Exploring Potential Intervention Strategies to Reduce Unsafe Youth Migration in Ethiopia: a Mixed Methods Study

There is dearth of evidence on what interventions help to reduce unsafe youth migration in sub-Saharan Africa. This study aimed primarily to identify interventions that may help to reduce unsafe youth migration from the perspectives of key stakeholders in Ethiopia. A mixed methods study was conducted in eight migration hotspot areas. For the qualitative study, fifty-six in-depth interviews and thirty-two focus group discussions (FGD) were done to collect data. Quantitative data were collected from students (n = 800), teachers (n = 240), and parents (n = 160). We developed a structured questionnaire to collect data. Descriptive statistics, t test and one-way analysis of variance were used to analyze the data. We identified potential interventions to reduce unsafe youth migration and grouped them into five domains: awareness and attitudinal/behavioral changes, training and job opportunities, enhancing governmental and parental roles, improving the education system, and law enforcement and migration management. More than 80% of the survey respondents agreed or strongly agreed that the interventions identified are potentially relevant and effective. Respondents who are older compared to those who are younger and respondents who are more educated compared to those who are less educated were more likely to endorse the interventions. Teachers and parents were more likely than students to endorse the interventions. The study indicates that the qualitative study helped to identify contextually relevant intervention strategies that would potentially be effective to reduce unsafe youth migration in Ethiopia. We suggest that there is a need to prioritize and empirically test the effectiveness of these intervention strategies with experimental studies.


Background
Migration is a global phenomenon which has been part of the human history at all times and across countries (Bariagaber 2014). Migration is a highly heterogeneous process, both in terms of quality and quantity (Bhugra 2004); it may involve one individual, a family, or a group of families (Bariagaber 2014;Uchehara 2016). Although accurate figures on the prevalence of international migration are impossible to determine, the International Labor Organization (ILO) estimated that there are 244 million crossborder migrants in 2015, of which 150 million are labor migrants (ILO 2015). Around 12% of international migrants (roughly one out of eight) are youth (defined as those between 15 and 24 years of age) (Cortina et al. 2014). The number of international migrants (persons living outside their country of birth) has continued to grow worldwide rapidly over the past 15 years (Nwalutu 2014) reaching 244 million in 2015, up from 173 million in 2000 (Mulugeta and Makonnen 2017). In 2019, the number grew to 272 million which is 3.5% of the world's population (IOM 2019). The number of migrants through illegal means is also on the increase and it will continue to hike in the future owing to the existing humanitarian situations in various parts of the world (Minaye 2012).

The Situation of Migration in Ethiopia
Ethiopia is one of the major sources of migrant flows particularly to the Middle East and South Africa . The number of international Ethiopian migrants is estimated to be 1.5 million, and each year, close to 120, 000 Ethiopians migrate (Minaye 2012). There are three main routes of migration in Ethiopia (Gezie et al. 2019). The first is the North West Route via Metema Yohannes in the border with Sudan. In this route, destinations include several countries in the Middle East and Europe. Sudan is both a transit to the move to Europe through Libya or to migrate to the Middle East. The second route is the Southern Route through Moyale near Kenya and the destination is South Africa by crossing several countries like Kenya, Tanzania, Zimbabwe, Malawi, and Mozambique. The third one is the Eastern Route through Galafi near the Djibouti border and the destinations are several countries in the Middle East where Djibouti and Yemen are the transits. The migration in Ethiopia, particularly the unsafe migration, involves particular skills, age group, and gender (Demissie 2018). In terms of skill, migrants from Ethiopia mostly involve domestic work and other nonprofessional works (e.g., driver, waiter/waitress, and barista) (Zewdu 2018). In terms of age group, most migrants are young men and women (roughly from early twenties to early thirties) (Anteneh 2011). Gender wise, it differs in terms of route and destination. For instance, migrants to the Middle East are mostly women and to South Africa are mostly men (Demissie 2018).

Push and Pull Factors
Studies show that poverty, overpopulation, unemployment, environmental degradation, war, and natural and man-made disasters are major push factors for migration worldwide (Deotti and Estruch 2016;Roth and Hartnett 2018;Joly 2019;Kazemi et al. 2018). Particularly in Africa, lack of livelihood opportunities, negative attitudes attached with low-paying informal jobs, and poor work ethics are considered major causes for migration (Jamie and Tsega 2016;Sparreboom et al. 2019;Minaye and Zeleke 2017). However, lack of employment opportunities is consistently cited as the major one. It is generally well-documented that poverty; system failure; actions of organized criminal groups like traffickers, smugglers, and brokers; family; and peer pressure act as causes for the high prevalence of legal and illegal migration in Africa in general and in Ethiopia in particular (Fernandez 2011;Zewdu 2018). Insecurity in Somalia, instability in Eritrea and the Ethio-Eritrea conflict are also considered important factors in determining the migration dynamics in Ethiopia (Gibson and Gurmu 2012).
Positive attitudes towards migration, lower risk perception, readiness to migrate, and peer and family pressure are also important emerging reasons for unsafe youth migration . Several studies from Ethiopia also found that youth unemployment/underemployment, conflict, poverty, income inequality, household indebtedness, population pressure, lack of good governance, landlessness, low agricultural productivity and agro-climatic disasters, low wages, and insecure informal economic activities are important causes for the youth to leave their homes (Mulugeta and Makonnen 2017;Minaye 2012;Gibson and Gurmu 2012;Zewdu 2018). There are also several pull factors for migration. These include attractive working conditions and demand for cheap labor in destinations, the growth of the service industry, aging population in developed countries, booming oil economy in the Middle East, and globalization and modernization that have improved communication, transportation, and social connection (Dessiye and Emirie 2018;Roberto and Moleiro 2016).

Consequences of Migration
International migration has both positive and negative consequences for the sending and receiving countries. Migration contributes for the development of countries of origin (sending countries) in terms of remittance, job opportunities, and technology and skill transfer; whereas for destination countries, migration contributes in terms of labor supply, cultural exchange, and skill transfer (Minaye 2012;Dako-Gyeke 2016;Urzúa et al. 2020). It brings about an increased level of urbanization, and a means of achieving economic efficiency and equity (Mulugeta and Makonnen 2017). Migration is regarded as a good cause for social diversity as well as a necessary condition for the creation and strengthening of a sense of nationhood (Cortina et al. 2014). Migration, through creation of diverse workforce with a wide range of skills and experiences, is more likely to benefit receiving countries (Tarisayi and Manik 2019;Minaye and Zeleke 2017).
The problem comes when migration, particularly cross-border migration, is unsafe. One cannot give a single and agreed upon definition for the concept of unsafe migration as it is broad and can be seen from different perspectives. Some defined it as "movement that takes place outside the regulatory norms of the sending, transit, and receiving countries" (Davies et al. 2009). Another definition for unsafe migration is "irregular movement that takes place outside regulatory norms of the sending, transit, and receiving countries, and during which a migrant suffers from or risks of severe physical, mental, or emotional injury/harm" (Sarrica 2010). Still, others defined it as migration with no adequate preparation from the perspective of the migrant in terms of language, culture, and skills in the destination country (Zewdu 2018). In this study, the concept of unsafe migration includes those migrants who, in many ways, are exposed to abuse, exploitation, and harm as a result of migratory experience, including victims of human trafficking and smuggled migrants. Unsafe migration brings about adverse impacts on the migrants, sending households, source countries, transit countries, and the host communities in the destination countries (Mulugeta and Makonnen 2017;Hare 1999). The most critical consequences of unsafe cross-border migration include security and economic aspects (ILO 2015). Organized crime in the sending, transit, and destination countries is also another dangerous consequence of unsafe migration (Rachlis et al. 2008;Baker and Aina 1995). Unsafe migration is a cause and consequence of inequality and unequal development as the immigrants sometimes are unable to access, at least temporarily, the existing services such as housing in their destinations (Cortina et al. 2014).
Unsafe and stressful migration affects migrants, their families and source and receiving countries in general. Unsafe migration is found to have impact on migrants physical, psychological, social and economic well-being (Minaye 2012;Salama et al. 2017), which include drowning, trafficking, sexual exploitation, labor exploitation, organ harvesting, degrading treatment, discrimination, physical attack and denial of medication (Shi et al. 2012;Carballo and Nerukar 2001). For instance, qualitative studies of female Ethiopian migrants to the Middle East found experiences of exploitation, enforced cultural isolation, undermining of cultural identity and thwarted expectations (Anbesse et al. 2009;Dessiye and Emirie 2018). Another qualitative study on migrant returnees from the Middle East also indicated that conditions for Ethiopian workers abroad are marked by frequent exploitation, neglect, and physical and sexual abuse (Busza et al. 2017). A recent survey of migrant returnees from the Middle East and South Africa found a high burden of mental health problems resulting from adverse migration experiences, with 27.6% respondents reporting symptoms of common mental disorders (Habtamu et al. 2017). Considering the context that migration is not positive or negative by itself, we need to enhance the positive and reduce the negative consequences of migration. This requires effective migration management which is informed by empirical evidence.

Intervention Strategies to Reduce Unsafe Migration
As documented above, there is ample evidence on what determines (risk and protective factors) unsafe migration. However, research on what interventions may help to reduce unsafe youth migration is lacking, particularly from sub-Saharan Africa. A qualitative study among Ethiopian migrant returnees from Middle East countries found that knowledge, skills, interpersonal attributes, and resources foster safer migration (Busza et al. 2017). Busza and colleagues recommended pre-migration information and training programs to reduce unsafe migration.
A systematic review of available studies, mostly from South East Asia and South America, identified most commonly used interventions to reduce negative migration outcomes (Zimmerman et al. 2016). These include awareness campaigns, informational education material distribution, capacity building of local partner organizations and government agencies, employment placement, skillbuilding, migrant resource centers, hotlines, home visits to migrant families, legal assistance, network building for partners and beneficiaries, peer educator training, pre-departure orientation seminars, and psychosocial support. Nevertheless, the theoretical and empirical basis of these interventions is not clear. More research is needed to identify specific contextually relevant strategies that may help to reduce unsafe youth migration.

Aims of the Study
This study was conducted as part of a larger thematic research project that aimed to examine the nexus among unsafe youth migration, quality of education, and gender. As part of this project, a qualitative exploration and quantitative measures of effectiveness of contextually identified interventions to reduce unsafe youth migration in Ethiopia were included. The overall aim of the study is, therefore, to identify specific contextually relevant interventions that may help to reduce unsafe youth migration from the perspectives of key stakeholders. The study primarily aimed at exploring contextually appropriate interventions that would be feasible, acceptable, and effective to deal with unsafe migration in Ethiopia. The study also aimed at determining the extent to which survey respondents endorse the usefulness of the interventions identified and the extent to which the endorsement differ in terms of basic background characteristics including sex, age, religion, number of years of education completed, occupation or respondent type, and study site.

Study Design
A mixed methods study, specifically exploratory sequential study (Creswell and Plano Clark 2017), was conducted because it was appropriate to address our research questions. The study was intended first to identify interventions (which may help to reduce negative migration outcomes) and determine their contextual relevance from the point of view of teachers, parents, and students. The first research question fits with the qualitative approach and the second to a cross-sectional survey. The qualitative study was conducted from February to April 2017; whereas the quantitative study was conducted from January to March 2018.

Study Sites
This study was carried out in eight migration hotspot areas (Dessie, Jimma, Assela, Atsbi Wonberta, Butajira, Hossana, Shoa Robit, and Kemissie) from four regional states in Ethiopia. Dessie, Shoa Robit, and Kemissie are from Amhara Regional State; Jimma and Assela are from Oromia Regional Sate; Butajira and Hossana are from Southern Nations, Nationalities and Peoples' Regional Sate; and Atsbi Wonberta is from Tigrai Regional State. These study sites were selected based on evidence from previous studies which confirmed that migration is highly prevalent in these hotspot areas (Minaye 2012;Anteneh 2011).
The size of the population in the study sites ranges from around 25,000 (Kemissie) to around 140,000 (Dessie) (CSA and ICF 2016). So, there is variation in population size across the study sites. Dessie and Jimma are relatively large towns with population each more than 125,000. Assela and Hossana are in the middle, with population 60,000 and 76,000, respectively. Atsbi Wonberta, Butajira, Shoa Robit, and Kemissie are relatively small with population ranging from 25,000 to 37,000. In terms of economic and business activities, small-and large-scale trading in towns and agriculture in the surrounding rural villages are common. Coffee is well-known in Jimma and khat is common in Kemissie, Shoa Robit, Butajira, and Hossana. Assela, Dessie, Shoa Robit, and Kemssie and their surrounding are known for cereals, such as wheat, teff, and barley. Overall, adult literacy rate in Ethiopia is around 51% (Gelana 2014), one of the lowest in Africa. No significant difference in the median number of years of schooling across the study sites (CSA and ICF 2016). There are many primary schools and at least one secondary school in each of the study sites. There are two or more secondary schools in Hossana, Jimma, Assela, and Dessie. Assela, Dessie, Hossana, and Jimma have each one university. Unemployment rate in urban areas in Ethiopia is very high (around 17%) (Mago 2014), and this is similar in all the study sites (Lercari et al. 2018).

Participants
In order to select the participants for the qualitative study, multistage sampling method was used. First, eight migration hotspot areas were selected purposively as research sites. In the second stage, one school (which has both junior and general secondary grades) was selected in each study site for accessing students, teachers, and parents as data sources. In the third stage, 16 students from grades seven to ten, eight teachers teaching from grades seven to ten, the school principal of the selected schools, and five parents who were working as members of the Parent-Teacher-Student Association (PTSA) of the schools were purposively selected and involved in the study. Those students and teachers who had knowledge and experience on the problems of migration and were able to provide consent were invited to participate in the study. Those participants from schools are assumed to have direct contact with potential migrants. Most of the youth migrants drop out from school (junior secondary and secondary schools and also technical and vocational colleges). Heads of local government officials representing different sectors and directly or indirectly involved in managing migration were involved in the study. Accordingly, heads of police, justice, women and children affairs, youth affairs, labor and social affairs, education, and medium and small manufacturing enterprise in each study site were invited to participate in the interviews. Our assumption is that these officers are directly or indirectly involved in managing migration (training, prevention, rehabilitation etc.). In addition, a community representative and a religious leader (either from Christian or Muslim) were interviewed from each study site as they are opinion leaders and have better access to information in the community related to unsafe migration.

Data Collection Methods and Procedures
In-depth interviews and focus group discussions (FGD) were used to collect data. Fiftysix in-depth interviews (seven in each study site) were carried out with heads of offices of the local government and the school principal. We were not able to use FGD for these participants as it was difficult to bring them together from different offices. The experiences and power hierarchies of these participants are also different, and running a comfortable discussion was difficult. So, FGDs were rather conducted with teachers, students, and parents. Accordingly, thirty-two FGD (four in each study site) were conducted. Eight FGDs were conducted with teachers (one in each site). Sixteen FGDs were held with students (one FGD with students from grades 7 and 8 in each site and another with students from grades 9 and 10). Then, eight FGDs were run with parents (one in each site); these parents are members of the Parent-Teacher-Student Association. FGD was used with all the three groups (students, teachers, and parents) to elicit more information through their interaction and to encourage sharing of their experiences (Patton 2014). In each FGD, eight to ten respondents were involved except in one of the FGDs with parents, where only five participants were involved.
Almost all of the in-depth interviews and FGDs were conducted in Amharic, the official language of Ethiopia. Researchers of the larger thematic project, where this study is nested in, conducted all FGDs and in-depth interviews (all are faculty members at Addis Ababa University). All of these researchers have training and several years of experience in qualitative data collection. While these researchers were moderating FGDs, their respective note takers or field assistants summarized the discussions and noted the non-verbal communications. In-depth interviews with heads of local government offices and the school principal were conducted in their respective offices; whereas the interviews with community representatives and religious leaders were carried out in a comfortable place in each town including their home, cafeteria, hotels, and churches/mosques. The FGDs with students, teachers, and parents were carried out in a free classroom in the selected schools. Privacy and confidentiality of participants of the interviews and FGDs were assured at all times.
All interviews and FGDs were tape-recorded, with the consent of the participants. Generally, in-depth interviews lasted between 40 and 80 min, whereas FGDs lasted between 60 and 120 min. As part of the main study, participants were asked four relevant issues (unsafe youth migration, quality of education, gender, and the nexus among these three). The migration component of the questions dealt with the conceptualization, trend in terms of magnitude, causes, consequences of unsafe migration in the study site, and the intervention strategies that may help to reduce unsafe migration Annexures II. Planned probes, which included questions, related to causes, consequences, and interventions for unsafe migration were identified from the existing local and international literature (Busza et al. 2017;Habtamu et al. 2017;Mulugeta and Makonnen 2017;Chinyakata and Raselekoane 2018;De Haas 2010;Minaye and Zeleke 2017). The topic guide was developed iteratively as data collection and preliminary analysis continued (Patton 2014).

Data Analysis
The interviews and FGDs were transcribed verbatim by experienced transcribers. Track changes were used to facilitate data management. Data analysis was undertaken in parallel with data collection, with frequent discussion of the emerging themes among members of the research project. Thematic analysis (Braun and Clarke 2006) was used to identify the prominent issues from the data. Themes and categories were compared between the different respondent groups. The first author coded two transcripts, and coding schemes were discussed with the second author. The first author coded the remaining transcripts applying the already identified codes and drawing upon additional codes where the data required. This was done by frequently discussing with the second author. Higher order codes were derived from the primary codes with thorough discussion between the first and second authors. Similarly, overarching themes were developed from the higher order codes. Illustrative quotes were selected for each theme.

The Quantitative Study
Participants Quantitative data were collected from students, teachers, and parents. The purpose of the quantitative study was to determine the extent to which intervention strategies to reduce unsafe migration, identified through the qualitative study, are endorsed and considered relevant by a large sample of respondents. The study sites and the criteria used to select them are the same as the qualitative study (as described above).
We used multistage sampling in order to select the participants of the study. In the first stage, eight migration hotspot areas (Dessie, Jimma, Assela, Atsbi Wonberta, Butajira, Hossana, Shoa Robit, and Kemissie) were purposively selected to be research sites. In the second stage, one school (with grades seven to twelve) and one technical and vocational education and training (TVET) college were again purposively selected in each site. Then, one section from each grade level (grades seven to twelve) and a section from the TVET College were randomly selected. One hundred students (10 from grades seven, eight, eleven, and twelve and the TVET college and 25 students from grades nine and ten) in the selected sections were randomly selected and invited to take part in the study. More numbers of students were selected from grades nine and ten because students at this level are highly vulnerable to migration. Those below grade nine are too young for migration and those at preparatory level and those in TVET have better hope of joining university and thus they are less likely to migrate. Thirty teachers teaching in grades seven through twelve were randomly selected and invited to participate. In addition, 20 parents who had at least one child in the school were recruited through the school principals and members of the PTSA.
Overall, 150 respondents (100 students, 30 teachers, and 20 parents) from each site, and 1200 from all the study sites (800 students, 240 teachers, and 160 parents) were involved as participants for the survey in this study. Missing data were less than 5% in all of the items in the questionnaire.

Measures
We developed a structured self-administered questionnaire for the larger thematic research project to assess the situation of unsafe migration, quality of education, and gender in the study sites. We used findings of the qualitative study and extensive review of the existing literature to develop the questionnaire. Specifically, the questionnaire was designed to gather data related to trends, causes, consequences of unsafe migration, and potential intervention strategies to reduce unsafe migration; input, process, and output indicators of quality of education; nexus between education and migration; and nexus between gender and migration. The draft questionnaire was reviewed by experts who have training and experience on developing research instruments. We pilot-tested the questionnaire with small sample in one of the study sites and improvements were made incorporating feedback from the experts and findings of the pilot study. Internal consistency reliability was determined and some vague or unclear items were rewritten. A few unnecessary items were removed and other items were added as per the suggestion of the experts. The questionnaire took an hour, on average, to complete (Annexures I and II).
For the study reported in this paper, data collected using section IV of the questionnaire (strategies to mitigate youth unsafe migration) was used Annexures I. Similar to the other sections of the questionnaire, this section of the questionnaire was developed being informed by findings of the qualitative study and extensive review of the existing literature. The questionnaire presented fifteen potential intervention strategies and asked participants to rate the extent to which they agree or disagree about the contextual relevance, feasibility, and effectiveness of those intervention strategies in order to reduce unsafe youth migration. The fifteen items measuring potential intervention strategies were grouped into three meaningful domains through expert discussions involving migration and measurement professionals: working on values and behavior, enhancing access to resources and jobs and law enforcement and migration management. Exploratory factor analysis confirmed the expert decision and each item fairly loaded to the respective domain with cross-loading of few of the items. Internal consistency reliability of items of the overall scale was 0.91; whereas internal consistency reliability of items of the sub-scales working on values and behavior, enhancing access to resources and jobs, and law enforcement and migration management was found to be 0.74, 0.80, and 0.83, respectively. Aggregate scores were computed for each sub-scale to facilitate further statistical analysis.
A structured background characteristics questionnaire was developed and used to collect data on the sex, age, education, religion, and occupation of the respondents. Experienced field workers, who have at least diploma level education, were employed to collect the data. Researchers from the larger thematic research project supervised the field workers (one supervisor was assigned for each study site). Training was provided for a day to the field workers focusing on the content of the questionnaire, ethical issues to be observed and ways of approaching respondents.

Data Analysis
Data were checked for completeness and consistency by the field supervisors during data collection. Data entry was carried out using Statistical Packages for the Social Sciences (SPSS version 22). Data were entered by a trained and experienced data entry clerk. The first author cleaned the data after data entry and before data analysis began using frequencies and logic checks. Descriptive statistics (frequency and percentage) were used to summarize the background characteristics of the participants. Descriptive statistics were also used to determine the extent to which respondents endorsed and considered the intervention strategies as relevant to reduce unsafe youth migration. One-way analysis of variance (ANOVA) and t test were used to determine the extent to which endorsement of the potential intervention strategies differ in terms of sex, education, religion, and respondent type (teachers, students, and parents) as appropriate. Statistical tests were set at α = 0.05 for significance.

Ethical Considerations
A technical committee established by the Office of the Vice President for Research and Technology Transfer (VPRTT) at Addis Ababa University reviewed and approved the study protocol for the larger thematic study. To that end, we secured a support letter from the VPRTT to collect data from the study sites. Permission to collect data was obtained from the concerned offices in all the study sites by presenting a cooperation letter written from Addis Ababa University. The concerned offices are local government offices mandated to work on migration, education, gender, security, and related issues, which include Labor and Social Affairs Office, Justice Office, Women and Children Affairs Office, Youth Affairs Office, Education Office, Medium Manufacturing Enterprise, and the Police. Participation was voluntary, and verbal informed consent was obtained from all the participants after the nature of the study was fully explained to them. Verbal informed consent was preferred to written informed consent just to put respondents at ease since informants may not be comfortable to put their signature on paper in the Ethiopian socio-cultural context. Respondents were explicitly informed that they could withdraw at any time from the study and cease to respond to any question they felt uncomfortable. Information obtained from all the participants was anonymized and confidentiality was assured throughout the data collection process.

i) Findings from the Qualitative Study
In responding to the question "what shall be done to reduce unsafe youth migration," participants suggested a long list of intervention strategies that would help to deal with unsafe youth migration and the consequent undesirable migration outcomes in Ethiopia. These specific intervention strategies identified are grouped into five broad categories. The first is creating awareness and bringing attitudinal/behavioral changes, the second is creating job opportunities and providing training, the third is fulfilling governmental and parental roles, the fourth is improving the education system, and the last is law enforcement and managing migration. In describing the different intervention strategies, which would help to reduce unsafe youth migration, participants from the three groups in different study sites focused on the need for cooperation among stakeholders. It was emphasized that there should be cooperation among different local government offices mandated to manage migration, elders, teachers, parents, religious institutions, schools, the community, and non-governmental organizations.

Awareness and Attitudinal and Behavioral Changes
All groups of participants across the study sites emphasized the importance of creating awareness regarding the negative consequences and harms of unsafe migration and subsequently changing attitudes and behavior of the youth, their parents, and the community at large. In connection with awareness creation activities, the participants suggested to educate the youth, their parents, and the community through several formal and informal structures. Interview and FGD participants particularly advised to use migrant returnees as witnesses. This requires supporting returnees to share their experiences. Some participants suggested teaching with the support of video films (showing the negative experiences of those who migrated illegally). During the focus group discussions, making religious institutions to teach their followers and using local and informal social structures such as "Idir" (informal insurance groups) to educate parents and the community were reiterated. Participants mentioned the importance of establishing formal and regular discussion forums at the community level to make the awareness creation activities sustainable. Regarding the need for awareness creation activities, a FGD participant noted the following.
It is critical to give education [about illegal migration] to parents and the community in schools through clubs and make students aware of the problems. Schools do have potential to create awareness among students and the community in general. (FGD with Parent-Teacher Association Members, Atsbi Wonberta) Participants from all the study sites underlined the need to bring about attitudinal and behavioral changes, particularly among the youth and their parents. It is vital for the youth to understand that it is possible to work and change their life here within the country. In addition, attitude change would bring difference in the decision to migrate unsafely if the youth understand the harms, risks, and negative consequences of migration particularly unsafe and illegal migration. A prosecutor from one of the sites reflected his view as follows: We have to exert all our efforts at the grassroots level to create awareness so that the youth would change their attitude to work in their home country. I know there are laws, policy and different banning directives on migration in Ethiopia; and we are also working some tasks as justice officers on migration; but still there is an increasing number of migrants in our locality. This is not because there is no law or punishment of illegal migrants, but unless we work on the attitude and change the mind of the youth, we cannot solve the problem. (In-depth interview with a prosecutor from District Justice Office, Jimma) Some participants pointed out that the concerned bodies should work hard and educate the youth in order to improve work ethics and work culture among the youth. They added that there is a need to make the youth love their country and have sense of citizenship. Similarly, awareness creation activities using the right media, at the right setting, should be done to change the youth so that they would not undermine locally available jobs and consequently work whatever job they get here within the country. Continuous and sustained awareness creation activities have to be done to make the parents understand the risks of unsafe migration and change their attitudes since much of the pressure for the youth to migrate come from the parents and other family members.

Creating Opportunities for Job and Training
Participants from every study site emphasized that the major reason for the youth to choose to leave their country is lack of job opportunities which subsequently lead to poverty. Accordingly, interview and FGD participants suggested that creating job opportunities, particularly expansion of manufacturing industries, would be a sustainable solution to deal with unsafe youth migration. One of the participants in the FGD held with high school students (grades nine and ten) at Kemissie described this as: "… to reduce unsafe youth migration the government and concerned non-governmental organizations have to organize the youth and create job opportunities for them." In addition, it was highly recommended for the government to organize the youth fairly (without discrimination), arrange loans with low interest rate, and make the loan process less bureaucratic. Participants also indicated that arranging employment opportunities for new university graduates would be important to address the increasing low value being given to education by students which in turn can help to reduce school dropout and the lower motivation and interest of students towards education. All these interventions were believed to reduce youth unemployment and joblessness, which are the root causes of unsafe youth migration in Ethiopia.
Related to creating job opportunities and expanding manufacturing industries, participants underlined the necessity of providing vocational and life skills training for the youth. The concerned government offices were advised to provide focused, hands-on, and practical training for the youth in collaboration with other stakeholders, such as technical and vocational training colleges, private firms, and non-governmental organizations. This would help the youth to increase their motivation for work and entrepreneurial skills. Some participants from a few of the study sites suggested that the government should plan facilitating skilled migration by providing potential migrants with marketable skills. This includes offering intensive and compulsory pre-migration training so that migrants would be competent and successful. One of the FGD participants with teachers commented as follows.
There must be training-based legal migration in the country. If potential youth migrants are well trained, they will be useful to the country. To make migrants competent, they should be provided with vocational skills training. There is also a need to provide life skills training for the youth properly and need to educate them about how they would realize their dream. (FGD with teachers, Shoa Robit)

Enhancing Governmental and Parental Roles
Participants identified several roles that the government should play and responsibilities that it should discharge as important intervention strategies to deal with unsafe youth migration in Ethiopia. Interview and FGD participants strongly advised the government to effectively and efficiently implement migration-related policies and plans and optimize the performance of the executive body at different levels to put these provisions into effect. Moreover, participants suggested the government to work on and bring about change on different aspects of good governance, including providing equal opportunity, avoiding discrimination, ensuring gender equality, and most importantly reducing corruption. The government also needed to do its best in terms of creating jobs and eventually reducing poverty, both at the household level and at the population level. Some participants also advised the government, at different levels, to encourage the youth to use the legal way for migration by way of making the process easier and less expensive. Conversely, the participants advised the government to discourage illegal migration through follow-up and strict monitoring. A parent from Dessie who participated in the FGD reflected his view as follows: The major thing is the role of the government. The government has to close borders and make serious patrolling. It has to support migration only through legal means. There should be serious control of the illegal brokers. The government has to also create job opportunities, particularly for women. The government has created jobs and made things suitable for rich investors, but, it has to prioritize and make jobs available for the poor and for those who have no jobs. (FGD with parents, Dessie) In addition, most of the participants mentioned roles of the government such as establishing a separate mandated institution that would manage migration, increasing salary for civil servants (particularly of primary school teachers and health extension workers), creating political stability in the country, and providing land and other resources primarily to the youth rather than to investors. It would also be useful if the government build the capacity of district officers who are working in areas related to job creation, providing loan, organizing the youth, and managing migration. Some participants emphasized that alongside creating jobs for those who have no job, the government should create a conducive or decent work environment for those who are already employed. A school principal from Butajira described this as "…the government needs to seriously consider the salary and work condition of government employees; large numbers of teachers and health extension workers leave their job and go abroad since their salary is not enough for fulfilling their basic needs." Participants underlined the role of parents in reducing unsafe youth migration. First and foremost, parents need to advise and educate their children about the risks of unsafe migration, the value of education, and the importance of developing work culture. Parents need to support and give continuous encouragement in the education and work life of their children. The other role that parents needed to play in terms of reducing unsafe youth migration was follow-up and monitoring of their children, particularly in relation to their education. Participants said that parents should control and monitor their children until they finish their education. Specifically, some participants advised parents to see their sons and daughters as equal and provide them equal opportunities. Regarding the role parents might play in reducing unsafe migration, the education expert from Butajira district described it as: To solve this problem [unsafe youth migration], it requires everyone to exert its effort. Parents need to be concerned about the safety of their children and they need to send their children to school. Parents may provide small seed money for their children to start business. If they invest on the education of children, there will be a lot of change. So, parents are expected to advise their children well. Parents do give money for their children to finance migration and they are expected to stop this. (In-depth interview with District Education Office Head, Butajira)

Improving the Education System
Interview and FGD participants, from the different study sites, stressed that one of the leading causes for youth migration is related to access and quality of the education system in the country. Hence, in order to deal with unsafe youth migration, participants advised the government to make university education and technical and vocational education and training (TVET) more accessible. The Ministry of Education (MoE) and other stakeholders should give attention and work on improving the quality of education. Specifically, the following were some of the issues respondents mentioned to be improved: large class size, shortage of teaching materials including Internet resources, poor teacher quality, lack of expertise in vocational training, lack of partnership between schools and the industry, and less opportunity for apprenticeship.
When quality of education declines, students would not attend their education properly; drop out increases; and students' interest, motivation, and performance decrease, all of which in turn would contribute for the youth to consider leaving their country. Therefore, participants advised to make the school environment as comfortable as possible (example: classrooms and playing ground). To improve the quality of education participants suggested increasing teachers' interest and performance through increasing their salary and improving the respect and value being given to teachers and the teaching profession. A FGD participant with students from grades 7 to 8 commented as follows.
There is need to work on quality of education to deal with unsafe youth migration and illegal migration. Students drop out when the quality of education declines. Students dislike education when quality declines. Their performance will also decrease when quality of education is compromised. So, there is a strong link between quality of education and migration. (FGD with students, Assela) Moreover, participants suggested that, to bring about attitudinal and behavioral changes with regard to unsafe migration, it would be helpful to establish migration clubs in schools and strengthening existing ones. Participants also suggested incorporating migration in the school curriculum.

Law Enforcement and Managing Migration
All groups of participants across the study sites stressed hunting illegal brokers and bringing them to court as a critical strategy to deal with unsafe youth migration. The Police have to work hard to identify, control, and punish illegal brokers. Coordinator of the District Small and Medium Manufacturing Industry Office from Atsbi Wonberta described this as: "The government has to control and take a serious action against illegal brokers and others who engage in facilitating unsafe youth migration. There should also be a serious border patrolling." It would also be important to make the public aware about the consequences of being involved in smuggling and trafficking by portraying sample stories of the punishments on convicted illegal brokers, smugglers, and traffickers through several means such as the media. Illegal brokers have a complex network and the police and all the other security apparatus should work in cooperation to break down their networks. Some participants also suggested for a sustained and continuous patrolling and controlling around border areas as a means to find the brokers and amputate their networks. Furthermore, it would be important to make the community aware and consequently engage in exposing illegal brokers. In relation to law enforcement and managing migration, participants emphasized that there should be strict enforcement of existing legal instruments (national, regional, and international). Moreover, participants advised the need to inform the community that unsafe migration and illegal migration are often linked to smuggling of migrants and trafficking in persons which are serious crimes.

Background Characteristics of the Respondents
A total of 1187 respondents participated in the study (a response rate of 98.9%). Of these, 55.4% were males and more than half (59.4%) were below 20 years of age. Nearly 90% of the respondents had secondary or higher level education. In terms of religion, the majority were Orthodox Christian (61.3%) followed by Muslim (26.7%). Two-thirds of the respondents were primary, secondary, or TVET students, nearly onefourth were teachers and a little over 10% were parents. There was almost equal distribution of the sample across the study sites. See Table 1.
N, number of participants

Distribution of the Endorsement of Intervention Strategies Identified Using the Qualitative Study
We asked survey respondents to indicate their degree of agreement regarding the contextual relevance and effectiveness of the intervention strategies that participants of the qualitative study suggested as may be useful to reduce unsafe migration and improve the outcomes of migration. Accordingly, more than 80% of the respondents agree or strongly agree that all the interventions suggested would be contextually relevant and effective in terms of reducing unsafe youth migration as well as improving migration outcomes (Table 2). Only less than 10% of the respondents disagree or strongly disagree regarding the contextual relevance and effectiveness of these interventions. Specifically, changing the attitude/behavior of the youth, creating job opportunities, and law enforcement and managing migration effectively through punishing illegal brokers and having a national policy and a national agency that would give services on migration were considered important interventions to deal with unsafe youth migration. However, around 62% of the respondents agree or strongly agree for one of the interventions (that is for promoting legal migration). For this intervention, around 27% of the respondents disagree or strongly disagree about its contextual relevance and effectiveness to reduce unsafe youth migration and improving migration outcomes.
N, number of participants

Difference in Endorsement of Intervention Strategies in Terms of Background Characteristics
No statistically significant mean difference was observed in the degree of endorsement of respondents regarding the contextual relevance and effectiveness of all the three domains of intervention strategies (working on values and behavior, enhancing access to resources and jobs, and law enforcement and migration management) in terms of sex and religion. In terms of age, statistically significant mean differences were found with regard to the domains of enhancing access to resources and jobs (F = 4.40, P < 0.05) and law enforcement and migration management (F = 7.24, P < 0.05). Tukey post hoc test showed that respondents with age less than 15 were found to have significantly lower mean value on the enhancing access to resources and job domain compared to all other age groups. With regard to law enforcement and migration management domain, respondents with age less than 15 and age 16-19 had significantly lower mean values compared to those with age 20-29 and 30-39. Overall, these findings showed that older respondents were more likely to endorse intervention strategies more than younger respondents. We found statistically significant mean differences across levels of education of respondents in the enhancing access to resources and job domain (F = 6.06, p < 0.05) and the law enforcement and migration management domain (F = 7.70, P < 0.05). More specifically, in terms of mean value for enhancing access to resources and jobs, those who have less than 8 years of education were significantly lower than both those with education 8-12 and > 12 years. With regard to mean value for law enforcement and migration management, those who have greater than 12 years of education were found to have significantly higher mean value than both those with education less than 8 years and 8-12 years. These indicated that respondents with higher level of education were more likely to endorse intervention strategies more than those with lower level of education. We found statistically significant mean difference across respondent type only in one of the three domains (the law enforcement and migration management domain) (F = 7.36, P < 0.05). Tukey post hoc tests showed that the mean level of endorsement of intervention strategies in the law enforcement and migration management domain of students was significantly less than both teachers and parents. Mean values for level of endorsement of intervention strategies differed significantly across the study sites in all the three domains of interventions: values and behavior domain of interventions (F = 11.08, P < 0.01), access for resources and job domain (F = 14.27, P < 0.01) and law enforcement and migration management domain (F = 20.2, P < 0.01). It seemed generally that the level of endorsement of intervention strategies in the three domains was higher in Dessie and Butajira compared to the other sites although it was high in all of the study sites (Table 3). SD, standard deviation; F, test value for one-way analysis of variance; t, test value for independent sample t test

Discussion
In this study, we identified a large number of specific interventions that may help to deal with unsafe youth migration from the perspective of several stakeholders (local government office heads, religious and community leaders, teachers, parents, and students). These large numbers of specific interventions were grouped into five domains: awareness and attitudinal/behavioral changes, job opportunities and vocational and life skills training, enhancing governmental and parental roles, improving the education system, and law enforcement and management of migration. The need to work in cooperation among several stakeholders, including local government offices, schools, the community, and non-governmental organizations, was cross-cutting. Randomly selected survey respondents (teachers, parents, and students) were asked to indicate their level of agreement about the contextual relevance and effectiveness of these interventions in terms of reducing unsafe youth migration in Ethiopia in particular and improving migration outcomes in general.
We found that a great majority (more than 80%) of the respondents endorsed (agreed or strongly agreed) about the effectiveness and contextual relevance of these interventions. One of the interventions (i.e., promoting legal migration) was comparatively less endorsed. Around 27% of the respondents did not endorse this intervention (disagree or strongly disagree). Overall, the degree of endorsement of the interventions was high and statistically significant 27.4 (6.2) difference was not found across most of the background characteristics. Nevertheless, older compared to younger respondents were higher in the mean values of levels of endorsement of the two domains: "enhancing access to resources and jobs" and "law enforcement and migration management." For the domains "enhancing access to resources and jobs" and "law enforcement and migration management," respondents with higher level of education were more likely to endorse the interventions than those with lower level of education. The level of endorsement of the intervention strategies by students in the domain "law enforcement and migration management" was significantly lower than the level of endorsement by teachers and parents. With regard to study sites, the level of endorsement of intervention strategies in all the three domains was higher in Dessie and Butajira than the other study sites.
There is generally lack of research on effective intervention strategies to reduce unsafe migration and undesirable migration outcomes to compare our findings with. Studies, from different parts of the world, showed that there are several risk factors for unsafe youth migration as well as for undesirable migration outcomes. These include irregular migration (Chantavanich et al. 2016), lack of skill and knowledge gaps (Anteneh 2011), law enforcement and policy gaps (Zimmerman et al. 2016), family and social pressure (Betancourt et al. 2013), poor border control and patrolling, political uncertainty in the sending country, unemployment/joblessness and poverty (Jaisat et al. 2014), and discrimination and government corruption . Interventions to make migration safe are expected to deal with these risk factors. The specific intervention strategies we identified are more or less related to the above risk factors and hence may potentially be effective to reduce unsafe youth migration and broadly to improve migration outcomes, although this has to be empirically tested through intervention studies.
The intervention strategies identified in this study (e.g., awareness creation activities, creating job opportunities and law enforcement) are also more or less similar with the few interventions that have been previously tested with empirical studies. A non-peer-reviewed systematic review report (Zimmerman et al. 2016) identified interventions that have been tested and found to be effective in South East Asia and South America to reduce unsafe migration. These interventions include the following: awareness campaigns, informational education material distribution, capacity building with local partner organizations and government agencies, employment placement, skill-building, migrant resource centers, hotlines, home visits to migrant families, legal assistance, network building for partners and beneficiaries, outreach activities for beneficiary identification, peer educator training, pre-departure orientation seminars, and psychosocial support. However, some of these interventions (such as hot lines) may not be contextually relevant in the Ethiopian situation. Hence, the need for qualitatively identifying contextually appropriate interventions that would be feasible, acceptable and effective to deal with unsafe migration in Ethiopia is justified. Although most of the interventions identified in this study are similar to those reported in the literature, we also come across with a few interventions that are unique to the socio-cultural situation in Ethiopia which are not reported in the literature previously. These include interventions related to making sure that parents fulfill their roles in terms of advising, educating and supporting their children, and working on the education system.
More than 80% of the survey respondents indicated that the intervention strategies we identified with the qualitative study are contextually relevant and potentially effective to deal with unsafe youth migration. We also found that older respondents and respondents with more education endorsed these interventions more than respondents with younger age and those who have lower education. Overall, these findings suggest that the intervention strategies we identified would likely be feasible, acceptable, and effective to reduce unsafe youth migration. Hence, there is need to conduct further research to empirically test the effectiveness of these interventions. The findings of this study also indicate the need for modernization of the education system and cooperation with the business community to reduce unsafe youth migration.
One of the major strengths of this is study is that we used a mixed methods study to qualitatively identify the potential intervention strategies and then test their contextual relevance and potential effectiveness with a survey study using a large sample. We developed the survey instrument following a qualitative study and extensive review of the literature. Although we did not comprehensively determine the psychometric properties of the instrument, its face and content validity were evaluated through a series of expert consensus meetings. In addition, the instrument is found to have fairly good internal consistency reliability. The study is conducted in eight migration hotspot areas in Ethiopia which is likely to represent the context of the country.
This study has several limitations, which need to be noted. First, the crosssectional design of the study makes it difficult to determine changes over time and causal relationship between background characteristics and endorsement of intervention strategies. Second, the endorsement of intervention strategies by survey respondents is just their opinion, which cannot be taken for granted. Third, although we carefully developed the survey instruments, being informed by a qualitative study and extensive review of the literature, we did not properly validate it. The participants in our qualitative study were parents, teachers, students, government representatives and community and religious leaders. For the quantitative study, we included parents, teachers, and students. The findings we reported in this paper cannot, therefore, be generalized to other stakeholders, such as the business community and migrants or migrant returnees. We believe that there are limited possibilities of teacher and parent respondents to connect with local businesses in order to improve the students' labor market-related competencies. Finally, although our sample (both in the qualitative study and in the survey study) was heterogeneous, we were not able to include migrant returnees.

Conclusions
In this mixed methods study, we identified contextually relevant intervention strategies that would help to reduce unsafe youth migration and improve migration outcomes in Ethiopia. These specific interventions are related to creating awareness and bringing attitudinal/behavioral changes, availing job opportunities and providing vocational and life skills training, making the government and parents fulfill their roles, and improving the education system and law enforcement and proper management of migration. Most of these intervention strategies are endorsed by more than 80% of survey respondents in terms of their contextual relevance and potential effectiveness. There is a need to prioritize and systematize these intervention strategies and evaluate their feasibility and acceptability using qualitative studies. Experimental studies to test for the effectiveness of the prioritized and systematized intervention strategies are also warranted. There is also need for further research and more funding on how to improve the education system in Ethiopia and relate it to the demands of the labor market.
& What proportion of the youth in your community do you think is migrating or planning to migrate through illegal means?