From Age to Agency: Frame Adoption and Diffusion Concerning the International Human Rights Norm Against Child, Early, and Forced Marriage

The way many human rights advocates frame the international norm against child, early, and forced marriage (CEFM) has shifted in the past decade. While CEFM has historically been framed as driven by poverty and underdevelopment, advocates have more recently discussed the problem with a feminist sexuality frame. What leads advocates to change their framing about an international norm? We build an argument that stresses how (a) the nature of the frame, (b) the characteristics of the advocates, and (c) the characteristics of the discursive environment interact to determine whether and to what extent advocates will adopt an alternative frame. We use a multi-methods approach (machine-learning text analysis, process tracing, and interviews) to examine the process of frame adoption around an international norm. Our theoretical understanding of international norms requires a better grasp on the role of framing, especially framing among the very norm entrepreneurs that are critical to international norm diffusion.

The way many advocates frame the human rights norm against child, early, and forced marriage (hereafter CEFM) has shifted in the past decade. While CEFM has historically been framed as driven by poverty and underdevelopment, advocates have more recently discussed the problem with a feminist sexuality frame. 1 According to this new frame, CEFM is a manifestation of patriarchy. The practice exists because of gendered control of sexuality. Ending CEFM requires women's empowerment and a restructuring of patriarchal society. Even if new laws restrict marriage to only those over 18 and criminalize CEFM, without addressing the root problem of patriarchal control of sexuality, the lived experiences of those affected by CEFM will not change.
This new frame is a radical departure from the traditional advocacy approach, which often framed CEFM as mainly development or poverty-based issue. A human rights norm against CEFM has long been established, and advocacy concerning the issue dates back centuries (Cloward 2014;Heywood 2017). The United Nations (UN) Convention on the Rights of the Child (1989) and the Convention on the Elimination of All Forms of Discrimination against Women (1979) both include prohibitions of the practice. Despite this, CEFM has remained; one in five girls is married before the age of 18, often with devastating consequences for many aspects of human security. 2 Even after 18, women continue to be denied their rights related to marital choice and sexuality.
Recently, international attention to the issue has been reinvigorated (Shawki 2015;Vilán 2022). In 2015, the Sustainable Development Goals around gender equality included CEFM targets. There have been more international and domestic policy efforts to change practices, and the number of global news stories on the topic more than doubled in the past 4 years. 3 The recent reinvigoration of the issue corresponds with the new feminist sexuality frame. While some advocates still frame CEFM in a development frame, many are increasing their use of this alterative feminist frame. Although we make no claims that the new frame is responsible for the reinvigoration, we know that framing is important for the resonance and legitimation of an international norm (Price 1995;Payne 2001;Allan and Hadden 2017;Rosert and Sauer 2021).
What leads to a shift in advocacy framing for an international norm? While much international relations theory has focused on how international human rights norms diffuse to state actors, less attention has been paid to the reasons why some advocates shift their framing about an already-established human rights norm. Even in sociology, where there has been more attention to the role of framing in social movements, "systematic examination of the factors that account for frame variations is rare" (Snow et al. 2007, 388). Our understanding of international norms requires a better grasp of the role of framing and frame choice.
Drawing on existing theoretical work in international relations and sociology, we build an argument that stresses how (a) the nature of the frame, (b) 1 3 From Age to Agency: Frame Adoption and Diffusion Concerning… the characteristics of the advocates, and (c) the characteristics of the discursive environment or venue interact to determine whether and to what extent advocates will adopt an alternative frame. While there are multiple critical factors that are important for the adoption of a specific frame, we argue that a discursive environment or venue built on mutual trust among networked, transnational advocates may be the most important factor for broad shifts in the framing of an international norm (Checkel 2001;Deitelhoff and Müller 2005;Deitelhoff 2009).
We employ a unique multi-methods approach to assess the implications of our argument in the case of CEFM advocacy. First, we gathered 42,839 documents from organizations interested in CEFM and examined the contents of those documents quantitatively using machine-learning text analysis approaches. Second, we used process tracing methods to trace the genesis and evolution of the new feminist frame in the internal documents of one international non-governmental organization championing the new frame. Third, we conducted interviews with key advocates. Part of our research team was embedded in the organizations and advocacy space during the time of frame diffusion, providing both lived experiences and unique research challenges that we discuss in more detail below.
The main contribution of our work is empirical. Framing within advocacy movements is a research area where theories are complex and deep, but empirical investigations are somewhat shallow and often lack a crossnational approach (Snow et al. 2007(Snow et al. , 2014. As Snow et al. (2014) remarked, a "tremendous opportunity" for pushing the field forward "lies in empirical examinations of frames and framing within transnational social movements." (37). Our multi-method approach shows how the reframing of the CEFM norm would not have been possible without repeated interactions between trusted, networked actors in a discursive venue that allowed a broad coalition of transnational advocates to build a shared understanding of the issue and its preferred solution.

The Importance of Framing for International Norms
Our theoretical interest is on the framing choices of advocates working on issues related to international norms. The particular international norm we are focusing on, a human rights norm against CEFM, does not vary in our study. Instead, it is the frame of advocates around this international norm that does vary, with some organizations adopting the new feminist sexuality frame over time. In building our argument concerning when frames around an international norm change, we draw on both work on international norms, mainly within international relations, and work on framing, mainly within sociology.
We define a norm as a "standard of appropriate behavior for actors with a given identity" (Finnemore and Sikkink 1998, 891). As Winston (2018) formalizes, "contemporary international norms" have the following three-part structure: "if [problem,] [value] suggests [behavior]" (641). A norm also requires the identification of the actor for which this behavior is expected and a collectively shared belief among society that the problem, behavior, and moral principle identified are specific enough to identify a violator (Jurkovich 2020, 2-3).
A frame is a "schemata of interpretation" that allows us to "locate, perceive, identify, and label" complex information" (Goffman 1974, 21). Frames are critical in the "creation and public discourse about social problems" (Hallahan 2008, 2). Frames help set the agenda (Joachim 2003;Raymond et al. 2014), lead individuals to join nascent movements (Snow et al. 1986), and influence the depth of international consensus and policy formation about a specific international problem (Charnysh et al. 2015).
Framing is central to the international norm development and contestation process; without a persuasive frame, an international norm will not develop or diffuse from norm entrepreneurs to a broad audience (Payne 2001). A shared and evocative frame is essential for the international norm diffusion process that could lead to the adoption of new international human rights law or changes in beliefs and behavior (Finnemore and Sikkink 1998;Brysk 2013;Rosert and Sauer 2021).
Framing is a "verb"; frame construction "implies agency and contestation at the level of reality construction" (Benford and Snow 2000, 614). Framing is a dynamic process that can be discursive, strategic, and contested. Movements try to make their framing of a particular problem similar to existing issues of interest (frame bridging) and/or speak to the individual's values and beliefs (frame amplification). Advocates may adopt new frames on issues that a targeted individual already sees as important (frame extension). Finally, advocates may have to promote a completely new understanding of the issue (frame transformation): "new values may have to be planted and nurtured, old meanings or understandings jettisoned, and erroneous beliefs … reframed" (Snow et al. 1986: 473).
What makes a particular frame resonate with the larger advocacy community? First, we expect that the conceptual nature of the frame will matter. Some frames are consistent with "master frames" that are inclusive and flexible and largely resonate with individuals in a broader audience (Benford and Snow 2000;Benford 2013). For example, equal rights is a master frame that is consistent with advocacy around the civil rights movement, LGBT + rights, and women's rights, among other advocacy movements (Benford 2013). Subsequent frames that draw on master frames can create "cycles of protest" that allow collective action to flourish even under unfavorable political conditions (Benford 2013: 1).
Discussions of master frames in sociology are like discussions of norm "clusters" in international relations (Lantis and Wunderlich 2018). Norm clusters are "multiple combinations of conceptually interlinked but distinct values and behaviors" (Winston 2018, 638). A norm that is internally consistent with a larger cluster of norms is more likely to flourish. If these norm clusters have already been embedded in international law, a new norm that is understood as part of this norm cluster may be more resilient (Lantis and Wunderlich 2018). A powerful frame can bridge multiple norms, helping to create a norm cluster (Snow et al. 1986).
We expect that a frame will be more successful and adopted more frequently if it provides a clear interpretation of the problem, remedy, and who is to blame. As Brysk (2013) contends, "evocative frames tell a story of innocent victims abused by identifiable perpetrators in a familiar repertoire that violates widely shared norms" (105). Although agents can craft frames that more clearly assign blame, some issues and problems simply have an easier story to tell about who is culpable and why (Keck and Sikkink 1998;McCammon 2009;Jurkovich 2020).
Second, it is not only the message that influences whether a particular frame resonates; the characteristics of the advocacy messenger also matter for frame adoption. The messenger must have perceived legitimacy, authority, or expertise on the issue or problem (Finnemore and Sikkink 1998;Keck and Sikkink 1998;Wiener and Puetter 2009;Stroup and Wong 2017). The perceived expertise of the messenger may be based on policy experience but often needs to go beyond any top-down, international experience or knowledge (Acharya 2004;Schneiker 2017;Rosert 2019). Expertise must be grounded in local knowledge for the messenger to be seen as "credible" (Acharya 2004, 248). As such, successful advocates are typically well-networked, having built the insider-outsider transnational networks or coalitions designed to translate norms or norm clusters to diverse audiences (Shiffman 2016;Schneiker 2017;Rosert and Sauer 2021). For example, Joachim's (2003) account of successful framing in the UN focuses on the need for advocates that are based on a "heterogeneous international constituency," providing legitimacy and making it harder for counter agents to "discredit it as representing only the interests of certain groups" (252).
The advocacy network where the messenger is embedded is thus a critical determinant of the messenger's success as a frame entrepreneur. As Shiffman (2016) points out, "networks may differ in their capacities to discover frames that work" (i62). Actors that are part of diverse networks will be better equipped to find and translate frames to others; their individual legitimacy and perceived resources for use as a frame messenger will have expanded as a result of their networked position (Keck and Sikkink 1998;True and Mintrom 2001;Shiffman 2016;Shiffman et al. 2016). Of course, resources outside of the advocacy network also matter. In general, effective advocates are more likely to be those that are adequately staffed and financed, as well as be embedded in supportive political opportunity structures (Keck and Sikkink 1998;Joachim 2003;Bob 2005;McAdam et al. 2012;Murdie 2014).
Third, we expect that the nature of the discursive environment or venue will influence the success of any new frame. Within work on international norms, there have been quite a few studies on the importance of venue choice in facilitating norm diffusion (Checkel 2001;Coleman 2013;Rosert and Sauer 2021). It is critical that there is a venue where debate and disagreement is possible; deep persuasion is more likely when the venue is "less politicized" and debate occurs in a more "insulated, private setting" (Checkel 2001, 562-563). Through the institutionalized rules and discussion styles at some venues, it is easier for there to be negotiation and persuasion across actors, building trust among participants in a non-hierarchical fashion (Risse 2000;Deitelhoff and Müller 2005;Wiener 2004Wiener , 2014Deitelhoff 2009;Wiener and Puetter 2009;Passy and Monsch 2020).
The environment or venue can also encourage the innovation necessary for a new frame. Certain environments create the space for frame innovation, allowing for "slices of observed, experienced, and/or recorded 'reality' to be assembled, collated, and packaged" in new ways (Benford and Snow 2000, 623). Some venues facilitate "discursive interventions," where actors can purposively try to change the "values, norms, and rules" entailed in a given normative structure (Wiener 2004, 190). The discursive environment can also be influenced by the "world time" when conversations are taking place (Finnemore and Sikkink 1998, 909). Some historical events can "lead to a search for new ideas and norms" (Finnemore and Sikkink 1998, 909). These events can heighten innovation and persuasion in the discursive environment, providing the "context" and "precedents" for current debates (Sandholtz 2008, 103).
Finally, we expect that frames do not necessarily diffuse equally across advocacy actors; some advocates will be more open to a novel frame. Organizations vary in their openness to new frames and ideas (Checkel 2001;Bob 2005;Carpenter 2007;Stroup and Wong 2017). Network ties, resources, and political opportunity structures can influence whether actors are even in the discursive spaces to hear about a new frame (McCammon et al. 2001;Joachim 2003;McAdam et al. 2012;Shiffman 2016). Organizational resources and characteristics also influence whether the new frame's adoption is seen as consistent with the identity of the organization and its network (Klotz 1995;Soule 2004;Della and Parks 2014;Schneiker 2017).
Based on our review of the literature, we provide a summary of our relevant predictions in Table 1. Worth noting, many of these characteristics do not fit neatly within the common dichotomies often relied upon in the social sciences: there is a blurring of structure versus agency, international versus domestic, and material versus immaterial. We do not think this is accidental but rather a function of the networked and transnational nature of the determinants of frame adoption.
Further, many of the identified factors are endogenous and reinforcing to each other. For example, the characteristics of a particular novel frame will influence the networked coalition of messengers sharing the message and the nature of the discursive environment or venue where the message will be shared. These factors will also influence the characteristics of organizations invited to the venue to hear the novel frame. At the same time, the networked resources that advocates have at their disposal may encourage the very discursive environment that could lead to the innovation in framing in the first place. The distinction between message-receiver and message-provider is also blurry and endogenous, with networked resources often making today's message-receiver tomorrow's message provider. In line with the cross-disciplinary literature we have reviewed, the adoption of a new frame is a reinforcing process that depends on a combination of characteristics that get the correct messenger, receiver, and the message in a place where deep discussion and persuasion can occur.
Therefore, if one characteristic was to be most important, we think the existing literature implies that the characteristics of the discursive venue are key. Without a safe environment for cycles of debate and contestation, the characteristics of the frame cannot be shared and the multifaceted legitimacy of the persuader may be lost. We see the creation of discursive environment as the necessary condition for other factors to help spur on the adoption of a new frame.

Table 1
Empirical implications on when a new frame of an international norm will be adopted

Level of variation
Empirical implication

Case Selection
Our dependent variable is frame adoption among CEFM advocates. Our key independent variables concern the characteristics of the frame itself, the characteristics of the environment, and the characteristics of the persuaders and persuadees. Our unit of analysis is "organization -unit of time" or "advocacy movement -unit of time." In other words, we are examining whether and to what extent a new frame was adopted by an advocacy organization at a specific time.
We chose this case for a variety of reasons. First, there is variation on the dependent variable. Although we had anecdotal examples of a new feminist frame in CEFM advocacy prior to our analysis, many organizations continue to champion for age restrictions and take a more traditional frame that identifies the problem of CEFM as being driven by a lack of development and education. Cash transfers to parents for postponing the marriage of their daughters until they are 18 continues to be a well-established intervention, despite some evidence that they have limited long-term effects (Baird et al. 2011).
Further, the newness of this case is helpful for interviews and document retrieval. The recent time period allowed us to conduct interviews when memories were not completely clouded by the passage of time and/or the relative success of this new framing for the eradication of CEFM as a whole, which still remains to be seen.
Members of our research team had lived experience with this case. Our team of five included two researchers with social science Ph.Ds. who were working for different organizations in this issue space during the time under examination. Both researchers were well-versed in feminist methodologies. Their involvement in the project provides us with rich anecdotal and lived information to draw on for the analysis (Ackerly et al. 2006). At the same time, their involvement presents unique challenges for the perceived objectivity of the analysis. We address this issue with the following concrete research steps. First, our data mining and text analysis were conducted on documents obtained by academic researchers and with no immediate connections to the organizations involved with the new frame. Second, our qualitative, process tracing analysis was headed by researchers who were not part of the lived experience and were trained in qualitative methods within political science. Further, our interviews were done without those with lived experience present. Some of our interviewees had moved on and no longer work on the issue of CEFM. Their responses could be seen as especially objective considering that they had no immediate interest in the issue area. Finally, it is worth noting that our interview evidence is not conceptualized as factual but as narrative-based; the interviews are merely part of our triangulated research approach to understanding frame adoption.
Our unique approach draws on multiple methodological traditions, is transnational and process-based, combines advocates and researchers, and assesses some of the major implications of the cross-disciplinary theoretical literature on norms and framing. Nonetheless, we acknowledge some issues which we hope future work will examine. First, our process tracing only concerns an organization that does adopt the feminist sexuality frame. We hope future work will focus on process tracing within an organization that does not adopt the new frame. Second, we hope to interview more advocates for future work, perhaps oversampling those who did not adopt or otherwise resisted the new frame. Finally, our study focuses completely on the international norm against CEFM; future studies could benefit from comparisons of framing across different norms.

Scope Conditions
We focused our analysis on the years 2011-2019 and on a subset of organizations that were members of the umbrella group Girls Not Brides. Girls Not Brides was founded in 2011 by The Elders, a group started by Nelson Mandela in 2007, in response to a perceived need to bring smaller organizations together to collaborate on CEFM eradication. Any group can join Girls Not Brides; applications are brief and are submitted online. At the time of the data collection in late 2019 and early 2020, there were 1383 members of Girls Not Brides, most of whom are domestic civil society organizations. The membership is diverse, with 105 countries represented.
We see Girls Not Brides as a group of organizations taking multiple different frames of the international norm against CEFM. The umbrella group itself has historically taken more of a development frame than a feminist sexuality frame. Organizations wanting to join Girls Not Brides must endorse a mission statement and a statement of membership principles. Neither of these documents mention sexuality or feminism. Only in May 2019 were the membership principles statement changed to focus more on empowerment, as consistent with the group's recent theory of change.
Girls Not Brides events have been a venue where there has been much discussion on the issue of sexuality as an alternative framing for the international norm against CEFM. In 2018, for example, the global coordinator of Girls Not Brides acknowledged sexuality and the opening of a new discussion on CEFM: "We're starting to tackle some difficult discussions. We may not always agree on how to address some key aspects of child marriage, but that's not a bad thing because important learnings emerge through debate. However, we need to have the courage to discuss some of these difficult topics -including sexuality, social norms and power and race dynamics -openly and respectfully. We're not going to make progress on ending child marriage if we're not willing to take these conversations beyond the Global Meeting." 4 Because of its scope, its diversity, and its regularized global and regional meetings, Girls Not Brides provides a good cross-section of organizations for our sample. In addition, the years 2011 to 2019 encompasses both the work of Girls Not Brides and the heightened advocacy attention to CEFM.

The Characteristics of the Frame Itself
Before turning to our quantitative text analysis, it is useful to outline our conceptualization and operationalization of the new feminist sexuality frame and its comparison to the more traditional frame. The eradication of CEFM is not a new issue (Cloward 2014;Heywood 2017). Early advocacy centered on ideas of modernity and development; there were often colonial undertones to the discussion. The focus was primarily on extending legal age restrictions to marriage. These restrictions were first designed to stop those under puberty from marrying. As conceptualizations of childhood and children grew, later advocacy focused on restricting marriage to older adolescence (16 or 18, for example) (Heywood 2017).
Prior to the expanded advocacy of the last decade, CEFM eradication was championed mainly by child rights and protection organizations, although an age restriction was included in the Convention on the Elimination of All Forms of Discrimination against Women (1979). Research on the development and educational consequences of child marriage flourished in the early 2000s, with the United Nations Children's Fund (UNICEF) providing useful summaries that were widely disseminated. After this research, there was more attention to CEFM as an impediment to development.
What we consider the traditional frame of CEFM, with its focus on age restrictions, does have some characteristics that make the frame likely to resonate or diffuse. The traditional frame fits within a now well-established norm cluster concerning childhood and children as a special class for government protection (Heywood 2017). The traditional frame's focus on a clear-cut remedy to the problem (legal age restrictions) also likely helps the frame to resonate. However, growing awareness in the human rights field in the early 2000s of slippage between legality and practice could have limited the importance of this clear remedy in spreading the norm and eradicating the practice (Hafner-Burton and Tsutsui 2005). Additionally, child marriage legal restrictions often include exceptions that still allow the practice to continue, even in highly developed countries (Arthur et al. 2018).
The feminist sexuality reframing of CEFM fits within a larger master frame or norm cluster concerning women's empowerment. The last 30 years have seen a growing focus on many different problems facing girls and women, including female genital cutting, unequal access to education, human trafficking, and discrimination in the workforce. 5 The 1995 Beijing Platform and growing UN involvement in this area helped provide a network for the diffusion of national-level gender equality policies and practices (True and Mintrom 2001). Moreover, the 2015 UN General Assembly Sustainable Development Goals championed gender equality; the goals clearly identify CEFM as a "harmful practice" that must be eliminated for gender equality (UN A/Res/70/1).
The feminist sexuality frame provides a clear-cut violator (the "patriarchy," as identified by many of our interviewees) and remedy (empowerment or equality). However, the violator and remedy are still rather vague; some populations may question who exactly constitutes the patriarchy. Similarly, there could be questions about the definition of empowerment. As such, although we find some limited evidence of the importance of the characteristics of the frame itself in encouraging the diffusion of the feminist sexuality frame, our findings on this point are mixed.

Quantitative Text Analysis of the Frame
We use quantitative text analysis to help us assess whether specific organizations in certain environments were more likely to adopt the feminist sexuality frame. We gathered all publicly available documents for all listed Girls Not Brides members. With the help of a research assistant, we assigned a date to each document.
Most Girls Not Brides organizations do not have an active online presence. Many of them are dormant. We found online documents for 147 of the Girls Not Brides organizations (a little over 10%), with a final sample of 42,839 documents. We restrict our final sample to only these organizations. Including non-online organizations would have harmed our inferences by adding copious amounts of zeros for organizations without an online presence. Due to the variety of organizations with materials online, together with the low cost of a website during this time period, we do not expect systematic differences between the organizations in our sample and the organizations without materials online. Nonetheless, this is a limitation of our study, and the inferences from this portion of the study should only be extended to organizations with an online presence.
Most of the organizations in our sample had English-language websites or had English options available on their websites. Due to the lack of a large corpus of non-English language documents, we further restrict our analysis to only-English documents. Our sample simply did not have sufficient documents in alternative languages to conduct meaningful text analysis. Future studies that look beyond Girls Not Brides member organizations could expand to include alternative languages.
We built a dictionary with terms related to the new frame: sexuality, gender, and patriarchy. We also included the term "adolescent health," which we see as somewhat of a frame bridging term between adolescent sexual rights, which relates to the new frame, and health, which relates more to the traditional frame. We then added "development' and "education" to represent the traditional frame. Although this is a straightforward lexicon-based task, we took a few extra steps to make sure that we found all the important concepts related to our frames for interest. We cleaned up the unstructured texts to remove noise. We used synset (WordNet 2010) to identify all the possible variants of the key concepts of our interest. We also used n-grams (uni, bi, tri) for identifying other related terms. Further, we used a word embeddings approach (Pennington et al. 2014), especially designed to identify conceptually similar multi-word expressions. After building a curated dictionary, we analyzed the texts and created aggregate counts of the terms in documents also identified to be pertaining to CEFM. We use this process to create monthly, yearly, and per-organization datasets for each term concerning both the traditional and the new frame. Figure 1 provides an overview of the CEFM documents produced by all organizations each year that includes key terms related to the new framing. As shown, there has been a marked increase in CEFM documents related to sexuality and gender over time and a smaller increase for documents related to patriarchy and adolescent health. This pattern continues to hold when we weigh these numbers by the total number of CEFM documents produced each year. Figure 2 provides a similar graph focusing on how the sexuality term, as one key component of the new frame, relates to two terms indicative of the more traditional frame: development and education. CEFM discussion in general has increased and invoked multiple frames over time; starting in 2014, sexuality overtook development and education as the most prevalent of the three terms examined here.
While Figs. 1 and 2 illustrate the diffusion of the new frame over time, they hide variation in frame adoption across the different organizations in our sample. We expect certain organizational characteristics facilitate adoption of the new frame. To examine this, we separate out the text data by organizations in 4-month time periods. Our dependent variables are the counts of documents each organization produced in each time period that are about CEFM and include the terms that we flagged related to the new frame (sexuality, gender, patriarchy, and, to a lesser extent, adolescent health). 6 Our first analysis examines the effect of a time trend and variables that represent the political opportunity structure and resources of the organizations in our sample. Specifically, we include variables from the Varieties of Democracy (V-Dem) Dataset for women's empowerment and freedom from civil society organization (CSO) repression of the country where the organization is based. The V-Dem variables are based on expert surveys (Coppedge et al. 2019). The women's empowerment variable is a composite index of women's civil liberties, political participation, and civil society participation; the scale varies from 0 to 1; a higher number indicates more women's empowerment in the country. Similarly, the variable for freedom from CSO repression captures whether a government represses civil society; a higher score indicates less abuse. We expect more participation in the new framing from organizations in countries with open political opportunity structures, as captured by these V-Dem indices.
We include variables that capture the age of the organization and whether it has any affiliation with the United Nations. These variables provide a snapshot of the longevity and network ties of each group. 7 We also include a variable for the total number of CEFM documents with any frame each organization produced in the time period. The count of total CEFM documents produced is probably the best predictor of organizational resources and a favorable political opportunity structure for advocacy. Organizations producing many documents that can be found publicly have the human resources necessary to reframe an issue, are likely not fearful of government reprisal, and have the network ties and connections necessary for research and advocacy. Table 2 provides the results of these basic models. Two things are worth noting. First, we find organizational-level evidence of diffusion of this new frame over time. Compared to the beginning of 2011, when organizations are very unlikely to produce any CEFM documents using the sexuality terminology in any 4-month time period (prediction of 0.13 documents with a 95% confidence interval of 0.10 to 0.17, with all other variables set at their mean levels), organizations are predicted to produce 0.38 documents that use this new frame in each 4-month time period at the end of 2019 (95% confidence interval of 0.30 to 0.45). Although this figure may seem small at first glance, given that the average organization produces only around one CEFM document of any kind in any time period (mean of 0.92), we see this as substantively important evidence of the diffusion of this new frame over time.
Second, we find some evidence that resources matter. In almost all instances, the count of total CEFM documents produced is a positive predictor of the adoption of the feminist sexuality frame. The inclusion of this variable almost completely wipes Year Sexuality D evelopment Education out any relationship between the other indicators for resources and political opportunity structures and our dependent variable.
We conducted one more quantitative analysis which we think is novel in this research area. Our process tracing and interview evidence showed that there was a potential turning point in the diffusion of this new frame in March 2016. At that time, a 2-day gathering of international and local advocates was convened to discuss the sexuality frame and examine how issues of sexuality and gender can bring lasting change in the areas of CEFM and women's empowerment. The event was an opportunity for discussion in an insulated, non-political environment, with frame articulators presenting local expertise with trusted partners; the group's main work started after this gathering. There were around 40 attendees. We see this event as consistent with Wiener's (2004) discussion of a "discursive intervention" in that it was a meeting designed to increase discussion around the new frame. The "CEFM and Sexuality Programs Working Group" was created following this gathering. The Working Group was originally comprised of a heterogeneous group of international  Table 3 provides the results of models where the key variable is now a dichotomous indicator for whether the time period under examination is after the March 2016 event. As shown, we find evidence that this discursive intervention did increase the use of the new feminist sexuality frame.

Process Tracing and Interview Evidence
We use process tracing and interviews to help us investigate the process by which the feminist sexuality frame diffused over time to some organizations. It is difficult to convey the full, rich description of this evidence within the word limit of an academic article; the following analysis provides just a snapshot of evidence and narratives we reviewed, highlighting the common themes we identified. For the process tracing, we were given access to internal documents, emails, and public reports of an organization that is part of the CEFM and Sexuality Programs Working Group. Because of concerns of confidentiality, we are not naming the organization (instead labeling it "Organization A"), but we can state that it is a US-based organization that has a broad human rights and development mission. It is well-staffed and funded and was in existence for decades prior to the time period under examination. In 2013, Organization A drafted an initial concept note, which outlined a partnership with a private funder. Although Organization A had previously worked in many areas of women's rights, CEFM was largely new for the organization. Following the original concept note, Organization A developed their specific CEFM strategy, including a discursive advocacy strategy based on the input of feminists embedded in local contexts. Special attention was paid to understanding how patriarchy and gender inequality are the root causes of CEFM; age restrictions were seen as a temporary delay to larger problems associated with a lack of women's empowerment.
Organization A helped in coordinating a small meeting of advocates in March 2016. As mentioned, this meeting resulted in the creation of the working group dedicated specifically to addressing CEFM through the sexuality-based frame. After this meeting, the internal documents of Organizational A show diffusion of the sexuality frame to its partners and others in its network. As we learned through our interviews, Organization A did not give themselves credit for the diffusion of the frame; instead, the organization continues to view itself as part of a heterogeneous coalition of local and international actors. Organization A wanted to build a shared vision and saw alliances around the new frame as more important than credit-taking.
After tracing the development of the frame in Organization A's documents, we conducted 10 1-to 2-h-long semi-structured interviews with advocates. Our first interviews were with individuals identified in Organization A's internal documents; later interviews were the result of snowball sampling. The interviews highlighted several narrative themes as to why the feminist sexuality frame originated and spread, including: (a) the bridging of the CEFM issue and new frame to issue areas and frames where advocates had previous experience and interest, (b) the establishment of patriarchal norms as a common source of blame, (c) the empowerment of local partners, (d) the conversational space where the reframing occurred, and (e) a global time or environment ripe for the reframing.
First, we identified a "frame bridging" narrative in the staff of Organization A: their past experiences helped bring a feminist sexuality frame into the CEFM issue space. When developing Organization A's CEFM strategy, advocates drew on their previous experiences in the areas of LGBT + advocacy and women and child rights. Table 3 The effect of the discursive intervention, political opportunity structures, and resources on a sexuality frame in CEFM discourse # GEE Negative Binomial Models with AR (1) Correlation Structure and Time Fixed Effects, ## rare event logit.
Standard errors in parentheses. **p < 0.01; * p < 0.05. (1) (2) (3) One interviewee identified a connection between CEFM and their previous work on teenage pregnancy prevention in the USA. Their earlier work led this interviewee to see age-based strategies as ineffective for transformative change. Instead, they adopted a feminist framing of the issue primarily as one of structural injustices and a lack of women and sexual empowerment. Although it is easy to measure the number of teenage pregnancies over time, the outcome "was not meaningful for girls' lives" and represented a "fundamental ideological confusion between cause and consequence." 8 Drawing on this experience, this interviewee framed CEFM in a similar way: age of marriage is "not totally irrelevant" but was alone unable to address the core issues of gender inequality and patriarchy which were the root cause of CEFM. Their experience and the bridge they were able to articulate between teenage pregnancy and CEFM was influential as the organization first developed its conceptualization of the issue and its advocacy strategy. Organization A drew on local partners in ways that facilitated frame bridging. Through an initial mapping of local initiatives using the feminist sexuality frame, Organization A was able to "demonstrate that these programs exist around the world" and are not just being championed by "Western NGOs." 9 As one interviewee noted, although not mainstream, the feminist frame was not altogether new: there were "always people that had the feminist theme or frame." 10 Similarly, the organization's initial local landscape analysis helped in later bridging the CEFM issue to women's rights organizations that already had a feminist lens but lacked specific "language" and "grammar" concerning CEFM. 11 A second key narrative theme was establishing a "link between the patriarchy as a form of gender injustice" and the ultimate root cause of CEFM. 12 While there may be many factors influencing CEFM, as one interviewee stated, "the primary among them (root causes) being patriarchy." 13 Nonetheless, interviewees also highlighted that the feminist sexuality frame may cause difficulties for organizations only interested in measurable output. Although it is easier to measure progress in establishing age restrictions, "choices aren't expanded if [the girl is] still married at 18." 14 Interviewees noted that their funder "trusted us, came along on the journey" concerning the new frame. 15 Organization A was able to devote its resources to the feminist sexuality frame, even if it was more difficult to measure.
The third narrative theme drawn from the interviews was the importance of local network partners. Organization A sought to interact with on-the-ground partners in a partnership shaped by privileging local or lived perspectives. The goal was not to take a Northern/Western framing to local audiences; instead, the focus was getting the work and voices of local actors to an international audience. As one interviewee described, 9 Interview 1, Former Organizational Advocate. February 11, 2020. 10 Interview 3, Organizational Advocate. February 20, 2020. 11 Interview 9, Organizational Advocate. April 9, 2020. 12 Interview 2, Organizational Advocate. February 20, 2020. 13 Interview 6, Former Organizational Advocate. March 3, 2020. 14 Interview 4, Organizational Advocate. February 20, 2020. 15 Interview 2, Organizational Advocate.  "really engaging them (the local partners) in a way that's respectful is essential to our success." 16 Another interviewee shared how "we have been very conscious to front out partners on platforms that are discursive spaces," demonstrating the importance of local partners and a link between their local work and global discursive spaces. 17 Local network partners also created credibility at both the local and the international level, which in the words of one interviewee, "you can't buy it. You can't buy real trust and credibility… there's nothing good that happens that isn't based on trust and relationships in terms of transformational change." 18 The new frame diffused as a result of the analysis and work of local network partners: "The first step, I think was to make sure, looking back that there was a program match (in the nation of focus) so that the global discussions and analysis were fully grounded in reality." 19 As one interviewee reported, "we didn't do anything without local actors." 20 The new framing was not without challenges, as it required partners to think and adopt policies in a more "nuanced, complicated way." 21 According to our interviewees' descriptions, local network partners provided contextualized language which allowed other organizations to meaningfully engage with their new framing of CEFM. They sought out "grassroots voices, feminist voices" to partner and support so that these organizations "had a seat at the table." 22 Through supporting local groups, the organization "brought the voices…into the public domain" in a way that was critical for success. 23 This process did not appear to be extractive; multiple interviewees, including interviewees from local partner organizations in the Global South, highlighted that local groups were encouraged to speak for themselves.
The empowerment of local groups was bolstered by the funding organizations had received. Grant funds allowed Organization A to "enter into rooms" they may have otherwise been left out of. 24 Funding allowed the organization to "make grants" to local partners, creating and maintaining an "informal tribe" of like-minded and empowered groups. 25 When the growing network then went to major international meetings, there was a "core group bolding using the language" of the new frame. 26 As hypothesized, the nature of the discursive environment or venue greatly helped the new frame diffuse to additional organizations and advocates; this was the fourth narrative theme identified through our interviews. The creation of a "fluid" 27 and "trusted" 28 space where "honest conversations" 29 could be had in confidence between advocates was repeatedly identified as critical for the diffusion of the new frame. The "discussion among peers" was essential for "assimilation" that was "at a different level." 30 Previous trust and work with local feminist organizations allowed for open dialogue about CEFM, even though many feminist organizations had not seen the issue as a particular one of importance prior to the discussions. 31 The new frame was not always welcome; interviewees noted that they were occasionally left out of global discussions. Repeated interactions and active engagement helped create "new spaces where these conversations could occur." 32 Organization A's reputation appears to also have aided in the diffusion of the frame and the creation of supportive discursive venues. Although they were not the biggest player, they were seen as "edgy, intellectually interesting, and honest" in creating the space for these conversations. 33 The organization's edginess was critical to its strategy and adoption of the alternative framing in the first place. 34 Finally, we identified global timing as an important narrative theme in the interviews. In the early 2010s, growing advocacy concerning LGBT + rights and health allowed for a bridging of issues around adolescent sexuality and an increased rightsbased approach to sexual health. Although the implications were different, there was a growing awareness that the issues were "all interlinked." 35 In addition, there were many different movements and actors that were "starting to get a sense" of what "girls represent" in relation to sexuality and patriarchy. 36 The time period had an increased focus on adolescent girls as a "constituency," 37 reflecting a broader idea that issues of girl's rights had to be addressed to create a "community of future leaders" 38 within the feminist movement. Table 4 provides a summary of the empirical evidence we found using our mixed methods approach. Our quantitative evidence shows that a feminist sexuality frame grew concerning CEFM over time, especially after a discursive intervention in March 2016. To varying degrees, our qualitative evidence indicates that the nature of the frame, the messenger, the actor receiving the message, and the discursive environment all influenced the adoption of the new frame by some organizations. It appears that the interaction of the discursive environment and the networked characteristics of the messenger were the most important in encouraging frame adoption. Organization A's characteristics and efforts in developing a heterogeneous, transnational network helped to facilitate the discursive environment and venue necessary for frame adoption. Let us briefly address potential alternative explanations. In line with traditional international relations theory, an alternative line of reasoning would suggest that state power drove the frame diffusion. If a powerful state or intergovernmental actor was championing the frame, this could lead organizations and advocates to adopt the frame in response. We find no evidence of this explanation. Although the issue of CEFM and the frame was later adopted by some states in this issue space, the most powerful being Canada, this frame adoption came later in the process, well after the key moments identified in our process tracing analysis.

Discussion
Another potential alternative explanation concerns donor funding: was a donor driving this diffusion? In other words, was it the anticipation of funding that led advocates to adopt the new frame strategically? This type of reasoning would be consistent with Bob (2005): organizations change their behavior to attract funding, ultimately leading to distortions of the principled ideals that would otherwise separate these actors from for-profit organizations. We find no evidence of this behavior. Although multiple interviewees, including someone from a funding organization, mentioned that the frame diffusion was aided by the resources that Organization A had, the causal arrow was clear: the organization, in consultation with its local partners, together championed a specific theory of change and the feminist sexuality frame. Although we acknowledge that all actors were strategic, we observed no evidence that the adoption of the frame was tactical or that it was predominately driven by donor desires.
Finally, there could be concerns that our analysis simply captured changes in attention to CEFM overall and not reframing. We do not see much validity in this statement. Our quantitative analyses account for the overall attention each organization was devoting to the issue in each time period. Even given increases in CEFM advocacy, we still identify increased use of the new frame over time. Nonetheless, the new frame has not yet completely overshadowed the traditional development or health frame. Future work drawing on our analyses could examine whether the new frame drove subsequent attention to the issue of CEFM and whether the development frame continues to subside. Multiple interviewees indicated that they "feel like there has been a sea change" 39 in CEFM advocacy and were "very proud" 40 about the role that the new frame had in increasing attention to the issue. As one interviewee put it, "you can't believe the change we are seeing" as a result of the sexuality frame. 41

Conclusion
Advocacy concerning child, early, and forced marriage has recently shifted. Unlike a traditional view of CEFM as a development or educational issue that can be stopped by instituting legal age restrictions to marriage, the new frame focuses on the agency 39 Interview 3, Organizational Advocate. February 20, 2020. 40 Interview 2, Organizational Advocate. February 20, 2020. 41 Interview 6, Former Organizational Advocate. March 3, 2020. of the individuals involved, especially affected girls and teens, and the need for a broader social transformation. Patriarchy and gender inequality are the root causes. Empowerment, including addressing control of adolescent girls' sexuality, is seen as key to eradication of the practice. This reframing was the result of repeated interactions between trusted, networked actors. The new frame diffused over time because of a combination of the characteristics of the frame, the message, the messenger, and the environment where the discursive intervention took place.
Our project shows the value of mixed methods research. Without both qualitative and quantitative evidence, we would not have been able to develop such a rich picture of how the feminist sexuality frame diffused. Further, our project was enriched by involving those from both academia and the advocacy community, allowing us access and perspectives that might have otherwise been missed. Finally, we draw on both the sociological literature on framing and the international relations literature on norms. Taken together, a much deeper understanding of the causal processes driving shifts in framing emerges.
Our project provides several implications and calls for future work. First, this study focused explicitly on framing in the advocacy space around the international human rights norm against CEFM. We made no attempt to theorize or examine whether the same characteristics also apply to frame diffusion or norm adoption in the general public. We predict that characteristics of trust and social environments drive diffusion writ large; this has been largely established in traditional political psychological literature (Levi and Stoker 2000). Future experimental work, like the one done by McEntire et al. (2015), may help us better understand how frames diffuse to the general public. Careful identification of the ways in which diffusion to the general public differs from diffusion among advocates can help us better understand whether advocates have certain social or psychological characteristics which separate them from the general public (Passy and Monsch 2020).
Further, we did not examine an environment where there was active advocacy against the underlying human rights norm. Our theoretical underpinnings and empirical examination did not focus on interactions with actors that were actively countering not only the reframing but the underlying international norm. Local advocates championing the frame did have to deal with resistance to the norm, and the theories of change we examined address how this new frame is best situated to deal with this resistance. Nonetheless, as the new frame continues to diffuse, future examinations that focus more on the existence of those counter to the norm is necessary.
Finally, our study spoke to the importance of patience. Although the time period under investigation may seem relatively brief to academics, this represents multiple funding cycles and change within organizations. Both qualitative and quantitative evidence showed that repeated interactions and discussions were key. Trust cannot be built, and discourse cannot be fruitful, in short bursts of time. It was a timeconsuming process. However, as is well-known among advocates, many funders demand evidence of immediate success; organizational leadership would like indicators of annual or semi-annual improvement on an advocacy issue. If the goal is persuasion and shifts in the underlying logic about a world problem, we caution against focusing only on short-term changes. We hope future work continues to problematize the issue of slow-moving change in the advocacy process.