Profiles of Future Expectations Among Urban Adolescents in Cambodia

Abstract

Adolescents are tasked with navigating competing priorities, including whether to marry, have children, pursue a job/career, go to college, and contribute to society. The developmental task of building expectations for the future is especially complex for Cambodian adolescents living within a society that strongly prioritizes family obligations, yet increasingly provides educational and professional opportunities. The current study, guided by Seginer’s (2003) future orientation model, applied latent profile analysis (LPA) to explore patterns of Cambodian adolescents’ (N=580, 64% female, M_age=15.85) future expectations across key life domains and predictors of those patterns. LPA identified four profiles: Low Expectancy (low expectations across all domains; 12%), Family Focused (high expectations to get married and have children; 31%), Professional/Service Focused (high expectations across education, employment, and societal contribution domains; 27%), and High Expectancy (high expectations across all domains; 30%). Females were more likely than males to be in the Professional/Service Focused than High Expectancy profile. Adolescents with greater internal locus of control and family obligation were less likely to be in the Low Expectancy and Family Focused than High Expectancy profile, whereas adolescents in higher grade levels were more likely to be in the Family Focused than High Expectancy profile. Adolescents with closer relationships with mothers were less likely to be in the Professional/Service Focused than High Expectancy profile; adolescents with closer relationships with fathers were more likely to be in the Professional/Service Focused than High and Low Expectancy profiles. Findings elucidate configurations of adolescents’ future expectations, and factors distinguishing among adolescents with different configurations.

Research on adolescent development is largely based on US and Western European samples. Less is known about how adolescents make decisions about their futures in developing countries, especially those that are experiencing both extreme poverty and rapid economic and cultural shifts. Cambodia is such a country. Many living in Cambodia experience poverty, economic inequality, and health and wellbeing challenges following the Khmer Rouge Genocide that occurred roughly 40 years ago (Seponski et al., 2019). Despite these extreme difficulties, life in Cambodia is rapidly changing. Economic growth has created more and different work opportunities, especially for women (Heuveline & Hong, 2016) and those living in areas such as Phnom Penh and Siem Reap that have been more affected by economic change (National Institute of Statistics, 2015). With new employment opportunities there is a greater emphasis on improving access to, and quality of, secondary and higher education (Ogisu & Williams, 2016). However, these changes also create uncertainty and potentially pull at the fabric of traditional social structures that place older generations in positions of respect and authority (Peou & Zinn, 2015). This shifting economic, educational, and cultural landscape has important implications for adolescents as they imagine and make decisions about their futures.

With these challenges and changes in mind, the current study examined patterns of future expectations across multiple life domains and predictors of those patterns among adolescents from the Siem Reap Province in Northwest Cambodia. The first goal was to examine profiles of future expectations using data on adolescents’ perceived likelihood of achieving familial, educational, vocational, and community-oriented aspirations. We used latent profile analysis (LPA; Collins & Lanza, 2010), a person-oriented method, to explore the premise that expectations across these key life domains form in conjunction with one another, as expectations in one domain (e.g., familial) may have implications for another (e.g., vocational). Expectations across these domains will be weighed to different degrees and in different combinations by different adolescents. However, expectations across domains may cluster in meaningful ways, such that there are subgroups of adolescents who demonstrate similar future expectation patterns. The second goal was to identify factors associated with adolescents’ odds of possessing particular constellations of future expectations. We considered demographic characteristics (e.g., adolescent gender, current grade level, parent education), as well as adolescents’ internal locus of control, sense of family obligation, and relational closeness with mothers and fathers.

Prior to setting out the study’s specific hypotheses, we describe the conceptual foundation and review literature on how Cambodian adolescents navigate developmental decisions and how Cambodia’s historical and contemporary contexts impact these decisions. In cases where pertinent Cambodian findings are not available, we rely upon research from countries other than Cambodia. These instances are noted in the text.

Conceptual Framework

The current study examined a culturally-informed conceptual model of future orientation. Future orientation is the model or image one consciously develops regarding the future (Seginer, 2003). Seginer and colleagues (Seginer, 1995, 2000, 2003, 2005) developed a three-component model of future orientation: (1) motivation, emphasizing individuals’ tendencies to see value in, expect successful outcomes from, and feel a sense of internal control over the process of achieving future goals (Carver & Scheier, 2001, 2002; Seginer, 2003, 2009); (2) cognition, consisting of plans, hopes, and fears related to future goals; and (3) behavior, consisting of exploring ways to achieve one’s goals by researching, seeking advice, and committing to a specific option (Seginer, 2003, 2009). Taken together, one’s future orientation provides a foundation for aspirations/expectations (motivational), goal setting and planning (cognitive), and exploring options and making commitments (behavioral). Our analyses largely examine the motivational component of Seginer’s model (i.e., future aspirations/expectations), which is theorized to co-occur with cognitive and behavioral components as a necessary ingredient (Seginer, 2003, 2009). Thus, identifying expectations for the future during adolescence across the various domains of life examined herein may be useful for understanding (and potentially intervening upon) adolescents’ future plans and actions.

Importantly, individuals in cultures higher on collectivism (e.g., many Eastern Asian countries such as Cambodia) tend to be less concerned with self-focused expectations and more concerned with community-related needs; the reverse is true in more individualistic cultures (Seginer, 2008; Shepard & Turner, 2019). That adolescents from different cultural backgrounds tend to select different tradeoffs between self- and other-focused goals (i.e., these domains co-occur within individuals to varying degrees) underscores the interdependence of these domains. Nonetheless, future expectations related to education, vocation, family, and community/society are common to adolescents across different cultural backgrounds (e.g., Nurmi et al., 1995; Seginer & Halabi-Kheir, 1998; Trommsdorff, 1983).

Cambodian Adolescents’ Future Expectations

Guided by Seginer’s model of future orientation, adolescents’ future expectations serve as precursors of corresponding attainment in later adolescence and adulthood (at least in the US and UK, Eccles et al., 1998; Mello, 2008; Schoon, 2001; for a review see Yamamoto & Holloway, 2010). For example, US high school students who aspire to professional/managerial occupations or have generally positive future expectations are more involved in school (Dubow et al., 2001), more likely to complete high school (Powers & Wojtkiewicz, 2004), more protected against longterm effects of stress on self-competence (Wyman et al., 1993), and have stronger self-regulation abilities (Schmid et al., 2011). These broader developmental benefits may be important not only for educational and career attainment, but also for future family relationships and societal contributions.

Educational and Vocational Aspirations

Across history, options for most Cambodian youth have been limited; education beyond elementary school was rare, and early marriage and parenthood were obligatory (Peou & Zinn, 2015). Immediately prior to the genocide, due to expanded options for education beyond childhood and career opportunities, schooling began to become a viable pathway toward self-improvement and socioeconomic mobility (Ogisu & Williams, 2016). Following the genocide and the 10-year Vietnamese occupation that followed, the economic and educational situation of Cambodia has dramatically improved (Peou & Zinn, 2015). During the 1990s, for instance, Cambodia’s gross domestic product (GDP) grew rapidly, with growth increasing between 4.1% and 9.2% annually (see Heuveline & Hong, 2016). School enrollment also increased. By 2014, over 85% of both girls and boys between the ages of 6 to 14 and 43% of adolescents between the ages of 13 to 18 were enrolled in school. Due to these changes, by 2014 40% of women and 52% of men had secondary or higher education (National Institute of Statistics, 2015). Mirroring the changes in education, by 2014 80% of women were employed (National Institute of Statistics, 2015) and job opportunities had expanded to include manufacturing, banking, education, technology and engineering (see Heuveline & Hong, 2016; Ogisu & Williams, 2016). Prior to COVID, the trajectory of these changes had continued and present many Cambodian youth with increasing opportunities to go to college and obtain stable, well-paying, and fulfilling careers.

Community-Oriented Aspirations

Yet, decisions about the future are not made in isolation from the country’s collectivist norms (Seginer, 2008). For example, adolescents might make future educational and occupational decisions based upon a desire to assist their communities or country as a whole. In Cambodia, obtaining financial security and a professional career are key components of girls’ educational persistence, but girls also hold “collective aspirations” such as desires to financially support their families, share knowledge with younger generations, and address community and societal challenges (Rogers & Anderson, 2019). Widespread societal issues arising from the genocide—anxiety, depression, post-traumatic stress disorder (Seponski et al., 2019), physical injuries/disabilities, and overall lack of an education system (Williams et al., 2016)—may motivate adolescents to contribute to their country’s healing. Some may pursue opportunities in medicine to treat pervasive physical and mental health challenges, while others may engage in education or nonprofit sectors to improve the quality of education for future generations. Both would play a role in the country’s recovery process. Thus, educational and occupational goals can be intertwined with collective aspirations, at least among Cambodian females.

Familial Aspirations

Family considerations and traditions such as marriage, parenthood, and religion/spirituality remain salient to young people’s future decision making (Peou & Zinn, 2015). Policies instituted by the Khmer Rouge to undermine customary marital arrangements by parents (see Heuveline & Hong, 2016) had a lasting impact on marriage practices in Cambodia (LeVine, 2010). Pre-marital cohabitation is becoming slightly more common in Phnom Penh and potentially other globalized Cambodian cities (Heuveline & Hong, 2017). Despite some shifting from traditional norms, divorce remains rare (Heuveline & Hong, 2017), and most young people plan to adhere to a traditional life course (Peou & Zinn, 2015). Additionally, religion and spirituality are traditionally very important (Lewis & Seponski, 2012). Young people may think about practicing and passing on religious/spiritual traditions to children as important aspects of their futures. In sum, adolescents in contemporary Cambodia must integrate emerging opportunities for the life course with traditional preferences in public and private spheres when forming future outlooks.

Predictors of Adolescent Future Expectations

Family Obligation and Relational Closeness with Mothers and Fathers

There are likely many factors influencing Cambodian adolescents’ future expectations. Primary among these may be family influences. Families are important not only for directly supporting adolescents with financial and cultural resources, but also for connecting them to social networks and providing important advice, encouragement, and emotional support (Peou & Zinn, 2015; Rogers & Anderson, 2019). High parental expectations and strong encouragement promote youth’s educational aspirations and persistence both in the US (Hossler & Stage, 1992) and Cambodia (Eng et al., 2017). Adolescent future expectations are also influenced by a sense of family obligation. Both Southeast Asian and Asian American parents can simultaneously expect children to support the family, such as through helping in the family business or at home with siblings or older family members (Mathews, 2000; Ogisu & Williams, 2016; Ramisetty-Mikler, 1993), and bring the family pride and honor by succeeding academically and in their chosen careers (House & Pinyuchon, 1998; Mathews, 2000). Adolescents can show great respect to their family by adhering to these wishes for the adolescent’s future success (House & Pinyuchon, 1998; Mathews, 2000) and by providing financial support as a result of that success (Rogers & Anderson, 2019). In sum, both support from and obligation to the family may have implications for adolescents’ future plans.

Demographic Differences in Adolescents ‘ Future Expectations

Factors such as adolescents’ gender, current grade level, and parents’ education can also impact adolescents’ future expectations. Cognitive-developmental changes (e.g., abstract reasoning) may lead to differences in how youth across grades 7-12 think about the future (e.g., Steinberg, 2005). Further, parental educational attainment has been positively linked with US parents’ educational expectations for their children (Hossler & Stage, 1992; Lawrence, 2014; Spera et al., 2009; Wood et al., 2007) as well as with Cambodian adolescents’ retention in school (No et al., 2012). However, Cambodian parents may support adolescents’ endeavors regardless of their own educations, especially given that the genocide destroyed the country’s educational system and severely limited educational opportunities for ensuing generations (Edwards et al., 2016; Rogers & Anderson, 2019).

Regarding the role of gender, there are many barriers to girls’ education in Cambodia, including disproportionately high engagement in domestic chores, lack of parental support, pressure to marry early, insufficient positive female role models, and constraining discourse surrounding girlhood/womanhood (e.g., the Chbap Srei; Rogers & Anderson, 2019). Yet findings on gender differences in educational attainment have been mixed. Some studies find girls are more likely to drop out of school (Keng, 2004), whereas others report no gender differences in school dropout/retention (No, et al., 2012; No et al., 2016; Zimmermann & Williams, 2016). Recent data suggest a trend toward greater gender equity; but disparities remain, particularly at higher education levels (Williams et al., 2016). Considering gender’s influence on the constellation of future expectations simultaneously in a person-centered fashion, rather than only on a single domain independently, may clarify how gender influences adolescents’ future expectations.

Locus of Control

An explicit part of Seginer’s motivational component of future orientation, locus of control (LOC) refers to both an individual’s tendency to attribute control to either oneself or external sources (e.g., fate, God, luck) and degree of self-efficacy in navigating life events (Yorkovsky & Zysberg, 2021). Those with strong internal LOC believe that life events result from their actions and future goals will be realized. LOC is highly correlated with positive future orientation among adolescents in Romania (Cazan & Dumitrescu, 2016) and higher levels of education and academic success among Canadian youth (Ratelle et al., 2007). Personal attributes related to LOC (e.g., self-determination, self-efficacy) are part of Cambodian girls’ educational persistence narratives (Edwards et al., 2016; Rogers & Anderson, 2019) and motivate university attendance in combination with family obligations (Eng et al., 2017). Thus, a sense of internal control may shape adolescents’ future-oriented thinking and expectations.

The Present Study

Researchers have conventionally employed variable-oriented analytic approaches to examine factors associated with adolescents’ future expectations in certain domains (e.g., education) independently of other domains (e.g., helping others; see Pastor et al., 2007). Yet, considering future expectations across multiple life domains independently, without regard to interdependence among domains, only tells part of the story of how adolescents develop holistic outlooks on their futures. Person-oriented approaches are needed that can describe distinct sub-populations in the sample—that is, groups of individuals whose members are similar to one another across multiple domains and different from members of other groups. The current study applied one such approach: latent profile analysis (LPA; Collins & Lanza, 2010). The first goal was to identify subgroups of adolescents characterized by similar patterns of future expectations across nine domains: marriage, parenthood, raising children in a religious tradition, career fulfillment, job stability, college attainment, contributing to the country, helping the country heal, and helping the community. The second goal was to determine whether gender, grade level, parent education, LOC, family obligation, and parent closeness predicted profile membership.

We expected to identify at least three adolescent profiles: (1) Low Expectancy (low expectations in all domains); (2) Family Focused (high expectations to get married and have children but low expectations in all professional- and community-oriented domains); and (3) High Expectancy (high expectations in all domains). We also had exploratory hypotheses for two additional potential profiles: a professionally-focused profile (high expectations to have a fulfilling career, job stability, and go to college), and a community-focused profile (high expectations to contribute to the country, help the country heal, and help the community). These hypotheses were based on the youth in Siem Reap increasingly having opportunities to pursue professional careers and give back to their communities. Yet at the same time, Cambodia’s collectivist norms emphasize family obligations, and many adolescents continue to strongly prioritize family considerations and traditions such as marriage and parenthood (Peou & Zinn, 2015). We treated analyses aimed at addressing our second study goal as exploratory given the lack of prior research to formulate hypotheses about how factors of interest relate to the clustering of future expectations within adolescents.

Methods

Participants and Procedures

The participants were 580 students (64% female), age 10-22 (M_age = 15.85, SD_age = 2.05) in grades 7-12 from six participating schools (two government-supported schools and four schools supported by a local nonprofit) in and around Siem Reap, Cambodia. Descriptive statistics for demographic characteristics are presented in Table 1. Of note, primary school was the most common level of education obtained by both mothers (49.5%) and fathers (36.3%); however, 20.1% of adolescents’ mothers, and 32.4% of adolescents’ fathers, had a high school education or higher. Additionally, 76.9% of adolescents lived with their biological mother, 69.1% lived with their biological father, and 65.6% lived with both biological parents. The majority (88.9%) lived at home full time.

Table 1.

Descriptive Statistics for Demographics and Predictors of Profile Membership

Variable	Frequency (valid %) or mean (SD)
Gender (predictor)
Female	348 (63.7%)
Male	198 (36.3%)
Missing	34
Grade (predictor)
7th	98 (16.9%)
8th	86 (14.9%)
9th	93 (16.1%)
10th	92 (15.9%)
11th	99 (17.1%)
12th	111 (19.2%)
Missing	1
Mother education
No schooling	72 (12.8%)
Primary school	278 (49.5%)
Secondary school	99 (17.6%)
High school	56 (10.0%)
Above high school	57 (10.1%)
Missing	18
Father education
No schooling	38 (6.8%)
Primary school	202 (36.3%)
Secondary school	136 (24.5%)
High school	91 (16.4%)
Above high school	89 (16.0%)
Missing	24
Locus of control (predictor)	M = 3.94 (SD = 0.71) Min = 1, Max = 5 Missing = 30
Family obligation (predictor)	M = 4.41 (SD = 0.46) Min = 2, Max = 5 Missing = 8
Mother closeness (predictor)	M = 3.40 (SD = 0.67) Min = 1, Max = 5 Missing = 13
Father closeness (predictor)	M = 3.18 (SD = 0.77) Min = 1, Max = 5 Missing = 25

Notes. Min = minimum; Max = maximum.

From the study’s beginning, we were aware of the many challenges associated with conducting research in a cross-cultural context. We partnered with a Cambodian non-governmental organization (NGO) working in the education sector in Siem Reap throughout all stages of the research process. The NGO was fundamental to gaining access to the participants, providing feedback on the design of survey instruments, and tailoring data collection procedures. An initial version of the survey was sent to the NGO for feedback and was adapted based upon their recommendations. We also sought feedback from NGO leaders on preliminary findings during a follow-up visit to Siem Reap. All procedures were designed to minimize researcher bias and implement the study in a way that was acceptable to all participating schools and students. Members of the university research team traveled to Siem Reap and visited each school individually to conduct in-person data collection using paper-and-pencil surveys. Participating students were selected with the help of administrators from each school while trying to recruit roughly equal numbers of students across gender and grade level. One member of the research team was critical in facilitating data collection across linguistic and cultural barriers. He had strong ties with the NGO, as well as personal experience growing up in Cambodia and professional experience conducting research in Cambodian schools. He explained the study purpose and requirements with the students, and gave them the opportunity to review the survey, ask questions, and accept or decline participation. Students were assured that participation was voluntary and anonymous. All survey items were translated into Khmer, the students’ native language, by translators found through a service recommended to us by our NGO partner. The translated survey items were carefully reviewed by the study’s fourth author to ensure accuracy. Students took the survey home to complete and returned it to school the next day. Compensation for study participation included a monetary donation to each participating school that provided access to their students, as well as small gifts to each participating student in the form of school supplies. The compensation was deemed appropriate by both the participating schools and university researchers, as it provided a small incentive without coercing participants. All procedures were approved by the Pennsylvania State University Institutional Review Board (Title: “Balancing the Past, Present, and Future: Cambodian Adolescents' Understanding of Education and Life Choices in a Rapidly Changing World”; Submission ID: STUDY00009251). This study was not preregistered. Study materials, data, and analysis code are not publicly available, but can be obtained by contacting the study’s first author.

Measures

An overview of study measures is provided in Appendix A.

Latent profile indicators

All nine indicators were single items that assessed adolescents’ expectations for their futures. Each indicator was presented with the stem, “When you get older, how likely is it that you will…” with specific items including, “get married”, “have children”, “pass on a religious or spiritual tradition to your own children”, “have a fulfilling career”, “have a stable and well-paying job”, “go to college”, “contribute to your country”, “help your country heal old wounds”, and “help others in your community”. The response scale ranged from 1 (Extremely unlikely) to 5 (Extremely likely).

Predictors of profile membership

Grade level.

Grade level was assessed with a single item ranging from 7^th - 12^th grade.

Parent Education.

Mother and father education were assessed separately with a single item each. The response options were no schooling, primary school, secondary school, high school, or above high school.

Gender.

Gender was assessed with a single item and was coded as 1=Female, 0=Male.

Locus of control.

Locus of control (LOC) was assessed using seven items about the extent to which adolescents felt like they had control over themselves and the things that happen to them (also sometimes referred to as "mastery"; Pearlin & Schooler, 1978; Rotter, 1966). The response scale ranged from 1 (Strongly disagree) to 5 (Strongly agree). The scale performed poorly, however, when negatively-worded items were included, so a single index was calculated as the average of the two positively-worded items only, with higher scores indicating stronger internal LOC. These items were: “What happens to me in the future mostly depends on me” and “I can do just about anything I really set my mind to”. The Pearson correlation between the two items was r = 0.31. Given that Cronbach’s alpha is impacted by the number of items in the scale, the Spearman-Brown correction was applied (see Eisinga et al., 2013), revealing that lengthening the scale by a factor of three would result in a reliability of 0.73 (assuming the same relations between variables).

Family Obligation.

Family obligation was assessed using 13 items about the extent to which adolescents felt a sense of obligation to show respect and provide future support to their families (Fuligni & Pedersen, 2002; Fuligni et al., 1999). This measure has two subscales, “respect for family” (seven items) and “future support” (six items). The respect for family subscale included items such as, “how important is it to you that you… ‘make sacrifices for your family’, ‘respect your older brothers and sisters’, and ‘do well for the sake of your family’”. The future support subscale included items such as, “how important is it to you that in the future you… ‘help your parents/adult caregivers financially in the future’, ‘spend time with your parents/adult caregivers even after you no longer live with them’, and ‘help take care of your brothers and sisters in the future’. Both subscales used response ratings ranging from 1 (Very unimportant) to 5 (Very important). Factor analysis indicated that all 13 items loaded strongly on a single factor. Thus, all items were averaged to create a single measure of family obligation, with higher scores indicating stronger beliefs about the importance of family obligation. The Cronbach’s alpha achieved by this measure was α = 0.82.

Mother closeness.

Mother closeness was assessed using five items about the extent of closeness and connection adolescents felt with their mothers, adapted from the trust and communication scales of the Inventory of Parent and Peer Relationships (IPPA; Armsden & Greenberg, 1987). The response scale ranged from 1 (Completely untrue) to 5 (Completely true). Following previous research (Fosco et al., 2016), a single index measuring the quality of adolescent’s relationships with their mothers was calculated as the average of all items, labeled closeness. Higher scores indicated stronger feelings of closeness. An example item is, “I tell my mother about my problems and troubles”. Reliability for this measure was α = 0.66.

Father closeness.

As with mother closeness, father closeness was assessed using the same five items about the extent of closeness and connection adolescents felt, this time with their fathers. The response scale ranged from 1 (Completely untrue) to 5 (Completely true). A single index was calculated as the average of all items, with higher scores indicating stronger feelings of closeness. An example item is, “My father encourages me to talk about my difficulties”. Reliability for this measure was α = 0.78.

Analysis Plan

The first analytic step was to identify and describe latent profiles of adolescent future expectations. The second step was to examine the characteristics of individuals belonging to each of the identified profiles by testing whether profile membership prevalence rates differed according to a number of covariates.

Rather than examining the effects of single variables between individuals, LPA is a type of finite mixture model that divides a population into mutually exclusive latent profiles across multiple characteristics within individuals (Collins & Lanza, 2010). The term “latent variable” is referring to the latent categorical variable of cluster membership. The goal of LPA is to identify clusters of observations that have similar values on profile indicators (Pastor et al., 2007). Two sets of parameters are particularly important in LPA: membership probabilities and item-response means (and variances). Latent profile membership probabilities describe the distribution of the profiles in the population. Item-response means (and variances) describe the profile-specific item means (and variances). The patterns of item means are used to interpret and name the profiles. Importantly, LPA assumes that indicators are normally distributed within classes (Bauer & Curran, 2003). Three indicators in the current study, “have fulfilling career”, “have stable job”, and “go to college”, were severely negatively skewed. The skewed distributions each had a clear point at which the indicator could be discretized to “high” (score of 5 on the Likert scale) and “low” (score of 4 or lower on the Likert scale) levels. These items were included as binary indicators in a “mixed” indicator LPA. All other indicators were standardized prior to analysis. Consistent with prior research (LoBraico et al., 2020), this approach allowed us to retain all important indicators while also increasing the plausibility of the model assumptions.

Item-response variances were restricted to be equal across profiles by default to improve model identification (e.g., Fosco & Bray, 2016; Lobraico et al., 2020). Model parameters were estimated using maximum likelihood (ML) estimation via the EM algorithm, a common method in LPA (Pastor et al., 2007). Models with 1-6 profiles were compared using the Akaike information criterion (AIC; Akaike, 1974), Bayesian information criterion (BIC; Schwarz, 1978), sample-size adjusted BIC (a-BIC; Sclove, 1987), entropy (Celeux & Soromenho, 1996), and a bootstrapped likelihood ratio test (BLRT; McLachlan, 1987; McLachlan & Peel, 2005), as well as model stability and interpretability. Lower values for the AIC, BIC, and a-BIC indicated better model fit; higher values for entropy indicated a higher degree of accuracy with which models classified individuals into their most likely class; significant bootstrapped likelihood ratio test p-values indicated a statistically significant improvement in fit for the inclusion of one more profile, compared to a neighboring model with one fewer profiles. Model selection involved consideration of the research question, fit indices, the utility and theoretical interpretation of each solution, and model parsimony (Bauer & Curran, 2003). Model identification for all models was checked with 1,500 initial stage starts and 500 final stage starts; all models were estimated using Mplus version 8 (Muthén & Muthén, 1998-2017).

After model selection, predictors of profile membership were examined by simultaneously adding adolescent gender, grade level, mother and father education, LOC, family obligation, and mother and father closeness to the model using baseline-category multinomial logistic regression (Vermunt, 2010; R3STEP command in Mplus), controlling for the non-independence of observations due to nesting within schools. Gender and grade level variables were left in their original metrics, while all other continuous predictors were standardized prior to model entry to facilitate interpretation. Only significant predictors were retained in the final model and reported on below. Predictors’ effects on profile membership are expressed as odds ratios describing the increase in odds of membership in a particular profile compared to the reference profile, for one-unit increases in the predictor. Any profile may be selected as the reference profile. Missing data for covariate analyses were handled using listwise deletion. The amount of missing data on all covariates did not exceed 6%.

The analysis sample for the first analytic step included n = 553 adolescents (of the original 580 adolescents) who provided data on at least one of the latent profile indicators. Of these, n = 486 provided data on all of the demographic covariates for the second analytic step. Research examining sample size requirements for latent variable models has found that required sample sizes range from 30 to 460 cases, depending on the type of model, number of factors/profiles, number of indicators, strength of the indicator loadings, and the amount of missing data per indicator (Wolf et al., 2013). This work suggests that our sample size was appropriate for conducting LPA analyses. Additionally, the same four profile model emerged when tested in the reduced sample, suggesting latent profile membership was not associated with missing data on covariates.

Results

Descriptive Statistics

Descriptive statistics for adolescent demographic characteristics and predictors of profile membership are shown in Table 1. Of note, adolescents’ sense of family obligation was very high on average (M = 4.41, SD = 0.46 on a five-point scale). Adolescents also felt relatively close to both mothers (M = 3.40, SD = 0.67) and fathers (M = 3.18, SD = 0.77) on average. Descriptive statistics for latent profile indicators are shown in Table 2. Expectations were highest for work and education indicators; 70.3% of adolescents indicated it was “extremely likely” that they would have a fulfilling career, 60.6% that they would have a stable job, and 55.1% that they would go to college. Expectations to help others were also high on average, as indicated by adolescents’ perceived likelihood that they would contribute to society (M = 4.09, SD = 0.93), help their community (M = 4.01, SD = 0.87), and help their country heal (M = 3.85, SD = 0.97). Family-related goals and expectations were slightly more moderate on average, as indicated by adolescents’ perceived likelihood that they would get married (M = 2.81, SD = 1.22), have children (M = 2.85, SD = 1.19), and raise children in a religious tradition (M = 3.36, SD = 1.24).

Table 2.

Descriptive Statistics for Profile Indicators

Indicator	Frequency (valid %) or mean (SD)	Min	Max	Skewness	Kurtosis
Get marrieda	2.81 (1.22)	1	5	−0.02	−1.03
Have childrena	2.85 (1.19)	1	5	−0.10	−0.99
Raise religious childrena	3.36 (1.24)	1	5	−0.35	−0.81
Contribute to countrya	4.09 (0.93)	1	5	−0.83	0.19
Help country heala	3.85 (0.97)	1	5	−0.63	0.03
Help communitya	4.01 (0.87)	1	5	−0.63	0.08
Have fulfilling careerb	382 (70.3%)	0	1	--	--
Have stable jobb	331 (60.6%)	0	1	--	--
Go to collegeb	300 (55.1%)	0	1	--	--

Notes. Min = minimum; Max = maximum.

^aIndicators were standardized for analysis

^bIndicators were binary. Frequency values for binary indicators are the frequency of the “high” value.

Primary Results

Model fit information and model selection criteria are shown in Table 3. The AIC, BIC, and a-BIC were not minimized for any of the models with 1-6 profiles. Additionally, the bootstrapped likelihood ratio test remained significant for each model. However, practical decrements in the fit criteria slowed around the 4-profile model. Entropy ranged from .82 to .85, with values for larger models in the mid .80s. Thus, we considered models with 4 or 5 profiles. Upon examination of these models, the 4-profile model had profiles that were all reasonably sized and substantively interpretable. One of these profiles was split into two similar profiles in the 5-profile model, suggesting extraction of an additional profile that was redundant and theoretically uninterpretable. Thus, we selected the 4-profile model for interpretation and additional analysis.

Table 3.

Model Fit Information for Latent Profile Analysis

No. of profiles	Log- likelihood	No. of free parameters	AIC	BIC	a-BIC	Entropy	BLRT
1	−5701.55	15	11433.10	11497.83	11450.21	--	--
2	−5286.95	25	10623.91	10731.79	10652.43	0.82	<.001
3	−5090.61	35	10251.21	10402.25	10291.15	0.83	<.001
4	−4993.34	45	10076.68	10270.87	10128.02	0.85	<.001
5	−4920.51	55	9951.01	10188.36	10013.76	0.84	<.001
6	−4863.39	65	9856.78	10137.28	9930.94	0.85	<.001

Parameter estimates, including within-profile item means (for continuous indicators) and probabilities (for binary indicators), are presented in Table 4. Standardization of the indicators prior to analysis facilitates interpretation of these parameters as the number of standard deviations above or below the overall mean. Profile 1 (12% prevalence) was characterized by lower than average values on all profile indicators; we labeled them Low Expectancy. Profile 2 (27%) was characterized by higher than average expectations to have a fulfilling career, have job stability, go to college, contribute to their country, help the country heal, and help others in the community, but lower than average expectations to get married, have children, and raise religious children; we labeled them Professional/Service Focused. Profile 3 (31%) was characterized by higher than average expectations to get married and have children but lower than average expectations in all other domains; we labeled them Family Focused. Profile 4 (30%) was characterized by higher than average expectations in all domains; we labeled them High Expectancy.

Table 4.

Parameter Estimates for Four-Profile Model

		Within-profile mean
		1. Low	2. Professional/ Service	3. Family	4. High
Profile prevalence	Item mean/ probability	Expectancy .12 (n = 67)	Focused .27 (n = 147)	Focused .31 (n = 173)	Expectancy .30 (n = 166)
Get married	0a	−0.956c	−1.011c	0.531d	0.734d
Have children	0a	−1.035c	−0.984c	0.483d	0.770d
Raise religious children	0a	−0.431c	−0.404c	−0.033	0.556d
Contribute to country	0a	−0.749c	0.605d	−0.826c	0.615d
Help country heal	0a	−0.599c	0.559d	−0.806c	0.575d
Help community	0a	−0.463c	0.450d	−0.764c	0.571d
Have fulfilling career	0.703b	0.450c	0.913d	0.462c	0.865d
Have stable job	0.606b	0.138c	0.885d	0.353c	0.807d
Go to college	0.551b	0.234c	0.903d	0.204c	0.724d

Notes.

^aIndicators were standardized for analysis

^bIndicators were binary.

^cStatistically significantly lower than the overall item mean at p < .05.

^dStatistically significantly higher than the overall item mean at p < .05.

We then examined adolescent characteristics as predictors of profile membership. We evaluated profile membership prediction using the High and Low Expectancy profiles as the reference groups, respectively. Results are presented in Table 5. Females had greater odds than males of expected membership in the Professional/Service Focused profile compared to the High Expectancy profile (OR = 2.02). Additionally, adolescents with higher internal LOC had lower odds than adolescents with lower internal LOC of expected membership in the Low Expectancy (OR = 0.39) and Family Focused (OR = 0.46) profiles compared to the High Expectancy profile, as well as higher odds of expected membership in the Professional/Service Focused (OR = 1.98) and High Expectancy (OR = 2.57) profiles compared to the Low Expectancy profile. A similar pattern was found for family obligation: adolescents with greater family obligation had lower odds than adolescents with lower family obligation of expected membership in the Low Expectancy (OR = 0.44), Professional/Service Focused (OR = 0.66), and Family Focused (OR = 0.44) profiles compared to the High Expectancy profile, as well as higher odds of expected membership in the High Expectancy (OR = 2.27) profile compared to the Low Expectancy profile. Further, adolescents with a greater sense of closeness with their mothers had lower odds than adolescents with a weaker sense of closeness with their mothers of expected membership in the Professional/Service Focused profile compared to the High Expectancy profile (OR = 0.64). In contrast, adolescents with a greater sense of closeness with their fathers had higher odds than those with a weaker sense of closeness with their fathers of expected membership in the Professional/Service Focused profile compared to High Expectancy (OR = 1.64) and Low Expectancy (OR = 1.86) profiles. Finally, adolescents in higher grade levels had greater odds than adolescents in lower grade levels of expected membership in the Family Focused profile compared to the High Expectancy profile (OR = 1.29).

Table 5.

Effects of Adolescent Characteristics on Profile Membership

	1. Low Expectancy	2. Professional/Service Focused	3. Family Focused	4. High Expectancy
Variable	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]
Gender	1.63 [0.66, 4.02]	2.02* [1.10, 3.73]	1.49 [0.79, 2.74]	ref
Locus of control	0.39* [0.25, 0.60]	0.77 [0.54, 1.09]	0.46* [0.32, 0.66]	ref
Family Obligation	0.44* [0.27, 0.71]	0.66* [0.44, 0.99]	0.44* [0.29, 0.68]	ref
Mother Closeness	0.80 [0.50, 1.29]	0.64* [0.43, 0.95]	0.74 [0.50, 1.09]	ref
Father Closeness	0.88 [0.52, 1.48]	1.64* [1.14, 2.37]	1.19 [0.84, 1.69]	ref
Grade	1.05 [0.77, 1.43]	1.15 [0.93, 1.42]	1.29* [1.03, 1.63]	ref
Gender	ref	1.24 [0.48, 3.25]	0.91 [0.37, 2.19]	0.61 [0.25, 1.52]
Locus of control	ref	1.98* [1.31, 2.98]	1.17 [0.84, 1.65]	2.57* [1.68, 3.92]
Family Obligation	ref	1.50 [0.99, 2.26]	1.01 [0.71, 1.44]	2.27* [1.41, 3.64]
Mother Closeness	ref	0.80 [0.48, 1.31]	0.92 [0.58, 1.46]	1.24 [0.77, 2.00]
Father Closeness	ref	1.86* [1.06, 3.28]	1.35 [0.81, 2.25]	1.13 [0.67, 1.91]
Grade	ref	1.09 [0.80, 1.49]	1.23 [0.91, 1.67]	0.95 [0.70, 1.29]

Notes.

^*p < .05. OR = Odds ratio. CI = Confidence Interval. Odds ratios were obtained by exponentiating beta coefficients. CIs were calculated as the beta coefficient +/− (1.96(SE)), and were then exponentiated to obtain 95% CIs for the ORs.

Discussion

Consistent with Seginer’s (2003, 2009) model of future orientation, adolescents form priorities across multiple life domains, including whether to marry, have children, pursue a job or career, go to college, and contribute to society. This developmental task is especially complex for urban adolescents in Cambodia given that they live within a society that strongly prioritizes family obligations and religious traditions, and yet increasingly provides educational and professional opportunities. Guided by the notion that decisions across these key life domains are interdependent, the current study applied latent profile analysis (LPA) to examine patterns of adolescents’ future expectations and predictors of those patterns. Four profiles were identified: Low Expectancy (12%), Professional/Service Focused (27%), Family Focused (31%), and High Expectancy (30%). Adolescent characteristics, including gender, locus of control, family obligation, grade level, and closeness with mothers and fathers, differentiated among adolescents in different subgroups with distinct constellations of future expectations. The findings from this study provide important insights into the different configurations of expectations that adolescents living in an urban area of Cambodia hold for their futures, and the factors distinguishing among adolescents with different configurations.

Cambodian Adolescents’ Future Expectations

Consistent with hypotheses, we identified High Expectancy, Low Expectancy, and Family Focused profiles. The size of the High Expectancy profile is particularly noteworthy, indicating that nearly one-third of adolescents in the sample were most probably classified as anticipating that they would successfully balance multiple competing priorities across different life domains. In contrast, 12% of adolescents were most likely classified as Low Expectancy. Considering that adolescents’ future expectations are important precursors of corresponding attainment in later adolescence and adulthood (e.g., Eccles et al., 1998; Mello, 2008; Schoon, 2001; Yamamoto & Holloway, 2010), the Low Expectancy profile represents a small but meaningful subgroup who may be particularly at-risk for poor future outcomes. It is therefore important to identify the adolescents in this group and address the underlying factors contributing to their low expectations.

The third subgroup, Family Focused, was the largest profile representing 31% of the sample. Compared to other profiles, adolescents with expected classification in this subgroup were characterized by high expectations to get married and have children, but were similar to the Low Expectancy group in that they also had low expectations for pursuing goals such as higher education and job stability that would ultimately lead to social mobility. Collectively, nearly half of youth did not expect to pursue these long-term self-benefitting goals; for some (i.e., Low Expectancy group), this was due to a generally poor future expectation, while for others (i.e., Family Focused group), it was out of a sense of commitment or prioritization of their family. The size of the Family Focused profile suggests that despite shifting from traditional Cambodian norms and an increase in opportunities outside the home, family considerations such as marriage and parenthood remain central to young people’s future decision making (Peou & Zinn, 2015).

In contrast with the Family Focused profile, over a quarter of adolescents (27%) were classified as Professional/Service Focused. Similar to the High Expectancy profile, adolescents most probably classified in this subgroup had high expectations for a college education, career fulfillment, job stability, and helping others in the country and community; however, they were differentiated from the High Expectancy profile by having low expectations for getting married and having children. Taken together, over half of youth in the sample expected to pursue more individualistic goals. Some planned to focus solely on professional- and service-related goals (i.e., Professional/Service Focused subgroup), whereas others expected to balance these pursuits with marriage and parenthood (i.e., High Expectancy subgroup). The existence and size of the Professional/Service Focused subgroup likely reflects both the expanding formal education and career possibilities in Cambodia (Heuveline & Hong, 2016; National Institute of Statistics, 2015) as well as the fact that these adolescents lived in an urban area where the growth of such possibilities has been largely concentrated (Ogisu & Williams, 2016). These findings partially supported the exploratory hypotheses that we would find a professionally-focused profile and a service/community-focused profile. Perhaps most interesting is the fact that these two hypothesized profiles were actually combined into a single profile. The importance that Cambodian adolescents in this subgroup place on going to college and pursuing a stable job or fulfilling career appears intertwined with the importance of helping their country and communities. These findings offer quantitative support to previous qualitative research which has found that financial security and a professional career were key parts of Cambodian girls’ future goals, but that they also intended to use their skills and resources to financially support their families, assist their communities, and address societal issues (Rogers & Anderson, 2019).

Predictors of Adolescent Future Expectations

Understanding the factors that underlie adolescents’ patterns of expectations may help to illuminate part of the process involved in forming a holistic future outlook within the context of Cambodian society. Contrary to expectations, mother and father education were not associated with profile membership when all other predictors were considered, and these variables were thus removed from the final model. The lack of a significant effect may suggest, consistent with prior research in Cambodia, that parents encourage adolescents in their future endeavors regardless of the education they themselves obtained (Rogers & Anderson, 2019). It is also likely that the genocide’s targeting of highly educated citizens has affected the distribution of education among the parental generation. In contrast to parent education, adolescent gender, locus of control, family obligation, current grade level, and parent-adolescent relationship closeness were all uniquely associated with profile membership.

First, females were more likely than males to be in the Professional/Service Focused profile relative to the High Expectancy profile. It may initially seem counterintuitive that females would have a higher probability than males of belonging to a subgroup characterized by a stronger emphasis on educational and professional expectations, given traditional attitudes about the roles that girls and women should occupy in society (Yang & Thai, 2017) and the many barriers to girls’ educational and professional achievements (Rogers & Anderson, 2019). However, it is interesting to note that in prior work, girls have been found to place greater value than boys on higher education, work, and career across different ethnicities and cultural backgrounds (Seginer, 2009). When considered in light of this profile being characterized by a combination of education/career and service/community indicators, it may be that girls are more likely than boys to have educational and career expectations that are inextricably linked to helping others. That is, girls may be more motivated than boys by a desire to help others (as has been found in the US, Keenan & Shaw, 1997; Pursell et al., 2008), and may view education and career attainment as means by which to do so. The strong belief that girls need to support their family and other members of the community more broadly may also be contributing to this finding. Expectations placed on girls to provide disproportionately high amounts of support in the present (e.g., housework, sibling care, etc.) may pose barriers to educational and career trajectories, but could also promote continued achievement if educations and careers are viewed as means by which girls can provide additional support in the future.

Second, adolescents who had a higher (compared to lower) internal locus of control were less likely to be in Low Expectancy or Family Focused profiles relative to the High Expectancy profile, and more likely to be in the Professional/Service Focused and High Expectancy profiles compared to the Low Expectancy profile. Locus of control did not distinguish between Professional/Service and High Expectancy profiles. These findings align with Seginer’s conceptual model of future orientation and empirical research on the importance of perceived control over decision-making, which has generally found that a greater sense of control is related to higher levels of academic success among Canadian youth (Ratelle et al., 2007) and educational persistence among Cambodian youth (Edwards et al., 2016; Eng et al., 2017). Moreover, these findings underscore the importance of considering not only how external forces (e.g., gender norms, obligation to family) may shape or motivate adolescents’ expectations for their futures, but also adolescents’ own self-determination and sense of control over making their own decisions. A strong internal locus of control may be an important characteristic of adolescents with higher future expectations in work, education, and service domains that could ultimately lead to social mobility in the long-term.

Third, family obligation was a strongly held value by Cambodian adolescents overall, yet similar to findings for locus of control, adolescents with greater (compared to lesser) family obligation were less likely to be in the Low Expectancy, Professional/Service, and Family Focused profiles relative to the High Expectancy profile. It could be that having a stronger sense of obligation to respect and support their family gives adolescents a purpose that is motivating their future plans, and thus acts as a protective factor against having low expectations for one’s future across all domains. Further, although initially surprising that adolescents with greater family obligation would be less likely to be in the Family Focused profile relative to High Expectancy, it appears that family obligation promotes an orientation toward balancing professional expectations and societal contributions alongside marriage and parenthood, rather than focusing exclusively on either domain independently. Adolescents with high family obligation likely feel a stronger need to provide financial support to their families in the future, and may intend to do so by securing an income as a result of their educations and careers (e.g., Ramisetty-Mikler, 1993). Additionally, it may be that adolescents hope to bring pride and honor to their families by working hard academically and succeeding in their chosen careers while simultaneously marrying and having a family of their own (e.g., House & Pinyuchon, 1998).

Fourth, adolescents in higher (compared to lower) grade levels were more likely to be in the Family Focused profile compared to the High Expectancy profile. Although adolescents in more advanced grade levels might be expected to have higher expectations in education and career domains, there could be a number of interrelated factors contributing to their increased likelihood of prioritizing family concerns. For example, they may not have financial or other resources to continue pursuing educational, career, and service goals, they may have younger siblings who need the resources to continue their educations, or they may be more likely than younger adolescents to be needed at home (e.g., Ogisu & Williams, 2016).

Fifth, the degree of relationship closeness with mothers and fathers differentially distinguished between adolescents in High Expectancy and Professional/Service Focused profiles. Adolescents with greater (compared to lesser) closeness with mothers were less likely to be in the Professional/Service profile relative to High Expectancy, whereas adolescents with greater (compared to lesser) closeness with fathers were more likely to be in the Professional/Service profile relative to High Expectancy and Low Expectancy. Interpretation of these findings is more speculative; however, one possible interpretation could be that mothers and fathers have different desires for their children’s futures. Mothers may desire their children’s futures to be more balanced, such that they emphasize the importance of both family and professional/service domains of life. Adolescents who have closer relationships with mothers could develop similar values to their mothers and receive a greater investment of mothers’ time and resources, and ultimately form expectations for the future that align with their mother’s wishes. On the other hand, fathers may have a stronger desire for their children to go to college, obtain a job/career, and give back to their communities, and place less emphasis on the family-centered domains of marriage and childbearing. Having a closer relationship with fathers could thus facilitate adolescents being more likely to expect that they will attain professional/service goals, relative to having low future expectations across all domains or attaining a balance across personal-professional-communal domains.

Another possible explanation is social modeling (Bandura, 1977). Perhaps adolescents who are closer to their mother are more likely to want to model what she does (which is increasingly likely to include both family responsibilities and paid employment, see National Institute of Statistics, 2015; Rogers & Anderson, 2019), whereas those who are closer to their father are more likely to want to model what he does (which may include paid employment but fewer family responsibilities). In sum, the degree of parent-adolescent relationship closeness distinguished adolescents who had high expectations solely in professional/service-related domains from those who expected to balance these domains with marriage and parenthood.

Limitations

It is important to note a few limitations. First, the sample was accessed through a collaboration with a single NGO working in and around Siem Reap, Cambodia, the second-largest city in a country with a population that remains roughly 75% rural (United Nations, 2018). Adolescents living in and around Siem Reap have a relatively high level of resources and opportunities, as exemplified by the high percentages who expected to go to college and obtain professional and well-paying jobs. This segment of Cambodian society is not representative of the country as a whole. The life experiences and future expectations of adolescents living in rural and remote areas of Cambodia are likely very different, making results unlikely to generalize beyond the urban geographical context. Nonetheless, the youth in this study are among the first to negotiate substantially greater access to educational and professional opportunities against a backdrop of strong religious and family traditions during the adolescent period, given the country’s recent economic growth. Thus, despite limits to generalizability, Siem Reap’s unique combination of rich tradition and rapid change provides an important look at the potential implications of such rapid economic and social change during a critical developmental period for adolescents’ future orientation. Future research should build on this work by determining if samples drawn from rural and remote areas provide similar results.

Second, the measures used for this study were existing scales developed in English, previously validated on mainly US and Western European samples, and translated into Khmer. They were not validated with Cambodian samples or developed with Cambodian youth in mind. As such, not only may items not have been interpreted in the same fashion as they would by Western youth, but it may also be the case that the constructs these Western-validated scales assess do not exist in entirely the same fashion in Cambodian culture. Either of these issues could have led to the pattern of results we observed among the LOC items. This limitation reflects the dearth of research on adolescent development in developing countries. Results should therefore be interpreted with caution, and future work should attempt to validate these and other measures for youth in Cambodia specifically and other developing nations more broadly. Further, our measures were limited to adolescent self-report only. Interpretation of some findings, such as the role of perceived closeness with mothers and fathers and other family processes, would be strengthened by assessing parents’ expectations for their adolescent’s future and examining the degree of similarity/dissimilarity with adolescents’ expectations.

Finally, the profile indicators were all presented with the stem, “when you get older…,” limiting our ability to disentangle when and in what order adolescents expected to attain certain milestones. Relatedly, the data were cross-sectional. Future work should collect longitudinal data on Cambodian adolescents to determine whether their future expectations change over the course of development, or whether the expectations formed during adolescence predict later attainment.

Conclusion

Our person-oriented analyses offer insight into the complex nature of future planning processes among urban adolescents in Cambodia and elucidate several factors underpinning differences and similarities across configurations of future expectations. Future work should build on these findings by replicating with rural samples and following adolescents longitudinally to examine whether future expectations formed during adolescence predict later attainment. An understanding of subgroups can be used to more effectively meet the needs of adolescents with different sets of plans and expectations for their futures.

Appendix A. Study Measures.

Measure	Response Scale	No. of items	Example Item
Latent Profile Indicators
Get married	Extremely unlikely – Extremely likely	1	“When you get older, how likely is it that you will get married?”
Have children	Extremely unlikely – Extremely likely	1	“When you get older, how likely is it that you will have children?”
Raise religious children	Extremely unlikely – Extremely likely	1	“When you get older, how likely is it that you will pass on a religious or spiritual tradition to your own children?”
Contribute to country	Extremely unlikely – Extremely likely	1	“When you get older, how likely is it that you will contribute to your country?”
Help country heal	Extremely unlikely – Extremely likely	1	“When you get older, how likely is it that you will help your country heal old wounds?”
Help community	Extremely unlikely – Extremely likely	1	“When you get older, how likely is it that you will help others in your community?”
Have fulfilling career	Extremely unlikely – Extremely likely	1	“When you get older, how likely is it that you will have a fulfilling career?”
Have stable job	Extremely unlikely – Extremely likely	1	“When you get older, how likely is it that you will have a stable and well-paying job?”
Go to college	Extremely unlikely – Extremely likely	1	“When you get older, how likely is it that you will go to college?”
Predictors of Profile Membership
Gender	Girl/Boy	1	“Gender (tick one): Girl/Boy”
Grade	Open-ended	1	“Grade [Fill in]”
Mother education	No schooling – Above high school	1	“What is the level of education of your mother?”
Father education	No schooling – Above high school	1	“What is the level of education of your father?”
Locus of control	Strongly disagree – Strongly agree	2	“What happens to me in the future mostly depends on me”
Family obligation	Very unimportant – Very important	13	“In general, how important is it that you do well for the sake of your family?”
Mother closeness	Completely untrue – Completely true	5	“I tell my mother about my problems and troubles”
Father closeness	Completely untrue – Completely true	5	“My father encourages me to talk about my difficulties”