Educational resource use during COVID-19 school closures,child cocoa agricultural work,and learning outcomes in rural Côte d’Ivoire

Abstract

Education is a protective factor against child labor; however, COVID-19 school closures threatened educational attainment and increased child labor. In this 2-year longitudinal study in rural Côte d’Ivoire, we investigated children’s (N = 955, M_age = 10.45) learning outcomes, educational resource use, and agricultural work pre-, during, and post-COVID-19 school closures. The number of children participating in agricultural work increased during school closures and persisted once schools reopened. However, children randomized to an educational technology (EdTech) intervention (Allô Alphabet) and children who reported receiving family tutoring during school closures were more likely to report reduced agricultural work. Children working in cocoa agriculture during school closures showed poorer reading scores the following year; however, reading scores marginally improved for children who used EdTech during school closures. Results are discussed in terms of negative COVID-19 impacts and the positive role of educational resources on the attainment of the United Nations Sustainable Development Goals in Côte d’Ivoire.

Keywords

Education Educational technology child labor sub-Saharan Africa Côte d’Ivoire literacy COVID-19 school closures

Introduction

The COVID-19 pandemic has undermined the progress made on achieving the United Nations’ Sustainable Development Goals (SDGs), particularly in regions where poverty, child labor, and low learning outcomes are prevalent (Min and Perucci, 2020; United Nations, 2015). Specifically, SDG 4 (ensuring inclusive and equitable quality education and promoting lifelong learning opportunities for all) and SDG 8.7 (eradicating child labor by its targeted deadline of 2025) are far from being met. Moreover, widespread school closures have been linked to reduced learning outcomes and increased child labor (International Labor Organization (ILO) and United Nations Children’s Fund (UNICEF), 2021; Jakubowski et al., 2023), further undermining progress towards SDG 4 and 8.7.

In Côte d’Ivoire, the government implemented emergency health measures and closed public schools between March 16 and May 10, 2020 due to the COVID-19 pandemic (see Supplementary Materials for details). While school closures were an important pandemic response, child protection experts raised concerns that school closures may contribute to an increase in child labor(International Labor Organization ILO and United Nations Children’s Fund UNICEF, 2021). Côte d’Ivoire has been the world’s largest cocoa producer since 1977 (African Development Fund, 2018). The Ivorian cocoa sector is characterized by small household farming, where household financial insecurity, opportunity for income-generating activities via household-led cocoa farming, and the availability of working-age children have put children at high risk of engaging in child labor (UNICEF, 2019). As child labor is a widespread problem in Côte d'Ivoire (see Child Labor versus Work below for detailed discussion of the differences between child labor and child work; Andvig et al., 2001; Child Labor Cocoa Coordinating Group [CLCCG], 2020, International Cocoa Initiative [ICI], 2022, The United Nations Educational, Scientific and Cultural Organization [UNESCO] Institute for Statistics [UIS], 2018; Institut National de la Statistique [INS], 2005; Jacquemin, 2006). Economic instability and lack of access to school, both negatively impacted by the COVID-19 pandemic, are well-established predictors of children’s participation in child labor (Centre ivoirien de recherches économiques et sociales (CIRES) Cellule d’Analyse de Politiques Économiques, 2023; Naeem et al., 2011; Ray, 2002). For example, past long-term school closures, such as those caused by the Ebola epidemic, resulted in increases in child labor in West Africa (Risso-Gill and Finnegan, 2015; Smith, 2021). In order to address this global situation caused by COVID-19, additional approaches to supporting children’s learning and preventing child labor are needed (Min, 2021).

When schools closed worldwide due to the COVID-19 pandemic, children lost access to education—a protective factor against child labor—pushing many into work. The relationship between education and child labor is complex. It is generally understood that children’s participation in labor disrupts involvement in education, leading to negative learning and academic outcomes (Boutin and Jouvin, 2022). Conversely, a growing body of evidence shows that access to quality education is associated with reduced rates of child labor (International Cocoa Initiative (ICI), 2019; International Labor Organization (ILO), 2017b; Schlemmer, 2002; Wortsman et al., 2024a). This link between education and child labor suggests that access to educational resources during school closures may mitigate an increase in child labor (ILO, 2020). When formal schooling is unavailable, supplemental educational resources can help bridge the gap by providing additional and/or tailored support to vulnerable learners (e.g., those who lost access to education, have limited economic resources to support private education, and children with low academic scores or learning difficulties; El Bouwab, 2021). Learning loss during normal periods of school closure (i.e., summer break) is a well-documented phenomenon. There is limited research in sub-Saharan Africa (SSA) on educational resources being able to support this learning loss. For example, Slade et al. (2017) examined learning loss after “grade-transition break” (i.e., the period after one grade ends and before another grade starts) and found average reduction of 0.38 standard deviations across measures of reading and reading-related language skills. Additionally, Slade et al. (2017) show that a literacy intervention that trained and supported teachers and provided additional literacy materials to 10,704 Malawi students going from grades 1 to 2 and 2 to 3 between 2013 to 2016 did not improve literacy scores, and worse, the economic realities pressuring out-of-school children to work limited potential benefits of the intervention.

Educational technology (EdTech) has been increasingly introduced to support learning, particularly as an at-home resource for learners that families leveraged during school closures. In response to the COVID-19 school closures, we explored if and how supplementary educational resources, including EdTech, may have mitigated against an increase in child labor and supported the learning in children in rural Côte d’Ivoire. We leverage a longitudinal randomized control trial of Allô Alphabet, an experimental EdTech developed to support literacy learning at home for primary-school children in rural Côte d’Ivoire between 2019 and 2021 (Madaio et al., 2019a, 2019b; Jasińska et al., in preparation). The occurrence of COVID-19 mid-way through the Allô Alphabet study provided the opportunity to examine the impacts of COVID-19 school closures on child labor and learning as well as explore how supplemental educational resource use during school closures (i.e., the Allô Alphabet platform, as well as additional family support for learning at home) related to child labor and learning. In this longitudinal study, we examined the associations between COVID-19 school closures, children’s educational resource use during school closures, children’s cocoa-related work activities, and learning outcomes. We compared children’s cocoa-related work activities and learning outcomes pre-, during, and post-COVID-19 school closures. Additionally, we examined whether educational resource use during school closures (e.g., using EdTech at home, receiving support from parents, older siblings, and tutors for learning) may mitigate any increase in cocoa-related work activities associated with school closures. We focused on EdTech as a complementary option to traditional educational support. EdTech is becoming more relevant in an increasingly technologically advanced world. However, EdTech options, and research on the effectiveness of EdTech in low- and middle-income countries (LMICs) across the Majority World is limited (Evans and Mendez Acosta, 2021). Existing research on COVID-19 and educational impacts predominantly comes from high-income countries in the Minority World, despite the fact that LMICs in the Majority World have the largest burden of the global learning crisis. Given the existing educational issues enhanced by COVID-19, and gap in research, exploring the role of EdTech in providing education in LMICs during school closures is critical.

Child labor in Côte d’Ivoire

In Côte d’Ivoire, agriculture is the primary economic sector, with a significant emphasis on cocoa farming. The industry is predominantly comprised of informal workers consisting of small household farmers and their families (Andvig et al., 2001; Jacquemin, 2006). The National Institute of Statistics in Côte d'Ivoire (INS, 2005) indicated that, nationally, 22% of children aged 5 to 17 were economically occupied in the agricultural sector. In cocoa-growing regions in the south of Côte d'Ivoire, 78% of the one million children in small cocoa farming households are economically active (African Development Fund, 2018). Approximately 38% of these children face hazardous working conditions (Sadhu et al., 2020).

The most common form of child work in Côte d’Ivoire is non-hazardous agricultural cocoa work on a family farm (UNICEF, 2019). Importantly, in many rural areas in Africa, child work is a means of developing vocational skills, socialization, and transmission of intergenerational knowledge (Ango et al., 2022). Distinguishing between hazardous child labor, child labor, and child work has important implications for research and providing solutions, yet it is hard to distinguish under Ivorian work-related norms and the lack of reporting infrastructure (Abdullah et al., 2022).

Child labor versus work

It is important to differentiate between hazardous labor, child labor, and child work. Not all working children are child laborers, and not all child labor is hazardous. The definition of child labor is internationally bounded by three United Nations conventions around two context-sensitive elements: age and the nature, duration and conditions of productive activities performed by the child.¹ This definition is further defined by the System of National Accounts (SNA) as encompassing (i) any economically active child under the age of 12; (ii) children 12–14, engaged in productive activities outside permissible light work; (iii) children aged 17 and younger engaged in “hazardous” activities (those affecting the child’s safety, physical, and mental development) or the “worst forms of child labor” (e.g., commercial sexual exploitation and armed conflict; International Labor Organization ILO, 2018). Child work can help develop life-skills and knowledge, and is distinguished from child labor when the work does not interfere with their education and enjoyment of rights. On the other hand, child work can become child labor if the work falls within the SNA boundary; depriving children of their childhood, their potential and their dignity, and is harmful to physical and mental development (International Labor Organization ILO, 2017a). Further, work can be deemed hazardous if it meets one or more of the following categories: (1) land clearing, (2) carrying heavy loads, (3) exposure to agro-chemicals, (4) sharp tool use, (5) long working hours, or (6) work between 7 p.m. and 7 a.m. or during school/business hours (Sadhu et al., 2020).

Education and child labor

School enrollment rates of cocoa farming families estimate that 80% of children who attend school also engage in some form of cocoa agricultural work (Sadhu et al., 2020). Child labor that occurs during school hours decreases school attendance (Ango et al., 2022; Jensen and Nielsen, 1997; Kruger, 2007). Child labor is also related to late school enrollment and higher rates of school dropout (Canagarajah and Nielsen, 2001; Wortsman et al., 2024b). Studies by UNESCO (Lanoue, 2003; UNESCO UIS, 2015) on SSA and Côte d’Ivoire found that up to a third of children will start school late (i.e., older than age 6). Children who are older for their grade are more likely to do poorly in school, dropout, and participate in child labor (Gubbels et al., 2019; Kuepie et al., 2015; Oruko et al., 2015; Rosati and Rossi, 2003; Whitehead et al., 2024; Wortsman et al., 2024b).

To combat low enrollment rates, Côte d’Ivoire mandated compulsory school enrollment in 2015. Consequently, enrollment has been improving (World Bank, 2023). For example, in 2019 Cote d’Ivoire’s enrollment rate was 94%, up from 77% in 2011 (World Bank, 2023). However, only 74% of children actually complete primary school in Côte d’Ivoire, the remaining 26% may have never entered school or may have dropped out (UNESCO, 2022; Wortsman et al., 2024b). Low SES, low attendance, and poor school quality have been linked to drop out rates and threaten school completion (Ampiah and Adu-Yeboah, 2009; Kuepie et al., 2015; Wortsman et al., 2024b). Importantly, child labor and schooling are not mutually exclusive. For every child laborer who has reached compulsory school age but is excluded from schooling, two others struggle to balance the concurrent demands of school and work (Canagarajah and Nielsen, 2001; International Labor Organization (ILO) and United Nations Children’s Fund (UNICEF), 2021; Sangli et al., 2022).

Child labor has negative consequences for learning, academic outcomes, and child well-being (Ibrahim et al., 2019; ICI, 2022; ILO, 2022; UNESCO, 2020). Currently, Côte d'Ivoire’s high rates of child labor are paired with an alarmingly high level of illiteracy; as 82% of Ivorian 10-year-olds and 53% of the entire population over the age of 15, are illiterate (Azevedo et al., 2021; UNESCO UIS, 2018). The physical effects of working long hours, including poor growth and malnutrition, have compounding negative effects on the overall well-being and learning of child laborers (Ibrahim et al., 2019). The direct impacts of missing school, as well as the indirect impairments to physical and mental well-being, leave little to no room for educational activities beyond school (i.e., homework and revision) and attentive learning in class (Guarcello et al., 2008). School attendance and educational activities are compromised in favor of work, or learning is inefficient because the negative effects of work hinder learning.

Education may also support positive outcomes by acting as a protective factor against child labor. Without access to quality education, participation in child labor increases (Ango et al., 2022; Canagarajah and Nielsen, 2001; Mohammed, 2023). In Ghana and Côte d’Ivoire, child labor rates are lower in communities with a school, and child labor rates increase with increased distance to the nearest school (ICI, 2019, 2022). School quality is also an important factor; children in Côte d’Ivoire who attended higher quality schools had a lower probability of working on a cocoa plantation relative to peers attending lower quality schools (Wortsman et al., 2024a). In Ghana, teachers protect against child labor involvement by encouraging parents to keep children in school and prioritize school-related work over labor (Mohammed, 2023). In Burkina Faso, a 5-year program providing educational support (i.e., awareness raising and social protection, increasing access to education, and strengthening institutional capacity and policies) lowered child labor rates by 35.59% (United States Department of Labor, 2017).

Poverty, Child labor, and education during the COVID-19 pandemic

Because poverty is a primary driver of child labor, economic shocks (e.g., cocoa price changes, climate change, pandemics) can exacerbate child labor rates and hinder access to education (Canagarajah & Nielsen, 2001; Cogneau & Jedwab, 2012; ICI, 2020). Households whose income decreased during a cocoa price drop crisis in 1990 experienced an increase in child labor and a decrease in school enrollment (Cogneau and Jedwab, 2012; ICI, 2020). Child labor is an expected response against the widespread school closures and economic instability caused by the COVID-19 pandemic (International labor Organization (ILO), 2020, International Labor Organization (ILO) and United Nations Childrenâ€™s Fund (UNICEF), 2021).

The COVID-19 pandemic both exacerbated an underlying risk factor for child labor—poverty—and removed one of the protective factors against child labor—education (Ango et al., 2022; ICI, 2019, 2020). In Côte d’Ivoire, schools were closed due to the COVID-19 pandemic for several weeks between March 2020 and May 2020, resuming in September 2020 after the summer break. The Ivorian Ministry of Education followed the OECD’s recommendations, making remote learning services available, and advising students to access educational materials via internet, television, radio, and SMS (Reimers et al., 2020). In particular, the Ivorian government launched “My School at Home” (Mon école à la maison), a distance learning service that provided prerecorded video and audio content for pre-primary, primary, and secondary students in accordance with their national curriculum (UNICEF, 2023). However, in Côte d’Ivoire, remote learning services were not widely accessible. Approximately half of the population lives in rural communities (47%); less than half of that rural population has access to electricity (45.2%), and even fewer have access to the internet (Global System for Mobile Communications [GSMA], 2017; World Bank, 2020). Without these essential components, children and families were left, in practice, without access to schooling. Indeed, in a scoping literature review of distance learning for SSA primary school students during COVID-19 closures, Wawire et al. (2023) found that most SSA countries emphasized multimodal distance learning solutions (i.e., radio instruction, television programs, mobile apps, electronic textbooks, and e-learning portals). They found most distance learning solutions were inaccessible to the marginalized populations due to a lack of resources and that “those lacking access to interventions consequently exhibited a greater magnitude of learning loss and were at-risk of lagging further behind their peers” (Wawire et al., 2023).

The COVID-19 pandemic was not an isolated incident: school closures due to epidemics, political turmoil, conflict, or teacher strikes/absenteeism are not uncommon in Côte d’Ivoire and other regions of SSA (Kizilcec et al., 2021; Verwimp and Van Bavel, 2014; Villegas et al., 2021). Evidence from previous school closure periods in West Africa (e.g., the 2013–2016 Ebola pandemic) indicates that children from lower socioeconomic status (SES) households and resource-poor areas face increased financial and familial pressure to work (Risso-Gill and Finnegan, 2015; Smith, 2021). The increases in child labor during the Ebola outbreak in West Africa, linked to school closures, had long-lasting negative effects on children’s academic outcomes and resulted in the highest rates of school dropouts post-pandemic (Smith, 2021).

When schools—a protective factor against child labor—are closed, educational resources, including family or external tutoring and/or EdTech can potentially fill the gap and support learning outcomes (Kizilcec et al., 2021; Madaio et al., 2020a). In other words, EdTech can accomplish what distance learning seeks to do: provide students with learning activities while schools are closed. Due to the design and focus on particular skills, EdTech may effectively complement or substitute for existing learning activities (Evans and Mendez Acosta, 2021). Kizilcec et al. (2021) found that families relied on EdTech platforms to supplement education during school closures due to political turmoil in Kenya and teacher strikes in Côte d’Ivoire. Indeed, there is strong parental support for using educational resources, including EdTech, in Côte d’Ivoire (Madaio et al., 2019a).

Current study

Our study aimed to investigate the links between children’s use of educational resources, including EdTech, child cocoa agricultural work, and learning outcomes during and in the aftermath of COVID-19 school closures. Although there is a growing body of evidence suggesting education is a protective factor against child labor, little work has explored whether educational resources, including EdTech, during school closures may replicate the protection that schools typically provide against child labor. Here, we examined children’s cocoa agricultural work pre-, during, and post-COVID-19 school closures. We further investigated children’s use of educational resources during school closures, including family reading support, private tutoring, and an EdTech application, Allô Alphabet, designed for low-literacy environments.

Allô Alphabet is an interactive voice response (IVR) system designed to provide literacy support to children and parents using simple mobile phones (which are widely used and accessible in rural regions; Jasińska and Guei, 2022; Madaio et al., 2020a). Madaio et al. (2019a, 2019b) developed and implemented Allô Alphabet, to support the literacy development of Ivorian children in rural cocoa farming communities. Allô Alphabet leveraged ubiquitous availability of mobile phones along with parents’ desires to help and enabled them to do so via interactive voice responses despite low adult literacy levels, unlike many EdTech which are often designed solely for the child, rely on literate parents, or depend on high-tech devices and access to data (i.e., computers or smartphones). Children call Allô Alphabet via a simple mobile phone and receive voice- and text-based reading lessons that build decoding skills (Madaio et al., 2019b).

Prior to the COVID-19 pandemic, a randomized control trial of Allô Alphabet was underway to investigate how simple, accessible technologies support literacy development in rural Côte d’Ivoire (Madaio et al., 2020b; Jasińska et al., in preparation). Children in the treatment group were given a phone with a SIM card to access Allô Alphabet, given training on how to use the system, and instructed to use Allô Alphabet five days a week for two years. This ongoing study revealed that the extent of Allô Alphabet usage positively predicted improvements in literacy over the two years (Jasińska et al., in preparation). Leveraging the same dataset, we focused on child cocoa agricultural work and examined whether and how children’s use of EdTech and other educational resources (family reading support and private tutoring) during Côte d’Ivoire’s COVID-19 school closures predicted agricultural work and learning outcomes. We asked three research questions:

(1) Were COVID-19 school closures associated with increased child cocoa agricultural work in rural settings?

(2) How did children’s literacy skills change between 2019 and 2021? Were COVID-19 school closures associated with learning loss?

(3) How did educational resources used during school closures relate to child cocoa agricultural work during school closures and children’s learning outcomes after schools reopened?

To address these research questions, we examined (1) child cocoa agricultural work pre- (Fall 2019), during (Fall 2020), and post- (Spring 2021) COVID-19 school closures, (2) children’s use of educational resources, relying on child reports of educational activities during school closures and randomization to the Allô Alphabet literacy intervention, and (3) children’s learning outcomes (pre- and post-COVID-19 school closures), specifically focusing on literacy skills. We hypothesized that educational resource use during school closures, or lack thereof, would be related to child cocoa agricultural work and that children’s academic outcomes would be related to both cocoa agricultural work and use of educational resources. To that end, we predicted an increase in children’s agricultural work during COVID-19 school closures and that educational resource use during school closures would be associated with lower levels of child agricultural work. We further predicted that children who reported educational resources use and lower levels of cocoa agricultural work will have better literacy scores following the COVID-19 school closures. Lastly, we explored the impact of school closures on growth in literacy skills over two academic years (2019–2020—school closures due to COVID-19; 2020–2021—schools fully open).

In order to meet SDG 8.7, it is imperative that detriments to progress be mitigated and potential consequences of the pandemic be explored. If education can be supplemented by additional resources (i.e., family support, tutoring, and EdTech), it may provide a critical protective factor against child labor particularly during periods of heightened risk (Guarcello et al., 2008; ICI, 2019; Jasińska and Guei, 2022; Madaio et al., 2019a; Wortsman et al., 2024). The findings of this study may therefore provide insights on solutions and policies that are needed to address the cycle of poverty, poor educational quality, child labor, and their intersections, as well as support children’s learning recovery in the aftermath of the pandemic.

Methods

Participants

955 primary-school children between the ages of 8 and 14 years from 8 rural communities and 32 schools in the La Mé region of Southern Côte d’Ivoire participated in the study (N_girls = 465, M_age = 10.45, SD_age = 1.28; N_boys = 489, M_age = 10.79, SD_age = 1.24). La Mé is a rural and agricultural cocoa-producing region. Children were enrolled in a 2-year pre-registered randomized-control trial (RCT) impact evaluation of an EdTech intervention, Allô Alphabet (Madaio et al., 2020b), aimed at supporting literacy development in rural school children who work in cocoa agriculture (N_intervention = 614; N_control = 340). Children were enrolled in CM1 (the equivalent of fifth grade) in Fall 2019. In Spring 2021, 835 children were enrolled in CM2 (the equivalent of sixth grade) and 108 children were repeating CM1 (grade data was missing for 11 children). Children with psychological disorders and physical, auditory, or visual impairments were excluded. It is uncommon for schools to screen for or keep records of children with developmental or neurological disorders in rural Côte d’Ivoire. Children who markedly suffer from such conditions rarely attend school. Therefore, no children were excluded due to these criteria. The study was approved by the University of Toronto Research Ethics Board, the Ivorian Ministry of Education, and Comité de Gestion d'Ecoles (COGES; the parent-teacher group in Côte d'Ivoire).

Measures

Children were assessed prior to the COVID-19 pandemic (Time 1: Fall 2019) at the start of the 2019–2020 academic year (Y1) and after schools re-opened (Time 2: Fall 2020 and Time 3: Spring 2021) in the 2020–2021 academic year (Y2). See Figure 1 for data collection timeline.

Figure 1.

Data collection timeline.

At Time 1 (Fall 2019), children completed a battery measuring child agricultural work and literacy skills. At Time 2 (Fall 2020), our team implemented a remote phone-based literacy assessment to reduce potential for COVID-19 transmission to the interior of the country. Only a brief literacy assessment was conducted with a subset of children (n = 690) and no other data were collected (see Sobers et al., [2023] for details of the remote phone-based data collection). At Time 3 (Spring 2021), children completed a battery measuring child agricultural work and literacy skills. At Time 3, children were also asked to report retrospectively about their experiences during the period of school closures (March–May 2020). Due to lockdown and travel restrictions, we did not interview children between March and May 2020.

Background questionnaire

A background questionnaire was administered pre-school closures (Time 1). The questionnaire included demographic and socioeconomic information including child basic demographics (i.e., age), literacy status of other family members, and family SES. Family SES was measured using a 15-item household inventory (e.g., children were asked to indicate whether their homes had running water, electricity, etc.) that accompanies the RTI International (2015) Early Grade Reading Assessment (EGRA) and has been previously used with primary-school children in Côte d’Ivoire (Jasińska et al., 2022a; 2022b; Khan et al., 2024; Wortsman et al., 2024a). See Supplementary Materials for discussion of household inventories as SES in low-resource contexts.

Post-school reopening (Time 3), children were asked about the effects of the pandemic and school closures on their activities (e.g., whether time spent studying or playing or doing household chores increased, decreased, or remained the same), agricultural work (e.g., whether time spent working on a cocoa plantation increased, decreased, or remained the same), and where they spent their time (e.g., mostly at home, mostly in the field) during school closures. Children were also asked about the pandemic generally (e.g., whether they understood why schools were closed and whether they followed COVID-19 protocols).

Child cocoa agricultural work

Tulane University's (2015) Survey Research on Child Labor in West African Cocoa Growing Areas, which has previously been used in Côte d’Ivoire (Lichand and Wolf, 2025; Wortsman et al., 2024a), was administered pre-school closures (Time 1) and after schools reopened (Time 3) as a measure of children’s cocoa agricultural work. The survey categorizes work activities into economic, domestic, and agricultural (cocoa). Children provided “yes/no” responses to whether they had engaged in cocoa agricultural activities in the last 12 months.

Cocoa agricultural activities including land preparation (e.g., clearing fields, felling and chopping trees, burning, digging tree stumps, and cutting stakes), planting (e.g., planting seedlings, aligning and planting stakes, and preparing/digging/planting nurseries), farm maintenance activities (weeding, using agrochemicals, transporting water, maintaining and pruning cocoa trees), harvesting (e.g., plucking pods, gathering and heading pods, and breaking cocoa pods), and post-harvest activities (e.g., fermenting and transporting cocoa beans, and drying the cocoa beans).

Post-school reopening, children were also asked to retrospectively report changes in their work responsibilities during school closures (i.e., whether the amount of time spent working in the fields increased, decreased, or stayed the same).

Educational resource use

Children completed a questionnaire about the forms of educational support they received during COVID-19 school closures at Time 3. The questionnaire asked whether children used EdTech resources during school closures; child-reported EdTech usage was corroborated with log data recorded in the Allô Alphabet platform. The questionnaire also asked whether children received any tutoring from family (parents, siblings) or a hired tutor. Our earlier work in Côte d’Ivoire found that parents in rural communities commonly hired private tutors for their children (Madaio et al., 2019a).

Literacy assessments

Children’s literacy skills in French were assessed pre-school closures (Time 1) and post-school reopening (Time 2 using remote phone-based assessment and Time 3 in person) using the French language Early Grade Reading Assessment (EGRA), a widely used measure of literacy skill in Majority World contexts. The French language EGRA has been previously used with Ivorian primary-school children (Ball et al., 2022; Brice et al., 2024; Jasińska, Akpe, et al., 2022; Jasińska, Zinszer, et al., 2022; Sobers et al., 2023; Whitehead et al., 2024; Zinszer et al., 2023a, 2023b) and in other Francophone West African countries (Senegal, Mali). The EGRA consisted of grapheme (e.g., oi, ch, T), word (e.g., ami, été), and pseudoword (e.g., doupé, zode) reading subtasks (Gove and Wetterberg, 2011; RTI International, 2015). Children were asked to read the items quickly and audibly for each subtask (100 graphemes, 50 words, 50 pseudowords; short passage) with each item only being used once. Graphemes were unigrams (e.g., “a”) and bigrams (e.g., “ch”) that constitute single orthographic units in French. Words were familiar words, confirmed by native speakers of Ivorian French. Pseudowords were mono- or disyllabic strings that conformed to French syllabic and phonotactic structure, but do not exist in French. The maximum time to complete each subtask was 60 seconds. The subtask was terminated if the first ten items (grapheme) or first five times (word, pseudoword, passage) in each subtask were read incorrectly (Gove and Wetterberg, 2011; RTI International, 2015). After reading the short passage, children answered 6 comprehension questions. Each subtask began with practice with example items where participants were provided with feedback and scaffolding.

Procedure

Children were interviewed and completed assessments one-on-one during school hours with trained researchers pre-COVID-19 school closures (Time 1: Fall 2019) and post-COVID-19 school reopening (Time 3: Spring 2021). Children also completed a remote phone-based literacy assessment (Time 2: Fall 2020). Post-school reopening, children were asked questions about their experiences during COVID-19-related school closures (Spring 2020). Data were collected in accordance with ethical guidelines and approval from the Ivorian Ministry of Education and COGES. All study procedures were conducted by researchers who spoke Ivorian French and the local languages of the community. All researchers had at minimum a Bachelor’s degree in a relevant field (e.g., psychology, education, and sociology), and many were graduate-level students at the time of assessment. Prior to beginning work in the communities, the research team presented and discussed the study with community leaders (village chiefs, school administration, parents’ organizations). Informed consent was verbally obtained from village chiefs, elders, and community members following culturally appropriate procedures (see Jasińska and Guei [2018] for details about informed consent procedures for low-literacy rural communities in Côte d’Ivoire). Children were compensated with small gifts (e.g., book, school supplies). Research Electronic Data Capture (REDCap), an online data collection platform, was used for the collection and maintenance of data (Harris et al., 2009; 2019).

Data analysis

All data analysis was done in R (R Core Team, 2023); see Supplementary Materials for specific packages used.

To address our first research question, whether COVID-19 school closures are linked to increased child cocoa agricultural work in rural settings, we conducted a χ² test comparing the proportion of children who reported working in cocoa agricultural pre-, during, and post-school closures, and a multilevel logistic regression predicting child-reported cocoa agricultural work by age, gender, SES, maternal literacy, and time (pre-, during, post-school closures), with a random intercept for school.

To address our second research question regarding whether COVID-19 school closures were linked to learning loss, we conducted longitudinal analyses of literacy outcomes (letter reading, word reading, pseudoword reading, passage comprehension, passage reading) measured before COVID-19 school closures at Time 1 (in person) and after schools reopened at Time 2 (remote) and Time 3 (in person). Model fit of linear and non-linear models was compared using likelihood ratio tests (LRTs) for each literacy outcome (letter reading, word reading, pseudoword reading, passage comprehension, passage reading) and the better fitting model was retained (see Supplementary Materials for additional details). Linear and non-linear mixed-effects growth models were estimated for each literacy subtest, with time, treatment, time × treatment, age, gender, and SES as fixed effects, and random intercepts for child and school.

To examine whether literacy growth rates differed across demographic characteristics, a two-stage approach was used, child-level linear growth slopes were first extracted with random intercepts and slopes for time, and then regressed on age, gender, SES, and treatment condition. Finally, change in scores between Time 1 and Time 2 (Δ T2–T1), and Time 2 and Time 3 (Δ T3–T2) were calculated for each literacy assessment, where Δ T2–T1 represented change in literacy skill during an academic year (2019–2020) that encompassed COVID-19 school closures and Δ T3–T2 represented change in literacy skill during an academic year (2020–2021) without school closures. Pairwise t-tests were conducted on Δ T2–T1 and Δ T3–T2 to determine if change in literacy skills differed across the two academic years and was used as an indicator of potential COVID-19-related learning loss during the 2019–2020 academic year. Only children who were tested at all three time points were included in descriptives, growth models, regression models, and change score analyses (n = 498–596; dependent on reading subtest, see Table 1).

Table 1.

Sample characteristics and descriptive results for agricultural work and literacy outcomes.

Variable		Mean	SD	Min	Max
Age		10.63	1.3	8	14
Gender (% male)		51.40%
Maternal literacy (% literate)		53.70%
SES (15 items)		6.59	2.73	0	15
Child work activities
Pre-school closures	Domestic work (# of activities)	4.78	1.17	0	7
	Economic work (# of activities)	4.03	1.77	0	9
	Cocoa agricultural work (# of activities)	5.55	6.67	0	24
	Cocoa agricultural work (% of children)	49.60%
During school closures	Cocoa agricultural work (% of children)	57.20%
Post-school reopening	Cocoa agricultural work (% of children)	54.40%
Literacy scores
Pre-school closures (Time 1)	Letter reading (100 items; n = 585)	24.99	19.53	0	89
	Word reading (45 items; n = 587)	17.51	14.64	0	45
	Pseudoword reading (45 items; n = 584)	13.63	13.85	0	45
	Passage reading (45 items; n = 596)	15.93	17.41	0	45
	Passage comprehension (6 items; n = 498)	1.06	1.93	0	6
	Reading composite (% correct; n = 558)	31.3	26.0	0	96
Post-school reopening (Time 2)	Letter reading (100 items; n = 585)	37.88	22.71	0	100
	Word reading (45 items; n = 587)	21.54	15.54	0	45
	Pseudoword reading (45 items; n = 584)	17.97	15.79	0	45
	Passage reading (45 items; n = 596)	20.69	18.46	0	45
	Passage comprehension (6 items; n = 498)	1.93	2.33	0	6
Post-school reopening (Time 3)	Letter reading (100 items; n = 585)	54.48	23.72	0	100
	Word reading (45 items; n = 587)	27.24	15.65	0	45
	Pseudoword reading (45 items; n = 584)	23.21	15.74	0	45
	Passage reading (45 items; n = 596)	25.73	19.47	0	45
	Passage comprehension (6 items; 498)	2.49	2.54	0	6
	Reading composite (% correct; n = 558)	55.7	29.2	0	100

Note: Time 2 was a subset of children assessed via phone due to COVID-19 travel restrictions (see Sobers et al., 2023 for details). Literacy descriptives are based on subtest-specific complete-case samples. Children were included for a given literacy outcome if they had data at all three time points for that subtest.

To address our third research question regarding how educational resources use during school closures relates to child cocoa agricultural work and children’s learning outcomes, we first conducted a multilevel multinomial regression predicting increase, decrease, or no change in cocoa agricultural work by age, gender, SES, pre-COVID-19 cocoa agricultural work, family/tutor educational resources, and EdTech educational resources or Allô Alphabet intervention assignment with a random intercept for school. Two models were evaluated: Model 1 included child-reported EdTech educational resource use during school closures (“EdTech educational resources”); Model 2 included Allô Alphabet intervention assignment (treatment vs. control). Both predictors were not included in the same model given that children assigned to the Allô Alphabet treatment condition had access to an EdTech intervention and therefore reported higher EdTech educational resource use; however, both models were conducted given that children assigned to treatment may not have used the intervention.

To address how educational resources use during school closures related to children’s learning outcomes, we conducted a multilevel linear regression to predict literacy scores post-school reopening by age, gender, SES, family/tutor educational resources, EdTech educational resources, and cocoa agricultural work during closures with a random intercept for school. A composite literacy score of letter, word, and pseudoword reading was computed and used in this analysis as passage comprehension scores were low (Time 1 passage comprehension floor effect = 72.4%, skew = 1.81). Pseudo-R² or R² values were computed for all models. We treat the coefficient on treatment assignment (Allô Alphabet) as causal due to randomization at the school level; all other predictors (e.g., self-reported EdTech use, family/tutor support, and work during closures) are interpreted as associational.

Results

Descriptive results

Descriptive results are shown in Table 1. The majority of children knew that their school was closed due to the COVID-19 pandemic (88%), were aware that COVID-19 was a disease (72%), and practiced protective measures (87%; see Figure 2).

Figure 2.

Children’s COVID-19 knowledge and experiences. Note: Panels report children’s knowledge of the reasons for school closures, practice of preventive health measures, educational activities during COVID-19 school closures, and COVID-19 impacts at home.

Children’s educational activities during school closures

Children reported that school closures reduced their well-being (e.g., playtime, study time, availability of study space, and food security) but increased parental tutoring (see Figure 2). About 50% of children reported doing some educational activities during school closures, including schoolbook exercises, lesson review, and reading, with their parents, siblings, or a tutor. Children who reported doing schoolbook exercises, lesson review, and reading with their parents, siblings, or a tutor were more likely to have a literate mother (b(SE) = .601(.217), z = 2.77, p = .006).

The control and treatment groups were tested for equivalence at Time 1, and the treatment group was found to have lower scores in French language and literacy, and higher rates of participation in child cocoa agricultural work. As the focus of the current paper is not the intervention impact per se, the full analysis of Time 1 equivalence can be found in Jasińska et al. (in preparation). The table of Time 1 equivalence on the measures reported on in this study can also be found in the supplementary materials of this paper. However, as the differences were all in the opposite direction to the effects reported in this study, they, if anything, bolster the conclusions.

Research Question 1: Were COVID-19 school closures associated with increased child cocoa agricultural work?

Number of children reporting working in cocoa agriculture

Pre-school closures, 49% of children reported working in cocoa agriculture. During school closures and at post-school reopening, the number of children reporting working in cocoa agriculture increased to 57% and 54%, respectively (ꭓ²(2, N = 954) = 12.116, p = .002). Post-hoc chi-squared tests showed that the number of children reporting working in cocoa agriculture was significantly higher at post-school reopening and during school closures relative to pre-school closures (pre- vs. post-school closure: FDR-adjusted p = .04; pre- vs. during school closure FDR-adjusted p = .002), but the difference observed between post-school reopening and during school closures was not significant (post vs. during school closure FDR-adjusted p = .229).

A logistic regression found that children who were older, male, from lower SES backgrounds, or whose mother was not literate were significantly more likely to work in cocoa agriculture. This model also found that working during school closures and post-school reopening was greater than pre-school closures, but no significant differences between working post-school reopening and during school closures were found (see Table 2).

Table 2.

Predictors of child- reported cocoa agricultural work.

Predictor	OR	b(SE)
Age	1.241	0.216(0.051)***
Gender (ref = F)	2.017	0.701(0.129)***
SES	0.955	−0.046(0.025).
Maternal literacy	0.77	−0.262(0.132)*
Post-school reopening (ref = pre-school closures)	2.343	0.852(0.122)***
During school closure (ref = pre-school closures)	2.62	0.963(0.123)***
Post-school reopening (ref = during school closure)	0.894	−0.112(0.121)
Treatment (ref = control)	−0.908	0.096(0.129)
R²	0.275

Note. ***p < .001, **p < .01, *p < .05, and p < 0.1.

Number of children reporting increase in cocoa agricultural work during school closures

43% of children in the sample reported an increase in the amount of agricultural work during school closures, while 57% reported a decrease or no change (see Table 3). Of the children who reported an increase in the amount of agricultural work, 54% reported spending at least half of their time working on a cocoa plantation. Of the children who reported a decrease in the amount of agricultural work, 67% reported spending the majority of their time at home. Of the 67% who reported spending the majority of their time at home, 38% reported an increase in the amount of domestic chores, suggesting the decrease in agricultural work may have corresponded to an increase in domestic chores.

Table 3.

Increase vs. decrease in agricultural work and time spent on plantation vs. at home.

Agricultural work	%	Time spent on plantation vs home	%
Increase in agricultural work	43%	At least half the time spent on cocoa plantation	54%
Increase in agricultural work	43%	At least half the time spent at home	46%
Decrease in agricultural work	57%	At least half the time spent on cocoa plantation	33%
Decrease in agricultural work	57%	At least half the time spent at home	67%

Research Question 2: How did children’s literacy skills change between 2019 and 2021? Were COVID-19 school closures associated with learning loss?

Across all literacy outcomes, scores improved from Time 1 to Time 3 among children with complete data at all three time points for the respective literacy subtest (n = 498–596; see Figure 3 and Table 1). This pattern indicates improvement over time, although the magnitude of growth varied across subtests and between time intervals. For example, mean letter reading scores rose from 24.99 (SD = 19.53) at Time 1 to 37.88 (SD = 22.71) at Time 2 and 54.48 (SD = 23.72) at Time 3, representing an overall increase of 118.0%. For four of the five subtests, percentage gains were larger from Time 1 to Time 2 than from Time 2 to Time 3. Passage comprehension showed the same pattern, with larger percent gains from Time 1 to Time 2 (82.1%) than from Time 2 to Time 3 (29.0%). Similar patterns were observed for pseudoword reading (31.8% vs. 29.2%) and passage reading (29.9% vs. 24.4%). Word reading was the exception, showing slightly larger percentage gains from Time 2 to Time 3 (26.5%) than from Time 1 to Time 2 (23.0%), though the overall percent gain remained substantial (55.6%).

Figure 3.

Trajectories of mean literacy scores across three time points

Results from the mixed-effects growth models (see Table 4) indicated significant improvement over time across all literacy outcomes; letter reading (b = 9.667), word reading (b = 3.427), pseudoword reading (b = 4.779), passage reading (b = 4.865), and passage comprehension (b = 1.083) (all p and q < .002). A non-linear (versus linear) model provided a better fit for letter reading, word reading, and passage comprehension; linear models were retained for pseudoword reading and passage reading (see Supplementary Materials for model fit comparisons). Letter reading showed a small but significant quadratic effect of time (b = 2.214, p = .028), suggesting accelerated improvement from Time 1 to Time 3; however, this association did not survive FDR correction (q = .084). Age was negatively associated with literacy performance across all measures (all p < .001), and gender and SES were not significant predictors for any reading outcomes (gender: all p ≥ .522; SES: all p ≥ .302). Treatment effects were negative across all subtests (consistent with the Time 1 non-equivalency reported in Jasińska et al. [in preparation]), indicating lower Time 1 performance in the treatment group; however, these findings were not statistically significant. Time × treatment interactions were not significant for any subtest (all p ≥ .275), and Time² × treatment interactions were likewise non-significant (all p ≥ .508). Regression models predicting growth rates (per-child slope) across literacy subtests (see Table 5) showed that older age was associated with slower growth for all reading outcomes (all p and q < .001). Gender, SES, and treatment condition were not significantly associated with growth rates for any subtest. Paired-samples t-tests comparing change in literacy scores across adjacent intervals (ΔT2–T1 vs. ΔT3–T2) indicated that gains were significantly larger during the 2020–2021 academic year when children did not experience COVID-19 school closures (Time 2 to Time 3) for letter and word reading, and these differences survived FDR correction (q = .022 for both; see Table 6). No significant differences were observed for pseudoword reading or passage reading (all p ≥ .175). However, passage comprehension showed the opposite pattern, and gains were significantly larger from Time 1 to Time 2 than from Time 2 to Time 3 (p = .001, q = .007).

Table 4.

Mixed- effects growth models predicting literacy outcomes across time (children nested within schools).

Predictor	Letter			Word			Pseudoword			Passage reading			Passage comprehension
Predictor	b(SE)	p	q	b(SE)	p	q	b(SE)	p	q	b(SE)	p	q	b(SE)	p	q
Time	9.667 (2.097)	<.001***	<.001	3.427 (1.097)	.002**	.002	4.779 (0.291)	<.001***	<.001	4.865 (0.364)	<.001***	<.001	1.083 (0.184)	<.001***	<.001
Time²	2.214 (1.007)	.028*	.084	0.512 (0.527)	.332	.332	—	—	—	—	—	—	−0.148 (0.088)	.094	.140
Treatment (ref = control)	−4.224 (3.473)	.232	.232	−3.539 (2.349)	.141	.232	−2.915 (2.298)	.213	.232	−3.540 (2.751)	.207	.232	−0.681 (0.348)	.058	.232
Age	−3.672 (0.552)	<.001***	<.001	−2.628 (0.420)	<.001***	<.001	−2.512 (0.422)	<.001***	<.001	−2.822 (0.510)	<.001***	<.001	−0.333 (0.068)	<.001***	<.001
SES	0.077 (0.271)	.778	.900	−0.026 (0.205)	.900	.900	0.055 (0.209)	.792	.900	0.122 (0.251)	.627	.900	0.034 (0.033)	.302	.900
Gender (ref = F)	0.908 (1.418)	.522	.980	0.316 (1.077)	.770	.980	−0.268 (1.086)	.805	.980	−0.033 (1.306)	.980	.980	−0.067 (0.171)	.698	.980
Time × treatment	3.285 (3.008)	.275	.953	−0.128 (1.573)	.935	.953	0.179 (0.419)	.669	.953	−0.031 (0.521)	.953	.953	−0.181 (0.269)	.500	.953
Time² × treatment	−0.939 (1.445)	.516	.774	0.501 (0.755)	.508	.774	—	—	—	—	—	—	0.010 (0.129)	.939	.939
R² (marginal)	.26			.12			.12			.09			.14

Note: ***p < .001, **p < .01, *p < .05, and .p < 0.1. q = false discovery rate-adjusted p-values. Non-linear models are reported for letter reading, word reading, and passage comprehension; linear models are reported for pseudoword reading and passage reading.

Table 5.

Linear regression models predicting individual literacy growth slopes (children nested within schools).

Predictor	Letter			Word			Pseudoword			Passage reading			Passage comprehension
Predictor	b(SE)	p	q	b(SE)	p	q	b(SE)	p	q	b(SE)	p	q	b(SE)	p	q
Age	−0.540 (0.083)	<.001***	<.001	−0.261 (0.061)	<.001***	<.001	−0.305 (0.05)	<.001***	<.001	−0.410 (0.11)	<.001***	<.01	−0.076 (0.021)	<.001**	<.001
SES	0.012 (0.041)	0.777	0.777	−0.045 (0.030)	0.134	0.362	−0.041 (0.02)	0.154	0.362	−0.069 (0.05)	0.217	0.362	−0.008 (0.010)	0.451	0.563
Gender (ref = F)	0.149 (0.217)	0.491	0.819	−0.121 (0.159)	0.447	0.819	0.030 (0.14)	0.839	0.839	−0.227 (0.29)	0.436	0.819	−0.014 (0.054)	0.802	0.839
Treatment (ref = control)	0.144 (0.217)	0.506	0.844	0.338 (0.186)	0.079.	0.396	0.075 (0.18)	0.686	0.852	0.077 (0.40)	0.852	0.852	−0.088 (0.074)	0.241	0.603
R² (marginal)		0.07			0.04			0.05			0.03			0.04

Note: ***p < .001, **p < .01, *p < .05, and p < 0.1. q = false discovery rate-adjusted p-values. Pseudoword reading and passage reading estimated as linear; letter reading, word reading, and passage comprehension estimated as non-linear (quadratic).

Table 6.

Paired T- tests comparing change in scores across waves (ΔT2–T1 vs ΔT3–T2).

Measure	n	Δ T2–T1	Δ T3–T2	t	df	p	q
Letter reading	585	12.89	16.60	2.49	584	0.013*	0.022
Word reading	587	4.03	5.71	2.49	586	0.013*	0.022
Pseudoword reading	584	4.34	5.24	1.36	583	0.175	0.219
Passage reading	596	4.76	5.04	0.39	595	0.695	0.695
Passage comprehension	498	0.87	0.56	−3.22	497	.001***	.007

Note: ***p < .001, **p < .01, and *p < .05. Δ indicates change.

Research Question 3: How did educational resources use during school closures relate to child cocoa agricultural work during school closures and children’s learning outcomes?

Children’s educational activities during school closures and cocoa agricultural work

Children with access to EdTech (Allô Alphabet; treatment group) were significantly more likely to report a decrease versus increase in cocoa agricultural work. Children who reported family and tutor educational resources during school closures (i.e., reading and lesson review with a parent, sibling, or tutor) were marginally more likely to report a decrease in cocoa agricultural work during school closures. Male children and children who reported higher numbers of Time 1 cocoa agricultural work activities were more likely to report an increase versus decrease in cocoa agricultural work during school closures (see Table 7).

Table 7.

Predictors of child- reported increase, decrease, or no change in cocoa agricultural work during COVID- 19 school closures.

	Increase vs. decrease in cocoa agricultural work	Increase vs. no change in cocoa agricultural work	Decrease vs. no change in cocoa agricultural work
Predictor	b(SE)	b(SE)	b(SE)
Panel A
Age	0.124(0.079)	0.163(0.108)	−0.039(0.107)
Gender (ref = female)	0.701(0.208)***	0.193(0.287)	0.508(0.275).
SES	−0.009(0.038)	−0.027(0.052)	0.018(0.051)
Time 1 # of cocoa agricultural tasks	0.025(0.015)	0.006(0.02)	0.019(0.02)
Family/tutor educational resources	−0.561(0.236)*	0.081(0.293)	−0.642(0.3)*
Treatment (ref = control)	−0.594(0.211)**	−0.298(0.284)	−0.296(0.282)
R²	0.201	0.062	0.131
Panel B
Age	−0.742(1.233)	0.014(0.015)	−0.705(1.217)
Gender (ref = female)	0.112(0.079)	0.001(0.02)	0.111(0.079)
SES	0.157(0.108)	0.076(0.201)	−0.045(0.107)
Time 1 # of cocoa agricultural tasks	0.675(0.206)**	−0.045(0.276)	0.678(0.207)**
Family/tutor educational resources	0.184(0.287)	−0.549(0.242)*	0.491(0.276).
EdTech educational resources	−0.009(0.038)	0.063(0.304)	−0.009(0.038)
R²	0.158	0.047	0.122

Note: ***p < .001, **p < .01, *p < .05, and .p < 0.1. Panel A shows the model that includes Treatment Status, whereas Panel B shows the model that included children’s reported EdTech educational resource usage.

Children who self-report using EdTech resources during school closures also had higher recorded EdTech usage, as indicated by the log data from the Allô Alphabet platform, indicating that children’s self-reported educational activities during school closures were overall accurate. Relative to children who did not report using EdTech resources during school closures, children who self-report using EdTech resources completed a marginally significantly greater number of Allô Alphabet literacy exercises between March 16 and May 10, 2020—the period when Ivorian schools were closed (self-reported EdTech users: M = 7.52, SD = 19.2; non-EdTech users: M = 4.60, SD = 11.18; t(363) = -2.018, p = .054). We note that 3% of children were recorded as having completed Allô Alphabet exercises between March and May 2020, but had reported not using the platform during their interview post-school reopening.

Children’s access to educational resources, child cocoa agricultural work, and academic outcomes

Children who had poorer literacy scores pre-school closure were children who were older for their grade (i.e., enrolled in school late and repeated more grades) and who reported working in cocoa agricultural work during school closures. These children predictably showed significantly poorer literacy scores post-school reopening. There was a marginally significant positive effect of using EdTech educational resources during school closures on post-school reopening literacy scores (see Table 8).

Table 8.

Predictors of children’s literacy scores at time 3.

Predictor	b(SE)
Time 1 literacy	0.927(0.035)***
Age	−2.08(0.608)***
Gender (ref = female)	1.733(1.546)
SES	−0.249(0.293)
Maternal literacy	−1.539(1.566)
Cocoa agricultural work pre-school closures (# of activities)	−0.131(0.121)
Family/tutor educational resources	2.936(1.923)
EdTech educational resources	3.101(1.724).
Cocoa agricultural work during school closure	−4.012(1.562)*
R²	0.569

Note: ***p < .001, **p < .01, *p < .05, and .p < 0.1.

Discussion

In this two-year longitudinal study, we examined (1) whether COVID-19 school closures were associated with increased child cocoa agricultural work and (2) literacy outcomes in rural settings, and (3) whether educational resource use, including EdTech, during school closures was related to child cocoa agricultural work and children’s learning outcomes. As predicted, we found that COVID-19 school closures were associated with a significant increase in the number of children reporting involvement in cocoa agricultural work. Overall literacy scores increased over the study period, albeit literacy gains were significantly lower during the 2019–2020 academic year when COVID-19 school closures occurred relative to the 2020–2021 academic year that remained uninterrupted in Côte d’Ivoire. Random assignment to receive access to the Allô Alphabet EdTech intervention (relative to control group) as well as family and tutor educational resources use during school closures increased the likelihood that children reported a decrease in agricultural work during school closures. Importantly, cocoa agricultural work during school closures was associated with lower literacy scores the following year, while access to EdTech was marginally linked to better literacy scores once children resumed schooling regardless of reported agricultural work activity.

COVID-19 school closures and children’s agricultural work

Previous research has shown that COVID-19 school closures reduced students’ average daily learning time and exacerbated inequalities in home learning environments (Grewenig et al., 2021). However, research regarding the impact of COVID-19 on educational outcomes has predominantly come from high-income Minority World countries, where children typically received some form of continuing education from their school systems via online learning. In contrast, in many LMICs in the Majority World such as Côte d’Ivoire, students did not, and most often could not, access education when school closures took effect. Corroborating previous work on the relation between access to education, poverty, and child labor (Ango et al., 2022; Canagarajah and Nielsen, 2001; Mohammed, 2023; United States Department of Labor, 2017), we found that during COVID-19 school closures, the number of children who reported working in cocoa agriculture significantly increased and persisted even once schools reopened. These findings are in line with estimates that the COVID-19 pandemic would lead to increases in child labor (International labor Organization (ILO), 2020, International Labor Organization (ILO) and United Nations Children’s Fund (UNICEF), 2021), and previous work demonstrating links between school closures and higher rates of children participating in economic activities and labor (Elston et al., 2017; Villegas et al., 2021).

Other studies from SSA show that families from lower SES backgrounds may be more likely to involve their children in economic work (Enebe et al., 2021; Vadivel et al., 2023). Not surprisingly, we found that children with more risk factors (lower SES backgrounds, maternal illiteracy) were more likely to engage in cocoa agricultural work during school closures. Boys were more likely to report being involved in cocoa agriculture and to report an increase in agricultural work relative to girls. Our results align with previous research showing that boys had more reduced learning time and more frequently replaced learning time with detrimental activities during COVID-19 school closures (Grewenig et al., 2021) and findings of child work trends in SSA that show the amount of domestic work is higher for girls while boys are more likely to work outside home (Canagarajah and Coulombe, 1999; Dumas and Lambert, 2005; Ilahi, 2001). Still, considering gender is relevant to combatting child labor as domestic work is considered to be a hidden form of child labor because it is unpaid and often goes unreported (e.g., Gibbons et al., 2005; Webbink et al., 2012). Despite differences in type of work, gender had no significant effect on Time 3 literacy. In many contexts, girls are disproportionately affected during crises due to increased domestic responsibilities, reduced access to schooling, or heightened vulnerability to dropout (UNESCO, 2020). The absence of meaningful gender disparities in literacy growth suggests that pandemic-related learning loss and recovery operated similarly for boys and girls in this sample. Rather than exacerbating gender gaps, the disruption appears to have produced broadly shared setbacks.

COVID-19 school closures and children’s literacy outcomes over time

Overall literacy skills improved across the study period; however, growth was smaller during the 2019–2020 academic year (with the exception of passage comprehension) when children’s schooling was disrupted by the COVID-19 pandemic in comparison to the 2020–2021 academic year that was uninterrupted. This pattern is consistent with the widespread instructional disruption caused by the COVID-19 pandemic, during which school closures, reduced contact hours, and limited access to learning materials altered typical learning trajectories (Angrist et al., 2021). School closures were relatively short in Côte d'Ivoire; schools were closed on March 25, 2020, partially reopened on May 6, 2020, and the subsequent academic year resumed in full in September 2020. Potential learning losses may have been mitigated by the relatively short period of school closures. However, high rates of student absenteeism and lower quality education undermine learning gains even when schools are fully open. Additionally, we found that older children showed less growth in literacy over the study period. Many older students in low-resource settings are “over-age for grade” due to late school entry, interruptions in schooling, or grade repetition, factors consistently associated with weaker foundational literacy and numeracy skills. Existing research shows that children who are over-age for their grade show poorer literacy skills (Jasińska et al., 2024; Whitehead et al., 2024).

Educational resources use during COVID-19 school closures, child agricultural work, and learning outcomes

About 50% of the children reported doing some educational activity (i.e., receiving support from a family member or tutor or using EdTech) during school closures. Our hypothesis that children’s use of educational resources was related to a reduction in child agricultural work during the COVID-19 period of school closures and positive learning outcomes was supported; children enrolled in the Allô Alphabet intervention and children who reported family support or tutoring during school closures were more likely to report a decrease in agricultural work during school closures. This finding is in line with a growing body of research that has established a link between access to quality education and lower rates of child labor (ICI, 2019; Wortsman et al., 2024a). Further, this finding supports previous work on the potential of supplemental educational resources such as EdTech during school closures (Kizilcec et al., 2021) and has implications beyond the COVID-19 pandemic given that school closures due to epidemics, civil unrest, or teacher strikes are not uncommon across SSA (Abadzi 2009; Kizilcec et al., 2021). Within Côte d’Ivoire, and the broader context of SSA, school closures coupled with the lack of electricity and internet to support distance learning in rural areas (GSMA, 2017; World Bank, 2023) can render access to quality education scarce. Children are less likely to enter the labor force at a young age when they have access to quality schooling (Ango et al., 2022; Canagarajah and Nielsen, 2001; Mohammed, 2023).

Working in cocoa agriculture during school closures was significantly associated with poorer literacy skills one year later (post-school reopening). Importantly, the negative link between working in cocoa agriculture during school closures and literacy held even when controlling for Time 1 literacy skills and child cocoa agricultural work at the start of the previous school year (pre-school closures). It is well understood that children enter school with widely different literacy abilities impacted by SES and other risk factors that are outside of schools (Lonigan and Shananhan, 2009; Stanovich, 2009; Waldfogel, 2012). Often referred to as the “Matthew Effect,” the model predicts that initial literacy gaps widen for children who face incipient experiences or difficulties with reading and writing, creating a rich-get-richer and poor-get-poorer achievement gap. Greater attention is now being paid to how out-of-school inequalities are contributing to literacy gaps and the role of the Matthew Effect. Research points to the interrelated roles of parents, home environment, and provision of reading materials as critical factors (Lonigan and Shananhan, 2009; Stanovich, 2009; Waldfogel, 2012). In the context of COVID-19 school closures that exacerbate an existing learning crisis among Ivorian children (Jasińska et al., 2023), the Matthew Effect paints a desperate picture for already vulnerable children. Predictably, we found children who had poor literacy scores prior to the pandemic showed significantly poorer scores after the school closures, suggesting educational disparities may be widening.

Interestingly, EdTech use during school closures was marginally associated with better literacy outcomes. Several meta-analyses of educational intervention in SSA have found that investments in instructional technologies had the largest impact on children’s learning outcomes in comparison to other investments, including nutritional and health interventions, reducing class sizes, or cash transfers conditional on school attendance (Castillo et al., 2022; Conn, 2017; Jasińska & Guei, 2022; McEwan, 2015). Allô Alphabet was designed to supplement schooling, rather than replace it, suggesting that even in the best case, usage during school closures is likely insufficient to meaningfully increase literacy skills (Kizilcec et al., 2021). In our related work, we report the impact of Allô Alphabet use over two academic years on literacy skills (Jasińska et al., in preparation). Children who utilized Allô Alphabet more showed a significantly greater increase in literacy skills; a 10% increase in the number of literacy activities completed translated to more than half a standard deviation greater increase in performance in literacy tasks over the two years of the intervention (Jasińska et al., in preparation). Although the effect of EdTech use during school closures on literacy outcomes was marginal, these findings add to the evidence that educational resources, including EdTech, can potentially provide some support when schooling is unavailable. This is particularly relevant in Majority World contexts like Côte d’Ivoire, which can face frequent school closures beyond COVID-19, and for vulnerable populations, such as child laborers, who have limited access to school. Additionally, access to Allô Alphabet in the 2020–2021 academic year may have also boosted learning recovery after potential learning loss due to school closures in the preceding academic year. For example, Lichand and Doria (2024) showed that remedial education programs implemented in the aftermath of the pandemic boosted learning recovery in Brazil.

Importantly, Allô Alphabet was designed for all families regardless of parental literacy level. Previous research on educational resource usage in Africa found that illiterate mothers were less able to directly support their child (i.e., read with them and lesson review; Lloyd and Blanc, 1996; Namatende-Sakwa et al., 2022), and children’s participation in educational activities has been linked to their home environment and parental literacy (Jasińska et al., 2022; Madaio et al., 2019a). While illiterate mothers were less able to support their children with traditional educational resources, children used EdTech despite the high prevalence of maternal illiteracy in our sample. Promisingly, this underscores EdTech’s potential in contexts where maternal literacy is low or where there are fewer traditional education supports available and is in line with previous findings that Allô Alphabet is an effective self-directed educational resource (Madaio et al., 2020a; Jasińska et al., in preparation).

Limitations and future directions

A limitation of our work is related to children’s conception of work and our timing of data collection. Children were asked about changes in their work-related activities; however, we did not obtain a detailed report of children’s activities and cannot definitively state whether or not children worked in a range that would qualify it as child labor (i.e., excess hours). We asked children about changes in their child cocoa activities pre-, during, and post-school closures; however, these times were all at different points in the year. Depending on the cocoa harvest season, fluctuations in the amount of work may influence children’s responses. For example, the large cocoa harvest takes place each fall—a period when children’s agricultural workload would be highest. Bøås and Huser’s (2006) report on child cocoa agricultural work in Côte d’Ivoire and Ghana found that Fall (September) is the main cocoa-harvesting period and most time-consuming. These findings align with previous research on child cocoa production in SSA that found a significant increase of hazardous work and absence from school during cocoa-harvesting seasons (Luckstead et al., 2019; Tulane, 2015). Similarly, Angula (2024) found that in rural Colombian farming regions students miss school during harvesting seasons. Policy from Ghana, prohibiting children to be withdrawn from school during the harvesting season (Tulane, 2015), speaks to the strength of the economic pulls from children’s education.

We asked children about cocoa agricultural work during school closures which took place in the spring, which would suggest that the number of children reporting agricultural work would decrease. In fact, we found the opposite pattern, suggesting children’s agricultural work increased despite the overall lower demands associated with the non-harvest season. While we observed an increase in agricultural work during school closures relative to pre-COVID-19 pandemic, agricultural work did not significantly decrease once children resumed schooling. Additional research needs to be done to investigate parental decisions and post-pandemic economic needs to understand why agricultural work did not decrease once children resumed schooling. Persistent agricultural work after schools reopened may be due to the proclivity for older children in SSA to work International Labor Organization (ILO), 2017a; United Nations Population Fund (UNFPA) and Population Reference Bureau (PRB), 2012. A study by Diallo (2002) shows that in general the older a child gets the greater their physiology and physical fitness, making them more suitable and likely to work. Indeed, the children in this study were almost two years older post-school reopening relative to the start of the study and may explain, at least in part, the greater number of children reporting working. While this age-related increase in work is expected, the number of children who reported working in cocoa agriculture post-school reopening was numerically lower than the number who reported working during school closures, although non-statistically significant. This suggests that the increase in work reported during school closures is likely attributed to school closures rather than age-related increases in work.

Relatedly, reports of child agricultural work during COVID-19 school closures were all retrospective, which likely introduced errors as well as limited direct comparisons between the two other time points (collected in person). While there was no possibility to assess children’s agricultural work directly during the period of school closures (i.e., due to global travel restrictions, and without significantly increasing risk of COVID-19 transmission in communities with limited health infrastructure), this remains an important limitation of the work and our conclusions remain cautious. Children’s self-report of agricultural work has been shown to be reliable and valid within this population; children are less likely to report with social-desirability bias than their parents, and when compared to certifier data in previous studies children’s self-reported data was more accurate than parent report (Lichand and Wolf, 2025). While the participants of our study are likely child laborers given the average age (under 11 years old) combined with the temporal and physical demands of cocoa agricultural work, measuring child labor (i.e., precise amount of hours worked) was beyond the scope of this study. Therefore, we refer to the children’s agricultural activities as work. Therefore, while this study applies to child work, it also has implications for child labor.

Further studies should also investigate how educational resources can best be used to help reduce child labor and combat the learning crisis. There is still very limited research on access to quality education as a protective measure against child labor. Our study found that about 50% of children used some type of educational resource during school closures, meaning only half are receiving any potential academic and protective benefits. Research into factors that predict use of educational resources may provide additional insights into how to best leverage educational resources for children in this context.

Conclusion

Child labor exists within a complex cycle of poverty, cultural norms, and access to quality education (Ango et al., 2022; Canagarajah and Nielsen, 2001; ICI, 2019; Jensen and Nielsen, 1997; Kruger, 2007; Mohammed, 2023). Families from lower SES backgrounds may be more likely to involve their children in economic work and have less access to educational resources both contributing to lower academic outcomes (Enebe et al., 2021; Vadivel et al., 2023). Thus, during COVID-19 school closures, children from lower SES backgrounds may be more likely to face a dual-negative impact of working more and having less educational resources to support learning. School disruptions in SSA are common either due to teacher strikes, natural disasters, or epidemics (i.e., Ebola and Malaria), and COVID-19 was not the first time that students experienced a break from traditional schooling (Kizilcec et al., 2021). Our findings suggest that the COVID-19 school closures were associated with increased rates of agricultural work and decreased learning outcomes. However, educational resources, including family tutoring and access to EdTech, may help mitigate these negative effects, as their use was associated with decreased agricultural work.

Our longitudinal study investigated COVID-19 school closures and whether using educational resources (i.e., family and tutoring support, EdTech—Allô Alphabet) mitigated negative impacts of school closures on the resurgence of child labor and learning outcomes in agricultural regions dominated by cocoa production, pre-, during, and post-COVID-19 pandemic. The results show that school closures were significantly associated with an increase in child cocoa agricultural work that persisted even once schools reopened. In vulnerable communities, school closures may fuel a surge in child labor. Cocoa agricultural work during school closures was strongly associated with poorer learning outcomes, even after controlling for Time 1 literacy and agricultural work. However, educational resource use during school closures was linked to marginally higher literacy skills.

Children growing up in impoverished communities (i.e., vulnerable background and lower education) are most at risk of child labor and becoming entrenched in the poverty cycle. Educational resources designed for low-literacy, low-SES families may provide a solution. Investing in high-quality educational supports is an equitable and productive use of resources, especially in a context of limited resources, as it may help protect against child labor (Wortsman et al., 2024a) and can provide needed support during extended periods of school closures. This investment is likely to lead to better educational outcomes and in turn increase intergenerational access to education and resources, potentially disrupting the negative cycle between education, child labor, and poverty (OECD, 2012). These findings are not limited to the COVID-19 school closures but are applicable for other school disruptions and closures which occur frequently in SSA (Kizilcec et al., 2021). Our findings are relevant to the United Nations Target 8.7, eradicating child labor (United Nations, 2015) by furthering our understanding of the drivers and protective factors of child labor. Achieving SDGs requires hand-in-hand strategies that target multiple SDGs simultaneously (United Nations, 2015). It is necessary to invest in quality education, including educational resources, otherwise the all too common school disruptions risk hindering child learning, surging participation in child labor, and entrenching children poverty.

Supplemental material

Supplemental material - Educational resource use during COVID-19 school closures, child cocoa agricultural work, and learning outcomes in rural Côte d’Ivoire

Supplemental material for Educational resource use during COVID-19 school closures, child cocoa agricultural work, and learning outcomes in rural Côte d’Ivoire by Joshua Lim, A. Micaela Ostry, Yomna Elshabrawy, Stephanie Brogno, Nana N’Goh, Hermann Akpé, Fabrice Tanoh, Henry Brice, Amy Ogan, and Kaja K. Jasińska in Research in Comparative and International Education.

Footnotes

Acknowledgments

This work was funded by the Jacobs Foundation through a research grant (PIs: Jasińska, K. K., Ogan, A.). We would like to thank the Ivorian Ministry of Education for their support of our research efforts as well as for the exchanges and information meetings that took place during the project. Our thanks also go out to the Transforming Education in Cocoa Communities (TRECC) program for their support. Finally, we are exceptionally grateful to the families and children who participated in this study, and extend a special thanks to the village chiefs and elders, school principals and teachers, and members of the Comité de Gestion des Établissements Scolaires (COGES).

ORCID iD

Kaja K. Jasińska

Ethical considerations

The study was approved by the University of Toronto Research Ethics Board (Protocol #39740), the Ivorian Ministry of Education, and Comité de Gestion d'Ecoles (COGES; the parent-teacher group in Côte d'Ivoire).

Funding

This work was funded by the Jacobs Foundation through a research grant (PIs: K. Jasińska, A. Ogan) and a science capacity building grant to support Ivorian scholars (2015-1184; PI: K. Jasińska; Allô Alphabet).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Data is available at

Significance statement

The COVID-19-pandemic-related school closures were associated with an increase in child work in rural Côte d’Ivoire. Assignment to an educational technology intervention developed for low-literacy and low SES environments during school closures resulted in decreased child agricultural work during school closures. Child agricultural work during school closures was associated with poorer learning outcomes after schools reopened.

Supplemental material

Supplemental material for this article is available online.

Note

Author biographies

Joshua Lim is a PhD student in the Department of Curriculum, Teaching and Learning, specializing in Comparative, International and Development Education, at the Ontario Institute for Studies in Educationn, University of Toronto. His research focuses on alternative education for impoverished indigenous, and incarcerated children in Bolivia.

A. Micaela Ostry is a PhD student in the Department of Applied Psychology and Human Development at the Ontario Institute for Studies in Education, University of Toronto. Her research focuses on children's academic skills and cognitive development in diverse learning contexts.

Yomna Elshabrawy is an undergraduate student at the University of Toronto. Her research explores children's learning and socio-emotional development in diverse contexts.

Stephanie Brogno is a Clinical Research Coordinator at the Stanford University School of Medicine working in the Autism and Developmental Disorders Research Program. She completed a BA in Psychology and a B.Ed. at York University

Dr. Nana N'Goh completed a Ph.D. in Sociology of Economy and Innovations rom Félix Houphouët-Boigny University. He is a lecturer and researcher affiliated with the Laboratoire d'Analyse Regionale et d'Expertise Social in Cote d'Ivoire.

Dr. Hermann Akpé completed a Ph.D. in Social Sciences and Human Capital Development from the Institute Development and Anthropology (ISAD) at Félix Houphouët Boigny University. His research focuses on education and the quality of professional development at private institutions of higher education. He is a member of the National Committee for Early Childhood Integrated Development (CNDIDJE) in Cote d'Ivoire.

Dr. Fabrice Tanoh is an assistant professor of economic sociology in the Faculty of Social Sciences, Department of Sociology at Péléfero Gon Coulibaly University in Korhogo, Côte d’Ivoire. His research focuses on issues related to children's education and women's socioeconomic empowerment in relation to household livelihoods in cocoa-producing communities.

Dr. Henry Brice is a postdoctoral fellow at the Applied Pscyhology and Human Development Department at the University of Toronto. His research examines how individual differences in reading are impacted by individual cognitive factors and environmental factors.

Dr. Amy Ogan is an Associate Professor in the Human-Computer Interaction Institute at Carnegie Mellon University, with an appointment at CMU Africa. She is an educational technologist focusing on ways to make learning experiences more engaging, effective, and culturally relevant in emerging economies. Her research encompasses areas including: context and infrastructure in education, engendering rapport and relationships with social technologies to aid learning and supporting teacher professional development through novel technologies.

Dr. Kaja Jasińska is an Associate Professor in the Applied Psychology and Human Development Department at the University of Toronto. Her research examines how childhood experiences shape neurocognitive development and learning, focusing on understudied environments characterized by high poverty-related risks.

References

Abadzi

(2009) Instructional time loss in developing countries: concepts, measurement, and implications. The World Bank Research Observer 24(2): 267–290. https://doi.org/10.1093/wbro/lkp008

Abdullah

Huynh

Emery

, et al. (2022) Social norms and family child labor: a systematic literature review. International Journal of Environmental Research and Public Health 19(7): 4082. https://doi.org/10.3390/ijerph19074082

African Development Fund (2018) Republic of Côte d’Ivoire cocoa sector governance project (PAGFIC) appraisal report. https://www.afdb.org/fileadmin/uploads/afdb/Documents/Project-and-Operations/ECGF_COTE_D_IVOIRE_APPROVED_COCOA_SECTOR_GOVERNANCE_SUPPORT__PAGFIC_.pdf

Ampiah

Adu-Yeboah

(2009) Mapping the Incidence of school dropouts: A case study of commnuities in Northern Ghana. Comparative Education 45(2): 219–232. Available at: https://doi.org/10.1080/03050060902920625

Andvig

Canagarajah

Kielland

(2001) Child Labor in Africa: Issues and Challenges. World Bank. https://hdl.handle.net/10986/9791

Ango

Börjeson

Wisborg

, et al. (2022) Coffee, child labor, and education: examining a triple social–ecological trade-off in an Afromontane Forest Landscape. International Journal of Educational Development 95: 102681. https://doi.org/10.1016/j.ijedudev.2022.102681

Angrist

de Barros

Bhula

, et al. (2021) Building back better to avert a learning catastrophe: estimating learning loss from COVID-19 school shutdowns in Africa and South Asia. International Journal of Educational Development 84: 102397. https://doi.org/10.1016/j.ijedudev.2021.102397

Angulo

(2024) Books and Bushes: Schooling Decisions and Coca Production in Colombia.

Azevedo

Goldemberg

Montoya

, et al. (2021) Will every child be able to read by 2030? Defining learning poverty and mapping the dimensions of the challenge. The World Bank. https://doi.org/10.1596/1813-9450-9588

10.

Ball

Curran

Tanoh

, et al. (2022) Learning to read in environments with high risk of illiteracy: the role of bilingualism and bilingual education in supporting reading. Journal of Educational Psychology 114: 1156–1177. https://doi.org/10.1037/edu0000723

11.

Bøås

Huser

(2006) Child labour and cocoa production in West Africa. In: The Case of Côte d’ivoire and Ghana. Fafo-report, 522 .

12.

Boutin

Jouvin

(2022) Child labor consequences on education and health: a review of evidence and knowledge gaps. IDEAS Working Paper Series from RePEc. https://www-proquest-com-s.web.bisu.edu.cn/docview/2717532832?pq-origsite=primoandsourcetype=Working_Papers

13.

Brice

Zinszer

Kablan

, et al. (2024) Individual differences in leveraging regularity in emergent L2 readers in rural Côte d’Ivoire. Scientific Studies of Reading 28: 1–20. https://doi.org/10.1080/10888438.2024.2317934

14.

Canagarajah

Coulombe

(1999) Child Labor and Schooling in Ghana. The World Bank. https://doi.org/10.1596/1813-9450-1844

15.

Canagarajah

Nielsen

(2001) Child labor in Africa: a comparative study. The Annals of the American Academy of Political and Social Science 575(1): 71–91. https://doi.org/10.1177/0002716201575001005

16.

Castillo

Adam

Haßler

(2022) Improving the impact of educational technologies on learning within low-income contexts. In: Wagner

Castillo

Grant Lewis

(eds) Learning, Marginalization, and Improving the Quality of Education in low-income Countries. Open Book Publishers, pp. 113–148.

17.

Centre ivoirien de recherches économiques et sociales (CIRES) Cellule d’Analyse de Politiques Économiques (2023) Impacts des politiques publiques liées à la pandémie de la COVID-19 sur le secteur informel, les femmes et les jeunes: Cas du Burkina Faso, du Cameroun, de la Côte-d’Ivoire et du Sénégal-desktop review. CIRES. Available at: https://idl-bnc-idrc.dspacedirect.org/items/b910da80-4ff2-44c0-bf20-d4650b7a6fac.

18.

Child Labor Cocoa Coordinating Group [CLCCG] (2020) CLCCG report: 2010-2020 efforts to reduce child labor in cocoa. Child Labor Cocoa Coordinating Group. Available at: https://www.dol.gov/sites/dolgov/files/ILAB/reports/CLCCG-Ten-Year-Report.pdf.

19.

Cogneau

Jedwab

(2012) Commodity price shocks and child outcomes: the 1990 cocoa crisis in Cote d’Ivoire. Economic Development and Cultural Change 60(3): 507–534. https://doi.org/10.1086/664017

20.

Conn

(2017) Identifying effective education interventions in sub-Saharan Africa: a meta-analysis of impact evaluations. Review of Educational Research 87(5): 863–898. https://doi.org/10.3102/0034654317712025

21.

Diallo

(2002) L'influence des facteurs familiaux, scolaires et individuels sur l'abandon scolaire des filles de l'enseignement fondamental en milieu rural de la région de Ségou au Mali. Doctoral thesis. Université Laval, 4064.

22.

Dumas

Lambert

(2005) Patterns of Intergenerational Transmission of Education: The Case of Senegal. IDEAS Working Paper Series from RePEc .

23.

El Bouwab

(2021) Utilisation des TICE et des technologies numériques pour favoriser l’apprentissage de la langue amazighe: ressources éducatives en ligne, produits didactiques multimédias et applications éducatives. Revue des Études Amazighes 5(1): 25–40.

24.

Elston

JWT

Cartwright

Ndumbi

, et al. (2017) The health impact of the 2014–15 Ebola outbreak. Public Health (London) 143: 60–70. https://doi.org/10.1016/j.puhe.2016.10.020

25.

Enebe

Agunwa

, et al. (2021) Prevalence and predictors of child labor among junior public secondary school students in Enugu, Nigeria: a cross-sectional study. BMC Public Health 21(1): 1–1339. https://doi.org/10.1186/s12889-021-11429-w

26.

Evans

Mendez Acosta

(2021) Education in Africa: what are we learning? Journal of African Economies 30(1): 13–54. https://doi.org/10.1093/jae/ejaa009

27.

Gibbons

Huebler

Loaiza

(2005) Child Labour, Education and the Priciple of Non-Discrimination. Human Rights and Development: Towards Mutual Reinforcement. Oxford University Press.

28.

Global System for Mobile Communications (GSMA) . (2017). National overview: Côte d'Ivoire . https://www.gsma.com/mobilefordevelopment/wp-content/uploads/2017/04/CotedIvoireCountryOverviewExecSumEng.pdf

29.

Gove

Wetterberg

(eds) (2011) The Early Grade Reading Assessment: Applications and Interventions to Improve Basic Literacy. RTI Press.

30.

Grewenig

Lergetporer

Werner

, et al. (2021) COVID-19 and educational inequality: how school closures affect low-and high-achieving students. European Economic Review 140: 103920. https://doi.org/10.1016/j.euroecorev.2021.103920

31.

Guarcello

Lyon

Rosati

(2008) Child labor and education for All: an issue paper. The Journal of the History of Childhood and Youth 1(2): 254–266. https://doi.org/10.1353/hcy.0.0002

32.

Gubbels

van der Put

Assink

(2019) Risk factors for school absenteeism and dropout: a meta-analytic review. Journal of Youth and Adolescence 48: 1637–1667. https://doi.org/10.1007/s10964-019-01072-5

33.

Harris

Taylor

Thielke

, et al. (2009) Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. Journal of Biomedical Informatics 42(2): 377–381. https://doi.org/10.1016/j.jbi.2008.08.010

34.

Harris

Taylor

Minor

, et al. (2019) The REDCap consortium: building an international community of software platform partners. Journal of Biomedical Informatics 95: 103208. https://doi.org/10.1016/j.jbi.2019.103208

35.

Ibrahim

Abdalla

Jafer

, et al. (2019) Child labor and health: a systematic literature review of the impacts of child labor on child’s health in low- and middle-income countries. Journal of Public Health 41(1): 18–26. https://doi.org/10.1093/pubmed/fdy018

36.

Ilahi

(2001) Children’s Work and Schooling: Does Gender Matter? Evidence from the Peru LSMS. The World Bank. https://doi.org/10.1596/1813-9450-2745

37.

Institut National de la Statistique (INS) (2005). Enquête nationale sur le travail des enfants (ENTE 2005). République de Côte d'Ivoire. https://nada.anstat.ci/index/catalog/16

38.

International Cocoa Initiative (ICI) (2019) Annual Report 2019. In: ICI Secretariat in Switzerland. https://www.cocoainitiative.org/sites/default/files/reports/ICI_English_Annual_Report_2019.pdf

39.

International Cocoa Initiative (ICI) (2020) Changes in Hazardous Child Labor in Côte D’ivoire’s Cocoa Communities Before and after Covid-19 Partial Lockdown. International Cocoa Initiative. Available at: https://cocoainitiative.org/knowledge‐centre‐post/changes‐in‐hazardous‐child‐labor‐in‐cote‐divoires‐cocoacommunities‐before‐and‐after‐covid‐19‐partial‐lockdown/.

40.

International Cocoa Initiative (ICI) (2022) Linking child labor, schooling and child wellbeing. International Cocoa Initiative. Available at: https://www.cocoainitiative.org/knowledge-hub/resources/linking-child-labor-schooling-and-child-wellbeing.

41.

International Labor Organization (ILO) (2017a) Global estimates of child labor: results and trends. https://www.ilo.org/wcmsp5/groups/public/---dgreports/---dcomm/documents/publication/wcms_575499.pdf

42.

International Labor Organization (ILO) (2017b) Ending child labor by 2025: a review of policies and programs. https://www.ilo.org/ipec/Informationresources/WCMS_IPEC_PUB_29875/lang--en/index.htm

43.

International Labor Organization (ILO) (2018) Ending Child Labour by 2025: A Review of Policies and Programmes. 2nd edition. ILO. Available at: https://www.ilo.org/sites/default/files/wcmsp5/groups/public/%40ed_norm/%40ipec/documents/publication/wcms_653987.pdf.

44.

International labor Organization (ILO) (2020) ILO monitor: COVID-19 and the world of work. Part I: Latest Labor Market Developments: Continuing Workplace Closures, working-hour Losses and Decreases in Labor Income. ILO. Available at: https://www.ilo.org/sites/default/files/wcmsp5/groups/public/@dgreports/@dcomm/documents/briefingnote/wcms_749399.pdf.

45.

International Labor Organization (ILO) and United Nations Children’s Fund (UNICEF) (2021) Child Labor: Global Estimates 2020, Trends and the Road Forward. New York: ILO and UNICEF. Available at: https://www.ilo.org/wcmsp5/groups/public/---ed_norm/---ipec/documents/publication/wcms_797515.pdf.

46.

Jacquemin

(2006) Can the Language of Rights Get Hold of the Complex Realities of Child Domestic Work? the case of young domestic workers in Abidjan, Ivory Coast. Childhood 13(3): 389–406. https://doi.org/10.1177/0907568206066358

47.

Jakubowski

Gajderowicz

Patrinos

(2023) Global learning loss in student achievement: first estimates using comparable reading scores. Economics Letters 232: 111313. https://doi.org/10.1016/j.econlet.2023.111313

48.

Jasińska

Guei

(2018) Neuroimaging field methods using functional near Infrared Spectroscopy (NIRS) neuroimaging to study global child development: Rural Sub-Saharan Africa. Journal of Visualized Experiments: JoVE 132: e57165. https://doi.org/10.3791/57165

49.

Jasińska

Guei

(2022) Ivory Coast: promoting learning outcomes at the bottom of the pyramid. In: Wagner

Castillo

Grant Lewis

(eds) Learning, Marginalization, and Improving the Quality of Education in low-income Countries. Open Book Publishers, pp. 343–360.

50.

Jasińska

Ogan

Brice

, et al. Impact of a Mobile Phone Literacy Intervention on Emergent Readers in Rural Côte d’ivoire. (in preparation).

51.

Jasińska

Akpe

Seri

BAD

, et al. (2022a) Evaluating bilingual children’s native language abilities in Côte d’Ivoire: introducing the Ivorian children’s language assessment toolkit for Attié, Abidji, and Baoulé. Applied Linguistics 43(6): 1116–1142. https://doi.org/10.1093/applin/amac025

52.

Jasińska

Zinszer

, et al. (2022b) Home learning environment and physical development impact children’s executive function development and literacy in rural Côte d’Ivoire. Cognitive Development 64: 101265. https://doi.org/10.1016/j.cogdev.2022.101265

53.

Jasińska

Ball

Guei

(2023) Literacy in Côte d’Ivoire. In: Handbook of Literacy in Africa. Springer International Publishing, pp. 235–254.

54.

Jasińska

Nematova

Brice

, et al. (2024) Neural activation during phonological processing in primary‐school children with limited reading experience: insights from rural Côte d'Ivoire. Mind, Brain, and Education 18: 307–322. https://doi.org/10.1111/mbe.12411

55.

Jensen

Nielsen

(1997) Child labor or school attendance? Evidence from Zambia. Journal of Population Economics 10(4): 407–424. https://doi.org/10.1007/s001480050051

56.

Khan

Worstman

Whitehead

Hannon

Tanoh

Sulik

Ogan

Obradović

Jasińska

(2024) Modeling the associations between socioeconomic risk factors, executive function components, and literacy among children in rural Côte d’Ivoire. Cognitive Development 70. Available at: https://doi.org/10.1016/j.cogdev.2024.101436

57.

Kizilcec

Chen

Jasińska

, et al. (2021) Mobile learning during school disruptions in Sub-Saharan Africa. AERA Open 7(1): 233285842110148. https://doi.org/10.1177/23328584211014860

58.

Kruger

(2007) Coffee production effects on child labor and schooling in rural Brazil. Journal of Development Economics 82(2): 448–463. https://doi.org/10.1016/j.jdeveco.2006.04.003

59.

Kuepie

Shapiro

Tenikue

(2015) Access to schooling and staying in school in selected Sub Saharan African countries. African Development Review 27(4): 403–414. https://doi.org/10.1111/1467-8268.12156

60.

Lanoue

(2003) Côte d'Ivoire. Paper Commissioned for the EFA Global Monitoring Report 2003/4, the Leap to Equality. United Nations Educational, Scientific and Cultural Organization UNESCO. https://unesdoc.unesco.org/ark:/48223/pf0000146800_eng

61.

Lichand

Doria

(2024) The lasting impacts of remote learning in the absence of remedial policies: evidence from Brazil. Proceedings of the National Academy of Sciences 121(22): e2316300121. https://doi.org/10.1073/pnas.2316300121

62.

Lichand

Wolf

(2025) Measuring child labor: The who’s the where’s, the when’s and the why’s. PLOS One 20(6): e0322987. Available at: https://doi.org/10.1371/journal.pone.0322987

63.

Lloyd Blanc

(1996) Children’s schooling in sub-Saharan Africa: the role of fathers, mothers, and others. Population and Development Review 22(2): 265–298. https://doi.org/10.2307/2137435

64.

Lonigan

Shanahan

(2009) Developing Early Literacy: Report of the National Early Literacy Panel. Executive Summary. A Scientific Synthesis of Early Literacy Development and Implications for Intervention. National Institute for Literacy. https://files.eric.ed.gov/fulltext/ED508381.pdf

65.

Luckstead

Tsiboe

Nalley

(2019) Estimating the economic incentives necessary for eliminating child labor in Ghanaian cocoa production. PLoS One 14(6): e0217230. https://doi.org/10.1371/journal.pone.0217230

66.

Madaio

Kamath

Yarzebinski

, et al. (2019b) you give a little of yourself. Proceedings of the 2nd ACM SIGCAS Conference on Computing and Sustainable Societies. New York, NY, USA: Association for Computing Machinery, 86–98. Available at: https://doi.org/10.1145/3314344.3332504

67.

Madaio

Tanoh

Seri

, et al. (2019a) Everyone brings their grain of salt”: designing for low-literate parental engagement with children’s literacy in Côte d’Ivoire. Conference on Human Factors in Computing Systems - Proceedings 1–15. Available at: https://doi.org/10.1145/3290605.3300695

68.

Madaio

Yarzebinski

Kamath

, et al. (2020a) Collective support and independent learning with a voice-based literacy technology in rural communities. In: Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, pp. 1–14. https://doi.org/10.1145/3313831.3376276

69.

Madaio

Zinszer

Jasińska

, et al. (2020b) Phone-based literacy curriculum intervention in Côte d'Ivoire. https://doi.org/10.17605/OSF.IO/N53W8

70.

McEwan

(2015) Improving learning in primary schools of developing countries: a meta-analysis of randomized experiments. Review of Educational Research 85(3): 353–394. https://doi.org/10.3102/0034654314553127

71.

Min

(2021) How COVID-19 has impacted the SDGs in Africa. United Nations: African Renewal. https://www.un.org/africarenewal/magazine/august-2021/how-covid-19-has-impacted-sdgs-africa

72.

Min

Perucci

(2020) UN/DESA Policy Brief #81: Impact of COVID-19 on SDG Progress: A Statistical Perspective. United Nations. https://www.un.org/development/desa/dpad/wp-content/uploads/sites/45/publication/PB_81.pdf

73.

Mohammed

(2023) Children's lives in an era of school closures: exploring the implications of COVID‐19 for child labor in Ghana. Children & Society 37(1): 91–106. https://doi.org/10.1111/chso.12611

74.

Naeem

Shaukat

Ahmed

(2011) Child labor in relation to poverty. International Journal of Health Sciences 5(2 Suppl 1): 48–49. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3533357/

75.

Namatende-Sakwa

Lewinger

Langsford

(eds) (2022) COVID-19 and Education in Africa: Challenges, Possibilities, and Opportunities. 1st edition. Routledge. https://doi.org/10.4324/9781003269625

76.

Organisation for Economic and Cooperative Development (OECD) (2012) Equity and Quality in Education: Supporting Disadvantaged Students and Schools. OECD Publishing. https://doi.org/10.1787/9789264130852-en

77.

Oruko

Nyothach

Zielinski-Gutierrez

, et al. (2015) He is the one who is providing you with everything so whatever he says is what you do': a Qualitative Study on factors affecting secondary schoolgirls’ dropout in rural Western Kenya. PLoS One 10(12): e0144321. https://doi.org/10.1371/journal.pone.0144321

78.

R Core Team (2023) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. https://www.R-project.org/

79.

Ray

(2002) The determinants of child labor and child schooling in Ghana. Journal of African Economies 11(4): 561–590. https://doi.org/10.1093/jae/11.4.561

80.

Reimers

Schleicher

Saavedra

, et al. (2020) Supporting the continuation of teaching and learning during the COVID-19 pandemic. OECD 1(1): 1–38. https://www.oecd.org/education/Supporting-the-continuation-of-teaching-and-learning-during-the-COVID-19-pandemic.pdf

81.

Risso-Gill

Finnegan

(2015) Children's Ebola Recovery Assessment: Sierra Leone. Save the Children, World Vision International, Plan International, and UNICEF. https://www.savethechildren.org/content/dam/global/reports/emergency-humanitarian-response/ebola-rec-sierraleone.pdf

82.

Rosati

Rossi

(2003) Children’s working hours and school enrollment: evidence from Pakistan and Nicaragua. The World Bank Economic Review 17(2): 283–295. https://doi.org/10.1093/wber/lhg023

83.

RTI International (2015) Early Grade Reading Assessment (EGRA) Toolkit.

84.

Sadhu

Kysia

Onyango

, et al. (2020) Norc final report: assessing progress in reducing child labor in cocoa production in cocoa growing areas of côte d’ivoire and Ghana. Technical report. https://www.norc.org/content/dam/norc-org/pdfs/NORC_2020_Cocoa_Report_English.pdf

85.

Sangli

Ouattara

Azianu

, et al. (2022) Les dessous de la présence des enfants sur les sites d’extraction artisanale de l’or dans les zones rurales au Burkina Faso. Revue Nigérienne des Sciences Sociales, revue internationale francophone(004): 109–120.

86.

Schlemmer

(2002) Paradigmes de l’enfance et de l’école, droit à l’éducation et droit au travail. Communications (Paris) 72(1): 175–194. https://doi.org/10.3406/comm.2002.2104

87.

Slade

Piper

Kaunda

, et al. (2017) Is ‘summer’ reading loss universal? Using ongoing literacy assessment in Malawi to estimate the loss from grade-transition breaks. Research in Comparative and International Education 12(4): 461–485. https://doi.org/10.1177/1745499917740657

88.

Smith

(2021) Consequences of school closure on access to education: lessons from the 2013–2016 Ebola pandemic. International Review of Education 67(1–2): 53–78. Available at: https://doi.org/10.1007/s11159-021-09900-2

89.

Sobers

Whitehead

N’Goh

KNA

, et al. (2023) Is a phone‐based language and literacy assessment a reliable and valid measure of Children’s reading skills in low‐resource settings? Reading Research Quarterly 58(4): 733–754. Available at: https://doi.org/10.1002/rrq.511

90.

Stanovich

(2009) Matthew effects in reading: some consequences of individual differences in the acquisition of literacy. Journal of education 189(1–2): 23–55. Available at: https://doi.org/10.1177/0022057409189001-204

91.

The United Nations Educational , Scientific and Cultural Organization (UNESCO) Institute for Statistics (UIS). (2022). Côte d'Ivoire/UNESCO UIS [Data file]. http://uis.unesco.org/en/country/CI

92.

The United Nations Educational, Scientific and Cultural Organization (UNESCO) Institute for Statistics (UIS) (2015) A growing number of children and adolescents are out of school as aid fails to meet the mark. https://unesdoc.unesco.org/ark:/48223/pf0000233610

93.

The United Nations Educational, Scientific and Cultural Organization (UNESCO) Institute for Statistics (UIS) (2018) Côte d'Ivoire | UNESCO UIS. [Data File]. https://uis.unesco.org/en/country/CI

94.

Tulane University (2015) Survey on child labor in Ghana and Cote d'Ivoire. https://www.cocoainitiative.org/knowledge-hub/resources/tulane-survey-child-labor-ghana-and-cote-divoire

95.

UNESCO (2020) COVID-19 school closures around the world will hit girls hardest. https://www.unesco.org/en/articles/covid-19-school-closures-around-world-will-hit-girls-hardest

96.

United Nations (2015) Transforming our World: The 2030 Agenda for Sustainable Development. United Nations. https://sdgs.un.org/2030agenda

97.

United Nations International Children's Emergency Fund (UNICEF) (2019) Final report - UNICEF Côte d’Ivoire: promoting the rights of children in the cocoa producing areas in Cote d’Ivoire. https://open.unicef.org/sites/transparency/files/2020-06/Cote-d-Ivoire-TP5-2018.pdf

98.

United Nations International Children’s Emergency Fund (UNICEF) (2023) Côte d’Ivoire builds a resilient and equitable education system during and after COVID-19. https://www.globalpartnership.org/blog/cote-divoire-builds-resilient-and-equitable-education-system-during-and-after-covid-19

99.

United Nations Population Fund (UNFPA) and Population Reference Bureau (PRB) (2012) Status Report Adolescents and Young People in Sub-Saharan Africa: Opportunities and Challenges. https://esaro.unfpa.org/sites/default/files/pub-pdf/Status_Report_Adolescents_and_Young_People_in_Sub-Saharan_Africa.pdf

100.

United States Department of Labor (2017) Independent external final evaluation of RCLES reducing childhood labor through education services in Burkina Faso. https://www.dol.gov/sites/dolgov/files/ILAB/evaluation_type/final_evaluation/BurkinaFaso_RCLES_feval.pdf

101.

Vadivel

Alam

Nikpoo

, et al. (2023) The impact of low socioeconomic background on a child’s educational achievements. Education Research International 2023: 1–11. Available at: https://doi.org/10.1155/2023/6565088

102.

Verwimp

Van Bavel

(2014) Schooling, violent conflict, and gender in Burundi. The World Bank Economic Review 28(2): 384–411. https://doi.org/10.1093/wber/lht010

103.

Villegas

Peirolo

Rocca

, et al. (2021) Impacts of health-related school closures on child protection outcomes: a review of evidence from past pandemics and epidemics and lessons learned for COVID-19. International Journal of Educational Development 84: 102431. https://doi.org/10.1016/j.ijedudev.2021.102431

104.

Waldfogel

(2012) The role of out-of-school factors in the literacy problem. The Future of Children 22(2): 39–54. https://doi.org/10.1353/foc.2012.0016

105.

Wawire

Barnes-Story

Kiru

, et al. (2023) Primary School Learning in sub-Saharan Africa During COVID-19: A Scoping Review of Responses and Recovery Initiatives Induced During School Closures.

106.

Webbink

Smits

de Jong

(2012) Hidden Child Labor: Determinants of Housework and Family Buisness Work of Children in 16 Developing Countries. World Development 40(3): 631–642. Available at: https://doi.org/10.1016/j.worlddev.2011.07.005, In press.

107.

Whitehead

Ball

Wolf

, et al. (2024) Variability in the age of schooling contributes to the link between literacy and numeracy in Côte d’Ivoire. Child Development 95(2): e93-109. Available at: https://doi.org/10.1016/j.worlddev.2011.07.005

108.

World Bank (2020) Cote d’ivoire 10th Economic Update: Cote d’Ivoire and the COVID-19 Pandemic. World Bank.

109.

World Bank (2023) School Enrollment, Primary (% Gross) - Cote d’Ivoire. UNESCO Institute for Statistics (UIS). https://data.worldbank.org/indicator/SE.PRM.ENRR?locations=CI

110.

Wortsman

Bhattacharya

Tanoh

, et al. (2024a) Accelerating progress towards eradicating child labor (SDG8.7) with quality education (SDG4): School quality is linked to reduced child cocoa labor in Côte D’Ivoire. Research in Comparative and International Education 19(3): 321–351. Available at: https://doi.org/10.1177/17454999241255182

111.

Wortsman

Capani

Brice

, et al. (2024b) Risk and resilience factors for primary school dropout in Côte d’Ivoire. Journal of Applied Developmental Psychology , 92: 101654. Available at: https://doi.org/10.1016/j.appdev.2024.101654

112.

Zinszer

Hannon

Kouadio

AÉ

, et al. (2023a) Does Nonlinguistic Segmentation Predict Literacy in Second Language Education? Statistical Learning in Ivorian Primary Schools. Language Learning 273(4): 1039–1086. Available at: https://doi.org/10.1111/lang.12603

113.

Zinszer

Hannon

Kouadio

AÉ

, et al. (2023b) Statistical learning and children’s emergent literacy in rural Côte d’Ivoire. Developmental Science 27(5): e13448. Available at: https://doi.org/10.1111/desc.13448

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.23 MB