Building an assessment of community-defined social-emotional competencies from the ground up in Tanzania.

Two studies were conducted in 2017 to investigate children's competencies seen as important by communities in Mtwara, Tanzania. Qualitative data from 95 parents (34 women) and 27 teachers (11 women) in Study 1 indicated that dimensions of social responsibility, such as obedience, were valued highly. In Study 2, the competencies of 477 children (245 girls), aged 4-13 years, were rated by teachers and parents. Factor analysis found the obedient factor explained the most variance in parent rating. In line with predictions, urban residence, parental socioeconomic status (SES), and parental education were all positively associated with ratings of curiosity, and parental SES was negatively associated with obedience and emotional regulation. Findings illustrate the need for culturally specific frameworks of social-emotional learning.

gross domestic product

social‐emotional learning

socioeconomic status

Western, Educated, Industrialized, Rich, and Democratic

Social‐emotional learning (SEL) is increasingly seen as important for children's academic success—in a review of studies largely from the United States (Durlak et al., 2011) as well as in Tanzania (Mulcahy‐Dunn et al., 2018)—and social adjustment. However, little scholarship has been devoted to SEL outside of Western, Educated, Industrialized, Rich, and Democratic (WEIRD) contexts (Henrich et al., 2010). This narrow perspective leads to concerns—common in the field of psychological science as a whole (Arnett, 2016; Rad et al., 2018)—that the current understanding of SEL is not representative of the human population. These concerns are particularly relevant to the continent of Africa, which is home to around a quarter of the world's children (United Nations, Department of Economic, & Social Affairs, Population Division, 2019) but is the subject of only 0.6% of articles in leading child development journals (Nielsen et al., 2017). The aim of this research was to identify and measure SEL competencies relevant to one sub‐Saharan Africa population—in southern Tanzania—through qualitative interviews and adult ratings of child behaviors. The research in Tanzania provided the opportunity to study SEL in two contrasting contexts: (1) rural areas with a long history of subsistence agriculture and with values that differ from WEIRD contexts, and (2) a growing urban, educated sector of society.

Current conceptualizations of SEL

Social‐emotional learning is defined as the “processes by which children and adults acquire and apply core competencies to recognize and manage emotions, set and achieve positive goals, appreciate the perspectives of others, establish and maintain supportive relationships, make responsible decisions, and handle personal and interpersonal situations constructively” (Osher et al., 2016, p. 645). The framework we use to characterize SEL (EASEL Laboratory, 2020; Jones et al. 2021) is based on Bailey and Jones’s (2019) model of the development of children's general regulatory skills, applied to three domains: (1) the cognitive domain, involving “skills required to manage and direct behavior toward the attainment of a goal” (Jones et al. 2021 p. 10); (2) the emotional domain, including “skills and competencies that help children recognize, express, and control their emotions as well as understand and empathize with others” (Jones et al. 2021, p. 11); (3) the social domain, including “social and interpersonal skills [that] support children and youth to accurately interpret other people's behavior, effectively navigate social situations, and interact positively with peers and adults” (Jones et al. 2021, p. 11). There is development from early childhood to adulthood in regulatory skills and how they are applied across the three domains. The application of skills and competencies to the three domains takes place in the context of a “belief ecology,” including three additional domains of values, perspectives, and identity. In total, therefore, the framework consists of six domains, which, together with 23 subdomains, are used by the Harvard EASEL Lab's Taxonomy Project to analyze SEL constructs across 43 frameworks, including the commonly used Collaborative for Academic Social and Emotional Learning (2021) framework. This analysis provides a “Rosetta stone” for mapping terms used to describe SEL constructs between frameworks. In our research, we used the EASEL framework and taxonomy to understand how locally identified SEL constructs overlap or differ from those in frameworks based on WEIRD literature.

Current approaches to measuring SEL across contexts

A review by Halle and Darling‐Churchill (2016) identified six SEL measurement tools having strong psychometric properties, all of which had been validated for use only in WEIRD contexts. More recently, two SEL assessments have been validated in populations of Syrian refugees—the International Social‐Emotional Learning Assessment measure of five domains of SEL in Iraq (D’Sa & Krupar, 2019), and the Social‐Emotional Response and Information Scenarios in Lebanon (Kim & Tubbs Dolan, 2019). Halpin et al. (2019) also found that the SEL domain in the International Development and Early Learning Assessment had the same factor structure across five countries: Afghanistan, Bolivia, Ethiopia, Uganda, and Vietnam. One limitation of these approaches, however, is that they take established SEL constructs and measurement approaches, developed in WEIRD contexts, as their starting point; thus, they do not allow for the possibility that some domains of SEL are conceptualized differently in non‐WEIRD contexts. Further, SEL domains that are important in non‐WEIRD contexts may be entirely absent from current frameworks and assessment tools. To address these limitations, there is a need for the development of SEL assessments based on local perceptions of SEL constructs rather than on existing tools.

Theoretical framework for the cultural conditioning of SEL development

We examine the applicability of SEL science, based largely on WEIRD contexts, through a framework of contrasting culturally conditioned developmental pathways (Keller, 2016). In this framework, autonomy and relatedness are viewed as basic human needs, with cultural differences in how those two constructs are interpreted and in the relationship between them. The cultural model that predominates among the Western middle‐class is described by Keller (2016) as “psychological autonomy.” In this model, autonomous action is based on an individual's preferences, mental states, and personality. Individuals see themselves as self‐contained, self‐reliant, separate, and unique (Markus & Kitayama, 1991). This cultural model focuses on mental states that support self‐enhancement and self‐expression. Relationships are self‐selected and between separate self‐contained individuals for mutual benefit. In this way, relatedness serves the needs of the autonomous self. This cultural model arises from child‐rearing strategies in WEIRD contexts (Greenfield, 2016; Keller, 2016) where available resources and a small family size allow for greater investment in the development of cognitive abilities of children in preparation for work in a knowledge economy; and where the nature of urban life means that encounters with strangers are a common form of interaction.

The current understanding of SEL based on research in WEIRD contexts may not be applicable to Tanzania, where the predominant cultural model is likely different from the one described above. The difference originates from the mode of economic production. In 2007, 76% of Tanzanians still relied on agriculture for their livelihood (Cleaver et al., 2010; United Republic of Tanzania, 2007). Keller (2016) describes the cultural model operating in subsistence agricultural communities as “hierarchical relatedness.” In this model, communal goals take precedence over personal preferences. Society members defer to others above them in a hierarchy largely defined by age. Individuals may have a high degree of autonomy but are motivated to act in the pursuit of communal goals. Such action confers a sense of well‐being (Keller, 2013) and belonging. Thus, relatedness predominates in this model and autonomy is employed in the service of an individual's needs for relationships with others and for a sense of belonging (Gaskins, 2020). This cultural model arises from child‐rearing strategies in farming communities (Greenfield, 2016; Keller, 2016), where economic productivity requires working toward communal goals, typically within a family; where large, multigenerational households live together in small villages; and where long‐term relationships are more important to daily social interaction than fleeting encounters with strangers.

We anticipated that this cultural model would predominate in Tanzania, as in other subsistence agricultural communities. In addition, the cultural model was evident in the ujamaa (translated directly as “familyhood”) political philosophy of Tanzania's first president after independence in 1964, Julius Nyerere. This philosophy was an attempt to revive a traditional African way of living, comprising cooperation among villagers, equality, and self‐reliance, and it remains influential today (Sakata et al., 2021).

The cultural models presented suggest that several SEL competencies described in WEIRD contexts may develop differently in other contexts. The differences are particularly apparent when children are required to regulate emotion or inhibit action at the request of an adult. For example, Lamm et al. (2018) found that the preschool children of Cameroonian farmers were more able than German preschoolers to delay gratification at the request of an adult. The authors argued that the Cameroonian children were comfortable following the instruction of an adult because they felt responsible to their community and understood their place in the social hierarchy. Haslam et al. (2019) argued that “collective regulation” may be a more relevant construct than self‐regulation in such contexts.

Different emphases are also placed on emotional regulation in the two cultural models. Where respect, obedience, and fitting in are valued, strong emotions are expected to be controlled because they are seen as a threat to social harmony (Chao, 1995). The development of self‐esteem is also expected to differ in the two cultural models. Western middle‐class parents spend time with their children, praising them and emphasizing their self‐worth (Miller & Cho, 2018). In subsistence communities, children are taught that they are not the center of attention (Keller, 2016). Shame is used as a strategy to regulate behavior (Lancy, 2014); children are instructed with moral codes and rarely praised (Keller, 2018).

In Western‐middle class households, adults respond to frequent “why” questions from children (Martini, 1996). In subsistence communities, questions are asked to seek information rather than explanations (Gauvain et al., 2013) and children's questioning of adults may be seen as rude (Lancy, 2014). Instead, children learn by engaging with parents’ routines. The greater prevalence of questioning and explanation (Lancy, 2014) in WEIRD societies, compared to subsistence agricultural communities, may reflect their stronger endorsement of curiosity as part of the implicit goal of developing cognitive skills (Greenfield, 2016).

The contrasting cultural models also help identify competencies that may be missing from established SEL frameworks. In several studies (Grigorenko et al., 2001; Jukes, Gabrieli, et al., 2018; Serpell, 1993), participants in African societies identified “social responsibility” as a dimension of intelligence, including the concepts of respect and obedience. These constructs receive little emphasis in current frameworks (Collaborative for Academic, Social, & Emotional Learning, 2021; EASEL Laboratory, 2020).

Commonality and specificity in SEL

The above theoretical framework informs our understanding of specificity and commonality in the development of social‐emotional competencies. With adaptations to the topic at hand, the Specificity Principle (Bornstein, 2017, p. 5) “asserts that specific setting conditions of specific people at specific times moderate specific domains of [social‐emotional development] by specific processes.” We interpret the specificity principle to apply to social‐emotional development at the societal level and within societies, based on contextual factors that give rise to the contrasting cultural models presented above.

The cultural model that predominates depends on the socialization strategies that are adaptive given the set of ecological and sociodemographic factors in a given context (Greenfield, 2016; Keller, 2016). At the societal level, we would expect that the pervasive influence of subsistence agriculture would cultivate the hierarchical relatedness cultural model in Tanzania. However, Tanzanian society is changing. The percentage of Tanzanians living in urban areas rose from 5% in 1960 to 33% in 2017 (World Bank, 2020), the gross enrollment ratio in primary schools rose from 33% in 1970 to 94% in 2018 (World Bank, 2020), and Tanzania is one of the fastest growing economies in Africa with a 6.3% average annual rate of growth in gross domestic product (GDP) this century (World Bank, 2020). The rise of education, urbanization, and commerce influences societies with a historic preference for the model of hierarchical relatedness (Jukes, Zuilkowski, et al., 2018; Kağitçibaşi et al., 2010), leading to a hybrid model of “autonomous relatedness” (Kağitçibaşi, 2017; Keller, 2016) in which autonomous values are adopted while a strong sense of relatedness continues to guide behavior in the family.

This shift in values is likely to take place first in a subsector of society, giving rise to within‐society specificity in SEL, because individuals have different experiences of the sociodemographic drivers of cultural change: urban life, commerce, education, and technology (Gauvain & Munroe, 2009; Greenfield, 2016; Jukes, Zuilkowski, et al., 2018; Jukes et al., 2013). Within these drivers, parental education may be a particularly strong determinant of an individual's developmental pathway (Keller, 2016, 2018) because educated parents have fewer children, later in life; live in smaller households; and are more likely to make the transition from farming to commerce and from village to city.

One aim of our studies was to investigate children's behavior and the competencies seen as important for success at home and at school. Because formal education is a driver of cultural change, schools are an interesting setting to study encounters between cultural models. Life in the classroom may embody values that differ from those at home. Teachers are typically outsiders, have a higher‐than‐average level of formal education, and are likely to have lived in an urban setting during their lives. It follows that teachers and parents may differ in their goals for children's development and that children's behavior, therefore, is guided by different forces at home and at school.

The current studies

The two studies presented here took place in Mtwara region on the southern Tanzania coast. The region is divided into nine districts, with a regional population of just over 1.2 million in the 2012 census (Tanzania National Bureau of Statistics, 2020). The region's economy is based on agriculture, with foreign investment in offshore oil and gas exploration in the coastal capital, Mtwara town. In 2018, Mtwara region was ranked 8th out of 25 regions with comparable data in primary school leaving examination pass rates (National Examinations Council of Tanzania [Baraza la Mitihani la Tanzania], 2018) and 12th out of 23 regions in GDP per capita (Tanzania National Bureau of Statistics, 2020). Mtwara is home to the Makonde, one of the five largest ethnic groups in the country, and, like much of coastal Tanzania, is predominantly Muslim.

The studies were mainly exploratory in nature, although they were guided by clear hypotheses regarding the role of culture and sociodemographics in social‐emotional development. They aimed to learn about the SEL competencies perceived as important by teachers and parents in the region (Study 1) and those that emerged from ratings of observable behaviors of children in preprimary and early primary grades (Study 2). The first aim of these studies was to understand the societal‐level specificity in social‐emotional development in the culture of predominantly subsistence agricultural communities. The second aim of the study was to understand within‐society specificity in social‐emotional development by investigating the role of sociodemographic variables—education, urbanization, and wealth—in shaping SEL competencies.

STUDY 1

The design of Study 1 emanated from our critiques, discussed above, that existing studies’ use of established SEL frameworks blunted their sensitivity to cultural differences. As a result, we used a grounded theory approach (Pidgeon & Henwood, 1996) to understand and measure SEL competencies from the perspectives of participant communities. The study aimed to understand societal‐level differences in valued social‐emotional competencies by comparing the views of respondents in this study with the EASEL framework. The study probed within‐society differences by comparing the views of teachers and parents.

Methods

The primary research questions were:

Which social‐emotional competencies do parents and teachers in Mtwara, Tanzania, see as important for children to develop in life?

Which social‐emotional competencies do parents and teachers in Mtwara, Tanzania, see as important for children to succeed in school?

We hypothesized that respondents would endorse competencies—such as respect and obedience—typically valued by subsistence agricultural communities. We further hypothesized that teachers would be more likely than parents to endorse competencies—such as curiosity—valued in societies with widespread formal education where questioning is encouraged (Lancy, 2014), particularly when discussing the competencies required for success in school.

Setting

Study 1 took place in May 2017 in rural Mtwara. Schools were selected using the approach of theoretical sampling (Pidgeon, 1996). According to the theoretical framework presented in the introduction, the views of rural villagers would produce the greatest contrast to the conceptions of SEL prevalent in literature from WEIRD countries. Consequently, Study 1 was based in three rural districts in Mtwara, Tanzania, with four schools selected at random to participate. The communities were predominantly from the Makonde ethnic group and were mostly Muslim, with some Christians. Most participants were subsistence farmers with some income from selling cashew nuts and sesame. Electricity was widespread in only two of the four villages, and all villages had mobile phone access.

Participants

The research team sampled and recruited teachers and children from the four schools, and parents of children attending the schools. Because of experience in other situations that women may be less open if they are expected to speak while men are present, men and women met in separate focus groups. The 12 parent focus groups involved 61 men (7 groups) and 34 women (5 groups), with an overall median focus group size of 8 parents and a range of 4–12. The research team interviewed 16 male teachers and 11 female teachers, one‐on‐one across the four schools. Parents were typically farmers, most aged approximately 30–45 years old with primary school education or less. Teachers were also aged approximately 30–45 years old, and all had secondary education or higher. The research team judged that young children were less able than adults to give opinions about the importance of competencies for their life. We used alternative methods with children which addressed the research question less directly and are not included here. The methods and results for interviews with children are described elsewhere (Jukes, Gabrieli, et al., 2018). The researchers monitored the adequacy of the sample size, assessing data saturation in the codebook of competencies populated by adult interviewee responses. When no new competencies were generated in interviews with the third and fourth schools, the team concluded that saturation had been achieved (Fusch & Ness, 2015) and terminated data collection.

Procedure

Interviews and focus groups

Following a standard protocol with each participant, all adults gave written informed consent. Data collection with teachers and parents was carried out by two teams of two researchers each. One person in the team would follow the interview protocol and ask most of the questions, while the other took detailed handwritten notes. Each team also made an audio recording of each interview for later reference. Given the variation in focus group size (4–12), researchers took care to elicit views from all members of the group.

Interviews and focus group discussions were structured around the two research questions concerning the competencies required for success in life and school respectively (full parent interview protocol in Supporting Information, Section A1). For each question, interviewees were asked to respond in reference to a child of approximately 8 years. Parents were asked about their child and teachers about their pupils, but otherwise the questions were identical. The interviewers’ goal was to guide the discussion to elicit as many qualities or competencies as possible from respondents, following an interview protocol with suggested probes. In addition to seeking an inventory of competencies, the interviewers asked subsequent questions that prompted participants to rate the relative importance of the competencies they had listed and to provide behavioral examples of each competency. The research team used two words in Kiswahili that most closely aligned with the goals of the research: sifa, which means character or reputation; and tabia, translated as behavior, character, disposition, or habits. The terms mwenendo (conduct) and matendo (actions) were also used, but less frequently. Interviewers did not specify the nature of the competencies (e.g., social‐emotional or cognitive) in their questions, to avoid influencing respondents with established definitions of SEL. However, if respondents focused on academic achievement when asked about the qualities required for success at school (e.g., “good at reading”), they were given prompts designed to shift their focus, by asking them, for example, to think of the qualities that a child brings to their first day at school (see Supporting Information, Section A1).

Analytical strategy

The main aim of the analysis was to identify unique competencies mentioned by participants and the definition of those competencies. Each competency (e.g., obedient) was associated with a set of behavioral examples (e.g., “follows instructions”), which were used to refine the definition of competency terms offered by participants. Commonalities in behavioral examples also helped to place competencies in three categories (see Supporting Information, Section A2, for detail).

Respondent‐generated definitions of competencies in the three categories are provided in Tables A1–A3 in Supporting Information. Here, we highlight constructs whose interpretation may differ from common usage in WEIRD settings. Discipline was used to refer to meeting social obligations and had many behaviors in common with respect and obedience. The term was not used to mean “self‐discipline”—for example, in managing time and tasks. Self‐belief and courage were used interchangeably, and both were valued more by teachers than by parents. Example behaviors included asking and answering questions, participating in groups, leading others, and doing voluntary assignments. Perhaps in contrast to how the term is used in WEIRD contexts, participants’ discussion of self‐belief did not mention confidence in one's abilities (self‐efficacy) or belief in one's own worth (self‐esteem).

Results

Frequency of competencies

The first analysis examined the frequencies with which competencies were mentioned across all questions in the interview, except where a competency was mentioned first by the interviewer. Eight competencies were mentioned by three or more parent focus groups and by three or more teachers and rated as important at least once by both parents and teachers: respectful, attentive listener, obedient, cooperative, disciplined, clean, self‐directed, and polite/calm (see Table 1). Responses were similar across the four schools; of the 15 most frequently mentioned competencies, all were mentioned at least once by parents or teachers in each school.

TABLE 1

Number of interviews or focus groups with parents, teachers, and pupils in which each competency was mentioned (minimum 7 mentions) and rated as important

English	Kiswahili	Mentioned in interviews		Rated as important a
		Parent FGD (13 FGDs)	Teacher (n = 23)	Parent FGD (10 FGDs)	Teacher (n = 21)
Respectful	Mwenye heshima	11	16	5	6
Attentive listener	Msikivu	11	17	4	5
Obedient	Mtii	11	15	8	3
Cooperative	Anayeshirikiana	8	14	2	4
Clean	Msafi	9	11	1	4
Polite and calm	Mpole/mtulivu	8	12	1	3
Disciplined	Mwenye nidhamu	7	13	5	11
Self‐directed	Anayejituma	5	15	1	10
Hardworking	Mwenye juhudi/bidii	8	10	0	3
Curious	Mdadisi	6	17	0	8
Clever	Mwerevu	6	7	0	1
Love of parent/teacher	Anayependa wazazi/waalimu	4	5	1	2
Trustworthy	Mwaminifu	9	1	4	0
Seeks understanding	Muelewa	9	8	2	0
Courageous	Aliye hodari	4	7	0	1
Confident, self‐belief	Anayejiamini	2	8	0	4
Persistent	Asiyekatatamaa	4	6	1	0
Sociable	Mchangamfu	4	3	0	3
Has fear of God	Mwenye hofu ya Mungu	2	5	0	1
Careful	Makini	3	4	0	0

Abbreviation: FGD, focus group discussion.

Importance ratings were introduced to the procedure after the first few interviews in the first school.

Different responses were evident for teachers and parents depending on whether they were questioned about competencies important for schooling or for life in general. Parents and teachers gave similar suggestions for the competencies seen as important for life in general. One exception was that parents showed more concern about trustworthiness than teachers, with the construct mentioned in four parent focus groups, but in no teacher interviews.

When asked about the competencies important for school, parents gave similar responses to the question about competencies important for life in general. Some parents explicitly stated that the competencies required to succeed in school were those also required for life in general. By contrast, teachers mentioned several competencies more frequently in relation to schooling. These included curiosity (mentioned by 17 teachers in relation to schooling vs. 10 in relation to life in general), being an attentive listener (16 vs. 10), being self‐directed (15 vs. 6), being hardworking (10 vs. 2), and having self‐belief (7 vs. 4). These competencies were also mentioned more frequently and valued more by teachers than by parents.

Certain competencies were seen as increasing children's success at school by helping them to build relationships with others. Respondents said that respectful and trustworthy children were favored by both teachers and fellow students and could receive more help in class as a result. Similarly, teachers said they gave more attention to children who showed curiosity in class, because such children make their work easier.

Most competencies were seen as important for both boys and girls, with a few exceptions. Parents, but not teachers, said that girls should be more obedient than boys. Several parents said that boys should be more socially active and hardworking than girls. In general, participants said that children developed competencies with age, although some teachers said that younger children can be more confident, motivated, and hardworking. Opinion was divided as to whether competencies were innate or developed through upbringing or schooling.

Study 1 Discussion

Our first research question concerned how parents and teachers responded to the question: “What are the qualities that you would like your children/pupils to develop in life?” Based on commonalities in behavioral examples, two groups of competencies emerged. The first group shared the characteristics of social responsibility—fitting in and working toward a common goal—and included five of the eight competencies mentioned most frequently and rated as important by both teachers and parents. The second category included prosocial competencies, which also shared the goals of social responsibility. The remaining competencies were ones that could be demonstrated independently of others, such as self‐directed.

Our second research question concerned the qualities parents and teachers saw as important for success in school. Parents did not differentiate between school and life in general in their responses, saying that the competencies in the social responsibility category were also most important for school success. Teachers, however, were more likely to mention being curious, an attentive listener, self‐directed, hardworking, and having self‐belief in response to the question about schooling compared to the question about life in general. Note that some of these competencies, such as curiosity and self‐belief, are those endorsed by societies where formal education is widespread. The teachers’ focus on these competencies may reflect both the experience of their own formal education and their familiarity with competencies required for children's success in the school where they teach.

Overall, the findings of Study 1 suggest that participants in rural Mtwara value social responsibility highly. This conclusion contrasts with existing SEL frameworks, which do not place an emphasis on social responsibility. As such, these findings provide initial support for societal‐level specificity in SEL in Mtwara in the domain of social responsibility. There were also initial findings relating to within‐society specificity in SEL in that teachers had more values in common with WEIRD societies than parents, perhaps because of their higher level of formal education.

STUDY 2

The value placed on social responsibility by communities in rural Mtwara may point to societal‐level specificity in the development of social‐emotional competencies. In Study 1, we used behavioral examples to understand each competency. In Study 2, we took this process further by examining parents’ and teachers’ quantitative ratings of children's behavior in a larger sample of schools. The aim was to use the behavioral rating data to identify underlying constructs representing key observable competencies in children. The second aim of Study 2 was to investigate within‐society specificity in the development of social‐emotional competencies by documenting key sociodemographic variables (gender, parental education, socioeconomic status [SES], and urban residence) and assessing their association with constructs identified in the data. To this end, Study 2 included urban districts in Mtwara in addition to the rural districts participating in Study 1. The specific research questions of Study 2 were:

What latent SEL constructs can be derived from parents’ and teachers’ behavioral ratings?

How do latent constructs (competencies) relate to sociodemographic variables: gender, parental education, SES, and urban residence?

We hypothesized that the latent constructs derived in Study 2 would support findings from Study 1 of the importance of social responsibility for communities in Mtwara. We also anticipated that children of parents who were more educated, were wealthier (higher SES), and lived in urban areas would be more likely to be raised with WEIRD values. As such, we hypothesized that these children would be rated more highly for competencies associated with the values of WEIRD societies—such as curiosity—and would be rated more poorly for competencies associated with the values of subsistence communities—such as emotional regulation and obedience.

Sample

Study 2 took place in August 2017 in five of Mtwara's nine districts: Mtwara Rural, Mtwara Urban, Newala, Nanyumbu, and Masasi. A sample of 25 schools (5 schools per district) was randomly selected from those that were accessible using public transportation and that had at least two registered teachers across the three target grades (preprimary, Standard 1, and Standard 2—equivalent to kindergarten, Grade 1, and Grade 2), to increase the possibility that teachers knew their students well. Schools were classified as urban/peri‐urban (henceforth “urban”) or rural based on their proximity to a tarmac road or an area of development, including a bus station, hotel, markets, shops, or restaurants. Using these criteria, 12 schools were classified as rural and 13 schools were classified as urban. Students did not take part in any data collection for this study but were sampled as the basis for selecting parents and teachers to interview. Students were selected randomly from class lists in preprimary, Standard 1, and Standard 2. Students were replaced in the sample if teachers did not know them well. In total, 477 students were sampled: 164 preprimary students, 153 Standard 1 students, and 160 Standard 2 students. Of the 477 students, 245 were girls and 232 were boys. Students’ ages ranged from 4 to 13 years, with almost 85% of the students between 5 and 9 years old. The median age of students was 5 years in preprimary, 7 years in Standard 1, and 9 years in Standard 2. Parent and teacher characteristics from Study 2 are described in Supporting Information, Section A3.

Assessment instruments

The list of competencies to be assessed was determined by the findings of Study 1 and consisted of the competencies identified most frequently by participants, with the following few exceptions: Example behaviors associated with discipline overlapped entirely with obedient or attentive listener. Consequently, discipline was dropped from the list of competencies. The competency of being clean was also dropped because this construct was not considered to be a social‐emotional competency. The constructs of empathy and persistence were mentioned by Study 1 respondents but were not the most frequently reported or highly valued constructs. However, we included them in the list of constructs because of their prominence in existing SEL frameworks and because we aimed to be conservative when identifying differences between competencies valued by respondents in the current studies and those included in existing frameworks.

Next, a list of questionnaire items was developed for each competency, requiring respondents to judge whether children engaged in a behavior exemplifying the competency. Where necessary, additional qualitative interviews were conducted to ensure there were 5–10 behavioral examples for each competency. For the constructs of empathy and persistence, there were insufficient respondent‐generated items. For these constructs, the research team created items, based in part on existing measures (Achenbach & Rescorla, 2001; Rieffe et al., 2010), to add to those submitted for pretesting. Two scales were developed and administered as part of this survey: one for parents and one for teachers. Both scales were administered in Kiswahili using Tangerine, data collection software designed for use on digital tablets.

Pretesting of the newly developed scales occurred first through cognitive interviewing (Beatty & Willis, 2007) with parents and teachers, and through pilot data collected in regard to 30 students. Problematic questions were identified as those showing little variance in scores or with an item–rest correlation of <.3. These questions were revised or dropped. The final parent questionnaire consisted of 71 questions measuring 13 different constructs (obedient, curious, respectful, courageous, cooperative, self‐directed, attentive, persistent, polite, sociable, careful, empathetic, religious) and 22 background questions. The final teacher questionnaire consisted of 40 questions measuring 12 constructs (all but religious—teachers were not aware of students’ religious practices) and 8 background questions.

Each behavior was assessed using two questions. The first question ascertained whether the child engaged in the behavior at all. For example: “Does (name of child) like to ask many questions?” If the answer was yes, respondents were asked: “More than other children, less than other children, or about the same as other children?” The two parts of the question were combined to produce a single score ranging from 0 (does not engage in the behavior) to 3 (engages in the behavior more than other children).

Parents were asked several questions to assess SES including the materials used for the walls and roof of their house, whether they had electricity at home, and whether they owned a television, a bicycle, and a motor vehicle. To create an SES index, a single component was extracted from these six variables using principal components analysis. Households were then grouped into tertiles based on this component.

Data collection

Data collectors were trained and assessed using an Assessor Accuracy Measure, which involved having them complete a questionnaire based on prerecorded interviews. On average, the data collectors recorded 97% of responses correctly. Data collectors worked in five teams of three. One team was responsible for data collection in one school, with each data collector being assigned a grade (i.e., preprimary, Standard 1, or Standard 2) and conducting all parent and teacher interviews respective to that grade. Each teacher gave one interview for each of the students from the standard(s) they taught.

Analysis

Analyses were conducted separately for parent and teacher ratings. Latent constructs in the data—each representing a social‐emotional competency—were identified through an exploratory factor analysis followed by a confirmatory factor analysis. The factor structure was refined based on model fit indices, rotated factor loadings, potential cross loadings, and interpretability. Multivariate regression analyses were used to determine the relation between background variables and each competency. Analysis details are provided in Section A4.

Factor analysis results

Parent factors

Six constructs were identified by factor analysis of parent ratings (see Table 2). Three parent factors—obedient, curious, and religious—mapped closely onto intended constructs. The conscientious factor included elements of self‐direction, persistent, and careful. The prosocial factor included elements of being cooperative, polite, sociable, and empathetic. The items included under emotional regulation all involved regulating anger, frustration, or behavioral responses previously included under other competencies, such as being polite. The obedient factor explained the largest proportion of variance in children's behavioral ratings, at 29%. Curious and conscientious explained reasonable proportions of variance at 7% and 6% each (see Table A8).

TABLE 2

Regression estimates indicating the relation between background variables and parent‐rated competencies

	Obedient	Curious	Conscientious	Religious	Prosocial	Emotional regulation
Child characteristics
Age (years)	.04*	.06*	.07***	.01	.02	.00
Grade 1 a	.09	.20	.20	.22 †	.06	–.08*
Grade 2 a	.03	–.05	.13	.35 †	.01	.05
Child is female	.21**	–.03	.17***	.18*	.13***	.09**
Parent characteristics
Medium SES b	–.20*	.00	–.19*	–.16	–.06	–.09*
High SES b	–.04	.19 †	–.05	–.08	–.04	–.04
Primary education c	.17*	.30**	.18*	.31*	.15**	.02
Secondary education c	–.06	.43**	.01	.18	.09	.02
Urban setting	–.08	.00	–.03	.07	–.03	–.05
Married	–.02	–.08	–.03	.05	.03	.07 †
Living together	.04	.15*	.11	.37**	.05	–.06 †
Rater is female	–.07	–.13	.01	–.02	–.05 †	–.03

Abbreviation: SES, socioeconomic status.

Reference category—preschool.

Reference category—low SES.

Reference category—no formal education.

p < 0.1.

p < .05.

p < .01.

p < .001.

Teacher factors

Three constructs were identified by factor analysis of teacher ratings (see Table 3). Like the equivalent parent factor, conscientious included elements of self‐direction, persistent, and careful. The agreeable factor contained elements of being obedient and responding in socially appropriate ways—for example, when being asked a question or when others are talking. For teachers, the largest factor was conscientious, explaining 37% of the variance. Among all parent and teacher factors, the school‐level intra‐class correlation was above .1 only for teacher ratings of curious, suggesting that, in general, there was not strong clustering of ratings within schools (see Table A8).

TABLE 3

Regression estimates indicating the relation between background variables and teacher‐rated competencies

	Conscientious	Agreeable	Curious
Child characteristics
Age (years)	.11***	.06***	.15***
Grade 1 a	–.10	.02	–.10
Grade 2 a	–.20*	–.02	–.32*
Female	.13**	.16**	.12 †
Parent characteristics
Medium SES b	–.03	–.05	–.08
High SES b	–.05	–.13 †	–.02
Primary education c	.14 †	.11	.16
Secondary education c	.29**	.21*	.41***
Urban setting	.11	.08	.25*
Married	–.03	–.04	.03
Living together	.11*	.15*	.02
Teacher characteristic
Rater is female	–.03	–.04	–.28*

Abbreviation: SES, socioeconomic status.

Reference category—preschool.

Reference category—low SES.

Reference category—no formal education.

p < 0.1.

p < .05.

p < .01.

p < .001.

Additional detail on the scoring of social‐emotional competencies and factor analysis results are provided in Section A5.

Correlation among factors

For parent ratings, there were moderate‐to‐high (.19–.68) and statistically significant correlations among all factor pairs except between curious and emotional regulation. For teacher ratings, all correlations between pairs of factors were high (.49–.71) and statistically significant. Correlations between parent factors and teacher factors were moderate (.16–.32) and statistically significant except for correlations between teacher factors and the parent factor of emotional regulation, which were smaller and non‐significant. The full table of correlations among the six parent factors and three teacher factors appears in Supporting Information, Table A12.

Associations between SEL competencies and individual characteristics

Regression estimates for associations of SEL competencies and background variables are presented in Table 2 (parent ratings) and Table 3 (teacher ratings), with detailed presentation of unstandardized and standardized models in Tables A13–A16. We turn first to the variables hypothesized to influence within‐society specificity of social‐emotional competencies: parental education, SES, and urban residence. The largest effect in the data was the relation between parental education and parent ratings of curious. Parents who had completed secondary school rated their children as .53 SD higher (=.43/.81) on curious than the children of parents who had not completed primary school. Teachers also rated the children of parents with secondary education .49 SD higher (=.41/.84) than other children on curious. Ratings of curious were also higher among parents in the upper tertile of SES with borderline statistical significance (p = .052) and among teachers rating children in urban settings. Associations with parental education and SES were higher for curious than for each of the other five parent factors (Supporting Information, Table A17). Associations with parental education, but not SES, were higher for teacher ratings of curious than the other two teacher factors, with borderline statistical significance (p = .072 and .059 respectively). Having parents who completed primary school was associated with higher parent ratings of obedient, conscientious, and prosocial. Having parents who completed secondary school was associated with higher teacher ratings of conscientious and agreeable. In contrast to the above pattern of results, parent ratings of obedient, conscientious, and emotional regulation had negative correlations with SES.

There were significant associations with age or grade for all parent and teacher ratings except for parent ratings of prosocial. The strongest associations with age were found for teacher ratings of conscientious and curious, which both increased by .18 SD with every year of age. Girls’ advantage over boys was statistically significant for all competencies except for both parent and teacher ratings of curious. Female teachers gave lower ratings than male teachers for curious but otherwise there were no significant differences in ratings by gender of the adult rater. Interactions between gender of adult and child were not significant for any outcome and were not included in final regression models.

Study 2 Discussion

The first research question in Study 2 concerned the latent constructs identified in adult and teacher ratings of student behavior. Parent ratings identified three factors—obedient, curious, and religious—that mapped closely onto constructs identified in Study 1. The items loading on three other factors—conscientious, prosocial, and emotional regulation—each mapped onto more than one construct from Study 1. Teacher ratings included one factor—curious—that mapped onto a single construct from Study 1; and two factors—conscientious and agreeable—that mapped onto more than one construct.

The second research question concerned the relationship between sociodemographic variables and factor scores. In general, findings supported the hypothesis that these variables would be strongly correlated with curiosity. The positive association between parents completing secondary education and both parent and teacher ratings of curious was stronger than for any other factor. Similarly, high parental SES was associated with ratings of curious and—for parent ratings only—was stronger than for any other factor. Teacher ratings of curious were also higher in urban settings. Also in support of our hypotheses, higher parental SES was associated with lower parental ratings of competencies associated with compliance and social responsibility, including obedient and emotional regulation. We now consider the implication of these findings in the light of results from both studies.

GENERAL DISCUSSION

This research project aimed to investigate societal‐level and within‐society specificity of social‐emotional development in Mtwara, Tanzania. We hypothesized that competencies associated with a cultural model of “hierarchical relatedness” would be more evident in the subsistence agricultural communities of rural Mtwara. Two sources of evidence were used to identify locally valued competencies: the elicited opinions of participants in Study 1 and the latent constructs in adults’ ratings of child behavior in Study 2. Our second hypothesis was that the experience of formal education, higher SES, and urban settings would promote development of competencies associated with a cultural model of “psychological autonomy.”

Societal‐level specificity in social‐emotional development can be understood by examining how the SEL constructs identified in Mtwara, Tanzania, relate to those in established SEL frameworks. The most apparent difference was the emphasis placed on social responsibility by respondents in Tanzania, in line with predictions based on the culture associated with subsistence farming communities. In qualitative interviews (Study 1), the competencies of being respectful, obedient, disciplined, polite/calm, and an attentive listener were mentioned most frequently by participants and were rated as the most important competencies for children to develop. In Study 2, the obedient factor explained the most variance of parents’ behavioral ratings of their children.

The construct of social responsibility is underrepresented in current frameworks. Based on the EASEL Lab's Taxonomy Project, none of the 43 frameworks analyzed mentioned the word “obedience” (EASEL Laboratory, 2020). The culturally related differences in the emphasis placed on obedience are consistent with the theoretical framework of this article. Whereas obedience may be seen as an unwelcome restriction on one's autonomy in WEIRD settings, children in subsistence agricultural societies are motivated to be compliant by a desire to fit in with the collective (Lancy, 2014).

Of the seven constructs identified in the current study, neither obedient nor religious mapped to existing constructs in the EASEL Laboratory (2020) taxonomy (see Table A18). The closest match to a notion of social responsibility is included under Civic Values: “Understand[ing] one's connection and responsibility to family, classroom, school community, neighborhood, country, and world; understand[ing] the value of civic responsibility” (p. 10). Our findings suggest that social responsibility is a more fundamental, foundational competency in Tanzania than this definition implies. For example, even when describing competencies that were not overtly social in nature, respondents often situated them in terms of the interconnectedness of social groups (Jukes, Gabrieli, et al., 2018). Curiosity was described as important because curious students make teachers’ work easier and consequently teachers will favor them in class. The view of social responsibility as a foundational competency is consistent with several other studies in sub‐Saharan Africa (Grigorenko et al., 2001; Jukes, Gabrieli, et al., 2018; Serpell, 1993; Super, 1983), which portray social responsibility as integral to the definition of intelligence. Existing frameworks fail to embrace this understanding of social responsibility because few of them have been derived from research and thinking outside of WEIRD contexts. One exception is the framework developed by the International Institute of Islamic Thought (Nasser et al., 2019), which includes the construct of community mindedness—“seeing the self as interconnected with and acting for the benefit of an inclusive whole” (p. 14). This discussion also illustrates the need for a culturally grounded, conceptual understanding of superficially similar terms, such as respect, used in different contexts (see Malti et al., 2020).

Emotional and behavioral regulation and inhibitory control are key components of the EASEL Laboratory (2020) framework. These concepts were absent from the competencies valued by participants in Study 1, in line with the view that self‐regulation is less critical for successful functioning in non‐WEIRD societies (Haslam et al., 2019). Discipline was mentioned frequently by participants but was described in terms of obedience rather than self‐discipline. It was interesting that emotional regulation did emerge as a factor in Study 2 and that poor emotional regulation—i.e., frequently becoming angry or frustrated—was negatively associated with SES. It is possible that the construct of emotional regulation, while not being prominent in the perceptions of adults in communities in southern Tanzania, is becoming more important as Tanzanian society embraces education and urbanization.

In addition to the competencies discussed above, other SEL constructs identified in the current study are similar to those included in existing frameworks (Table A18). In particular, prosocial maps onto the EASEL Laboratory (2020) category of the same name and conscientiousness is an example of the EASEL Laboratory (2020) category of performance values. These findings do not preclude the possibility that there are culturally specific meanings, precursors, and developmental pathways of these constructs. It does, however, point to a degree of cross‐cultural similarity in the behaviors that exemplify these domains in preschool and the early years of primary school. In this way, we provide support for the domain‐specificity of cultural influences on social‐emotional development. Cultural models of the self have greater implications for some domains of social‐emotional development than others.

It is noteworthy that most competencies mentioned by respondents were in the category of social‐emotional, even though interviewers’ questions referred to competencies in general terms, without specifying them as social‐emotional. Other studies in Africa (Jukes, Gabrieli, et al., 2018; Serpell, 1993) have found that cognitive competencies, such as “cleverness,” are considered to have less social value and may have overtones of cheekiness and cunning. Such findings are consistent with the ecological framework (Greenfield, 2016) guiding this study, which posits that some cognitive skills have less value—and therefore are valued less—in subsistence agricultural communities.

The second aim of this paper was to examine within‐society specificity in social‐emotional development—how SEL competencies are influenced by the experience of formal education, higher SES, and urban life. The results of Study 2 supported our hypotheses. The clearest findings related to curiosity—a competency associated with WEIRD societies. There was a positive correlation between ratings of children's curiosity and urban settings and parental education and SES. Also in line with our hypotheses, ratings of obedient and emotional regulation—competencies associated with subsistence societies—were negatively correlated with parental SES. However, the pattern of associations was not entirely consistent with the hypotheses, in that only medium—but not high—SES was negatively correlated with these competencies. Also, parental primary education was positively associated with five of the six parent‐rated competencies and parental secondary education was associated with all three teacher‐rated competencies. The widespread positive associations with education were also not consistent with hypotheses.

The association between formal education and competencies valued in WEIRD societies was also supported by qualitative findings. In Study 1, we found that teachers were more likely than parents to identify curiosity, self‐direction, and self‐belief as important competencies for schooling. This result suggests that schools may act as settings that, through the goals of their activities and the cultural models of their teachers, influence children's social‐emotional development in a manner that is different from the influences in the home.

A key individual characteristic related to the development of SEL competencies was gender. Parents rated girls more highly on competencies related to compliance and cooperation in the culture of subsistence agricultural societies (being obedient and having emotional regulation) but not on those associated with WEIRD contexts (being curious). This finding is consistent with research in similar contexts (Whiting & Edwards, 1973) and aligns with the view of subsistence agricultural communities promoting ascribed gender roles, with girls being encouraged to stay at home and carry out chores (Palermo et al., 2020). All competencies increased with age except for parent ratings of prosocial. In general, however, developmental effects were smaller than effects due to gender, SES, and parental education.

The combined results of the two studies suggest several implications for practice. Given their different perceptions of the competencies required for learning, efforts to promote dialogue between teachers and parents on this topic could be beneficial. Our findings may also point to social‐emotional competencies that could be developed to improve students’ learning in school. For example, teachers want students to be curious and self‐directed, but these values may not be emphasized in the homes of rural families with little formal education. One response could be to develop programs or classroom activities to develop these competencies. One preschool preparation program in Tanzania (Education Quality Improvement Programme–Tanzania, 2017), for example, has had success in developing children's confidence to speak up in class.

The different methods used in the two studies offer some insights into how best to study community perceptions in order to extend our understanding of SEL and the cultural models that influence its development. Study 1 used qualitative methods and gave participants freedom to discuss any construct they valued. This approach helped identify several constructs that were not included in existing SEL frameworks, and different conceptualizations of existing constructs. Study 2 used quantitative methods to identify groups of related observable behaviors. This approach identified constructs—such as emotional regulation and conscientiousness—that were not identified as important by participants in Study 1. These findings illustrate the importance of combining qualitative and quantitative data in such investigations. In future studies, an additional approach worth exploring would be to directly question participants about key constructs in current frameworks and to ask them to rate their value in their own society.

It is important to acknowledge that our qualitative and quantitative data consist of adults’ perspectives of children's behavior rather than direct child assessments. With this in mind, one strength of our findings is that they give us insights into the perspectives of adults. Study 1 asked directly about caregivers’ socialization goals (Harkness & Super, 1992), whereas Study 2 investigated child behaviors that were most salient to adults, and therefore help to reinforce and extend our understanding of the behaviors to which caregivers pay attention. The focus on adult perceptions also limits the interpretation of findings in several ways. First, some behaviors may not be observable by adults. For example, showing curiosity by asking questions may be discouraged by adults and thus only exhibited in peer groups. Other behaviors may receive little attention from adults, yet have significance for children's development. For teachers in particular, there may be challenges in recalling behaviors of individual students in large classes.

Adult ratings of children may also be influenced by their individual characteristics. Thus, associations between parental education and children's competencies may represent an effect on parental perception rather than on children's behavior. Given this limitation, it is an important finding that parental education was related to both teacher and parent ratings of curiosity, providing some triangulation for this finding. However, additional work is needed to identify causality in such relationships.

Another set of limitations concerns the sample of parents and children. The respondents in Study 1 were those motivated to come to school on the day of interviews and were not necessarily representative of their communities. The sample of students was drawn from those attending school. Primary education is compulsory in Tanzania, but 17.7% of primary‐school‐age children are out of school nationwide (UNESCO Institute for Statistics, 2018). It is possible that some competencies associated with schooling were overrepresented in our sample. A final limitation is that our sample was not large enough to conduct exploratory and confirmatory factor analyses with separate sub‐samples. Further work is needed to confirm the validity of constructs identified in this research.

In conclusion, our findings provide some support for the hypothesis that the development of some social‐emotional domains is specific to the cultural model shaping a child's development. Our theoretical framework centered on two prototypical cultural models: psychological autonomy and hierarchical relatedness. Our findings support hypothesized societal‐level specificity in some domains of social‐emotional development to the extent that one cultural model is dominant in a society. Findings also support hypothesized within‐society specificity in social‐emotional development contingent on exposure to sociodemographic factors—education, SES, and urban settings—that foster the adoption of one cultural model or another. We recommend that future work aims to further specify cultural models—their variants, their hybrids, and their dynamic evolution—and the way they shape social‐emotional development. We call on more work—in research, measurement, and evaluation of SEL—to be explicitly driven by a framework of culturally conditioned development and to avoid unwarranted assumptions of universality. Such an approach has the potential to improve our understanding of social‐emotional development in WEIRD settings as well as in less‐studied contexts and would allow for a strength‐based approach to assessment (Rogoff et al., 2017) that would appreciate the logic and value of competencies in a cultural context.

Supplementary Material

Click here for additional data file.