Evidence-based approaches for promoting gender equity in global mental health research: Study protocol for social network analysis of researchers in Nepal

Abstract

There are increasing efforts for capacity building of researchers in low- and middle-income countries (LMIC) to foster local ability to conduct high quality research. However, female researchers remain underrepresented in scientific communities, particularly in LMIC where they have limited networking and mentorship opportunities. This protocol is for a Social Network Analysis (SNA) to evaluate if gender-sensitive, need-based capacity building can improve researchers' networking and mentorship opportunities in Nepal. The conceptual framework is informed by Social Cognitive Career Theory. Cross-sectional and longitudinal SNA are used to a) assess individual researchers’ network characteristics and their association with academic productivity; and b) examine if the association of network characteristics and academic productivity is mediated by self-efficacy and outcome expectations. Recruitment is designed to include early-career and senior researchers conducting mental health research, as well as students interested in pursuing a career in mental health research. The network characteristics will be mapped for approximately 150 researchers in working in Nepal. SNA characteristics in the network (individual density, homophily, and centrality) will be compared with academic productivity (total peer reviewed publications, h-index), including mediation effects via self-efficacy and outcome expectations. Ultimately, this study will generate information to design more evidence-based strategies for capacity building of a gender-equitable research workforce in global mental health.

1. Introduction

There is a growing need for evidence-based research in global mental health, especially in low- and middle-income (LMIC) settings where there is high burden of mental health conditions and limited evidence-based research (Patel et al., 2018; Shaji, 2013). Improving local research capacity in LMIC can generate evidence to advance health services and reduce health dispairities (Shaji, 2013). Researchers' strong contextual knowledge and training are important to understand what influences local health system development (Beran et al., 2017) and what interventions should be adopted and implemented in a setting (Chase et al., 2018). Despite its benefits, mental health research is still a neglected field in LMIC (Patel et al., 2018). Institutional challenges such as lack of research training opportunities (Oquendo et al., 2018), lack of peer support and mentorship, limited collaborations, inadequate research opportunities, lack of research culture (Ponka et al., 2020; Thornicroft et al., 2012) and low institutional capacity to conduct research (Hanlon et al., 2017; Schneider et al., 2016) limit research opportunities. The situation is even more restrictive for female researchers in LMIC who face the double burden of gender discrimination and resource-strained research environments. Globally, gender has been shown to be a limiting factor in research opportunities (Shannon et al., 2019). Only 30% of the world's researchers are women, the representaiton is even less (19%) in South and West Asia (United Nations Educational, Scientific and Cultural Organization, 2019). Female researchers publish fewer papers than their male counterparts and are less likely to have access to international collaborations (De Kleijn et al., 2020). This gender discrimination is an impediment to academic career development (Cochran et al., 2013; Ranieri et al., 2016).

Over the past decade, there have been increasing efforts to support mental health research capacity in LMIC, with a strong focus on research networks (Breuer et al., 2019; Collins & Pringle, 2016). This includes strengthening research networks' inclusion of governmental, non-governmental, and community-based stakeholders (Rahman et al., 2020), which support cultural acceptance and contextual understanding. Strengthened research networks contributes to in-country capacity to conduct implementation studies and generate scientific evidence in developing mental health programs globally (National Institute of Mental Health, 2021). Developing a truly global mental health workforce translates into innovation across cultures by cultivating an environment of collaborations across differently resourced global settings (Griffith et al., 2016). However, mental health attracts less funding than other fields of global health (Iemmi, 2020, 2021a, 2021b), particularly in LMIC where on an average only 1.6% of the government budget is allocated for mental health (Iemmi, 2021b). This can be a highly limiting factor for conducting capacity building activities.

An inclusive global mental health network is integral to lead studies among underrepresented populations, especially women and ethnic-minorities, who are disproportionately affected by mental health disorders (World Health Organization, 2021). In Nepal, as in many LMIC, studies have shown women to have higher rates of mental health disorders, such as anxiety (Kohrt et al., 2012, Kohrt and Worthman, 2009) and depression (Albert, 2015). Moreover, in Nepal, suicide is the leading cause of death among women in reproductive age (Bhardwaj et al., 2018, Hagaman, Khadka, Wutich, Lohani, & Kohrt, 2018, Hughes, 2012; Marahatta et al., 2017). Gendered expectations of child behavior also contribute to a lower threshold for characterizing girls as having a disruptive behavior disorder (Langer et al., 2019). Healthcare utilization has been historically low among ethnic-minorities in Nepal (Chaurasiya et al., 2019). Studies are highlighting risk factors of poor mental health including mental health stigma among ethnic-minorities (French, 2020, Kohrt, 2009, Rai, Adhikari, Acharya, Kaiser, & Kohrt, 2017).

The social and cultural challenges faced by women and early-career researchers can be uniquely understood and discussed in a supportive network. For example, compared to other gender combinations, female mentees with female mentors are more likely to agree that their mentors served a role modelling function (Ragins & McFarlin, 1990). Some women feel that advice from male colleagues is less applicable because male colleagues often lack experience with career-oriented women, and men find women less relatable when interacting with them in professional leadership roles (Bickel, 2004; Brown et al., 2004; Dutta et al., 2010). To address this, initiatives such as Women in Global Health (Women in Global Health, 2021) and WomenLift Health (WomenLift Health, 2021) are creating space to support transformative leadership in global health. Studies in the US have shown how mentor-mentee pairs who share common characteristics such as race, ethnicity, and gender can create supportive academic environment. Forced diversity, however worsens opportunities for minorities and should therefore be avoided when supporting networking among researchers (Dobbin & Kalev, 2016, Dobbin & Kalev, 2017, 2018). Based on the National Academies of Sciences, Engineering, and Medicine (NASEM) recommendations, effective mentorship teams should include some mentors who share identity-level similarities (e.g., race, ethnicity, gender, age, stage of life, family/parental roles) (National Academies of Sciences, Engineering, and Medicine, 2019). Women receiving mentorship from other women, including dyadic peer mentoring (women-to-women), has shown benefits for academic productivity (Varkey et al., 2012).

Mental health research networks can improve researchers’ academic productivity. Research networks influence publication productivity and quality of research (Adams, 2012; Kyvik & Reymert, 2017). Successful international teams are characterized by strong collaborative activities (Barjak & Robinson, 2007). Highly productive researchers not only have important connections but are also integral in creating connections among other researchers (Ebadi & Schiffauerova, 2015). Similarly, collaboration with stakeholders such as people with lived experience of mental illness has shown promising results particularly in improving cultural acceptance and reducing stigma associated with mental health in primary care settings (Kohrt et al., 2020; Lempp et al., 2018; Rai et al., 2018).

When designing capacity building activities that support health networks, an important consideration is inclusion of gender- and ethnicity-minorities. How, where, and for whom and by whom capacity building is conducted is especially important in acknowledgement of the global Black Lives Matter activism and associated calls for decolonializing global mental health and addressing barriers to a global representative mental health research workforce (Büyüm et al., 2020; Weine et al., 2020). Supportive networks can help local researchers with long-term academic productivity and professional growth (Collins & Pringle, 2016; Kohrt et al., 2016; Thornicroft et al., 2012). To improve inclusion in these networks, we need a clearer understanding of research networks as well as individual traits such as self-efficacy and outcome expectations that uniquely identify gender and ethnic minorities. Thus, in our study, we plan to understand the network characteristics of mental health researchers along with their individual attributes of self-efficacy and outcome expectations to see how these constructs influence researchers’ academic productivity.

1.1. Self-efficacy and outcome expectations influence academic productivity

Social Cognitive Career Theory (SCCT) is a career development model by psychologists Lent, Brown, and Hackett to understand how individual, contextual, and socio-cognitive factors influence formation of vocational interests, goals, and choices (Lent et al., 1994). SCCT states that person-environment interactions determine learning experiences which can then influence an individual's ability to perform career-related tasks (self-efficacy), and her/his expectations of pursuing a specific career (outcome expectations) (Bakken et al., 2006). Self-efficacy and outcome expectations in turn impact performance domains and attainment. For example, if a female researcher wants to network with senior researchers for her career development, her experience is shaped by her personal characteristics (e.g., sex) and her environmental factors (e.g., how gender is perceived in her society). These person-environment factors not only shape how and if she engages in networking but also her perception of her ability to network (e.g., I can introduce myself to a senior researcher if I meet him/her in a social gathering), and her expectations from networking results (e.g., networking will help me develop my career).

Self-efficacy defined as “people's judgments of their capabilities to organize and execute courses of action required to attain designated types of performances” (Bandura, 1986) has four informational sources: a) personal performance accomplishments refers to personal success stories or past experiences of success in carrying out a task; b) social persuasion is verbal encouragement to carry out an activity; c) vicarious learning is learning by observing what others in one's field are doing; and d) physiological and affective states are having low anxiety and a relaxed state to carry out an activity (Bandura, 1997). An inclusive capacity building activity will aim to include these informational sources, and provide an experience of social persuasion and vicarious learning through network members that share a common goal of academic productivity. One of the most important environmental predictors of career persistence, especially among diverse populations is mentoring (Bakken et al., 2006; Gloria & Kurpius, 2001). However, mentoring can also be challenging if the mentor is a different gender or race because gender and racial stereotypes can reduce performance (Chesler & Chesler, 2002; Steele, 1997) especially when there is a lack of overlapping characteristics between role models and mentees (Cora-Bramble et al., 2010; Price et al., 2005). Thus, successful mentoring goes beyond sufficient time to mentee. It should include sharing life experiences and technical expertise in a way that respects and more importantly empathizes with the mentee (Bakken et al., 2006).

Similarly, outcome expectations are beliefs of the results or consequences of performing certain activities. If self-efficacy answers the question “Can I do it?”, outcome expectation asks “If I do it, what will happen?” We will assess both positive and negative outcomes in terms of result expectations. For instance, a female researcher could have positive outcome expectations (e.g., if I attend the networking event, I will get to meet senior researchers in my field) or negative outcome expectations (e.g., attending the networking event means I will have to stay out late at night, which my family will not approve). People are more likely to attempt behaviors that will result in positive outcomes (Lent & Brown, 2006). Much like self-efficacy, positive outcome expectations are strengthened due to personal success experiences (e.g., networking has helped me increase my collaborations in the past); and outcome expectations are also shaped by exposure to successful role models, meeting a mentor in a professional setting, and social and verbal persuasive communications or verbal encouragement (Lent et al., 1994).

Self-efficacy influences performance and academic outcomes, as demonstrated by an association of self-efficacy scales with problem solving ability (Bouffard-Bouchard, 1990) and academic performance (Chemers et al., 2001). Self-efficacy in the domain of a research career directly correlates to academic productivity among graduate students as well as faculty members (Hemmings & Kay, 2010; Pasupathy & Siwatu, 2014). While there have been many studies that associate high self-efficacy with publication outputs, most of these studies have been conducted in the US and high-income countries, which have particular constellations of individual and environmental factors influencing academic productivity. The most prominent self-efficacy tools (Research Self-Efficacy Scales, Self-Efficacy Research Measure, and the Research Attitudes Measure) that are used to assess research self-efficacy have all been designed using responses from North American graduate students (Hemmings & Kay, 2010). However, the environmental influences in LMIC are different. Culture shapes gender differences in perception of public and professional roles, and perception of one's environment is considered to be greater determinants of individual's behavior than objective reality (Swanson & Fouad, 2010). For example, limited networking opportunities (objective environmental influences) for a female researcher are important to assess, but her individual perception of these limited networking opportunities and how it influences her academic productivity (perception of reality) are important especially when applying self-efficacy concepts to minorities (Lent & Hackett, 1987). How early career minority researchers perceive their environment and its limitations are different for the perception of non-minority early-career researchers in the same environment. Thus, researchers' productivity can be determined by their networks as well as through individual-level constructs such as self-efficacy and outcome expectations, which can limit with whom a researcher connects and how they can best use their network for professional development.

1.2. Assessing networks and its influences

Social Network Analysis (SNA) is a method used to map, measure, and analyze social relationships between individuals, groups, or organizations (Blanchet & James, 2012). It supports exploration of the types of relationships between actors (individuals, groups, organizations), considered “nodes” within the network. Each relationship (ties or links) between nodes is presented as a line drawn connecting them. Some of the relations that we want to explore in this study are density, homophily, and centrality.

Network density is a measure of the frequency of connections in the network compared to the maximum possible connections that could occur (Powell & Hopkins, 2015). It is calculated by actual connections within the network divided by all possible connections. The number ranges between 0 and 1, with the values closer to 1 for denser networks. Individual densities for each node consist of actual connection a node has divided by all possible connections in the network. Homophily measures the tendency of nodes to form ties with other nodes with shared characteristics such as gender, race, and age (Powell & Hopkins, 2015). Finally, degree centrality is the total number of connections a node has within the network, and can help identify the key nodes within the network (Steketee et al., 2015).

Fig. 1 shows a network of six individuals (A, B, C, D, E, F), with the first three as men (A, B and C squares) and the latter three as women (D, E, F circles). We anticipate Fig. 1a as our baseline network with fewer connections among researchers (as determined by the lines joining each node), and less men to women connections (more homophily). At the end of five years (Fig. 1b), we anticipate a denser network with more connections between nodes (higher density), more connections for each node (higher centrality), and more diverse networking, e.g., more connections across genders (lower homophily).

Fig. 1.

Connections between researchers before and after the capacity building activities. In 1(a), the male (squares) and female (circle) researchers have unidirectional (line with one arrowhead) or bidirectional (line with two arrowheads) connections. At the end of our capacity-building activities 1(b), we anticipate a higher density (actual connections of nodes per total potential connections), lower homophily (tendency of nodes to form ties based on similar characteristics) and higher degree centrality (total number of connections of nodes).

SNA allows analysis on the role and influence of the actors, making it a widely used methodology in both public health settings (Chambers et al., 2012) and organizational research (Valente, 2010). Although SNA as a methodology traditionally has been used descriptively, there has been more support to use SNA for evaluation of interventions (Chambers et al., 2012). Recently, SNA has been used as a methodology to monitor and evaluate capacity building activities using measures of connectedness such as in-degree centrality, betweenness, and eigenvector (Leppin et al., 2018; Popelier, 2018) and cohesion such as density and geodesic distance (Popelier, 2018). Prior studies have used SNA to map networking, mentorship and collaboration following capacity building activities (Luque et al., 2010; Petrescu-Prahova, 2015) and have incorporated social network analysis to assess the relationship of network properties with academic productivity. SNA has been used to determine degree centrality (how connected the individual is) among faculty members to map the relationship based on number of publications. The study was successful in mapping in-department and out-department collaborations and also in identifying key actors who were associated with well-connected members in the network (Royal et al., 2014).

Another study assessed the correlation of betweenness centrality (collaborations) with h-index (the Hirsch-index of academic productivity, which correlates to the h publications being cited a total of h times, e.g., an author with 10 publications that have been cited 10 times has an h-10) to confirm positive correlation of collaborations with research productivity (Contandriopoulos et al., 2018). There is also evidence of better research performance among researchers who have more co-authors, and particularly one frequent, primary co-author (Abbasi et al., 2012). Studies have also shown that international collaboration as assessed by connection of a researcher (node) with a collaborator from an institution in a different country is associated with greater academic productivity (number of papers published), particularly among early-career researchers (Hara et al., 2017). Assessing mentorship potential, especially among early-career researchers using network characteristics, is a relatively new field (National Academies of Sciences, 2019).

In this study, we plan to use SNA to determine if network influences academic productivity of mental health researchers in Nepal. Our aim is to identify if network attributes such as higher degree centrality and lower homophily correlate with academic productivity. We anticipate a heterogenous network with more connections between individuals of different attributes (e.g., sex, age, ethnicity) to have higher academic productivity than a homogenous one. Finally, we will determine if self-efficacy and outcome expectations will have mediating effect in establishing the relationship between network characteristics and academic productivity (Fig. 2) (See Textbox 1).

Fig. 2.

Conceptual framework: The relationship between increased density of academic networks and improvements in academic productivity is mediated by increased self-efficacy and increased outcome expectations. Network characteristics include density, degree centrailty and homophily) and academic productivity includes the Hirsch-index and number of peer-reviewed articles.

Textbox 1. Primary outcomes and hypotheses.

Cross sectional analysis

Primary hypotheses

Hypothesis 1a

There is a positive association between an individual's degree centrality (degree) and her/his academic productivity.

Hypothesis 1b

There is a positive association between an individual's density in the network and her/his academic productivity.

Hypothesis 1c

There is an association between lower homophily and higher academic productivity.

Secondary hypotheses

Hypothesis 2a

There is a mediation effect of self-efficacy and outcome expectations between the association of an individual's degree centrality in the network and her/his academic productivity.

Hypothesis 2b

There is a mediation effect of self-efficacy and outcome expectations between the association of an individual's density in the network and her/his academic productivity.

Longitudinal analysis

Primary hypotheses

Hypothesis 3a

There will be an increase of 10% in network density from baseline to endline.

Hypothesis 3b

There will be a decrease of 10% in homophily from baseline to endline.

Hypothesis 3c

Increase in degree centrality and academic productivity from baseline to endline.

Secondary hypotheses

Hypothesis 4a

There will be an increase in the number of researchers engaging with service users from baseline to endline.

Hypothesis 4b

There will be an increase in average self-efficacy and outcome expectations of health researchers.

Hypothesis 4c

There will be an increase in an average academic productivity (number of publications) from baseline to endline.

Alt-text: Textbox 1

2. Methods and analysis

2.1. Study setting

Nepal is a landlocked country with a population of 29 million. It has the third lowest human development ranking in South Asia, and it currently ranks 110th out of 162 countries in the Gender Inequality Index (United Nations Development Project, 2020). It is in the bottom third of the world's countries in the Gender Development Index (United Nations Development Program, 2020). The estimated gross national per capita income for women is $2910 compared to men's $4108 (United Nations Development Program, 2020). The literacy rate for women is 59.7%, while that of men is 78.6%. Only 18.4% of people in the roles of legislators, senior officials and mangers are women compared to men's 81.6% (United Nations Development Project, 2020). International collaborations such as PRIME (Program for improving mental health care) (Jordans et al., 2019) and EMERALD (Emerging mental health systems in low- and middle-income countries) (Semrau et al., 2015) have identified capacity building as an integral step forward to improve research capacity in Nepal, with a need to focus on implementation science, leadership and writing skills (Evans-Lacko et al., 2019). Building on findings from these prior initiatives, our study will focus on need-based, gender equitable capacity building that goes beyond knowledge enhancement. This study will have a strong focus on improving academic productivity of Nepali researchers through collaborations of Nepali with Nepali and Nepali with non-Nepali researchers, with a focus on women and ethnic minority researchers. We will evaluate the study by assessing network characteristics influencing academic productivity.

2.2. Study design

2.2.1. Social network analysis component

We plan to use SNA to map the network of mental health researchers in Nepal through descriptive and evaluative analysis. In descriptive analysis, we will map the researchers’ networking (among the individuals within the network whom have you reached out to in the last 12 months for professional reasons?), collaboration (who among the network members have you collaborated with for grants/papers/conference presentations?), and mentorship (who among the network members have you reached out to for a career advice?). We will assess the network density, degree centrality, and heterogeneity/homophily among the network members in Year 1 and Year 5. For evaluative purposes, we plan to assess if a need-based, gender sensitive capacity building activities can change network attributes of density, degree centrality and homogeneity, and whether these changes lead to improved academic productivity.

2.3. Recruitment and study participants

2.3.1. Social network analysis

To generate the sample included in the SNA, we will employ a nomination method identifying mental health researchers in Nepal. A group of mental health research experts in Nepal will help us identify ten individuals who are well-connected and representative of the actual mental health research group in Nepal. Each of these ten researchers will be asked to nominate five individuals based on the inclusion criteria. We will repeat this process in three cycles to reach wider groups of mental health researchers in Nepal. Because we are employing a snowball sampling method, findings from this network might not be representative of all the researchers in Nepal. However, our study is the first of its kind in Nepal where we plan to assess both individual (self-efficacy and outcome expectations) and network characteristics (density, homophily, and degree centrality). Furthermore, we plan to assess these individual and network-level characteristics among the same group of researchers for the next five years. These will draw meaningful conclusions on how the characteristics change over time, and how they influence academic productivity. In addition to a snowball sampling method, we will also include all the first-authors from mental health peer-reviewed publications in Nepal between 2015 and 2020. With the combination of snowball sampling and inclusion of first-authors from peer-review publications, we anticipate a group of approximately150 mental health researchers in Nepal.

The network will range from early-career to senior Nepali and international researchers currently conducting research in Nepal. The early-career researchers will also include students who are interested in pursuing her/his career in mental health research. Senior researchers can provide mentorship to young researchers within the network. Non-Nepali (international) researchers contribute significantly to the mental health research in Nepal at present with a number of research studies funded by academic institutions and non-profit organizations. Therefore, it is important that we include these international researchers in our network, to promote collaboration with and mentorship of early-career researchers.

2.3.2. Inclusion criteria for nomination to the network

• a)Nepali or non-Nepali researchers;
• b)researchers may be affiliated with a non-profit, governmental, or academic institutions conducting mental or behavioral health research in Nepal;
• c)Nepali students pursuing an undergraduate or a graduate degree in public health, medicine, psychology, sociology, or related discipline and interested in pursuing mental or behavioral health research; and
• d)Out of 5 persons, at least 3 early-career and 3 participants who identify as female. We define early career researchers as anyone who has completed their degree in the last 10 years.

2.4. Data collection

2.4.1. SNA data

We have designed a closed SNA tool to assess networking (who interacts with whom), collaboration (who is working together), and mentorship (who is helping whom in professional growth). In SNA, the data collection tool can be open-ended, where respondents are asked to enter the name of the individuals for a particular prompt such as “Whom have you reached out to in the last 5 years for professional suggestions?” We employ open-ended questions usually when the network boundary is not known (Borgatti et al., 2013; Petrescu-Prahova, 2015). This will be a closed network of mental health researchers who will be followed up for the next five years. The participants will also be asked metrics of professional development (grants submitted, degrees earned, job opportunities). Participant dropout is a major methodological issue in Social Network Analysis (Stadtfeld et al., 2020). To deal with participant dropouts, we will adopt the following – a) we plan to communicate the research duration very clearly with the participants, so we only enroll those that are highly motivated to participate in the study, b) our surveys will document multiple platforms to connect to participants (multiple email addresses, LinkedIn), so we have multiple ways to reach out to them even if they move organizations or countries, and, c) we will still include the dropped out participants in the network to obtain uni-directional information about them, particularly how frequently the existing network members mention the missing individuals in their networks.

Each of the 150 health researchers will be provided a list of names of other members within the network and give a 5-year time frame which is known to reduce recall bias and increase accuracy in SNA data collection (Borgatti et al., 2013). Table 1 provides details of the network attributes that will be assessed in the study.

Table 1.

Network attributes and SCCT constructs.

Attributes	Online survey
Network characteristics
Average Density – The average number of connections participants have divided by the total possible connections.	X
Homophily - Tendency of people to form positive ties with people similar to themselves on socially significant attributes e.g., race, gender, age, education, social class.	X
Average Degree centrality - Structural measure of a node's position in the network. Average number of ties of a node.	X
Individual characteristics
Degree centrality – The number of ties of each node.	X
Homophily – Individual tendency to form positive ties with similar people in terms of race, gender, age, education, social class	X

2.4.2. Academic productivity

We will assess academic productivity outcomes in terms of the peer-reviewed articles, grant submissions, and research career development. Table 2 provides a detailed overview of the academic productivity outcomes and data collection methods. We have chosen academic productivity as an indicator of professional development because this is a metric used by funding agencies and research institutions worldwide (Sarli & Carpenter, 2014). Peer-reviewed articles use authorship as an important indicator of academic productivity and disparity in publications (Filardo et al., 2016; Pinho-Gomes et al., 2020; Shah et al., 2021; Smith et al., 2014). Though number of peer-reviewed articles alone might not give a holistic picture of a researcher's productivity, it is still a common indicator of productivity. In our study, we plan to assess researchers' publications along with grant submissions, degrees obtained, and presentations delivered over the last five years. We will primarily ask respondents in the network to provide information on the academic productivity outcomes and will conduct a review of published articles, abstracts, presentations, and research projects to triangulate the information.

Table 2.

Academic productivity outcomes.

Academic Productivity Outcomes	Online survey	Secondary data
Primary Outcome
Total peer-reviewed articles publications	X	X
Secondary outcomes
h-index	X	X
i-index	X	X
Number of abstracts accepted for presentation	X
Participation in graduate/post-graduate training program (Masters, Ph.D., post-doctoral fellowship)	X
Current collaborative research projects	X	X
Number of grant application submissions	X

2.4.3. Self-efficacy and outcome expectations

We will assess if the association between network changes and academic productivity is mediated by researchers’ self-efficacy and outcome expectations. Because most of the research self-efficacy and outcome expectations tools so far have been used among researchers in high-resource settings, we will design our own tool that is culturally relevant by adopting the tool development techniques used in prior studies (Anderson et al., 2016). It consists of four steps - 1) literature review and item generation 2) focus groups, 3) expert review, and 4) pilot testing. We will design a self-efficacy and outcome expectation tool based on the recommendation by Lent and Brown on measuring self-efficacy (Lent & Brown, 2006) including items that rate their perceptions about their ability at present and their ability to perform activity over a period of time (Bandura, 1997). For outcome expectations, we will assess positive as well as negative outcomes, since contextual knowledge is integral to understand outcome expectations in our study (Lent & Brown, 2006). We anticipate negative outcomes being important contextually in Nepal due to discrimination and social disapproval, especially for certain ethnicity and gender based on their career choices.

2.4.3.1. Self-efficacy measures–

We will design our self-efficacy tools that will be relevant to the early-career researchers in Nepal. We will refer to the items from Self-efficacy in Research Measure (SERM) (Phillips & Russell, 1994), Career-Decision Self-Efficacy Measure (Betz et al., 1996) and self-efficacy tools used in prior study to assess academic publications (Anderson et al., 2016), including one among ethnic minorities (Byars-Winston et al., 2016). We have also reviewed research self-efficacy scale, research skills self-efficacy scales, and research training environment scales to identify relevant items to be incorporated in the study (Burke & Prieto, 2019; Lachance et al., 2020; Livinti et al., 2021).

2.4.3.2. Outcome expectation measures–

We will design our outcome expectations items based on the Research Outcome Expectation Questionnaire (ROEQ), a tool to assess academic persistence for historically underrepresented groups in science (Byars-Winston et al., 2016), and outcome expectation measures in scientific communication (Anderson et al., 2016). We have also added relevant items from Research Outcome Expectation Scale (Bieschke, 2000). Table 3 provides detailed description of self-efficacy and outcome expectations measures.

Table 3.

Self-efficacy and outcome expectations measures.

Outcomes		Online survey
Self-efficacy	people's judgments of their capabilities to organize and execute courses of action required to attain designated types of performances” (Bandura, 1986)	28-item scale currently in development (cognitive interviewing)
Outcome expectations	beliefs of the results or consequences of performing certain activities (Lent et al., 1994)	10-item scale currently in development (cognitive interviewing)

2.5. Data analysis

We will assess the networking, collaboration, and mentorship using SNA density, centrality, and homophily (Table 1). The same questionnaire will be administered to network members at follow-up to assess the improvement in networking, collaboration, and mentorship opportunities during the project duration. We will also assess the researchers’ academic productivity (Table 2), self-efficacy, outcome expectations, and knowledge (Table 3). We will use UCINET software for analysis (Borgatti et al., 2002).

The descriptive analysis includes network analysis (density, homophily, average degree centrality) and individual-level analysis (degree centrality, homophily, individual density). We will assess academic productivity to assess if changes in network attributes leads to improvement in academic productivity. We will determine if changes in social network attributes such as density, homophily, and degree centrality changes the academic productivity (number of peer-reviewed journals published, h-index). Finally, we will see if the association between network attributes and academic productivity is mediated by self-efficacy and outcome expectations.

3. Ethics and dissemination

Ethical approval has been obtained from the Nepal Health Research Council (#351/2020P) and the George Washington University Institutional Review Board (IRB # NCR202522).

Authors’ contribution

AP, DG, and BAK conceptualized the capacity building project. AP designed the social network component of the study. AP, DG, and BAK drafted the manuscript.

Funding statement

This study has been funded by the National Institute of Mental Health (R01MH120649). BAK is the primary recipient of the award. BAK and DG are supported by NIMH R01MH120649 and Medical Research Council MR/R023697/1. AP is supported by the NIMH T32 on Social Determinants of HIV (T32MH128395-01). The funding sources were not involved in the design of the study.

Declaration of competing interest

The authors do not have any competing interests.