “heterosexual DRV” measures assessed perceptions of DRV within heterosexual relationships (perpetration by girls and boys within one measure (Foshee et al., 2001; Reyes et al., 2016), boys’ perpetration (Nardi-Rodríguez et al., 2022; Price, 2002; Reed et al., 2011) or girls’ victimization (Nardi-Rodríguez et al., 2022).
Injunctive DRV norms
We identified 14 eligible measures of injunctive DRV norms from eight included reports (Table 1) (Enosh, 2007; Flisher et al., 2007; Hopper, 2011; Hunt et al., 2022; Kernsmith & Tolman, 2011; Nardi-Rodríguez et al., 2022; Pöllänen et al., 2021; Van Ouytsel et al., 2020). These ranged from two to 28 items (mean = 8, median = 6). Six specified a single reference group of respondents’ friends and six referred to multiple reference groups, one of which also assessed the importance of each (Kernsmith & Tolman, 2011). One measure did not specify a reference group (Pöllänen et al., 2021).
Measures asked respondents to report their perceptions of the views of reference group members, or the extent to which the respondent thought that DRV perpetration would “make me seem successful” (Pöllänen et al., 2021, p. 9). Nine (64%) asked about norms governing DRV perpetration alone (Enosh, 2007; Flisher et al., 2007; Kernsmith & Tolman, 2011; Nardi-Rodríguez et al., 2022; Pöllänen et al., 2021; Van Ouytsel et al., 2020). Most measures used Likert scale response options.
Half of the measures were gender specific (Kernsmith & Tolman, 2011; Nardi-Rodríguez et al., 2022; Pöllänen et al., 2021), and all but one measure (Kernsmith & Tolman, 2011) specified a single type of DRV. We identified three inductive categories of included measures (Table 1). Seven “respondent DRV” measures assessed injunctive norms governing DRV among survey respondents (Kernsmith & Tolman, 2011; Nardi-Rodríguez et al., 2022; Pöllänen et al., 2021) (e.g., asked to select a response for what will happen “[i]f I put pressure on my boyfriend or girlfriend to have sex. . .”) (Pöllänen et al., 2021, p. 9). Five “gender-neutral heterosexual DRV” measures combined DRV among girls and boys and focused on heterosexual partnerships (Enosh, 2007; Flisher et al., 2007; Hopper, 2011; Hunt et al., 2022). Two “mixed-or-unspecified DRV” measures assessed a combination of DRV perpetrated by girls and boys in heterosexual relationships and by young people responding to the survey (without specifying partner gender) (Flisher et al., 2007), or gender-neutral DRV perpetration (Van Ouytsel et al., 2020).
Descriptive gender norms
The review identified one eligible measure of descriptive gender norms from one included report (Table 1). This measure assessed perceptions of the prevalence of male-perpetrated sexual coercion of females (without specifying a dating/relationship context) among friends from the past year (Gagné et al., 2005).
Injunctive gender norms
The review identified six eligible measures of injunctive gender norms from three included reports (Table 1) (Shakya et al., 2022; Shamu et al., 2016; Wesche & Dickson-Gomez, 2019). Measures ranged from one to 15 items (mean = 6, median = 5) and where response options were described, measures used Likert scales. Four “gendered-violence” measures assessed injunctive norms governing male-perpetrated violence and violence against girls/women (e.g., “My family thinks that there are times when a woman deserves to be beaten”) (S. Shamu, personal communication, May 2, 2019), without specifying the context of adolescent dating/relationships. Two “gendered expectations” measures assessed social norms concerning broader gender roles (Shakya et al., 2022; Wesche & Dickson-Gomez, 2019), including sexual expectations of female gang members (Wesche & Dickson-Gomez, 2019) and gender roles within the family or household (Shakya et al., 2022).
Quality of Measures
Table 1 shows the quality of included measures by domain and inductive category. Further details on the evidence underpinning our quality assessment are available in Supplemental Appendices D (study samples, DRV outcome measures), E (summaries of initial development, reference group, content validity, reliability, and construct validity) and F (construct validity: analysis methods, results, and summary findings showing alignment between norm and outcome measures).
Descriptive DRV norms
Among the 19 included measures of descriptive DRV norms, three (16%) were informed by participatory development and all had defined reference groups. In all, 11 (58%) had good reliability and two (11%) had poor reliability. In total, 12 measures (63%) had good content validity. All showed a significant association between higher levels of perceived DRV prevalence and higher DRV risk. Eight measures (42%) also had other evidence of validity and 12 (63%) had statistically desirable properties.
Most measures were tested separately against DRV perpetration and/or victimization outcomes. Although most descriptive DRV norms measures were gender neutral, almost all were tested against gender-specific DRV outcomes, primarily standalone measures of girls’ victimization and/or boys’ perpetration.
The six gender-specific descriptive DRV norms measures were tested against DRV outcomes that matched the gender of the norms measure (Nardi-Rodríguez et al., 2022; Price, 2002; Reed et al., 2011) (i.e., norms concerning DRV among boys tested against DRV outcomes among boys). In all, 15 measures were tested against DRV outcomes that matched on type of DRV involvement (victimization or perpetration), and 11 were tested against outcomes that matched on type(s) of DRV (psychological, physical, and/or sexual). Five measures matched the DRV outcome against which they were assessed in all three dimensions, which focused on boys’ perpetration (Nardi-Rodríguez et al., 2022; Price, 2002) and girls’ experience (Nardi-Rodríguez et al., 2022) of psychological DRV.
Injunctive DRV norms
Of the 14 included measures of injunctive DRV norms, three (21%) were informed by participatory development. In all, 13 (93%) included a defined reference group, 11 (79%) had good reliability, and two (14%) had poor reliability. Nine measures (64%) had good content validity. In total, 11 (79%) showed a significant association between pro-DRV norms and higher DRV risk. Ten (71%) had other evidence of validity and 11 (79%) had statistically desirable properties.
Most injunctive DRV measures were tested against standalone DRV perpetration outcomes and against gender-specific outcomes. All seven gender-specific measures were tested against DRV outcomes specifying the same gender (Kernsmith & Tolman, 2011; Nardi-Rodríguez et al., 2022; Pöllänen et al., 2021). Ten measures were tested against outcomes that matched on victimization (Nardi-Rodríguez et al., 2022) or perpetration (Enosh, 2007; Flisher et al., 2007; Nardi-Rodríguez et al., 2022; Pöllänen et al., 2021; Van Ouytsel et al., 2020), and 13 were tested against DRV outcomes that matched on type of DRV (Enosh, 2007; Flisher et al., 2007; Hopper, 2011; Hunt et al., 2022; Nardi-Rodríguez et al., 2022; Pöllänen et al., 2021; Van Ouytsel et al., 2020). Six gender-specific measures aligned with assessed DRV outcomes in all three dimensions, focusing on girls’ and boys’ perpetration of sexual DRV (Pöllänen et al., 2021) and on boys’ perpetration and girls’ experience of psychological DRV (Nardi-Rodríguez et al., 2022).
Descriptive gender norms
The single measure of descriptive gender norms had a defined reference group and showed a significant association with DRV outcomes: Girls who reported more inequitable descriptive gender norms (i.e., more friends involved in sexual coercion) were significantly more likely to report DRV victimization (Gagné et al., 2005). The measure met no other quality criteria.
Injunctive gender norms
Of the six included measures of injunctive gender norms, three (50%) were informed by participatory development and five (83%) had a defined reference group. Two (33%) had good reliability and three had poor reliability. Five (83%) had good content validity. For five measures, inequitable gender norms were significantly associated with higher DRV risk. Two measures had other evidence of validity and five had statistically desirable properties. Five were tested against gender-specific DRV outcomes (Shakya et al., 2022; Shamu et al., 2016).
Discussion
Summary of Key Findings
Our findings suggest that social norms measures relating to DRV that are valid and reliable among young people can be developed, but that measurement is inconsistent and evidence supporting the quality of existing measures is limited. We found no eligible measure used more than once, limiting comparability across studies. Geographic diversity was also limited, with more than half of included studies taking place in the Region of the Americas. We found no eligible measures used in the South-East Asian, Eastern Mediterranean, or Western Pacific Regions (World Health Organization, 2021).
Most measures reviewed had evidence of construct validity, assessed as a significant association between pro-DRV/inequitable gender norms and increased DRV risk. Measures were typically tested against gender-specific DRV outcomes, most commonly girls’ victimization and boys’ perpetration. Psychological, physical, and sexual DRV all featured frequently among the behavioral outcomes explored. Evidence on reliability and on other types of validity was mixed.
Though under a third of included measures had evidence of being informed by participatory development with young people, nearly all specified a defined reference group. However, all reference groups were pre-defined; no measure asked respondents to identify who held the most influence over them in relation to the assessed norms (Costenbader et al., 2017) and only one assessed the importance of each reference group to the respondent (Kernsmith & Tolman, 2011).
Two-thirds of gender norms measures asked about the respondent’s friends and/or family, two groups that are particularly influential in gender socialization (Kågesten et al., 2016). However, several measures combined items asking about multiple reference groups, including unbounded groups of “others” and “people important to you” (Flisher et al., 2007, p. 622): features that limit their usefulness for gathering valid data about norms among a clear, coherent group and the relationship between these norms and DRV. Only two measures of injunctive norms referenced social sanctions, both without specifying the reference group applying these (Flisher et al., 2007; Pöllänen et al., 2021).
Several measures specified norms within heterosexual partnerships. Though sexual-minority youth face significantly higher risk of DRV than their heterosexual peers (Dank et al., 2014; Luo et al., 2014; Young et al., 2017), no measures specified norms governing sexual-minority relationships and no studies explicitly explored associations between included measures and DRV within sexual-minority relationships. Little is known about social norms contributing to DRV among same-sex partners and the key reference groups among which these norms are held. Some experts have suggested minority-stress theory (Dietz, 2019; Martin-Storey & Fromme, 2021; Reuter & Whitton, 2018) as a framework for understanding the elevated DRV risk among sexual-minority youth, which would suggest that homophobia, underpinned by gender norms (Solomon, 2015; Whitley, 2001), could play an important role. Formative research is needed to explore the social norms influencing same-sex DRV, and its findings should form the basis of social norms measures used with sexual-minority youth.
Considering measures of DRV norms, several studies explored the relationship between descriptive DRV norms and DRV outcomes, while fewer explored the relationship between injunctive DRV norms and DRV outcomes. DRV norms measures most commonly focused on DRV perpetration, and most were gender neutral. While studies usually explored DRV norms as predictors of gender-specific DRV outcomes, this was less common for measures of injunctive than descriptive DRV norms, despite evidence suggesting that predictors of DRV differ for girls and boys (Ali et al., 2011; Arriaga & Foshee, 2004; Capaldi et al., 2012; Foshee et al., 2001, 2011; Leen et al., 2013). A minority of DRV norms measures were tested against DRV outcomes focusing on the same gender, involvement (victimization/perpetration), and DRV type. This presents an important limitation to existing measures of DRV norms: social norms theorists hypothesize that norms relating directly to a behavior of interest (as the most salient at the time of the behavior) generally exert a stronger influence than do more distal norms (Cislaghi & Heise, 2018). Empirical literature suggests that this may be the case for DRV. DRV norms may affect DRV outcomes via gender-specific pathways (Foshee et al., 2001; Pöllänen et al., 2021; Shorey et al., 2018), and in Gagné et al.’s (2005) research physical DRV norms predicted physical and psychological but not sexual DRV. The relationship between attitudes and DRV outcomes has been more widely explored, finding that young people tend to view male-perpetrated DRV more negatively than female-perpetrated DRV (Exner-Cortens et al., 2016b; Reeves & Orpinas, 2012; Rogers et al., 2019), and that attitudes toward DRV vary by DRV type (Exner-Cortens et al., 2016b; Reeves & Orpinas, 2012), with attitudes most strongly predicting DRV outcomes of the same type (Exner-Cortens et al., 2016b). Omitting or combining genders, victimization/perpetration, and/or types of DRV in measures of DRV norms (and the outcomes these might predict) therefore risks missing important differences in norms and their influence.
Far fewer studies explored the relationship between gender norms and DRV. Compared to measures of DRV norms, gender norms measures tended to have less evidence of reliability and of validity assessed as an association with theoretically related constructs aside from DRV behaviors. As a strength, most gender norms measures were assessed for their relationship with gender-specific DRV outcomes. We identified only one measure of descriptive gender norms, which did not appear to be conceptualized as such given that only half of its items assessed this domain. Injunctive gender norms measures were more conceptually consistent, with the vast majority showing good content validity. However, both tended to focus on the social acceptability of violence by males and/or against females. This is a limitation to existing measures, as evidence points to the importance of separating gender norms from violence norms to avoid conflating the relationships between these distinct constructs and DRV behavior (Reyes et al., 2016). Only two measures asked about broader gendered expectations, assessing norms governing female sexual roles and gender roles within the family/household. No measures explored other gendered expectations that qualitative research suggests contribute to DRV, such as the social importance of sustained heterosexual relationships for girls (Barter et al., 2009; Marston & King, 2006) and of being sexually active for boys (Wood et al., 2011).
No eligible measures assessed bystander norms, reflecting limited evidence on the relationship between norms supporting protective, DRV-specific bystander behaviors and DRV outcomes. However, it is important to note that research with adolescents that reports on measures of DRV, gender, and bystander norms not assessed for their relationship to DRV outcomes, or on measures of related norms (e.g., sexual violence norms), can offer insights into norms measurement among this population.
Limitations
Like all reviews, this review might have missed eligible reports published after our search was completed. However, our database search was extensive and updated near the end of the study period, and no additional reports were identified through our expert requests. Eligible reports might also have been missed where abstracts did not indicate that relevant norms measures were used. However, we mitigated this risk by full-text screening evaluations of DRV interventions identified via reviews, and reports for which abstracts referenced “attitudes” or any terminology suggestive of norms.
We did not undertake dual data extraction, but worked with a second reviewer to check data extraction and identify and reconcile disagreements. We used a novel, tailored tool for quality assessment rather than an existing tool.
Implications
Findings from this review support a number of recommendations for practice, policy, and research (Table 2). We recommend that future research build on existing measures where evidence supports their reliability and validity among similar populations, and where measures distinguish between victimization/perpetration among girls and boys and focus on the DRV type(s) of interest; or where they can be adapted to do so. New measures should be informed by existing literature and participatory research with young people to develop and refine measures and to select reference groups (Costenbader et al., 2019). Researchers should report on the development, piloting, refinement, reliability, and validity of such measures, which in addition to enhancing social norms measurement in DRV research would also contribute to learning on best practices for social norms measurement among adolescents. Future research should synthesize this learning with findings from other areas of norms research among adolescents to inform methodological approaches with this population.
Implications for Practice, Policy, and Research.
DRV = dating and relationship violence.
Future research should inform the development of gender norms measures that predict DRV but are distinct from norms about gendered violence itself, including descriptive gender norms. New research is also needed to inform the development of measures of social norms influencing same-sex DRV, considering the higher risk of DRV among sexual-minority youth.
New measures should specify a bounded reference group (Ashburn et al., 2016), and where more than one reference group is pertinent, norms among each should be measured separately. Finally, future research should use valid and reliable measures to explore relationships between descriptive and injunctive DRV and gender norms and subsequent DRV outcomes, assess the impact of interventions on these norms and explore their role in reducing DRV.
Conclusions
Developing valid, reliable measures of social norms associated with DRV is possible, but measurement methods are currently inconsistent. Researchers should report on the development, reliability, and validity of such measures, which should be gender-specific where norms exert gendered influence, consider sexual-minority relationships, and assess gender norms beyond gendered violence.
Supplemental Material
sj-docx-1-tva-10.1177_15248380231155526 – Supplemental material for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research
Supplemental material, sj-docx-1-tva-10.1177_15248380231155526 for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research by Rebecca Meiksin, Chris Bonell, Amiya Bhatia, G.J. Melendez-Torres, Nambusi Kyegombe and Anjalee Kohli in Trauma, Violence, & Abuse
Supplemental Material
sj-docx-2-tva-10.1177_15248380231155526 – Supplemental material for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research
Supplemental material, sj-docx-2-tva-10.1177_15248380231155526 for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research by Rebecca Meiksin, Chris Bonell, Amiya Bhatia, G.J. Melendez-Torres, Nambusi Kyegombe and Anjalee Kohli in Trauma, Violence, & Abuse
Supplemental Material
sj-docx-3-tva-10.1177_15248380231155526 – Supplemental material for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research
Supplemental material, sj-docx-3-tva-10.1177_15248380231155526 for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research by Rebecca Meiksin, Chris Bonell, Amiya Bhatia, G.J. Melendez-Torres, Nambusi Kyegombe and Anjalee Kohli in Trauma, Violence, & Abuse
Supplemental Material
sj-docx-4-tva-10.1177_15248380231155526 – Supplemental material for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research
Supplemental material, sj-docx-4-tva-10.1177_15248380231155526 for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research by Rebecca Meiksin, Chris Bonell, Amiya Bhatia, G.J. Melendez-Torres, Nambusi Kyegombe and Anjalee Kohli in Trauma, Violence, & Abuse
Supplemental Material
sj-docx-5-tva-10.1177_15248380231155526 – Supplemental material for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research
Supplemental material, sj-docx-5-tva-10.1177_15248380231155526 for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research by Rebecca Meiksin, Chris Bonell, Amiya Bhatia, G.J. Melendez-Torres, Nambusi Kyegombe and Anjalee Kohli in Trauma, Violence, & Abuse
Supplemental Material
sj-docx-6-tva-10.1177_15248380231155526 – Supplemental material for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research
Supplemental material, sj-docx-6-tva-10.1177_15248380231155526 for Social Norms About Dating and Relationship Violence and Gender Among Adolescents: Systematic Review of Measures Used in Dating and Relationship Violence Research by Rebecca Meiksin, Chris Bonell, Amiya Bhatia, G.J. Melendez-Torres, Nambusi Kyegombe and Anjalee Kohli in Trauma, Violence, & Abuse
Footnotes
Acknowledgements
The authors gratefully acknowledge Manika Garg, Ben Pelhan and Nicola Pocock for their contributions to and support for this research.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This article was supported with funding from the Passages Project. Passages is made possible by the generous support of the American people through the United States Agency for International Development (USAID) under the terms of Cooperative Agreement No. AID-OAA-A-15-00042. The contents are the responsibility of the authors and do not necessarily reflect the views of Georgetown University, USAID, or the United States Government.
Supplemental Material
Supplemental material for this article is available online.
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
