Shared perceived causes of suicide among young men and violence against young women offer potential for co-designed solutions: intervention soft-modelling with fuzzy cognitive mapping

ABSTRACT

Violence against young women (VAW) and suicide among young men are serious concerns in Botswana and elsewhere. We examined the overlap in locally perceived causes of these two forms of violence in Botswana using the results from separate studies that used fuzzy cognitive mapping (FCM) to explore perceived causes of the two outcomes. FCM depicts perceived causes of an outcome and their links to the outcome and each other, with weights denoting the perceived strength of each link. The two studies engaged groups of young women, young men, older women, and older men in rural communities. We grouped related concepts into broader categories, then combined category maps for each outcome into a single map including both forms of violence. Based on social network analysis, we calculated the out-degree centrality of each category indicating its influence within the network. Intervention soft-modelling explored the effects of removing individual categories on suicide and VAW. Of 24 causal categories in the combined map, six were shared between both outcomes, 10 were for suicide only, and seven were for VAW only. The six shared categories accounted for 60% of cumulative influence of all categories in the combined map. The three most influential shared categories were financial difficulties, conflict in relationships, and parenting and family issues. Based on local perceptions, avoiding conflict in relationships could reduce suicide by 4.8% and VAW by 18.5%. Eliminating parenting and family issues could reduce suicide by 3% and VAW by 5.4%. Preventing financial difficulties could reduce suicide by 9.3% and VAW by 2.9%. The findings support the idea that some interventions might reduce both personal and interpersonal violence among youth. Analysis of stakeholder perceived causes and soft-modelling of potential interventions could inform community-led co-design of strategies to reduce youth suicide and violence against young women in Botswana.

KEYWORDS:

• Soft models
• violence against women
• suicide of young men
• Interpersonal violence
• fuzzy cognitive mapping

Background

Suicide among young men and violence against young women are both serious public health problems, especially in low- and middle-income countries (LMICs). Botswana, now an upper-middle-income country but with high-income inequality, is no exception. In 2019, there were estimated 21.6/100,000 (pcm) deaths by suicide among young men aged 20–24 years in Botswana, and 14.2 pcm among those aged 15–19 years (World Health Organization, 2022). Studies in Botswana have reported high rates of experience of physical intimate partner violence in the last one year among adult women: 19% in 2002 (Andersson et al., 2007) and 13% in 2011 (Machisa & van Dorp, 2012). In 2017, 21% of marginalized young Batswana women (15–29 years) disclosed physical IPV in the last year (Cockcroft et al., 2018).

An extensive literature reports factors related to violence against young women and a separate literature examines factors related to suicide and suicide ideation among young men and women. Many studies are from high-income countries, but several recent reviews of the literature about factors related to these two violent outcomes have focused on LMICs or sub-Saharan Africa (Quarshie et al., 2020; Shamu et al., 2011; Vyas & Watts, 2009). Cross-sectional studies of the two forms of violence have described some factors in common, but we have not found published studies that examine factors related to both outcomes together. Interventional studies focus on one or the other outcome separately. There are many reported interventions for prevention of violence against young women in LMICs and sub-Saharan Africa, with combined individual, community and structural approaches reported to be more effective (Keith et al., 2023; Yount et al., 2017). Interventions for suicide prevention in LMICs often identify youth at risk or those with depression or other mental health issues, offering support to these subgroups (Davaasambuu et al., 2019).

In the context of participatory research and stakeholder engagement (Andersson, 2011, 2018), our question is whether, based on stakeholder perceptions, suicide among young men and violence against young women have shared causes. If they do share perceived causes, community mobilisation interventions to tackle these shared causes could reduce both outcomes. Two recent studies used fuzzy cognitive mapping (Andersson & Silver, 2019) to document separately locally perceived causes of suicide among young men (Sarmiento et al., 2023a) and violence against young women in rural Botswana (Sarmiento et al., 2023b). The maps produced in these two separate exercises suggested considerable overlap between the local community-perceived causes for the two outcomes. This paper describes these shared perceived causes, their influence within the networks of the causes of the violent outcomes, and, based on soft-modelling, the potential impact on both outcomes if these shared causes could be addressed. This analysis is not a predictive exercise but will provide the information input for local stakeholder dialogues about how some interventions might tackle both forms of violence in Botswana.

Methods

Fuzzy cognitive mapping

Fuzzy cognitive mapping (FCM) generates diagrams of perceived causes linked, directly and indirectly, with an outcome (Kosko, 1986). Arrows link each cause with its consequence(s), often with weights to indicate the relative strength of influence of each link (Andersson & Silver, 2019). The maps are soft models of how stakeholders understand causal relationships that can be used across multiple cultural backgrounds (Sarmiento et al., 2020). In two separate exercises, we used FCM to depict local perceptions about the causes of violence against young women and suicide among young men in rural Botswana. Two other papers describe details of the methods and findings of these two studies (Sarmiento et al., 2023a); Sarmiento et al., 2023b). In each study, four stakeholder groups (young women, young men, older women, older men) from several rural settings created fuzzy cognitive maps of their knowledge of causes of the outcome and weighted the influence of each link from 5 (strongest) to 1 (weakest). A local facilitator supported each session, and a local reporter took notes of the accompanying discussion. Following digitisation of the maps, fuzzy transitive closure analysis (Niesink et al., 2013) calculated the influence of each cause on the outcome, taking all other perceived causes into account. An inductive analysis of the causal factors combined them into broader categories. Summary category maps showed the findings across all stakeholders and for individual stakeholder groups.

Combining the maps for the two outcomes

Based on the overall category maps for perceived causes of suicide among young men and violence against young women, we created a combined map that included both outcomes (Dickerson & Kosko, 1994; Sarmiento et al., 2022). The net influence of each category level relationship in the combined map was the average of the net influence of this relationship in the separate outcome-specific maps. We reviewed the categories from the separate maps for the two outcomes and used a common name for those categories that included the same, or very similar, factors in the map of both outcomes. The combined map had causal categories of three types: those that appeared only on the map for suicide among young men, those that appeared only on the map for violence against young women, and those that appeared on the maps of both outcomes.

Social network analysis

Fuzzy cognitive maps represent networks, with causal categories and outcomes as nodes and the links between them as edges within the network. Social network analysis allows estimation of the role of nodes within a network, including through centrality measures (Knoke & Yang, 2019; Papageorgiou & Kontogianni, 2012). Outdegree centrality is a measure of the number of outgoing edges from a node in a network. A high outdegree centrality indicates a node that is influential as a cause within the network. We calculated the outdegree centrality of each perceived causal category within the combined map as the sum of the absolute weight of all its outgoing edges (arrows). We normalised this sum by dividing it by the maximum outdegree centrality, thus putting all values in a range between 1 for the most influential category and 0 for a category with no influence. We calculated the sum of all category outdegree centrality values in the combined map and the proportion of this cumulative influence contributed by the shared categories.

Intervention soft-modelling using what-if scenarios

We used a technique described by Kosko et al (Dickerson & Kosko, 1994; Kosko, 1993) to calculate the potential effect of changing one category in the map on all other categories in the map. The initial activation level of every node (category) in the map is between 0 and 1, where 0 means the node is not present at all and 1 means it is fully operational. For the first iteration, the level of activation of each node in the network is then a function of the sum of the level of activation multiplied by the weight of the link from each other node linked to it, also with a value between 0 and 1 (Özesmi & Özesmi, 2004). The modelling repeats the calculations until the further change in activation levels is zero, when the model has reached a stable state. We calculated the effect on the model of keeping the level of activation of certain nodes to zero throughout the iteration; in other words, the scenario if the potential causal category was removed. For each scenario, we calculated the reduction in the activation of nodes compared with the activation in the full model.

Results

Figure 1 shows the combined category map of suicide among young men and violence against young women. The two outcomes shared six perceived causal categories: financial difficulties, conflict in relationships, parenting and family issues, men being stressed and insecure, alcohol and drug use, and a culture of violence. A further 10 perceived causal categories appeared only on the map of suicide among young men, and seven appeared only on the map of violence against young women. For clarity, Figure 1 only shows the links of shared perceived causal categories. Based on outdegree centrality, three shared categories had the highest influence within the network: financial difficulties (outdegree centrality 1.0), conflict in relationships (0.9) and parenting and family issues (0.7). Overall, the six shared categories accounted for 60% of all the influences in the combined map. Financial difficulties accounted for 18% of all the influence in the combined map, conflict in relationships for 17%, and parenting and family issues for 12%. Appendix 1 shows all relationships in the combined map for the two violence outcomes. Appendix 2 shows the individual factors within each of the perceived causal categories. Appendix 3 shows the final activation levels in the full model for the two outcomes and the various causal categories in the combined map.

Figure 1. Combined fuzzy cognitive map of categories affecting suicide of young men and violence against young women.

Legend: Blue boxes correspond to categories which appeared only in the map of causes of suicide of men in Botswana. Yellow boxes are categories that appeared only in the map of violence against young women. The grey boxes in the middle are categories both maps shared. Black arrows correspond to relationships coming from shared categories to the two main outcomes, and grey arrows come and go from the rest.

Table 1 shows the decrease from the full model activation value of the two outcomes and other categories on the combined map, removing each of the shared causal categories in turn. The biggest potential change was from removing conflict in relationships, potentially associated with 4.8% decrease in suicide among young men and 18.5% decrease in violence against young women. Removing parenting and family issues was potentially associated with 3.0% decrease in suicide of young men and 5.4% decrease in violence against young women. Removing financial difficulties was associated with the largest potential decrease in suicide among young men (9.3%), as well as 2.9% decrease in violence against young women. It also had important potential effects on intermediate causal categories, with 8.7% reduction in men’s involvement in crime, 4.7% reduction in transactional and intergenerational sex, 10.8% reduction in alcohol and drug use, and 9.6% reduction in men being stressed and insecure.

Table 1. Reduction in the activation level of nodes in the maps with six theoretical scenarios for removing the causal categories appearing on maps of both outcomes

	% reduction of baseline activation of other nodes if:
	No family Issues	No alcohol or drug use	No conflict in relationships	No men’s insecurity	No culture of violence	No financial difficulties
Nodes only present on the map about causes of suicide among young men
Suicide of young men	3.0	1.2	4.8	3.3	0.2	9.3
Unemployment	0.0	0.0	0.0	0.0	0.0	0.0
Badly treated by society	0.8	0.0	0.7	0.0	0.0	0.8
Being bewitched	0.0	0.0	0.0	0.0	0.0	0.0
Effective correctional services	0.0	0.0	0.0	0.0	0.0	0.0
Health problems	2.2	0.7	1.7	0.5	0.0	2.0
Lack of self-esteem and self-worth	1.8	0.4	0.5	0.1	0.0	2.4
Laws protecting children’s rights	0.5	0.0	0.0	0.0	0.0	0.0
Men do not discuss their problems or seek help	1.0	0.0	0.0	0.0	0.8	0.0
Men’s involvement in crime and consequences	3.6	1.5	1.9	1.3	0.8	8.7
Negative influence of friends and peers	0.0	0.1	0.5	0.1	0.0	2.0
Religion and spirituality	0.0	0.0	0.0	0.0	0.0	0.0
Nodes only present on the map about causes of violence against young women
Violence against young women	5.4	3.1	18.5	2.5	0.0	2.9
Male involvement	0.0	0.0	0.0	0.0	0.0	0.0
Transactional/intergenerational sex and related factors	2.7	0.1	3.6	0.2	0.0	4.1
Young women putting themselves at risk	0.0	0.0	0.0	0.0	0.0	0.0
Ethical behaviour/good choices	0.0	0.0	0.0	0.0	0.0	0.0
Women being disrespectful/not cooperative	0.2	0.1	6.7	0.2	0.0	0.3
Laws and customary laws	0.0	0.0	0.0	0.0	0.0	0.0
Teaching youth about violence- providing role models	0.0	0.0	0.0	0.0	0.0	0.0
Nodes present on both maps
Parenting or family issues	-	0.3	0.7	1.3	0.0	0
Alcohol and drug use	2.3	-	4.4	1.9	0.6	10.8
Conflict in relationships	3.2	0.8	-	2.9	0.1	3.9
Men are insecure and stressed	3.6	1.3	4.9	-	0.5	9.6
Culture of violence	0.1	0.4	0.1	0.0	-	1.6
Financial difficulties	2.7	2.7	5.4	2.1	0.0	-

Cells in grey are the top five reductions for each scenario.

Discussion

Our analysis revealed substantial overlap between perceived causal categories for suicide among young men and violence against young women. This supports the idea that these are not separate problems that need to be investigated and tackled separately. From the standpoint of community perceptions and engagement in prevention, both forms of violence are partly due to shared causes, including parenting and family problems, relationship conflicts, and economic difficulties. Many participatory research authorities note that communities respond best, or only respond, to mobilisation efforts when these are consistent with their own beliefs (Andersson, 2018; Israel et al., 1998; Macaulay et al., 1999). Based as it is on community perceptions of causality, the results from this analysis combining fuzzy cognitive mapping for the two violence outcomes and soft-modelling potential interventions might focus community energies on actions that could lead to reduction in both suicide in young men and violence against young women.

The issues influencing both forms of violence highlighted in our analysis have been reported in previous separate-outcome studies. Financial difficulties were a prominent perceived influence on both outcomes. Suicide is associated with poverty in LMICs (Iemmi et al., 2016), male perpetration of partner violence is related to poverty and food insecurity (Awungafac et al., 2021; Fulu et al., 2013), and women in poverty have higher risks of intimate partner violence (Jewkes, 2002). Our soft modelling suggested a potentially important effect of reducing conflict in relationships for both outcomes, especially violence against young women. Other authors have reported that relationship factors (including having multiple partners) are associated with male perpetration of partner violence (Fulu et al., 2013), and women’s experience of partner violence (Abramsky et al., 2011). A review reported associations between difficulties in intimate partner relationships and suicidal ideation (Kazan et al., 2016). The potential role of parenting and family issues surfaced in our analysis aligns with reported associations between childhood abuse, trauma and witnessing violence in the home and experience and perpetration of partner violence (Abramsky et al., 2011; Fulu et al., 2013; Kimber et al., 2018) and reported associations between family relationships and suicidal phenomena in adolescents (Evans et al., 2004).

Our intervention soft model suggested only minor impacts on the two outcomes if there was no alcohol or drug use. Other authors have reported important associations of alcohol and drug use with experience and perpetration of partner violence, including in Botswana (Abramsky et al., 2011; Barchi et al., 2018; Laslett et al., 2021), and with suicidal behaviour (Amiri & Behnezhad, 2020; Evans et al., 2004).

Limitations and strengths

The intervention soft-modelling is not a predictive exercise using empirical effect measures. Based explicitly on locally perceived causes of the two forms of violence, we present the stakeholder perception of potential reductions with different interventions as an input into discussions with those stakeholders about prioritising prevention strategies. There is support for these strategies in the international literature considering the two outcomes separately. Discussing them as possibly influencing both outcomes, based on community perceptions, is likely to generate local community buy-in.

Our analysis is based on a few maps; findings from the intervention soft-modelling might be more stable with more maps. Different individuals mapped the two outcomes, although they were the same types of people (young women, young men, older women, and older men) in the same types of communities (rural communities close to the capital). If the same people created the two maps, their experience creating the first map might have influenced the second.

The research team grouped perceived causal factors into broader categories and our prior beliefs and biases may have influenced categorisation. We mitigated this possibility by involving the fieldworkers who facilitated the mapping sessions in the categorisation and consulting the field notes from the sessions. We assessed potential reductions in outcomes after removing whole categories of causes. Each category includes many factors and in practice it would clearly not be possible to remove a category totally.

The findings reflect the situation in rural communities in southern Botswana and may not be generalizable to community interventions outside this context. But a strength of the study is that it is based on local knowledge and therefore locally relevant and with a good potential for guiding local prevention discussions. While the findings may be specific to a particular context, the procedural and analytic approach could have much wider applicability.

Conclusions

This analysis supports the idea that at least some of the causes of youth suicide and violence against young women are shared in the context of rural Botswana. This could inform discussions and community mobilisation for joint interventions to address both forms of violence.

Ethics review

This study is part of a Grand Challenges Canada project (Grant number R-ST-POC-1909–28463), which received ethical approval from the Botswana Ministry of Health under the Health Research and Development Division IRB (Reference HPDME 13/18/1).

Acknowledgments

We thank the men and women who contributed their time and knowledge in the FCM sessions. The members of Participatory Research at McGill (PRAM) kindly participated in earlier discussions of our results and intervention soft-modelling.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data supporting this study’s findings are available with the publication.

Additional information

Funding

Grand Challenges Canada funded this work under Grant number R-ST-POC-1909-28463

Notes on contributors

Anne Cockcroft

Anne Cockcroft is a professor of family medicine in CIET-Participatory Research at McGill (PRAM) with a background in respiratory and occupational medicine. Over the last 25 years, she has undertaken large scale community-based participatory research projects in some 20 countries. She works with vulnerable populations to document their access to and experience of health and other services, and with service providers and policy makers to use evidence to develop equitable and effective services. In the last decade, her work has focussed on co-designing interventions, implementing them, and measuring the impact. Her current work includes participatory research to improve adolescent sexual and reproductive health and community responses to the impacts of COVID-19 pandemic in Bauchi State, Nigeria, and a study of community-based interventions to reduce youth personal and interpersonal violence in Botswana.

Ivan Sarmiento

Iván Sarmiento is an independent researcher at CIET, a member of the Groups of Studies in Traditional Health Systems, and the program administrator of Participatory Research @ McGill (PRAM). He has over two decades of experience collaborating with local and Indigenous groups in Colombia. His main interest is in promoting intercultural dialogue between Indigenous traditional medicine and Western medicine, particularly for primary health care. He has contributed to developing procedures for participatory modelling of health issues using fuzzy cognitive mapping, applying these methods in over 20 projects across eight countries.

Neil Andersson

Neil Andersson is a Professor of Family Medicine, director of the amalgamated CIET and Participatory Research at McGill (PRAM) and co-director of the McGill Institute of Human Development and Well-being. His main focus is on method development for large scale participatory approaches that address different health issues. He is particularly concerned about reproducible and culturally safe techniques to build stakeholder voices into systematic reviews, research conceptualization and co-design, intervention development, implementation and analysis. Dr. Andersson’s current interest is in community-led randomized trials of older adult participation in dementia prevention.

Appendix 1.

Adjacency matrix of the category map of causes of suicide among young men and violence against young women

Rows and arrows correspond to the nodes on the map. Each cell indicates the possible relationship between two nodes, from the row to the column. The value in the cell is the weight of the causal relationship. A weight of 0 indicates that participants did not report the relationship. Weights closer to one indicate stronger causal relationships. Negative and positive signs correspond to negative and positive relationships between cause and effect, respectively. Each category has a corresponding code with S for the categories that were present in the map of causes of suicide among men and V for the categories from the maps of violence against women, with a consecutive number according to the order of categories in the maps.

Table

Name of the category\Category code		S01	S02	S04	S05	S07	S08	S10	S11	S13	S14	S16	S18	V01	V03	V04	V06	V07	V08	V11	V13	V02 S09	V05 S03	V09 S17	V10 S15	V12 S12	V14 S06
Suicide of men in Botswana	S01	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Unemployment	S02	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Badly treated by society	S04	0.2	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.1	0.2	0.0	0.2
Being bewitched	S05	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0
Effective correctional services	S07	−0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	−0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	−0.1	0.0	0.0	−0.1	0.0	−0.1
Health problems	S08	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0
Lack of self-esteem and self-worth	S10	0.3	0.0	0.0	0.0	0.0	0.0	−0.1	0.0	0.0	0.1	−0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.4	0.0	0.0
Laws protecting children’s rights	S11	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.1	0.0	0.0
Men do not discuss their problems or seek help	S13	0.3	0.0	0.1	0.0	0.0	0.1	−0.1	0.0	0.0	0.1	−0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.1	0.2	0.2	0.0	0.0
Men’s involvement in crime and consequences	S14	0.3	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.1	0.2	0.0	0.1
Negative influence of friends and peers	S16	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0
Religion and spirituality	S18	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Violence against young women and girls	V01	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Male involvement	V03	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	−0.1	0.0	−0.1	0.0	0.0	−0.1	0.0	0.0	0.0	0.0	−0.1	0.0	0.0	0.0
Transactional/intergenerational sex and related factors	V04	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.2	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Young women putting themselves at risk	V06	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Ethical behaviour/good choices	V07	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	−0.1	0.0	−0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	−0.1	0.0	0.0	0.0
Women being disrespectful/not cooperative	V08	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.4	0.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.0	0.2	0.1	0.0	0.0
Laws and customary laws	V11	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	−0.2	0.0	−0.1	−0.1	0.0	0.0	0.0	0.0	−0.1	−0.1	0.0	−0.1	0.0	0.0
Teaching youth about violence- providing role models	V13	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	−0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	−0.1	0.0	0.0	0.0	0.0	0.0
Parenting or family issues	V02S09	0.5	0.0	0.0	0.0	0.0	0.1	0.1	0.0	0.0	0.2	0.0	0.0	0.4	0.0	0.1	0.0	0.0	0.0	0.0	0.0	0.1	0.1	0.2	0.4	0.0	0.2
Alcohol and drug use	V05S03	0.2	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0	0.2	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.1	0.0	0.2
Conflict in relationship	V09S17	0.5	0.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.1	0.0	0.0	1.0	0.0	0.1	0.0	0.0	0.2	0.0	0.0	0.0	0.2	0.5	0.4	0.0	0.3
Men are insecure and stressed	V10S15	0.3	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.1	0.1	0.0	0.1
Culture of violence	V12S12	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0
Financial difficulties WOMEN, mostly – OF THE VICTIM	V14S06	0.8	0.0	0.0	0.0	0.0	0.1	0.1	0.0	0.0	0.3	0.1	0.0	0.1	0.0	0.1	0.0	0.0	0.0	0.0	0.0	0.0	0.4	0.1	0.7	0.0	1.0

Appendix 2.

Individual factors within each of the combined causal categories

Table

Category	Factors from the maps about violence against women	Factors from the maps about suicide of young men
Suicide of men in Botswana		Suicide of men in Botswana.
Violence against young women and girls	Violence against young women (YW) and girls
Financial difficulties	Risk Young women live in poverty. Youth lack employment. Young women struggle to find ways to be financially independent. Financially dependent on boyfriend.	Risk Debts and financial problems. Fear of having to pay for child support. Fear of pregnancy-responsible payments. Financial lessons from parents. Financial pressure from family. Financial pressure from girlfriend. Good financial sense. Man has multiple children. Poor living standards. Reckless spending. Spending too much on friends. Spending too much on the partner. Loss of job. Unemployment.
	Protective Government economic empowerment programmes.	Protective Teach good financial sense. Having work.
Conflict in relationship	Risk Romantic/sexual relations at a young age. Young women lie about their age. Children mature at young age. Stealing from your partner. Partner has a child from a previous relationship. Men not accepting children from women’s previous relationships. Young women having multiple children, especially if children are from different partners. Low self-esteem. Men don’t accept being dumped. Not giving your partner attention while on social media. Conflict about phones (if she uses the phone or if she complains because he is using it). Taking and checking man’s phone. Hearing people talk about your partner and believing them. Men believe rumours about girls. Conflict with boyfriend caused by his friends. Not telling your partner you have an STI, and he finds out. Lack of trust in relationships. Young women lie a lot. Young women losing interest in relationship. Not respecting each other in a relationship.	Risk Cheating by partner. Cohabitation with partner. Confusion about sexual orientation. Denial of sex by partner. Girlfriend reveals his secrets. Men’s infertility. Multiple relationships (both the woman and the man). Pressure to support other children of the partner. Previous partners negatively influence new relationship. Rejection by woman/partner. Relationships at a young age. Sexual dysfunction.
	Protective Avoid keeping secrets from partners. Girls should not share the same man. Young women having fewer sexual partners. Avoid dating at a young age because women are more vulnerable and lack support. Giving the boyfriend more time and ignoring the phone. Everyone should keep their own phones (not snooping phones). Young women spending more time in church. Men should not believe their friend’s stories about their partners. Not letting other parties get between those in relationships. Partners trust each other in a relationship. Partners are faithful to each other. End relationships if there are problems, instead of cheating. Leave a relationship with an abusive partner. Good communication between couples. Couples go for counselling. Young women asking for help from government workers (counselling).	Protective Stop giving priority to girlfriends. Teaching young women not to be promiscuous.
	Poor communication. Fighting over money with your partner. Living with a man who does not provide enough money. Partner does not give you money. Young women having multiple partners. Young women not replying to men. Dating your partner’s friends. Youth (men and women) are promiscuous. Paying attention to other men – girl. Cheating on your partner. Male partner cheating and women complaining about that. Cheating/dumping because men are broke. Leaving partners for immigrant workers.
Parenting or family issues	Risk Leave school early so girls are not at school. Young women leave home. Leaving home young to live with a partner. Orphaned girls. Lack of support from parents. Lack of good parenting. Parents encourage transactional/transgenerational sex (usually the mother). Parents allow abuse on girls (because the man provides income). Parents do not pay attention to their children’s relationships – encouraging the girl to ask for more money or ‘catch more fish’. Vulnerable Girls. Uncles abuse YW girls. Mothers contrasting discipline (not disciplining girls). Conflict between siblings (male sibling is violent against female sibling). Children do not listen to their parents.	Risk Bad treatment by family. Being single parent. Death of parents. Deserting family. Discrimination by family. Domestic abuse of young men. Family does not provide support. Fear to communicate with father. Loss of support from parents. Misunderstanding in the family. Deserting family. Not listening to elders’ advice. Raised by single mothers. Unable to survive because they were spoilt.
	Protective Parents interfere in children’s relationships. Parents talk to children about relationships. Parents teach YW about contraceptives. Family elders refer to social workers and counsellors when girls suffer violence. Elders are strict with the youth. Mothers cooperating when disciplining girls. Parents do not cover up for children. The strict upbringing of girls.	Protective Elders advise young men on relationships. Engaged parents. Fathers form a close relationship with sons. Good parental guidance. Cooperative parenting. Family support.
Men are insecure and stressed	Risk Women should not go to the bar with their partners (can cause jealousy). Women refusing to come to boyfriend’s house (incites jealousy and suspicion). Male partners being very insecure. Stressed men are more aggressive.	Risk Anxiety. Depression. Lack of emotional control. Stress. Stress from losing court cases.
	Protective Counselling for stressed men.	Protective Emotional control.
Alcohol and drug use	Risk Alcohol abuse (both men and women). Drug abuse (both men and women). Drug and alcohol abuse during youth. Men drinking too much. Young women drinking too much alcohol. Women are disrespectful when drunk.	Risk Alcohol and Drug abuse.
	Protective Young women avoiding alcohol abuse. Youth do not use alcohol and drugs.	Protective No use of alcohol or drugs abuse.
Culture of violence	Risk Men are stronger than women. They learn violence at school (culture of violence and peer pressure). Bad influence/example from other men. Disrespect of women by men. Being a barren women (she becomes target because men want to prove they can make her pregnant).	Risk Men are violent. Violence against partner.
	Protective Accepting barren women	Protective Learn to communicate without violence.
Badly treated by society		Risk Badly treated by society. Bullying. Defamation of character. Discrimination due to poverty. Humiliation on social media. Men are not understood. Pressure from society. No awareness in society about the problem of suicide. Trivialise men’s problems.
Being bewitched		Risk Being bewitched.
Effective correctional services		Risk Calm approach by police to offenders. Equal treatment of men and women by police. No – Law more favorable to women. Make men and women equal under the law. Police and social workers’ intervention. Rehabilitation centre. Suicide awareness by police.
Health problems		Risk Health problems HIV Positive
Lack of self-esteem and self-worth		Risk Comparing themselves to peers. Comparing themselves to siblings. Competing with peers. Easily give up on themselves. Failure at school. Failure in life. Lack of progress compared to peers. Men showing off. Poor decision-making skills. Lack of self-discipline and self-respect.
		Protective Accepting your situation. Being hopeful and positive. Being self-dependent. Confident man. Encourage young men to be active. Having thick skin. Being self-dependent. Self-discipline and self-respect.
Laws protecting children’s rights		Risk Local chiefs (dikgosi) cannot overrule child protection laws. Rights hinder parental discipline. Softer laws for child rights.
Men do not discuss their problems or seek help		Risk Men do not share problems. Not using government support programs. Unable to communicate with friends.
		Protective Asking for advice from family. Asking for advice from friends. Conversations with community. Creating a safe place for young men. Encourage young men to attend kgotla meetings. Guidance and counselling. Men’s groups. Asking for relationship advice. Seeking professional counselling. Seeking professional help.
Men’s involvement in crime and consequences		Risk Crimes like rape and murder. Criminal acts and consequences. Fear of arrest after a crime. Fear of arrest/pressure after committing murder. Fear of being arrested after rape.
		Protective Law fearing man.
The negative influence of friends and peers		Risk Being used/abandoned by friends. Gangster groups. Lack of support from peers. Poor advice from friends.
		Protective Choosing friends wisely.
Religion and spirituality		Risk Encourage young men to attend church. Spiritual interventions.
Male involvement	Risk Boys grew up in a female-headed family (women struggle to teach the boy how to be a man. The older boy disrespects his mother because he did not have limits).
	Protective Male involvement in sexual education (fathers talk to girls about using sex for money). Fathers are good role models to boys. Good parenting with male involvement. Men involved in discussing sexual issues. Upfront talking to boys about sex (find a direct way to talk about sex).
Transactional/intergenerational sex and related factors	Risk Sleeping with older men. Taking money from older men. Using men as a source of income. Prostitution. Taking alcohol from a man and running away.
Young women putting themselves at risk	Risk Girls should respect their bodies by not wearing revealing clothes. Women put themselves in a risky situation by walking alone at night. Girls are going after men with cars (getting into a risky situation). Sex with strangers. Young women dressing provocatively.
Ethical behaviour/good choices	Protective Girls should evaluate their choices. Youth should make ethical choices (girls and boys). Youth should value their relationships. Youth should stand up for themselves. Youth should be true to themselves.
Women being disrespectful/not cooperative	Risk Men’s agenda is always sex. Young woman trying to control the relationship. Disrespecting your co-habiting partner. Women are disrespectful of ex-partners when they find new partners. Young women using disrespectful words. Failure to show up for appointments/dates. Disrespect towards men. Young women talking nonstop. Young women must be humble – avoid upsetting/confronting their partner. Girls refusing to wear what their partner wants. Men do not want to be corrected/contradicted. Refusing to do basic chores. Young women not allowing their partner’s relatives to visit/not attending to them well. Women leave their kids alone, and then their partner gets angry at them. Women go to bars and their kids follow them. Women not letting their partners choose their friends. Young women love money (only care about money). Young women refusing sex with their partners. Making young women leave at night when they refuse sex – women are put in a risky situation. Women reject men’s advances.
	Protective They should stop giving excuses for not having sex. Accept your partner as he is. Young women should respect their parents and stay at home.
Laws and customary laws	Risk Men are not taken seriously if they report violence by a woman.
	Protective Chiefs should have the power to protect co-habiting women. Chiefs should be strict about how young women should dress. Chiefs should be strict about neglectful parents. Bars for older men should not let young women enter. Return to arranged marriages. Bring back initiation schools. Bringing back culture in the community and respecting it. Strong legislation against violence. Assistance from the police. Men do not want to be arrested (fear of law). Arrest old men having sex with underage women. Law must be equal to men and women – men’s abuse should be taken seriously. Government to punish parents who neglect children/bad parenting. Government to be strict on young people who leave their homes.
Teaching youth about violence- providing role models	Risk Lack of male role models
	Protective Youth clubs that will educate them about violence. Men groups. Youth must listen to elders’ advice. More communication about sex. Chiefs teach youth about violence. Chiefs and other male leaders should lead. Government to teach about violence.

Appendix 3.

Final activation levels in the full model for the two outcomes and the various causal categories in the combined map

The table shows an iterative sequence of calculations for the activation level of each category, starting with the initial status (t0), assuming all the categories are fully activated (value of 1). The system reached a stable state after iteration t5 when the activation levels no longer changed. Categories with activation levels closer to 1 are considered fully activated. This means they would occur if the relationships described in the map occur. Activation levels closer to 0 indicate categories that might not occur even if all the influences described in the map occur.

Table

	t0	t1	t2	t3	t4	t5	t6
Suicide of men in Botswana	1	0.969	0.921	0.911	0.909	0.909	0.909
Unemployment	1	0.500	0.500	0.500	0.500	0.500	0.500
Badly treated by society	1	0.540	0.524	0.523	0.523	0.523	0.523
Being bewitched	1	0.485	0.492	0.492	0.492	0.492	0.492
Effective correctional services	1	0.507	0.504	0.504	0.504	0.504	0.504
Health problems	1	0.589	0.558	0.556	0.555	0.555	0.555
Lack of self-esteem and self-worth	1	0.495	0.505	0.504	0.503	0.503	0.503
Laws protecting children’s rights	1	0.515	0.508	0.508	0.508	0.508	0.508
Men do not discuss their problems or seek help	1	0.542	0.523	0.522	0.522	0.522	0.522
Men’s involvement in crime and consequences	1	0.705	0.648	0.641	0.640	0.640	0.639
Negative influence of friends and peers	1	0.493	0.501	0.501	0.501	0.501	0.501
Religion and spirituality	1	0.500	0.500	0.500	0.500	0.500	0.500
Violence against young women and girls	1	0.881	0.808	0.794	0.791	0.790	0.790
Male involvement	1	0.500	0.500	0.500	0.500	0.500	0.500
Transactional/intergenerational sex and related factors	1	0.500	0.518	0.514	0.514	0.514	0.514
Young women putting themselves at risk	1	0.475	0.488	0.488	0.488	0.488	0.488
Ethical behaviour/good choices	1	0.500	0.500	0.500	0.500	0.500	0.500
Women being disrespectful/not cooperative	1	0.550	0.540	0.537	0.536	0.536	0.536
Laws and customary laws	1	0.500	0.500	0.500	0.500	0.500	0.500
Teaching youth about violence- providing role models	1	0.500	0.500	0.500	0.500	0.500	0.500
Parenting or family issues	1	0.527	0.522	0.521	0.520	0.520	0.520
Alcohol and drug use	1	0.725	0.675	0.666	0.664	0.663	0.663
Conflict in relationship	1	0.781	0.714	0.702	0.700	0.699	0.699
Men are insecure and stressed	1	0.956	0.890	0.880	0.878	0.877	0.877
Culture of violence	1	0.535	0.523	0.522	0.522	0.522	0.522
Financial difficulties WOMEN, mostly – OF THE VICTIM	1	0.895	0.843	0.829	0.826	0.825	0.825