Childrens well-being in diverse migration contexts: Goals, design - - PDF document

Children’s well-being in diverse migration contexts: Goals, design and preliminary findings from the FAMELO project. Jennifer E. Glick, Victor Agadjanian, Natalie D. Eggum-Wilkens Dirgha Ghimire, Sarah Hayford, Carlos Santos, Scott Yabiku Abstract: Migration is an increasingly prevalent demographic behavior that has important consequences for families and communities around the world. Families and households play a central role in shaping migration decisions; in turn, migrants can produce important economic returns to the households from which they originate. Both migration decisions and eventual remittances have important implications for children’s development and future opportunities. Yet, we know comparatively little about the dynamic role migration may play in children’s lives. A core challenge in understanding commonalities and differences in the way family migration context is linked to children’s development is the difficulty in comparing associations across studies that use different definitions of migration and different conceptualizations of children’s development. The Family Migration Context and Early Life Outcomes (FAMELO) project begins to fill these gaps by conducting comparable longitudinal surveys of children and their caregivers in households with and without migrants in three traditional sending areas: Jalisco, Mexico; Gaza, Mozambique; and Chitwan, Nepal. This paper describes the conceptual framework, preliminary field work and initial analyses of pilot data collected for the FAMELO project. Introduction: Migration is an increasingly prevalent demographic behavior that has important consequences for families and communities around the world. Families and households play a central role in shaping migration decisions. In turn, migrants can produce significant economic returns to the households from which they originate. Both migration decisions and eventual remittances have consequences for left-behind children’s development and future opportunities. Yet, we know comparatively little about the dynamic role migration may play in children’s lives in origin

• communities. To better understand the cumulative and intergenerational impacts of migration,

research must go beyond a focus on the economic determinants and outcomes of migration to consider the ways familial migration itself alters children’s options, aspirations and subsequent pathways to adulthood. Of course, the impacts of migration on families and children left behind are most likely not

• universal. The diversity of family structures, motivations for migration, and human and social capital

embedded in migrant households along with the gender, religious, and ethnic stratification of their

1

• rigin societies may all play important roles in how children fare when migration occurs. But findings

from existing studies are remarkably mixed with some reporting that migration enhances children’s well-being and others reporting worse outcomes among children of migrants (Cebotari & Mazzucato, 2016; Creighton & Park, 2010; Hu, 2012; Jordan & Graham, 2012; Lee, 2011; Van Hook et al., 2012; Vogel & Korinek, 2012). Further, much of our understanding of children’s developmental trajectories are based in countries and communities with higher levels of infrastructure and development with far less attention to contexts with fewer resources and more constraints for children’s development (Bornstein et al., 2012). This paper reports on a new project just underway to compare these impacts among children in three very diverse contexts of migration: Jalisco, Mexico; Gaza Province, Mozambique and Chitwan District, Nepal. This new project is focused on addressing unanswered questions about migration’s influence on left-behind children’s social competence and adjustment problems, aspirations and plans, and the key transitions children make as they move into adolescence and early

• adulthood. This paper relies on existing data and new pilot data to demonstrate variations in children’s

schooling and activities as well as caregivers’ views of parenting and socio-emotional development among families engaged in migration and those without current migration from their households. These preliminary analyses have helped set the stage for the full “Family Migration and Early Life Outcomes” (FSMELO) study that is now underway in three focal settings. This paper first reports on the motivations for FAMELO and describes our focal settings. The paper then turns to results of preliminary analyses and pilot work. We conclude with a discussion of the lessons learned and a description of the full data collection activities now underway. Motivating FAMELO: Role of Migration on Children’s Well-Being. Existing studies of migration’s impacts on families and households report very mixed findings for returns to children with some studies reporting that migration enhances children’s well-being and others reporting worse outcomes among children of

• migrants. These mixed findings are likely domain dependent. Theoretically, there are reasons to

2 expect positive outcomes for children in migrant-sending households and families on some measures. Much of the early research on children in migrant-sending communities focused on the economic returns to households that send migrants with the expectation that migrant remittances can improve children’s access to resources. Indeed, research has generally accepted that positive economic returns to households tend to benefit children left behind in terms of improved nutrition or access to schooling More recently, research has pointed to both non-economic benefits and costs for migrant- sending households that may yield less positive results for children. Migrants and migrant networks can transmit information, values and ideas that can alter the expected timing of transitions throughout the life course (Levitt, 1998). These ‘social remittances’ will likely influence children’s own

• rientations towards education, family formation and their own migration. Migration changes social

norms, roles and expectations in participating households (Levitt & Jaworsky, 2007). These changing

• rientations may encourage higher aspirations for education or, alternatively, may reduce educational

commitment and encourage children to become migrants themselves (Kandel & Kao, 2001; McKenzie & Rapoport, 2011). Children may also experience negative outcomes. For example, children may be called upon to substitute for missing migrants’ labor and have less time to engage in schooling potentially negating the economic gains that make schooling more affordable. Parental absence or stress within migrant households could reduce child well-being and discourage children’s aspirations (Jordan & Graham, 2012). In short, the psychological and human capital investment consequences of migration can mean worse outcomes for children left behind. Part of the difficulty determining the extent to which migration yields positive or negative

• utcomes for left-behind children stems from the lack of comparative data and measures. We may

expect some similar relationships of migration and outcomes across settings. For example, socioeconomic gradients in child development appear across a variety of settings (Paxson & Schady, 2010). Thus, household assets and home environments may mediate the relationships between migration and outcomes in similar ways, regardless of the setting (Vogel & Korinek, 2012). There is also variation such that not all children may similarly benefit depending on age, gender and other

3 socially ascribed statuses. In Mexico, for example, there is little gender variation in school enrollment

• r educational attainment (Creighton & Park, 2010); thus, we would not expect significant variation in

the returns to migration on children’s schooling among boys and girls. In settings where gender differences in school access are large, migration may play a role in reducing gender gaps when economic returns accrue to origin households or children may benefit differentially from migration by gender (Vogel & Korinek, 2012). There are reasons to expect the role of migration on children’s well-being will also vary because contexts of development vary considerably in the resources available in the broader community and these can make a difference in the physical, cognitive and social development of children in the poorest families (Gauvain & Munroe, 2009; Paxson & Schady, 2010). A core challenge then in understanding commonalities and differences in the way family migration context is linked to children’s development is the difficulty in comparing across studies in different settings that use different definitions of migration and different conceptualizations of children’s development. Data needs. Increasing scientific interest in migration has resulted in more sophisticated surveys of migrants and their social networks, but few of these sources include indicators of child well-being or even the presence of children in origins and destinations (Fussell, 2004; Levitt & Jaworsky, 2007; Schoumaker & Digne, 2010; Wilson, 2009). The majority of studies of migration and ‘left behind children’ focus on a single sending and/or receiving context (Antman, 2012; Cebotari, Mazzucato & Appiah, 2017; Dreby, 2010). A few well-known comparative migration studies exist (notably, the Child Health and Migrant Parents in South-East Asia (CHAMPSEA), Migrations between Africa and Europe (MAFE) projects, and the Latin American Migration Project (LAMP)). These studies share many strengths but generally make comparisons within a geographic region (e.g., migration from sub-Saharan Africa to Europe; migration from Southeast Asia) and have limited measures of children’s development, educational transitions and progress, and transitions through adolescence. These limitations make it difficult to identify the importance of migration for children or the extent to which migration can alter

4 children’s home environments, their expectations and aspirations, and their own transitions to adulthood across diverse contexts. Another challenge is that much of our understanding of children’s developmental trajectories is based on research in countries and communities with relatively high levels of infrastructure and development with far less attention to contexts with fewer resources and more constraints for children’s development. Other sources focused on children’s development and home environments in comparative work in less-resourced settings tend to contain fewer details on migration. There is global recognition of a need for high-quality data on parental involvement and early learning in low-income countries (Frongillo et al., 2017; Sun et al., 2016). Research efforts such as the Multiple Indicator Cluster Survey (MICS) and The Young Lives Study focus on multiple domains of parenting, home environments and children’s health and development across a wide variety of countries including many low- and middle-income countries (Barnett et al., 2013; Bornstein, et al., 2012; MICS Methodological paper 2017; McCoy et al., 2016; Reynolds et al., 2017). These studies have indeed

• ffered new insights into the importance of home environments for children’s well-being. Yet, they

contain less detail on the experience of having migrants into and out of the household and, in the case

• f MICS, can provide only cross-sectional snapshots.

The missed results from studies of left-behind children and the need for comparable data across settings motivated the new FAMELO project. FAMELO is being designed to fill existing gaps in research by conducting comparable longitudinal surveys of children and their caregivers in households with and without migrants in three sending areas: Jalisco, Mexico; Gaza, Mozambique; and Chitwan, Nepal. The project covers several domains of child well-being: children’s social competence and adjustment problems; their educational aspirations and experiences; and their early family formation and transition to adulthood under different conditions of migration.

5 The FAMELO settings: The three settings for the FAMELO project were chosen to represent areas with high levels of adult out-migration that also differ substantially in economic development, family structure, cultural context, and public support for families and children. Mexico represents one example of a well- established migration system with economic and social ties to the United States that perpetuates international migration. Ironically, migration to the United States from Mexico has been decreasing and return migration is becoming increasingly relevant for families there. FAMELO will be able to consider this dynamic bi-directional familial migration context going far beyond a dichotomized view

• f migration.

The second focal area, Gaza Province in Mozambique represents the area with the lowest economic status and weakest institutional infrastructure to support families and children. The province has one of the highest HIV prevalence rates in the world, with estimated prevalence rates at around 25% of adults (Ministry of Health, 2010). Located in southern Mozambique along the border with South Africa, the area is predominantly rural and labor migration from the region to South Africa as well as other urban areas within Mozambique is quite prevalent. Chitwan district in Southern Nepal represents a middle context in terms of infrastructure when compared to the other FAMELO sites. Labor migration has become an increasingly important economic activity and households in Chitwan send migrants both internally to larger urban areas such as the capital, Kathmandu, as well as internationally. International migration increased rapidly during the late 1990s and 2000s (Wagle, 2012). At the same time, the Chitwan region of Nepal has transitioned rapidly from a setting of subsistence agriculture and limited infrastructure to a setting with sufficient resources and labor to promote and support international migration. One of the first tasks for FAMELO is to typify children’s environments and activities in each

• f the three focal settings. To begin, we analyzed existing data from each country available in the

MICS surveys available through UNICEF. The Multiple Indicator Cluster Survey, MICS, collects comparable basic indicators of child health and well-being (Bornstein, et al., 2012). In addition to questions about children’s physical health, recent rounds of MICS surveys have included measures of

6 children’s school enrollment, their involvement in home labor and chores, parenting practices and indicators of their home environments. These surveys are conducted in many countries including the three we focus on in FAMELO. The data are cross-sectional and, in most cases, are designed to be nationally representative. The FAMELO project focuses specifically on children between the ages of 5 and 17. MICS provides information on all children in the household with some items and modules specifically designed for children in these age groups as well. The most recent data available from MICS round 5 (MICS5) were collected in 2014 in Nepal and 2015 in Mexico. The most recent available MICS data from Mozambique were collected as part

• f round 3 (MICS3) in 2008. All surveys include similar information on children’s activities and the

resources available to children in their homes. This provides a good base upon which FAMELO can build additional items to contrast children’s well-being in each of the three countries. For MICS5, one child is randomly selected from the household at the time of the survey. For MICS3, information on school enrollment and chores are available for all children. One drawback of the MICS data is that there is only limited information on the employment and migration of individuals in the household. For our purposes, we can identify children in MICS5 (Mexico and Nepal) who have a parent living abroad. For MICS3 (Mozambique) we are limited to identifying children with a parent living outside of the focal household but we cannot determine whether they are living in another country. Nonetheless, the analyses across each country help give us some preliminary expectations going forward into the FAMELO project. One question about migration and children left behind is the extent to which children’s own activities are influenced by the absence of migrants from the household. MICS allows us to explore this, at least cross-sectionally, in several domains. First, we explore the association between current migration of a parent (i.e., at least one parent lives abroad at the time of the survey) and children’s school enrollment with an indicator of whether the child attended school during the previous academic year. Next, we consider the association between parental migration and children’s own contribution to household labor with caregiver reports of the number of household tasks performed and the number of hours children spent on these tasks in the previous week. We rely on a few control

7 variables to adjust for child characteristics (age and gender) and household resources (an index for household assets and urban vs. rural location). We anticipate finding consistent socioeconomic gradients for our outcomes within countries with better outcomes among children in better-resourced households (Reynolds et al., 2017). In the MICS5 sample, over 92% of the children age 5-17 in Mexico and Nepal attended school in the previous school year. In the MICS3 data available from Mozambique, around 72% of children age 5-17 were enrolled in school. The MICS allow us to consider the importance of parental migration (or parental absence in the case of Mozambique) while controlling for some key covariates known to be associated with differential access to and persistence in school. We control for child age, gender, household wealth and urban residence. Logistic regression models predicting school enrollment are presented in Table 1. [Table 1 about here] The logistic regression models predicting school enrollment suggest some consistent patterns across the countries. For example, in all three countries, school enrollment is more likely as children get older and among children from households with greater wealth. There is not much consistency in the association of having a parent living abroad, or having a parent absent in the case of Mozambique with school enrollment. In Nepal and Mexico, a parent living abroad is positively associated with school enrollment, although this is not statistically significant for children in Mexico but an absent parent is negatively associated with school enrollment in Mozambique. Migration may also be associated with children’s other activities including their contributions to household labor. Theoretically, an absent migrant may increase the need for children’s labor in the

• household. On the other hand, migrant households may see improved living standards, require fewer

hours of labor performed or have other competing activities for children like school attendance. The MICS data include questions about the tasks children performed for the household in the last week and the total number of hours spent on those tasks. Here, we compare the hours of labor performed last week in order to keep the measure as comparable as possible across the three settings. On average, children in Mexico performed 5 hours of labor in the household, children in Nepal

8 contributed 7 hours on average and children in Mozambique were reported to perform 12 hours of household chores in the previous week. In Table 2, we present OLS regression models predicting hours spent in household labor in the prior week controlling for the same variables as above: child gender, age, household wealth and urban location. These models also include a variable for whether the child was enrolled in school in the prior year. [Table 2 about here] Overall, these analyses again suggest some consistency in the characteristics that predict children’s household labor across the three focal settings. Girls tended to perform more hours of household tasks in the previous week than boys, and children enrolled in school in the previous year also spent less time on household tasks. In Nepal, wealth is also associated with household labor such that children in wealthier households spent less time on household tasks; this coefficient is much smaller and not significant in Mexico or Mozambique. In Mexico and Mozambique, having a currently migrating or absent parent is positively associated with children’s time spent on household chores when compared to children without a currently migrating parent. We do not observe this pattern in Nepal. These snapshots of children’s activities are not sufficient for drawing strong conclusions about migration’s role in children’s lives. But, the preliminary view afforded by the MICS data suggests that there are considerable similarities in how these activities vary by child age, gender and household wealth/resources across the three countries that are the focus of the FAMELO project. Unfortunately, MICS include minimal information on parenting (i.e., child discipline in Mexico [MICS5], Nepal [MICS5]), and no information on children’s social competence or adjustment problems for most children. New modules from recent rounds of MICS do offer some cross-sectional measures of very young children’s socioemotional and cognitive skills but do not include any similar measures for older children or allow researchers to follow trajectories over time (McCoy et al., 2016). To understand the degree to which migration is altering children’s home environments and development, we clearly need more in depth examination with children from a wider age range that can follow their transitions over time. The FAMELO project is designed specifically to do this.

9 FAMELO pilot data. The challenge facing FAMELO is constructing comparable measures of child outcomes in the three settings as well as comparable measures of migration. The first year of the project has been devoted to testing new measures that may improve on existing instruments developed for more- developed settings. The FAMELO team has been working to consider how such measures can be correctly constructed to be applicable across settings with various social norms and expectations for children’s development and transitions through adolescence. This was accomplished through a culturally-grounded effort that involved focus groups, interviews, consultation with local experts, extensive attention to translation efforts and cultural adaptation of instruments across contexts, among

• ther efforts described below.

To develop the full protocols for the FAMELO project, we conducted a pilot study in each of the three focal sites. Key informants and focus groups in each site helped inform the development of questionnaire items used for subsequent interviews with 60 adults who were identified as primary caregivers (usually mothers) for at least one child between the ages of 8 and 17. Adult caregivers provided basic demographic information about themselves and their households. They were also asked specific items designed to measure their perceptions of the focal child’s social competence and adjustment problems, as well as their own aspirations and expectations for the child as they enter

• adulthood. One focal child of each caregiver was also interviewed. The children were administered

the same instruments to measure social competence and adjustment problems. They were also asked about their own aspirations, expectations, attachment to schooling and other behaviors. For adolescents, items included questions about their own romantic relationships and life expectations. The interviews were conducted with a relatively small and geographically concentrated sample in each site. Caregivers provided information on their own characteristics, household composition and current migration from the household. They also were asked a series of questions about the focal child’s education, socioemotional development, aspirations, and expectations for the

• future. Table 3 summarizes the characteristics of caregivers who participated in the FAMELO pilot in

each of the three sites.

10 [Table 3 about here] The majority of caregivers interviewed in the pilot in all three sites are women (by design, interviewers selected mothers as respondents unless the mother was not available). Caregivers in Nepal are slightly younger than those in Mexico or Mozambique. There is also considerable variation in the educational distribution of caregivers across sites. In Mexico, caregivers have much higher levels of education with all reporting that they completed at least primary schooling and nearly half reporting completed secondary education. This is in contrast with Nepal where 21.7% have no formal schooling and the majority reporting having less than a School Leaving Certificate (SLC) which represents fewer than 10 years of formal schooling. In Mozambique, nearly half of caregivers reported that they had not completed primary school. Caregivers also reported a fairly high level of current migration from the household in all three sites in response to the question, “Is there anyone who usually shares meals or stays here in your household but is now away for school, work or somewhere to look for work?”. In the pilot study, 30% of households in Mexico, 53% of households in Nepal and 65% of households in Mozambique reported a current migrant away from the household. We do not further disaggregate reasons for migration or relationship to the migrant here but we recognize that this combines many different types of migration (internal and international) and reasons for being away. The FAMELO pilot study interviewed one child per caregiver in each of the focal sites. The sample was designed to include an even distribution of children age 8-17 and an equal number of boys and girls. Children reported on items designed to assess their socioemotional functioning, their educational aspirations and expectations as well as their own hopes and aspirations for their future transitions to adulthood. Table 4 reports on the characteristics of the children participants in the pilot

• study. Because the design of the pilot study targeted children of a specific age and gender distribution,

these do not vary across sites. We observe some small variation in the levels of schooling completed by the participating children with more children reportedly beyond lower secondary schooling in Mexico and Nepal when compared with Mozambique. However, almost all of the participating children were currently enrolled in school in all three sites. We expect to observe more variation in

11 this in the full FAMELO sample which will include more geographic diversity and a wider age range than the pilot study. [Table 4 about here] Socioemotional development. There is very little guidance in the international measurement of children’s socioemotional development. Therefore, substantial attention in the pilot study was devoted to creating appropriate measures for each of our three research sites. The focus groups revealed many commonalities (e.g., obedience, manners, getting along with peers), but some differences in expected behaviors for children in each site. These discussions then helped inform the items included in the pilot surveys. We also conducted ‘cognitive interviews’ with a subsample of children of different

• ages. These children were asked what they thought we meant by translated terms in their respective

language, and their responses illustrated whether a construct translation worked as expected or not. This attention to translation beyond the usual translation-back translation process, highlights the ways in which our team is dedicated to a culturally-grounded approach to assess developmental competencies among children in each of these international sites. Items that were not well understood by children in the cognitive interviews or did not translate were reworded or omitted from consideration in the creation of measures of socioemotional development. Items that had zero variability also were omitted. The remaining items were subjected to confirmatory factor analyses by construct (e.g., prosocial behavior, social skills, internalizing, externalizing) within reporter and site. Items with non-significant loadings were not automatically omitted because of model susceptibility to sampling variability (due to low sample sizes); instead, we only omitted items if loadings were poor/odd for multiple samples (i.e., child and adult within country). Results were compared across sites to arrive at a final set of common (across site) items, as well as a few unique items at each site. Broadly, social competence items assess agency/responsibility, deference/compliance, family collectivism, prosocial behavior, social harmony/manners, and socio-emotional skills/social problem

• solving. Adjustment problems assess delinquency, aggression, anxiety, depression, and withdrawal.

Using these pilot data, we will be able to examine differences in children’s social competence and adjustment problems by migration status (i.e., family migrant status, number of migrants, relation of

12 migrant to child) and by demographic variables (e.g., sex, education) within sites. Analyses of these data are ongoing.

• Education. Another goal of FAMELO is to understand how migration alters children’s educational

trajectories in sending communities. Overall, school enrollment is very high among all of the children participating in the pilot study with only two children per site reported as not attending school. However, with the pilot data, we can go beyond school enrollment and consider caregivers’ and children’s expectations of future education. The pilot interviews included questions about schooling, educational expectations (‘how far in school do you think this child/you will go’?) and aspirations (‘how far in school do you want this child/you to go’?) reported by caregivers and children. To compare across research sites, we convert levels of schooling specific to each country into International Standard Classification of Education (ISCED) categories (Silke, 2013). For the most part, adults and children have high aspirations and expectations in the pilot interviews in Mexico and Nepal with slightly lower aspirations and expectations in Mozambique as shown in Table 5. Concrete educational expectations tend to be a little lower than aspirations suggesting respondents do indeed distinguish a difference between desires and expectations. This is consistent with studies conducted in the United States and other countries. [Table 5 about here] The pilot samples are small so there is limited power to detect statistically significant

• variation. We investigated several bivariate associations between caregivers’ characteristics and their

aspirations and expectations for the children. We find caregivers’ expectations and aspirations for children are positively associated with the caregivers’ own education (p < .05) when the data are pooled for all three sites. Caregivers with higher educational aspirations for the child also have children with higher educational aspirations for themselves. These patterns are not confined to aspirations (i.e., how far in school adults and children wish the child to go) but hold for the more concrete educational expectations (i.e., how far in school adults and children think the child will actually go). There is a clear positive correlation between child and caregiver expectations (r = .34 in Mexico, r = .59 in Nepal and r = .42 in Mozambique).

13 Overall, we do not observe much variation in educational expectations depending on current migration from the household (ex: 84% of respondents in migrant households in Mexico expect children to achieve a completed secondary education or higher vs. 86% of caregivers in non-migrant households). Exploratory ordered logistic regression models predicting caregiver’s educational expectations for children with a pooled sample across the three countries (n = 173) suggest that adult education is an important predictor of expectations. And, expectations are lower in Mozambique and Mexico than Nepal with caregiver education controlled in the model. These initial explorations suggest school enrollment and educational expectations are high in all three sites regardless of migration. They also point to considerable variation in expectations across

• sites. The MICS data also suggested limited consistency in the role of migration across the three

countries with larger and more representative data. The full FAMELO survey will allow us to explore this in more depth with a larger sample that is representative of regions of high migration. For example, we will be able to consider not only current migration out of the household but prior migration and the age of children when they are first exposed to a family member’s migration out of the household. Plans for adulthood. The final focus of FAMELO is on transitions in adolescence that may be influenced by migration from the household. In the pilot study, children in each setting are asked about their own life goals in order to determine if we can identify variations in priorities for the transitions to adulthood among caregivers and children. The pilot asks, “If you could pick only one thing to do, which is the most important? Complete secondary school, have a salaried job, or get married?” Answers to these questions are presented in Table 6. [Table 6 about here] In Mexico, the majority of children (64%) report that completing secondary school is the most important of these goals. In Nepal, we observe a larger proportion of children report that getting a salaried job is the most important goal (55% vs. 28% in Mexico). However, another sizable group report that completing secondary school is the most important (45%). Far fewer children reported their life goals in Mozambique. Among those who did, responses were almost evenly split between

14 completing secondary school or getting a salaried job. Only a few children in Mexico reported that marriage was more important than a job or completing school; none reported this as a most important goal in Nepal and Mozambique. There is a modest socioeconomic gradient in children’s goals. Children whose caregivers have more education are more likely to report that going to school is the most important goal. As with educational expectations, however, life priorities do not vary much among children living in households with current migrants when compared to their peers in households without a current

• migrant. Multinomial logistic regression models predicting life goals (similar to those examined for

educational expectations) suggest adult education is associated with greater priority placed on

• schooling. In Mozambique greater priority was placed on paid work than Nepal or Mexico but there is

little that remains statistically significant in the model (not shown). Because the full FAMELO survey is designed to provide more detailed migration history information and a considerably larger sample, we expect to be able to detect more variation in the completed full dataset. Nonetheless, the pilot study gives us substantial confidence that the measures developed for the full survey are meaningful in all three contexts. The future of FAMELO: There is a great deal of interest in the well-being of children left in origin communities when family members migrate. Beyond the economic returns to origin households, migrants may influence the expectations and aspirations of children and the way parents interact and transmit their own expectations for children. There are few studies, however, that directly measure children’s social development or their own expectations in diverse origin communities. The FAMELO project seeks to fill that gap by collecting high-quality longitudinal data on children and families. The preliminary analyses described here demonstrate our progress toward constructing comparable measures in order to carry out the larger goals of this ambitious three-site study.

15 The FAMELO project is now interviewing children and their caregivers in households in all three focal sites: Jalisco, Mexico; Gaza, Mozambique; Chitwan, Nepal. The target sample for each location is 2,000 households. Interviews include detailed information on the timing and type of migration both into and out of the household. Caregivers report detailed information about the focal children’s social development, school engagement and learning activities as well as parenting

• practices. Children are also interviewed to report on their own development, expectations and goals

for their own futures. Another important part of the FAMELO project is the plan to conduct follow-up interviews with these same individuals in two years. This will provide a longitudinal picture of migration and child well-being as children move through key developmental transitions and help further our understanding of the different trajectories children may follow depending on how migration unfolds in their own familial lives. The preliminary analyses presented here suggest we can expect some similarities in the importance of gender and socioeconomic status on children’s opportunities but also some significant variation in children’s lives and development and the role of migration in these processes across these

• settings. There are several innovations included in the full survey based on the preliminary pilot study

that will provide a greater understanding of these processes. First, we have developed culturally appropriate items for our focal outcomes: detecting children’s socio-emotional functioning and adjustment problems, site appropriate measures about school enrollment but also school engagement, and hopes and plans for the future. All of these refinements will provide more detailed and comparable measures for future analyses of children’s well-being. Second, the full survey extends prior work by allowing greater consideration of how migration might alter children’s developmental trajectories. For example, previous research strongly supports the hypothesis that the age at which children experience migration of family members can be very important for the impact on their schooling, happiness and general well-being (Antman, 2012; Cebotari, Mazzucato, & Appiah, 2017). The FAMELO interviews include retrospective and prospective measures of migration and should allow a more complete analysis by migration timing and type. The first wave of data collection, in the field now, includes a full household roster and

16 information on the location of any household members currently away from home. Further, the timing

• f the current migration spell and the first migration spell will allow us to determine just when in the

child’s life migration occurred and the duration of the migrant’s absence from the child’s household. The roster also includes information on children’s relationship to each migrant and the types of migration children are experiencing from their households such as temporary internal labor migration

• r long-term international migration. We will be able to consider the importance of gender of

migrants as well as gender of the left-behind children. Finally, the full survey extends prior research to better match household migration dynamics to children’s development and school trajectories and provide more information on the mechanisms through which migration might alter these trajectories. For example, we include questions about remittances but also about communication with migrants and children’s own feelings about the

• migrant. This should give us a better opportunity to investigate how migration can be both a positive

impact (i.e. through increased financial support) and a negative impact (i.e. through emotional distress) for children. The FAMELO project is a timely contribution to the growing body of scholarship on left- behind children’s well-being. It combines the strengths of previous work capturing detailed information on migration with other prior work analyzing multiple domains of child and youth

• development. Findings from the full study should help elucidate the coexistence of both positive and

negative consequences of familial migration in children’s lives.

17 References: Antman, F. M. 2012. Gender, educational attainment and the impact of parental migration on children left behind. Journal of Population Economics, 25: 1187-1214. Barnett, I., Proochista, A., Petrou, S., Penny, M.E., Duc, L., Glab, S., Woldehanna, T., Escobal, J.A., Plugge, E., & Boyden, J. 2013. Cohort profile: The young lives study. International Journal

• f Epidemiology, 42: 701-08.

Bornstein, M.H., Brittlo, P.R., Nonoyama-Tarumi, Y., Ota, Y., Petrovic, O & Putnick, D.L. 2012. Child development in developing countries: introduction and methods. Child Development, 83: 16-31. Cebotari, V.; Mazzucato, V.; Appiah, E. 2017. A longitudinal analysis of well-being of Ghanaian children in transnational families. Child Development, Cebotari, V., & Mazzucato, V. 2016. Educational performance of children of migrant parents in Ghana, Nigeria and Angoal. Journal of Ethnic and Migration Studies, 42: 834-56. Creighton, M. & Park, Hyunjoon. 2010. Closing the gender gap: Six decades of reform in Mexican

• education. Comparative Education Review, 54: 513-537.

Dreby, J. 2010. Divided by borders: Mexican migrants and their children. Berkeley CA: The University of California Press. Frongillo, E.A.; Kulkarni, S., Basnet, S., de Castro, F. 2017. Family care behaviors and early childhood development in low- and middle-income countries. Journal of Child and Family Studies, DOI 10.1007/s10826-017-0816-3. Gauvain, M., Munroe, R.L. 2009. Contributions of societal modernity to cognitive development: A comparison of four cultures. Child Development, 80: 1628-1642. Jordan, L.P. & Graham. E. 2012. Resilience and well-being among children of migrant parents in South-East Asia. Child Development, 85(5): 1672-1688.

18 Kandel, W. & Kao, G. 2001. The impact of temporary labor migration on Mexican children’s educational aspirations and performance. International Migration Review, 35: 1205-1231. DOI 10.1111/j.1747-7379.2001.tb00058.x Levitt, P. & Jaworsky, B.N. 2007. Transnational migration studies: Past developments and future

• trends. Annual Review of Sociology, 33: 129-156.

McCoy DC, Peet ED, Ezzati M, Danaei G, Black MM, et al. 2017. Correction: Early Childhood Developmental Status in Low- and Middle-Income Countries: National, Regional, and Global Prevalence Estimates Using Predictive Modelling. PLOS Medicine 14(1): e1002233. https://doi.org/10.1371/journal.pmed.1002233 McKenzie D, Rapoport H. 2011. Can migration reduce educational attainment? Evidence from

• Mexico. J Popul Econ 24(4):1331–1358

Paxson, C. & Schady, N. 2010. Does money matter? The effects of cash transfers on child development in rural Ecuador. Economic Development and Cultural Change, 59: 187-229. Reynolds, S.A., Andersen, C., Behrman, J., Singh, A., Stein, A.D., Benny, Ll, et al. 2017. Disparities in children’s vocabulary and height in relation to household wealth and parental schooling: A longitudinal study in four low- and middle-income countries, SSM - Population Health, 2017, ISSN 2352-8273, http://dx.doi.org/10.1016/j.ssmph.2017.08.008. Silke L. Schneider, (2013), The International Standard Classification of Education 2011, in Gunn Elisabeth Birkelund (ed.) Class and Stratification Analysis (Comparative Social Research, Volume 30) Emerald Group Publishing Limited, pp.365 - 379 Sun, J., Liu, Y., Chen, E.E., Rao, N. & Liu, H. 2016. Factors related to parents’ engagement in cognitive and socio-emotional caregiving in developing countries: Results from MICS 3. Early Childhood Research Quarterly, 36: 21-31. Van Hook, J., Baker, E., Altman, C.E., & Frisco, M. 2012. Canaries in a coalmine: Immigration and

• verweight among Mexican-origin children in the US and Mexico. Social Science &

Medicine, 74: 125-134.

19 Vogel, A. & Korinek, K. 2012. Passing by the girls? Remittance allocation for educational expenditures and social inequality in Nepal’s households 2003-2004. International Migration Review, 46: 61-100.

20 Table 1: Logistic regression predicting the log-odds of school enrollment among children age 5-17 in Mexico, Nepal and Mozambique Mexico Nepal Mozambique At least one parent lives abroad (vs. neither parent lives abroad) .21 .37**

• .34**

Child characteristics Male (vs. Female) .08 .09 .19*** Age .33** .41*** 1.8*** Age squared

• .03***
• .03
• .08***

Household wealth index .43*** .17*** .42*** Urban household (vs. Rural location)

• .46***

.05 .25*** Intercept N 3.5 11,680 3.4 16,469

• 9.5

22,740 Source: MICS5 data for Nepal and Mexico; MICS3 data for Mozambique. Note: For Mozambique, parental absence includes parents living outside of the household who may or may not be living

• abroad. *p < .05, **p < .01, ***p < .001, two-tailed tests

21 Table 2: OLS regression predicting hours spent on household chores last week by children age 5-17 in Mexico, Nepal and Mozambique Mexico Nepal Mozambique At least one parent lives abroad (vs. neither parent lives abroad) 1.6***

• .05

.62*** Child characteristics Male (vs. Female)

• .61***
• .71***
• 2.0***

Age .22** .27** 1.2** Age squared .002

• .004
• .02

Enrolled in school (vs. not enrolled)

• 1.34***
• 1.07***

.42 Household wealth index

• .01
• .80***

.13 Urban household (vs. rural location)

• .02
• .80**
• 1.4*

N 11,680 16,469 12,922 R square .05 .07 .08 Source: MICS5 data for Nepal and Mexico; MICS3 data for Mozambique. Note: For Mozambique, parental absence includes parents living outside of the household who may or may not be living

• abroad. *p < .05, **p < .01, ***p < .001

22 Table 3. Characteristics of caregiver respondents, FAMELO pilot data: Mexico Nepal Mozambique Gender (%) Male 11.7 15.1 Female 88.3 100.0 84.9 Mean age (sd) 41.1 (12.8) 35.5 (5.7) 39.1 (17.0) Married or cohabiting (%) 91.7 96.7 58.2 Education (% completed ISCED level of education) No completed primary 21.7 49.1 Completed Primary 11.7 23.3 18.2 Completed Lower Secondary 31.7 16.7 12.7 Completed Upper Secondary 46.7 15.0 7.3 Post-secondary or more 10.0 1.7 1.8 Education unknown 10.9 Migrant Household (% of HH with at least one current migrant) 30.0 53.3 65.4 Source: FAMELO Project Pilot Study (2016-2017). Mexico n = 60; Nepal n = 60; Mozambique n = 55.

23 Table 4. Characteristics of child participants in the FAMELO pilot study by site Mexico Nepal Mozambique Gender (%) Male 50.0 50.0 49.1 Female 50.0 50.0 50.9 Mean age (sd) 12.5 (2.9) 12.5 (2.9) 12.8 (2.9) Education (% completed ISCED level of education) No completed primary 35.0 38.3 38.2 Completed Primary 30.0 33.3 45.5 Completed Lower Secondary 31.7 20.0 16.4 Completed Upper Secondary 1.7 8.3 Post-secondary or more Enrolled in school (%) 96.7 96.7 96.4 Source: FAMELO Project Pilot Study (2016-2017). Mexico n = 60; Nepal n = 60; Mozambique n = 55.

24 Table 5: Panel a: Caregiver educational expectations for children, Mexico, Nepal and Mozambique (%) Mexico Nepal Mozambique Complete Primary (ISCED-1) 1.7 Complete Lower Secondary (ISCED-2) 15.0 1.7 Complete Upper Secondary (ISCED-3) 31.7 35.0 61.8 College (first post-secondary ISCED-6) 48.3 21.7 23.6 Masters (ISCED-7) 3.3 16.7 1.8 Doctorate (ISCED-8) 1.7 23.3 1.8 Panel b: Children’s educational expectations for children, Mexico, Nepal and Mozambique (%) Mexico Nepal Mozambique Complete Primary (ISCED-1) 6.7 3.3 5.5 Complete Lower Secondary (ISCED-2) 5.0 13.3 Complete Upper Secondary (ISCED-3) 33.3 43.3 78.2 College (first post-secondary ISCED-6) 35.0 18.3 12.7 Masters (ISCED-7) 3.3 13.3 Doctorate (ISCED-8) 1.7 8.3 Source: FAMELO pilot project, Mexico (n = 60), Nepal (n = 60), Mozambique (n=55).

25 Table 6. Children’s life expectations, FAMELO pilot study (percent) Mexico Nepal Mozambique Most important goal: Complete Secondary School 64.1 45.2 29.1 Get a salary job 28.2 54.7 30.9 Marriage 7.7 No response/missing 40.0 Source: FAMELO pilot project, Mexico (n = 60), Nepal (n = 60), Mozambique (n=55).