Mixed Methods
Mixed methods can be particularly appropriate for multidisciplinary social research such as that conducted by GAGE which brings together political scientists, economists, anthropologists, public health experts, and more. Greene et al.’s (1989) typology provides a helpful basis for understanding how mixed methods can be useful in the context of GAGE’s research. The use of qualitative and quantitative data can be triangulated to mutually corroborate findings; the different types of data also complement each other, with qualitative interviews allowing elaboration on survey data, for example. Qualitative findings may also be used to inform the design of quantitative studies as well as refine questions; Guglielmi et al. (this issue) find their interviews with Rohingya refugees enabled conversations to take place about the most appropriate terminology for particular concepts, which then informed the quantitative survey design.
The role of qualitative data is particularly important in the case of sensitive topics (e.g. experiences of child marriage, maintaining health and hygiene during menstruation, or utilizing sexual and reproductive health services) where the value of quantitative data is limited. Qualitative, semi-structured interviews create space for such topics to be explored or even be brought up by respondents themselves without direct probing through predetermined questions. For example, in Coast et al.’s article (this issue) qualitative research was able to add richness to data on topics such as abusive relationships; whilst the quantitative data indicates prevalence of consensual vs. non-consensual relationships, the qualitative research shows girls’ aspirations for their relationships and the stigma they have to navigate. The latter would be challenging to collect in a quantitative survey. Similarly, Abu Hamad et al. (this issue) observe that child marriage rates are underestimated in surveys, again likely due to stigma and illegality; but qualitative research created opportunities for adolescents to disclose their experiences as well as point to its drivers.
Extensive piloting enabled GAGE to develop and refine the instruments used for both methods, and ensure that they aligned with the wider objectives of inclusivity and diversity. Using surveys has expanded the breadth and range of the study to include data from a large number of participants in a short time, and across multiple cohorts. The inclusion of quantitative data allows for broader generalisation of results, and anchors the qualitative data in findings that can be considered representative of a given study population. This allows us to draw robust quantitative comparisons across characteristics of adolescents such as urban/rural, refugee/non-refugee, ever married/never married, disabled/non-disabled, older/younger, and male/female. The quantitative data also allows us to utilise multivariate regression analysis and other methods, providing a robust statistical foundation to support qualitative descriptive findings.
By piloting a variety of methods prior to large scale data collection, researchers in the six focus countries were able to explore what methods were most effective, and adapt existing tools to meet the conditions and requirements of different contexts and cohorts. This then informed subsequent stages of the research. An example of this can be found in the article by Bani Odeh et al. on the experiences of adolescents with disabilities, which outlines the training that researchers were given in the use of multiple methods including visual methods, drawing techniques and interviewing. This gave researchers a broad toolkit for generating data with adolescents with visual and hearing impairments on a ‘case-by-case’ basis according to the interests and adaptation requirements of adolescent participants themselves. Indeed, this flexibility is key to promoting innovative research practices (Hesse-Biber 2015, p. xxxiv) as well as centring participants within an ethic of care (Jones et al. 2020).
For the quantitative surveys, a number of approaches were taken to ensure inclusivity. First, cognitive interviews were conducted to ensure comprehension that also included the socially marginalised populations described above. Second, enumerators were extensively trained on working with socially marginalised adolescents, with a small subset specifically trained to work with adolescents with disabilities. Third, the research team utilized local expertise (e.g. disability support services) and incorporated team members with sign language expertise as needed.
Alderson and Morrow (2011) suggest that mixed methods can also be a way of including children and young people in research, actively and respectfully. For example, Clark and Moss (2001) describe using a multi-method or ‘mosaic; approach as a way of recognising young children’s voices and language. A mosaic approach enables the inclusion of a greater diversity of perspectives, recognising that not all methods will suit all young people, as in Bani Odeh et al.’s findings. Visual methods—for example, those which combine drawing, crafts, photography or other activities not premised on being able to verbalise—can be a powerful way to convey ideas (Alderson and Morrow 2011). However, whilst young people who have attended school may be familiar with these activities from their use in educational settings, this does not mean that they will necessarily enjoy using them (Christensen 2017), and young people who are out-of-school may not find these methods intuitive (Crivello et al. 2009). Having a range of ideas and methods as part of a toolkit can give more flexibility, and thus more scope for engaging a diversity of participants and voices in research (Jones et al. 2019; Małachowska et al., 2020).
A lack of adaptation of research tools can lead to the tokenistic inclusion of disadvantaged young people, including those with disabilities, who cannot use their participation as a means to express their opinions (Ottmann and Crosbie 2013). Developing tools which enable marginalised young people to both participate in research and use their participation to reflect on and explore their own experiences in a supportive setting is therefore an ethical imperative. The adaptations needed to include adolescents and youth with disabilities in research often benefit their non-disabled peers too (Wickenden and Kembhavi-Tam 2014), as has been the case with the use of participatory photography approaches in GAGE developed initially to facilitate the engagement of adolescents with hearing impairments and audio diary approaches to facilitate the participation of adolescents with visual impairments (Jones et al. 2017). Whilst during the baseline research drawn on throughout this special issue, adolescents with intellectual disabilities were not included, researchers are exploring the safeguarding and practical measures that will be needed to support their inclusion in future rounds—one of the further benefits of longitudinal research being the opportunity to refine methods to promote inclusive practices where their need emerges.
It is important to recognise that whilst mixing methods can strengthen findings, there are also research questions and domains which are better suited to exploration using particular methods, as well as occasions in this Special Issue where authors have elected to focus on findings from a particular methodlogy. The articles by Essaid et al., Emirie et al. and Bani Odeh et al. each use a range of qualitative methods such as focus groups and interviews to explore the violence of life in displacement, child marriage and disability-based exclusion respectively. These are complex and sensitive topics on which quantitative survey data from GAGE is available, yet where a qualitative analysis that centres the subjectivity and agency of marginalised adolescents can both nuance and enrich understanding of their experiences, and counter representations of such young people as passive victims of structural violence. There is no corresponding quantitative data in the qualitative article by Ghimire and Samuels on commercial sex work (CSW) in Nepal; further quantitative research could deepen their analysis by establishing numbers of adolescents participating in CSW, creating a better understanding of the scale and who is participating; disaggregate by districts of origin, contributing towards a clearer picture of drivers; and disaggregate by venues of CSW, enabling better targeting of interventions. Whilst quantitative data—which is the subject of analysis in Woldehanna et al.’s paper on educational attrition in Ethiopia—can be compared across sites, qualitative research generates data which is context-specific and rich in detail. A combination of both under a single conceptual framework which prioritises adolescents’ agency and works towards transformative policy and programming objectives is key to the mixed-methods approach championed by GAGE.
Qualitative and Participatory Approaches
Several articles in this Special Issue explore data generated through participatory methodologies (Jones et al.; Hamad et al.; Emirie et al.; Bani Odeh et al.). The ‘new social studies of childhood’ emphasises how children and young people are agentic actors and can construct and shape their social lives (Prout and James 1990; Davidson 2017). This approach asserts that the definitions of childhood which dominate popular discourse are socially constructed, culturally specific, and a product of various political and social actions over time (Hendrick 1997). While childhood is evidently affected by social, environmental and political change, children are active in this process and can resist these changes, show resilience to them, or rework them to their advantage (Katz 2004, p. xi). In line with the United Nations Convention on the Rights of the Child (UNCRC 1989) children have a right to be properly researched. Articles in the UNCRC on children’s right to express themselves and their opinions, to expect high-quality services, and receive protection from exploitation all have relevance for a participatory research agenda (Beazley et al. 2006; Warshak 2003).
Participatory research with young people has emerged from this tradition, demonstrating its utility for understanding young people’s agentic participation in their own lives. Bradbury-Jones et al. (2018, p. 80), in a systematic review of participatory research with children and young people across diverse geographies, highlight that participatory spaces can ‘recalibrate opportunities and attention given to marginalized and silenced groups’ such as young people with disabilities or survivors of abuse and neglect. Participatory research can also support young people to develop skills and exercise agency in otherwise adult-dominated decision-making spheres. This emphasis on youth participation has been evident within international development since the early 2000s, in part as a facet of the broader participation agenda, originating in critiques of ‘top-down’ development that became popularised within work and research during the 1980s and 1990s (Chambers 1983, 1997). A central aim of the participatory agenda was to challenge the power dynamics of development processes and build in mechanisms for those affected by these processes to be agentic in determining them (Mohan 2001).
Participatory methods have become particularly popular for researching marginalised and socially marginalised children because they name and seek to address power inequalities both between children and adults, and within groups of children. They can provide a means to address the structural and relational marginalisation of particular groups of people. In this framing of young people’s participation in research, children are not seen as different and in need of particular methods on the basis of their biological age (Christensen 2004). Addressing power relations is the central concern when designing methodology; rather than seeing participatory methods as ‘child-friendly’, participation should be reconceptualised as a way of ensuring that marginalised voices are represented and included. In terms of methodology, this means an awareness of power differentials and a politicisation of processes for generating data.
The extent to which these transformative and empowering aims have been realised by participatory approaches has been subject to critical reflection, most notably by Cooke and Kothari (2001). Work on young people’s participation has been similarly critiqued; White and Choudhury (2007, p. 539) suggest that the structure of the development industry itself can result in the co-option of young people’s voices into a pre-established agenda, often in ways that reinforce rather than challenge class and racial divides. Marco-Crespo et al. (2018) add to this the challenge of sustainability, as short-term funding cycles for research and programmatic interventions too often replace supporting young people in age-appropriate ways over the course of adolescence with participation in participatory research activities.
These important methodological caveats notwithstanding, there is a growing body of evidence that emphasises not only the importance of participatory research as a process that seeks to disrupt relations of power and disadvantage, but also underscores the quality and innovative nature of the insights that such research can generate. Engaging with young people through participatory research and evaluation can strengthen understandings of their lived realities and, in turn, strengthen the interventions designed to support their well-being (Marco-Crespo et al. 2018).
GAGE has used a number of qualitative methods that have a participatory element in that they create space for young people to decide how and what they want to talk about (see Jones et al. 2019 and Małachowska et al. 2020 for more information on these tools in the context of the broader GAGE research programme). The object-based interviews described in the articles by Hamad et al., Mitu et al. and Emirie et al. (this issue) which engage with the issue of child marriage included the use of the My Favourite Things tool. This tool was originally developed in the United Kingdom (UK) for exploring the trajectories of pregnant young mothers and, later, their children (Thomson et al. 2011; Thomson and Hadfield 2014). It is unstructured and open-ended, as it starts by encouraging adolescents to think of a favourite object and use this as a point from which to explore its significance in their lives. This tool was able to be adapted for use in urban and rural settings where people had fewer possessions, and with adolescents of varying abilities. Because it works with the capacities of adolescents and allows them to set the agenda for the interview, it can be seen as genuinely participatory and inclusive, as well as shifting the power of interpretation within an interview environment to centre on participants’ intentions and priorities. For example, in the case of some married girls who engaged with this tool, the household items that they selected allowed them to talk about their mixed emotions around gaining respectability in the community through marriage and in their new role as a homemaker, but the simultaneous challenges of shouldering adult burdens as a child with limited guidance and support and the loss of opportunities to socialise with unmarried peers who were now on a different life trajectory.
Intergenerational interviews can also be used to improve understanding of how adolescent decision-making has changed over time, by gender, and also potentially by disability. This format of interviewing was used by Hamad et al. (this issue) to explore drivers of and changing norms around child marriage in Gaza and Jordan and encourage reflection by caregivers on the practice. It was also used by Bani Odeh et al. (this issue) to understand how the social experience of disability has changed over time in Jordan, offering insights into entry points for effective norm change. Specific training and debriefing sessions were held to mitigate concerns around confidentiality and anonymity when interviewing within families, including, for example, on how to ensure neutrality and avoid expressing judgment regarding intra-family dynamics, developing tailored probes based on information learned in one interview to encourage in-depth and well triangulated findings, and reporting any child protection issues that arise. The tool has enabled adolescents to take the lead in interviewing parents and grandparents on topics such as the key challenges they perceive adolescents as facing—for example, child marriage and child work—and whether these are similar or different to the challenges they themselves encountered at the same age. This can help to unpack and elucidate complex interconnected drivers; for example, Abu Hamad’s article finds that whilst intergenerationally transmitted cultural norms may play a part in the ongoing practice of child marriage, the experience of displacement has amplified the prevalence of the practice for some families. For Bani Odeh et al. (this issue) intergenerational interviews gave adolescents with disabilities the opportunity to lead this aspect of the research. By putting the adolescent in a different role, it opens up opportunities for conversations that adolescents may be interested in having, but find difficult to initiate themselves, given family hierarchies and dynamics (see also Bani Odeh 2019).
There are challenges in determining the extent of adolescents’ participation, even where participatory methods are used. Indeed, this is a key debate within the literature on participatory methods. For example, Beazley et al. (2006) argue that participatory research should involve those who will participate in determining questions, generating data and analysing it. However, Morrow (2008) suggests that this can be particularly difficult; the differential power between adult researchers and young people becomes even more prominent when it comes to analysis and presentation. At this point, young people, particularly those from marginalised groups, cannot draw on their own knowledge due to the need to interpret data according to academic conventions, which can create feelings of disempowerment (Mayall 1994). Nonetheless, one of the benefits of a longitudinal approach in which cohorts are followed up over time through repeated interactions is that this allows space for further engagement and refinement of analysis and findings with those adolescents who have participated in the research. For example, in Ethiopia the second round of qualitative interviews about the cultural practice of ‘shegoye’, an adolescent-only dance where young people meet prospective marriage partners discussed in Emirie et al.’s article, revealed the deeply gendered nature of the practice, including physical and verbal gender-based violence towards girls to ensure their active participation. This did not emerge until later in the data collection, indicating the benefit of ongoing engagement and the building of relationships with participants so that their participation can be continuously advanced.
Social Networks and Peer Relationships
The social-ecological framework at the heart of GAGE research draws on an understanding of adolescence not only as a biological experience, but also a social one. The emotional and physical changes adolescents undergo change how their families, communities and societies view them (Patton et al. 2016). In many places, this leads to greater restrictions on mobility for adolescent girls due to fears around their safety (Harper et al. 2018), while adolescent boys are given greater access to food and educational opportunities when resources are limited. Outcomes for adolescents are highly dependent on their family, household, community, state and global contexts (Jones et al. 2018). The social-ecological framework means that GAGE research engages not only with adolescents, but with the contextual factors and individuals which shape their agency and capabilities. To explore these factors, several different methodological tools were used (Jones et al. 2019).
One tool used to explore relationships by Hamad et al. in Jordan and Gaza and by Jones et al. in Ethiopia (this issue), both articles that explore the experiences of adolescents who are known to experience social isolation and limited mobility, sees researchers supporting adolescents to draw a social support quadrant. Adolescents were given a pen and paper and shown how to divide the page into four sections: one each for people they liked to spend time with, people they avoided, people they went to with their worries, and who they went to for advice. The aim was to find out more about adolescents’ social resources—who they were able to go to and the types of support they could draw upon in different situations. One of the strengths of this approach is its flexibility; if respondents were reluctant to disclose their own social support networks, then questions could be asked in an abstract sense—for example, ‘who do boys of your age turn to’—to garner information on the sources of support they saw as available more generally to young people of the same and opposite gender.
A second tool used in the same articles is the social network hexagon, which draws on social network literature (Flynn et al. 2017; Tubaro et al. 2016; Vincent et al. 2018) to explore the range of people an adolescent interacts with in-person and virtually, and the quality of those interactions. The hexagon’s six segments include family, friends, neighbours, community members, romantic interests and intimate partners, and online peers and community. It is used to probe the relative importance of different individuals in each of these domains in the adolescent’s life, and why. The tool enables the adolescent to share the sorts of things they discuss or confide with a given person in their network, whether or not that person is a role model, as well as the extent to which adolescents form ties with similar or dissimilar others along lines of gender, age, ethnicity, religion, class or educational status. It also looks at the extent to which adolescent lives cross over in different spaces—community, school, neighbourhood and online. In particular, it facilitates a discussion around the quality of peer relationships—often problematised as sources of negative pressure for young people—may be very important and positive for them (Christensen 2017). Peer groups are a significant influence on the mental health of adolescents because this is a time of life when they are seen to increasingly turn to friends rather than parents for emotional support and advice (Patton et al. 2016). As such, we found that this tool is better suited to older adolescents for whom non-family peer and trusted adult relationships are becoming increasingly important, as highlighted for example in the paper by Jones et al. (this issue) in their exploration of the drivers of psychosocial wellbeing among adolescents with disabilities in Ethiopia.
Quantitative Approaches
Complementing these qualitative tools, GAGE uses three core quantitative instruments, with common as well as country-specific questions and modules. The survey instruments can be found here: https://www.gage.odi.org/types/method-tools-and-guides/. The primary female caregiver module involved interviewing the adolescent’s main female caregiver in each household (although secondary female caregivers or male caregivers were substituted as needed). Questions focus on the household composition, family background, assets, durables, dwelling, and access to safety nets. For primary female caregivers of younger adolescents (aged 10–12), there are questions on parenting, gender attitudes and norms, mental health, financial inclusion, time allocation, exposure to violence, fertility, marriage, and use of technology. GAGE drew on validated tools when they existed, adapting as appropriate to the local context. In addition, GAGE developed new measures on topics where measures did not exist, particularly measures specific to very young adolescents where current tools are often not appropriate. The quantitative surveys cite all measures used. The focus on parenting is particularly important as it allows us to look at parents’ influence over adolescent trajectories as we continue to follow the adolescents over time. Collecting detailed information on the primary female caregiver also allows us to investigate the intergenerational transmission of mental health, violence, and attitudes, among other issues.
The survey tool used directly with adolescent respondents explores education, time allocation, paid work, health and nutrition, mental health, mobility, voice and agency, social inclusion, programme support, financial inclusion, economic empowerment, technology, marriage and relationships, sexual and reproductive health, and violence. The questions are tailored to the respondent’s age and adapted to the local context. There are also specific skip patterns, based on (for example) refugee status, marital status and disability status that allow the enumerator to ask certain questions of certain populations. These surveys are closely linked to the GAGE conceptual framework to collect detailed data across all six GAGE capability domains, focusing on gender as a cross-cutting issue.Footnote 1
In the articles included in this Special Issue, quantitative data does several important things. Firstly, it can help to elucidate apparent contradictions that may emerge within qualitative data sets where subject experiences are centred. In the article by Sultan et al., the qualitative data suggests that high levels of corporal punishment and bullying puts adolescents off attending school, yet the quantitative findings on actual school attrition shows that this does not lead to them dropping out because it happens so extensively as to be normalised. This then generates further important questions about how to challenge norms. The quantitative findings also allow the exploration of group characteristics and their impact on outcomes. In Jones et al.’s article, the authors show the way that patterns in educational achievement by adolescents with disabilities across different sites could be explored by disaggregating data by impairment, gender and age, finding correlations between numbers of years of schooling attained and different types of impairment. In Coast et al.’s article, a quantitative survey being undertaken with both adolescent mothers and non-mothers found that adolescent mothers are significantly more likely to have experienced forced sexual encounters than non-mothers, emphasizing their particular vulnerability.
Longitudinal Research
The articles in this Special Issue focus on GAGE’s baseline data, which will serve as the first of up to four rounds of panel data collected over the course of our study. In 2016, The Lancet’s call for a stronger focus on adolescents highlighted the need for more long-term research with young people (Patton et al. 2016). The short-term basis of much of the available knowledge on adolescence is inadequate for understanding the consequences of development interventions, especially given that this period of life is characterised by such a significant transition. Longitudinal research can offer unique insights into patterns of change in a way that is simply not possible without following groups of people over time. Yet qualitative longitudinal research presents both practical and ethical challenges. Both the meanings attached to data and the contexts in which they are generated will change over time. Additionally, there may be challenges in maintaining consent in participation as young people become more mobile (Taylor 2015). The richness of detail built up through long-term, repeated interactions can also cause problems in assuring anonymity, especially when those represented in reports remain involved in projects (Neale 2013; Morrow et al. 2014). There can also be difficulties in maintaining contact with participants, especially in contexts of displacement or mobility. Developing relationships that can withstand this requires time and investment by researchers, which must be supported by funding (Adamson and Chojenta 2007). Because of these complexities and expenses, cohort studies such as GAGE are rare. Thus, the baseline findings highlighted in this Special Issue provide critical insights into the lives of socially marginalised adolescents who are under-represented in the literature on young people’s well-being, and foreshadow some of the challenges they are likely to face as they transition through adolescence and on to adulthood. For more detail on specific longitudinal research challenges, see Baird et al. (2021).