Abstract
This study examines the factors influencing dropout rates among girls in secondary schools in Lilongwe Rural East and West Districts, Malawi, utilizing Self-Determination Theory and Intersectionality Theory. Through a mixed-methods approach, the research integrates quantitative data from structured surveys and qualitative insights from semi-structured interviews. Quantitative analyses, including Exploratory Factor Analysis, Confirmatory Factor Analysis, Hierarchical Linear Modeling, logistic regression, path analysis, and discriminant analysis, identified significant predictors and constructs influencing dropout rates. Qualitative data provided a contextual understanding of the girls’ experiences. The findings reveal that perceived autonomy, competence, and relatedness are critical constructs impacting dropout rates. Distance to school, household responsibilities, and socioeconomic status significantly influence the likelihood of dropping out. The study underscores the importance of addressing both psychological needs and socioeconomic barriers to enhance educational retention. Implications for policymakers and educators include developing targeted interventions that foster intrinsic motivation and address intersectional disadvantages, contributing to the broader goals of gender equality and sustainable development.
1 Introduction
Education is vital for sustainable development, but gender disparities in developing countries, particularly Malawi, hinder progress. Despite an 8-4-4 education structure and numerous interventions, girls face significant challenges in accessing and completing secondary education, with dropout rates exceptionally high in rural areas like Lilongwe Rural District. These challenges are exacerbated by socioeconomic conditions, cultural norms, inadequate facilities, household responsibilities, and long distances to school (Banda et al. 2024e).
Malawi has undertaken efforts to reduce dropout rates through governmental, international, and community-led initiatives, focusing on removing financial and logistical barriers. However, the lack of attention to intersectional factors and psychological needs has limited their effectiveness, often leading to the underutilization of resources. The introduction of Free Primary Education (FPE) in 1994 increased enrollment but failed to address systemic barriers such as poverty, cultural norms, and household duties, particularly for low-income, rural girls (National Statistical Office & ICF 2017; UNICEF 2023). Similarly, the National Girls’ Education Strategy combated gender-based violence and dropouts but overlooked the compounded effects of socioeconomic status, ethnicity, and geography, limiting its support for marginalized groups (The Government of Malawi 2018; Chimombo et al. 2014).
International programs like UNICEF Malawi’s Girls’ Education Program provided scholarships, supplies, and infrastructure improvements but often neglected psychological factors like autonomy, competence, and relatedness, reducing their long-term impact. CAMFED’s financial and mentorship support addressed relatedness but lacked an intersectional framework, leaving unsupported rural girls who face greater cultural resistance and logistical barriers (CAMFED 2023). Grassroots initiatives, such as Mother Groups, addressed issues like early marriage and absenteeism but failed to tackle systemic challenges like poverty and cultural expectations. They also lacked strategies to foster autonomy and competence, leaving marginalized girls vulnerable to systemic inequalities. While these interventions have localized impacts, they struggle to create sustainable solutions for reducing dropout rates among the most vulnerable girls.
2 Overlooked Intersectionality and Psychological Needs in Educational Interventions for Girls
The primary shortcoming of these interventions lies in their insufficient focus on intersectionality and the psychological needs critical for preventing school dropout. By treating girls as a uniform group, these initiatives overlook how intersecting identities – such as gender, socioeconomic status, ethnicity, and rural residency – create unique educational barriers. This lack of specificity has resulted in interventions that may not effectively reach or support the most marginalized girls, who are often the ones most in need.
Many Malawian youths fall short of coping strategies when faced with challenges (Banda et al. 2024d). This could be partially due to low self-determination levels. Moreover, according to SDT, the psychological needs of autonomy, competence, and relatedness are crucial for fostering intrinsic motivation and sustained engagement in education. When these needs are unmet, girls are less likely to value or persist in their education, regardless of the material support provided. For instance, if a girl perceives herself as having no control over her educational choices (lack of autonomy) or as incapable of academic success (lack of competence), she is more likely to disengage from school, even if financial barriers have been removed.
The failure to incorporate intersectionality and address psychological needs in previous interventions has led to several unintended consequences. In some instances, girls have misused or underutilized the opportunities provided, such as scholarships or educational resources, because these interventions did not address the underlying motivations and barriers influencing their educational decisions. For example, a scholarship alone may not suffice to keep a girl in school if she faces pressure to marry early or feels unsupported in the school environment.
To achieve lasting impact, educational interventions must go beyond merely removing external barriers and focus on empowering girls to take ownership of their educational trajectories by cultivating self-determination skills. Without this empowerment, the effectiveness of these interventions is likely to be short-lived, as the challenges leading to dropout will resurface once external support is withdrawn. The current study addresses this significant gap by advocating for a dual approach that mitigates external barriers while enhancing girls’ internal capacities. By equipping girls with the skills necessary to navigate their educational journeys independently, this approach aims to ensure that interventions yield enduring benefits, enabling girls to remain resilient even in the absence of ongoing external support.
Thus, this research contributes to a more comprehensive understanding of how to achieve sustainable educational outcomes for girls in Malawi. By applying Intersectionality and Self-Determination Theories, the study explores the dropout rates among girls, identifying how intersecting identities and unmet psychological needs contribute to this issue. The findings highlight the importance of developing targeted interventions that address material needs and foster autonomy, competence, and relatedness tailored to the specific contexts of different groups of girls. This approach aims to create a more supportive and motivating educational environment, ultimately reducing dropout rates and improving educational outcomes for girls in Malawi.
Despite extensive efforts from the government, non-governmental organizations (NGOs), and international bodies to mitigate these barriers, the dropout rate among girls in secondary schools remains critically high in Malawi. This persistent issue undermines the goals of gender equality and sustainable development, as outlined in the Sustainable Development Goals (SDGs) 4 (Cameron et al. 2020; UNESCO Education Sector 2030 2017), Quality Education (Banda et al. 2024e), and 5, Gender Equality (A4ID 2022).
This study underscores the critical link between gender equality, female education, and sustainable development by uncovering girls’ multifaceted barriers in accessing and completing secondary education. Through an intersectionality framework, the study investigates the factors influencing girls’ dropout rates in rural districts of Malawi. It aims to propose targeted interventions that address these specific needs, thereby enhancing retention rates and contributing to sustainable educational development. The study is structured to achieve the following objectives:
Understand multi-level influences on dropout rates, including individual and school-level factors.
Determine the probability of girls dropping out of secondary school based on various predictor variables.
Specify and test relationships between psychological constructs and dropout rates.
Discriminate between dropouts and non-dropouts based on variables.
By addressing these objectives, this research provides a more holistic understanding of the dropout phenomenon among girls in rural Malawi. The findings will enrich the existing literature by offering a nuanced analysis considering the interplay between psychological needs and socio-cultural and economic barriers. This significant study will inform policymakers, educators, and NGOs on developing more effective strategies that address socioeconomic and cultural challenges and enhance girls’ psychological resilience and motivation. Ultimately, this approach aims to improve their educational outcomes and contribute to the broader goals of gender equality and sustainable development in the Global South.
The rest of this study is structured as follows: the literature review section is followed by methodology and design. After that is the results section, followed by a discussion. The conclusion includes limitations, proposed future studies, and policy recommendations.
3 Literature Review
Intersectionality, introduced by Crenshaw (1989) and further developed by Collins (1991, 2015), provides a framework for understanding how intersecting social identities – such as gender, race, and socioeconomic status – shape unique educational challenges. These factors significantly influence students’ academic outcomes, as research demonstrates the compounded impact of these identities on educational achievement and experience (Collins 1991; Crenshaw 1989; Gershenson et al. 2021). Cultural norms and stereotypes reinforce systemic inequalities, limiting opportunities for marginalized groups (Boris 2008; Collins 2015; Hancock 2004). Girls, in particular, face barriers stemming from poverty, limited access to resources, and cultural devaluation of their education, which negatively affects their perceived competence and academic performance (Mamie Martin Fund 2016, United Nations Development Programme (UNDP) 2022). Intersecting disadvantages lower self-efficacy and increases dropout rates, with perceived competence being critical for academic engagement. Intersectionality identifies these barriers, guiding interventions to improve girls’ perceived competence. Perceived relatedness, a sense of belonging in school, is also shaped by intersecting social identities. Girls in rural Malawi face compounded challenges, including cultural norms prioritizing boys’ education and household responsibilities, which hinder their experiences and increase dropout risks (Gershenson et al. 2021; Goodenow and Grady 1993; Graham 2018; OECD 2019). Minority girls often report lower belonging due to discrimination, which negatively impacts their academic outcomes (Ghavami and Peplau 2018). Addressing these issues requires strategies encompassing girls’ identities and experiences (Good, Rattan, and Dweck 2012; McCall 2005). Cultural norms and lack of supportive relationships in rural areas exacerbate isolation, emphasizing the need for inclusive policies (Chimombo et al. 2014).
Perceived autonomy, or control over educational choices, is similarly constrained by traditional gender roles in rural Malawi, where household duties limit girls’ independence. Strategies to address these barriers include community engagement, policy reforms, and targeted educational programs (Deci and Ryan 2012). Integrating Intersectionality and Self-Determination Theories (SDT) provides a robust framework for understanding dropout factors. SDT emphasizes autonomy, competence, and relatedness as critical to intrinsic motivation, while intersectionality reveals how systemic inequalities shape these needs (Deci and Ryan 2012). Girls’ low perceived autonomy and competence, rigid school structures, and unsupportive relationships increase dropout rates (Chimombo et al. 2014; Deci and Ryan 2012). Addressing these needs improves engagement and reduces dropouts (Chimombo et al. 2014; Deci and Ryan 2012).
Socioeconomic status significantly impacts educational attainment, with economic hardship and the need to contribute to household income being key barriers in Malawi, where poverty drives high dropout rates . The school environment, including resources and distance, further influences outcomes. Inadequate facilities and long commutes contribute to dropouts, while supportive, resource-rich schools reduce them (Ang et al. 2024; Edgerton and McKechnie 2023; UNICEF 2023). Academic self-efficacy, a component of competence, is vital for success. Lack of support and feedback diminishes self-efficacy, increasing dropouts, while high self-efficacy fosters persistence, emphasizing the need for supportive academic environments.
4 Methodology and Design
This study utilized a mixed-methods design to examine the factors contributing to girls’ dropout rates from secondary schools in Lilongwe Rural District, Malawi. By combining qualitative and quantitative approaches, the research aimed to analyze the multifaceted nature of school dropout comprehensively. The qualitative component provided an in-depth exploration of participants’ lived experiences, offering insights into how autonomy, competence, social relationships, and economic barriers influence educational engagement and persistence (Banda et al. 2024c; Meissner et al. 2018). These context-specific findings were quantitatively validated to ensure generalizability to a larger population, enhancing the study’s robustness (Tashakkori and Teddlie 2010).
The mixed-methods design facilitated triangulation by cross-verifying data collected through different methods, such as interviews and surveys, thereby enhancing validity and reliability (Banda et al. 2024b; Bryman 2006). This approach balanced the inherent limitations of each method, providing a nuanced understanding of the psychological, socioeconomic, and cultural factors affecting educational outcomes. For example, statistical analysis established significant correlations between economic status and dropout rates, while qualitative data illuminated the underlying reasons for these relationships, offering actionable insights for policymakers (Greene 2007). Integrating qualitative and quantitative data during analysis made the findings statistically robust and contextually meaningful, forming a solid basis for culturally and contextually relevant interventions.
Semi-structured interviews were conducted with 30 participants, comprising 15 dropouts and 15 current students, selected using a stratified purposive sampling technique. This ensured diverse representation in terms of geographic location, socioeconomic status, grade level, dropout status, household responsibilities, and cultural norms, capturing the complexity of the dropout phenomenon (Patton 2015; Creswell and Plano Clark 2017). Interviews revealed detailed insights into the interplay of these factors and their contribution to dropout rates among rural Malawian girls. Although some participants initially hesitated to share their experiences, reassurance of confidentiality and anonymity encouraged openness. Table 1 summarizes the demographic characteristics of the interviewees.
Characteristics of the interviewees.
Category | Sub-category | Number | Description |
---|---|---|---|
Geographic location | Rural | 15 | Participants from typical rural areas |
Semi-urban | 15 | Participants from semi-urban areas | |
Socioeconomic status | Low | 12 | Participants from very low-income households |
Middle | 10 | Participants from middle-income households | |
High | 8 | Participants from high-income households | |
Grade level | Form 1 | 7 | Participants in their first year of secondary school |
Form 2 | 8 | Participants in their second year of secondary school | |
Form 3 | 8 | Participants in their third year of secondary school | |
Form 4 | 7 | Participants in their final year of secondary school | |
Dropout status | Current students | 15 | Girls currently enrolled in school |
Recent dropouts | 15 | Girls who have recently dropped out of school | |
Household responsibilities | High | 11 | Participants with significant household duties |
Moderate | 9 | Participants with moderate household duties | |
Low | 10 | Participants with minimal household duties | |
Cultural norms | Traditional ethnic groups | 22 | Participants from traditional ethnic communities |
Diverse/non-traditional | 8 | Participants from diverse or non-traditional communities |
4.1 Survey
The sample size (N = 300), determined via power analysis, included current students and dropouts to analyze factors influencing retention and dropout. Participants from diverse socio-demographic backgrounds spanning rural and semi-urban areas were selected to capture challenges like distance and resource availability. Stratified purposive random sampling ensured representation across schools, grades, and socioeconomic groups, highlighting economic, cultural, and domestic barriers to education (Banda, Banda, Banda, Hlaing, et al., 2024a; Lloyd and Mertens 2018; Calleja 2009). Among non-dropouts, 208 students were randomly selected, while 92 dropouts were identified using attendance registers, snowball, and convenient sampling (Banda et al. 2024c; Banda et al. 2024e). This approach provided comprehensive insights to design targeted, intersectional interventions (Patton 2015). Table 2 presents the demographic characteristics of the survey sample.
Demographic characteristics of respondents to the survey.
Age group (by 2019) | Form/class | Dropout year | Reachable dropouts (N = 92) | Non-dropouts (N = 208) | Gross dropouts |
---|---|---|---|---|---|
18–19 | 4 | 2014 | 5 | 52 | 19 |
17–18 | 4 | 2015 | 5 | 16 | |
16–17 | 4 | 2016 | 6 | 21 | |
18–19 | 4 | 2017 | 10 | 24 | |
17–18 | 4 | 2018 | 12 | 28 | |
16–17 | 4 | 2019 | 12 | 26 | |
15–16 | 3 | 2014 | 4 | 52 | 12 |
14–15 | 3 | 2015 | 5 | 18 | |
15–16 | 3 | 2017 | 8 | 15 | |
14–15 | 3 | 2018 | 8 | 28 | |
16–17 | 2 | 2014 | 3 | 52 | 10 |
15–16 | 2 | 2015 | 4 | 10 | |
14–15 | 2 | 2016 | 5 | 10 | |
12–13 | 1 | 2014 | 1 | 52 | 3 |
12–13 | 1 | 2015 | 2 | 4 | |
12–13 | 1 | 2018 | 5 | 6 |
4.2 Data Analysis
Thematic analysis, conducted using NVivo 12, systematically organized and analyzed interview data. After thoroughly reviewing transcripts, initial codes were generated and managed efficiently with the software. Focused coding refined these into broader categories, while NVivo 12 facilitated pattern visualization and theme identification. Iterative comparisons ensured that themes accurately reflected participants’ perspectives. Word clouds highlighted prominent terms, further guiding thematic refinement.
Quantitative data were analyzed using advanced statistical techniques to provide actionable findings for tailored interventions. Exploratory Factor Analysis (EFA) identified the underlying constructs of autonomy, competence, and relatedness, while Confirmatory Factor Analysis (CFA) validated these constructs, ensuring reliability and theoretical soundness (Tashakkori and Teddlie 2010). Together, these analyses established a robust foundation for subsequent quantitative methods.
Hierarchical Linear Modeling (HLM) accounted for the nested data structure, analyzing individual and school-level influences on dropout rates. This approach allowed for the simultaneous examination of contextual factors, such as school resources and distance to school, alongside personal predictors, offering a multi-level perspective on structural inequalities (Meteyard and Davies 2020). Logistic Regression (LR) modeled the probability of dropout as a binary outcome based on predictors like perceived autonomy, competence, relatedness, economic status, and household responsibilities. This method quantified the individual-level factors influencing dropout risk and complemented the multi-level insights from HLM (Kumar and Reddy 2020).
Path Analysis (PA) examined direct and indirect relationships among psychological constructs, academic performance, and dropout rates. This technique revealed how autonomy, competence, and relatedness influence educational trajectories, providing deeper insights into causal pathways (Chaitanya et al. 2024). Discriminant Analysis (DA) identified variables distinguishing dropouts from non-dropouts, classifying students based on predictors like competence and economic status (Eardi et al. 2024). Cluster Analysis (CA) further grouped students with shared characteristics, uncovering patterns highlighting distinct risk profiles and intervention needs.
The sample of 300 participants, including 92 dropouts and 208 non-dropouts, ensured adequate cases for all techniques. Stratified sampling captured diverse socio-demographic backgrounds, meeting EFA, CFA, HLM, LR, PA, DA, and CA methodological requirements. This methodological framework comprehensively analyzed dropout predictors, informing culturally and contextually relevant interventions.
4.3 Dependent Variables and Alignment
The primary dependent variable in this study is the dropout status of female students in rural Malawi, defined as a binary outcome (“dropouts” or “non-dropouts”), making it suitable for Logistic Regression (LR) (DeMaris 1995; Fitzmaurice and Laird 2015; Hosmer, Lemeshow, and Sturdivant 2013; Kumar and Reddy 2020; Stoltzfus 2011). A student was classified as a dropout if she permanently left school before completing the designated level without transferring to another institution. This status was verified through school records and corroborated by interviews with administrators, teachers, and families. Non-dropouts included enrolled students or those who had completed their current educational level.
The dependent variable was tailored to the objectives of each statistical analysis. In EFA and CFA, autonomy, competence, and relatedness were analyzed as key psychological constructs. In HLM, dropout status was the dependent variable, with autonomy and school-level factors like resource availability as predictors. LR used dropout status to quantify the impact of psychological and contextual predictors, while PA examined the mediating role of academic performance. DA and CA utilized dropout status to identify and cluster predictors most strongly associated with this outcome.
This study utilized secondary data from a prior research project that had received ethical approval from the second author’s institution’s Institutional Review Board (IRB). The original data collection complied with Helsinki’s ethical standards, including informed consent, confidentiality, and minimizing risks to participants. The data was used in alignment with the objectives of the original study, maintaining consistency with the purposes for which participants had given informed consent (Tripathy 2013).
5 Results
EFA and CFA were conducted to validate the constructs of perceived autonomy, competence, and relatedness. The EFA results in Table 3 revealed strong factor loadings for the items associated with each construct, indicating that the items effectively represented their respective factors. Subsequently, the CFA results supported these findings by demonstrating high standardized factor loadings, thereby confirming the reliability and validity of the constructs within the context of this study.
Factor loadings from Exploratory Factor Analysis (EFA-FL) and standardized factor loadings from Confirmatory Factor Analysis (CFA-SFL) for various items.
Construct | Item | EFA-FL | CFA-SFL |
---|---|---|---|
PA | I have a say in decisions regarding my education | 0.78 | 0.82 |
I can choose the subjects and activities I participate in | 0.75 | 0.8 | |
My opinions are considered at school | 0.72 | 0.78 | |
PC | I feel confident in my ability to succeed in my school subjects | 0.81 | 0.84 |
My teachers provide helpful feedback and support | 0.79 | 0.82 | |
I generally perform well in my school subjects | 0.76 | 0.79 | |
PR | I have good relationships with my teachers | 0.82 | 0.85 |
I feel accepted and supported by my classmates | 0.8 | 0.83 | |
I participate in extracurricular activities at school | 0.78 | 0.8 |
-
PA, perceived autonomy; PC, perceived competence; PR, perceived relatedness.
5.1 Multi-Level Influences
HLM was employed to assess the multi-level influences on dropout rates, incorporating individual and school-level factors. This analysis aimed to determine the extent to which perceived autonomy, competence, and relatedness at the individual level and contextual variables at the school level influence the likelihood of dropout. The results indicate that perceived autonomy significantly negatively affects the outcome, with a path coefficient of −0.35 (SE = 0.08), meaning that as perceived autonomy increases, the outcome variable decreases. This effect is statistically significant with a p-value of less than 0.01, and the 95 % confidence interval [−0.5068, −0.1932] suggests that the estimate is reliable and precise. Table 4 is the summary.
Hierarchical Linear Modeling (HLM) and logistic regression results. β stands for coefficient, and SE stands for standard error.
Variable | β | SE | p-Value | 95 % CI | |
---|---|---|---|---|---|
Individual-level factors | Perceived autonomy | −0.35 | 0.08 | <0.01 | [−0.5068, −0.1932] |
Perceived competence | −0.38 | 0.07 | <0.01 | [−0.5172, −0.2428] | |
Perceived relatedness | −0.32 | 0.09 | <0.01 | [−0.4964, −0.1436] | |
School-level factors | Distance to school | 0.22 | 0.1 | <0.05 | [0.024, 0.416] |
School resources | −0.26 | 0.12 | <0.05 | [−0.4952, −0.0248] |
These results show that perceived competence significantly reduces the outcome (β = −0.38, SE = 0.07, p < 0.01, CI [−0.5172, −0.2428]), as does perceived relatedness (β = −0.32, SE = 0.09, p < 0.01, CI [−0.4964, −0.1436]). At the school level, greater distances to school increase the outcome (β = 0.22, SE = 0.10, p < 0.05, CI [0.024, 0.416]), while better school resources decrease it (β = −0.26, SE = 0.12, p < 0.05, CI [−0.4952, −0.0248]). These results highlight the roles of individual and school-level factors, with consistent confidence intervals confirming the reliability of these effects.
5.2 Predictive Factors Related to Dropout Rates
LRA was conducted to assess the probability of dropout based on psychological and contextual predictors, including perceived autonomy, competence, relatedness, household responsibilities, and economic status. Perceived autonomy was found to significantly reduce the likelihood of dropout, with an odds ratio of 0.65 (SE = 0.09), p < 0.01. The confidence interval [0.5459, 0.7780] confirms the reliability and precision of this finding, indicating a robust inverse relationship between perceived autonomy and the probability of dropout, as detailed in Table 5.
Logistic regression results.
Variable | Odds ratio (OR) | Standard error | p-Value | 95 % CI |
---|---|---|---|---|
Perceived autonomy | 0.65 | 0.09 | <0.01 | [0.5459, 0.7780] |
Perceived competence | 0.62 | 0.08 | <0.01 | [0.5298, 0.7254] |
Perceived relatedness | 0.68 | 0.1 | <0.01 | [0.5591, 0.8276] |
Household responsibilities | 1.55 | 0.12 | <0.01 | [1.2251, 1.9618] |
Economic status | 0.75 | 0.11 | <0.05 | [0.6046, 0.9308] |
Perceived competence significantly reduces the likelihood of dropout, with an odds ratio of 0.62 (SE = 0.08), p < 0.01, and a confidence interval of [0.5298, 0.7254], indicating a robust inverse relationship. Similarly, perceived relatedness decreases the odds of dropout, with an odds ratio of 0.68 (SE = 0.10), p < 0.01, and a confidence interval of [0.5591, 0.8276], confirming the reliability of this effect.
Household responsibilities, in contrast, positively influence the likelihood of dropout, with an odds ratio of 1.55 (SE = 0.12), p < 0.01, and a confidence interval of [1.2251, 1.9618], demonstrating a strong positive association. Economic status, however, shows a protective effect, with an odds ratio of 0.75 (SE = 0.11), p < 0.05, and a confidence interval of [0.6046, 0.9308], indicating that higher economic well-being reduces the odds of dropout. These findings highlight the interplay between psychosocial and contextual factors influencing the outcome.
5.3 PA and DA
Path and discriminant analyses examined the direct and indirect effects of perceived autonomy, competence, and relatedness on dropout rates, alongside the roles of household responsibilities and economic status. PA revealed that perceived competence had direct and indirect effects on dropout rates mediated by academic performance (β = −0.15, p < 0.05), highlighting the critical role of academic self-efficacy. DA demonstrated that perceived autonomy, competence, relatedness, household responsibilities, and economic status reliably differentiated between dropouts and non-dropouts. Table 6 summarizes the combined influence of psychological and contextual factors on educational outcomes.
Path analysis and discriminant analysis results.
Variable | PC (β) | SE | Sig | 95 % CI | DFC | SCDFC |
---|---|---|---|---|---|---|
Direct effects | ||||||
Perceived autonomy | −0.3 | 0.07 | <0.01 | [−0.4372, −0.1628] | 0.52 | 0.45 |
Perceived competence | −0.35 | 0.08 | <0.01 | [−0.5068, −0.1932] | 0.58 | 0.48 |
Perceived relatedness | −0.28 | 0.09 | <0.01 | [−0.4564, −0.1036] | 0.5 | 0.42 |
Indirect effects | ||||||
Perceived competence (via academic performance) | −0.15 | 0.05 | <0.05 | [−0.248, −0.052] | N/A | N/A |
Other factors | N/A | N/A | N/A | N/A | 0.65 | 0.55 |
Household responsibilities | N/A | N/A | N/A | N/A | 0.65 | 0.55 |
Economic status | N/A | N/A | N/A | N/A | 0.48 | 0.4 |
-
PC, path coefficient (beta); SE, standard error; Sig, significance; DFC, discriminant function coefficient; SCDFC, standardized canonical discriminant function coefficient.
The model fit indices for the PA indicate an excellent fit: Chi-Square (χ 2) = 98.45 (p < 0.01), CFI = 0.95, TLI = 0.94, and RMSEA = 0.042. Perceived autonomy has a significant negative direct effect on the outcome (path coefficient = −0.3, SE = 0.07, p < 0.01, CI [−0.4372, −0.1628]), indicating that increased autonomy reduces the outcome variable. Similarly, perceived competence exhibits a significant negative direct effect (path coefficient = −0.35, SE = 0.08, p < 0.01, CI [−0.5068, −0.1932]), emphasizing its vital role in reducing the outcome. Perceived relatedness also shows a significant adverse effect (path coefficient = −0.28, SE = 0.09, CI [−0.4564, −0.1036]), highlighting the collective impact of psychosocial factors.
The model identifies a significant indirect effect of perceived competence mediated by academic performance (path coefficient = −0.15, SE = 0.05, p < 0.05, CI [−0.248, −0.052]), indicating that competence affects the outcome both directly and indirectly through academic performance. Household responsibilities and economic status also significantly influence the outcome, with DFC and SCDFC values of 0.65 and 0.55 for household responsibilities and 0.48 and 0.4 for economic status, respectively, underscoring their substantial contributions to outcome variance.
5.4 Profiles and Latent Classes
LCA identified distinct profiles and latent classes based on students’ survey responses. These analyses revealed groups with varying levels of perceived autonomy, competence, and relatedness corresponding to different dropout risks (Table 7).
Cluster analysis results.
Cluster | Number of students | Characteristics |
---|---|---|
Cluster 1 | 120 | High perceived autonomy, competence, and relatedness; low dropout risk |
Cluster 2 | 100 | Moderate perceived autonomy and competence; high perceived relatedness; moderate dropout risk |
Cluster 3 | 80 | Low perceived autonomy and competence; low perceived relatedness; high dropout risk |
Clustering analysis identifies three student groups based on autonomy, competence, and relatedness. The first group (120 students) with high levels of all three shows low dropout risk, reflecting strong perceptions of control, skill, and connection. The second (100 students) has moderate autonomy and competence but high relatedness, facing moderate dropout risk, as social support partly offsets lower autonomy and competence. The third group (80 students), with low levels of all three, faces a high dropout risk due to disempowerment, low confidence, and disconnection. Table 8 presents an LCA categorizing students based on these factors.
Latent Class Analysis (LCA) results, showing the sample proportion and characteristics of each class.
Latent class | Sample proportion | Characteristics |
---|---|---|
Class 1 | 40 % | High autonomy and competence, strong peer relationships, high academic performance |
Class 2 | 35 % | Moderate autonomy, competence, and relatedness; moderate academic performance |
Class 3 | 25 % | Low autonomy, competence, and relatedness; low academic performance; high dropout risk |
Three clusters were identified through latent class analysis, each representing distinct levels of educational engagement and dropout risk. The first cluster, comprising 40 % of the sample, includes students with high autonomy, competence, strong peer relationships, and high academic performance, indicating a well-rounded profile with low dropout risk. The second cluster, encompassing 35 %, consists of students with moderate autonomy, competence, relatedness, and academic performance, suggesting some resilience but weaker outcomes than the first group. The third cluster, representing 25 %, includes students with low autonomy, competence, relatedness, and poor academic performance, placing them at the highest risk of dropping out.
This clustering highlights a gradient of dropout risk, demonstrating the need to prioritize fostering autonomy, competence, and peer support to enhance academic outcomes and reduce dropout rates.
Table 9 presents seven distinct codes: Lack of Autonomy, Support Systems, Household Responsibilities, Economic Challenges, Cultural Norms, Relatedness, and Perceived Competence.
Details the coding process.
Word list | Code developed | Theme |
---|---|---|
“Choice,” “control,” “decide,” “restrict,” “freedom,” “agency” | Lack of autonomy |
|
“Help,” “support,” “encourage,” “teacher,” “mentor,” “peer,” “friendship,” “relationship” | Support systems |
|
“Housework,” “chores,” “family,” “siblings,” “responsibilities,” “duty” | Household responsibilities |
|
“Fees,” “cost,” “afford,” “money,” “expenses,” “uniform,” “books,” “poverty,” “poor” | Economic challenges | |
“Expectation,” “society,” “norms,” “culture,” “gender roles,” “tradition,” “pressure” | Cultural norms |
|
“Connected,” “belonging,” “community,” “involved,” “engaged,” “included” | Relatedness |
|
“Confidence,” “ability,” “skills,” “competence,” “perform,” “capable,” “achieve” | Perceived competence |
|
Figure 1 was instrumental in developing three overarching themes from the qualitative analysis: Lack of Autonomy and Its Impact on School Engagement, Importance of Supportive Relationships in Educational Persistence, and The Role of Economic and Social Barriers in School Dropout. Each theme encapsulates a critical aspect of the participants’ experiences, reflecting the complex interplay between psychological factors, socioeconomic challenges, and cultural expectations that influence girls’ educational outcomes in rural Malawi.

Word cloud highlighting key concepts from qualitative analysis.
5.5 Lack of Autonomy and Its Impact on School Engagement
A key theme from the qualitative data was the lack of autonomy in educational decision-making, with participants highlighting parental control, societal expectations, and restrictive school policies as significant barriers. This lack of autonomy was linked to disengagement and increased dropout risks, supported by quantitative findings showing that higher perceived autonomy significantly reduces dropout likelihood (β = −0.35, p < 0.01). One dropout expressed frustration: “My parents decided everything for me… I felt like I had no control over my life” (Participant A). Another noted, “The school rules were so strict… we had no say in anything” (Participant B). These accounts reveal how a lack of autonomy fosters feelings of powerlessness and disengagement.
This disconnection arises when students view education as imposed rather than self-directed. One participant shared, “When I wasn’t allowed to choose my subjects, I stopped caring about school. It didn’t feel like it was my life anymore” (Participant C). Another remarked, “It was hard to stay interested in school when everything was decided for me. I felt like a puppet” (Participant D). These narratives underscore the importance of autonomy in fostering intrinsic motivation, engagement, and educational persistence.
5.6 Perceived Competence and Academic Performance
Perceived competence strongly impacts academic performance and dropout rates, as confirmed by PA, demonstrating its direct and indirect effects on academic performance (β = −0.15, p < 0.05). Low self-efficacy often leads to academic struggles, declining motivation, and increased dropout risks. Students internalize failure, eroding confidence and fostering disengagement. One dropout noted, “I tried so hard, but every time I failed, it just confirmed I wasn’t smart enough. Eventually, I stopped trying” (Participant E). Another added, “I felt like I was always behind, and the teachers didn’t care. It made me feel like I didn’t belong” (Participant G). These narratives underscore how perceived incompetence undermines belonging, fostering disengagement and dropout.
Supportive relationships are critical anchors for students navigating educational challenges in rural Malawi. These connections mitigate isolation and foster resilience, with quantitative findings showing that perceived relatedness significantly reduces dropout risk (β = −0.32, p < 0.01). A dropout shared, “I was bullied a lot, and the teachers didn’t do anything. I felt so alone and hated school” (Participant G). In contrast, a current student remarked, “Having supportive friends and teachers makes a huge difference. It feels like a family, and you don’t want to leave” (Participant I). These contrasting experiences highlight the transformative role of support in fostering belonging and resilience, emphasizing the need for inclusive school environments that ensure all students feel valued
5.7 The Role of Economic and Social Barriers in School Dropout Rates
Economic challenges, household duties, and cultural norms significantly hinder education, with financial hardships often forcing girls to prioritize work or domestic tasks. LRA confirms economic status as a critical dropout predictor (OR = 0.75, p < 0.05). One participant shared, “My parents couldn’t afford my uniform or books, so I stopped schooling” (Participant N). Another said, “I cared for my siblings and did all the chores, leaving no time for school” (Participant M). These accounts highlight the compounded pressures leading to dropping out and the urgency for targeted financial support and interventions.
5.8 Household Responsibilities and Educational Engagement
Household responsibilities emerged as a significant barrier to educational engagement, consuming students’ time and energy while fostering a sense of disempowerment and reducing aspirations. Quantitative findings confirmed this, with household responsibilities identified as a significant predictor of dropout (OR = 1.55, p < 0.01). One dropout shared, “I had to care for my siblings and do all the chores, leaving no time for school” (Participant M, dropout), showing how heavy household duties cause disengagement and higher dropout risks.
5.9 Cultural Norms and Gender Roles
Cultural norms and gender roles limit girls’ education and increase dropout risks, as confirmed by the intersection of gender, poverty, and cultural norms. One dropout noted, “No female teachers to look up to, and male teachers didn’t care” (Participant H). Another shared, “I felt disconnected; no one cared if I was there” (Participant S). These accounts underscore the need for culturally sensitive interventions to promote gender equity in education.
6 Discussion
The results of this study provide a nuanced understanding of the factors influencing girls’ dropout rates from secondary schools in rural Malawi through the lens of Intersectionality Theory. This discussion situates the findings within the broader literature and the specific context of Malawi, highlighting corroborating and contradictory evidence from previous studies.
There was a critical trend in the educational journey of girls in Malawi, where dropout rates significantly increased as students progressed to higher forms, particularly in Forms three and 4. These findings corroborate studies by the Ministry of Education, Science, and Technology (2019), National Statistical Office, and ICF (2017), as well as the regional trends in the Global South, as reported by UNICEF (2023). This trend, which sees the highest dropout rates among girls aged 16–17 years (29 %), reflects the growing socioeconomic and cultural challenges they face as they near the completion of their secondary education. Despite the financial and social interventions by the government and NGOs, the statistics remain substantive (UNICEF 2023). Conversely, the dropout rates are markedly lower among younger students, with only 2 % of dropouts occurring among those aged 12–13 years in Form 1. This distribution aligns with the broader context of girls’ education in Malawi, where economic pressures, cultural norms, and educational barriers intensify in the later years of schooling, contributing to higher dropout rates among older girls.
Economic hardship is a predominant factor driving this trend. As girls grow older, the expectation to contribute to their family’s income becomes more pressing, often at the expense of their education. In many cases, families in Malawi are forced to make difficult decisions, prioritizing immediate economic survival over long-term educational attainment (UNICEF 2018). This economic imperative is particularly acute in rural areas, where opportunities for secondary education are limited, and the costs associated with schooling – such as transportation, uniforms, and school fees – are prohibitively high.
Cultural factors, especially early marriage and pregnancy, also play a significant role in the high dropout rates among older girls. These practices are deeply ingrained in many communities, particularly in rural areas, where girls are often seen as ready for marriage as soon as they reach puberty. As girls approach the ages of 16–17, the pressure to marry or adhere to traditional gender roles becomes overwhelming, leading to their withdrawal from school (National Statistical Office and ICF 2017). This trend is reflected in the data, which shows a sharp increase in dropouts in the upper forms, as girls are more likely to leave school for marriage or due to pregnancy.
The availability and quality of educational resources further exacerbate the problem. As students progress to higher forms, the academic content becomes more challenging, and the number of schools offering Forms 3 and 4 diminishes, particularly in rural areas. This scarcity of resources, combined with the difficulty of the curriculum, creates a situation where many girls struggle to keep up academically and are eventually forced to drop out (Ministry of Education, Science, and Technology 2019). The higher dropout rates in the later forms indicate the challenges girls face in accessing and completing secondary education, particularly in environments with limited educational support.
These findings corroborate a plethora of extant studies and reports. According to UNICEF (2018), dropout rates among girls in Malawi increase significantly as they approach the end of secondary school, driven by a combination of economic pressures, cultural expectations, and limited access to educational resources. Similarly, the Malawi Ministry of Education et al. (2019) has documented higher dropout rates in Forms three and 4, underscoring the impact of academic and logistical challenges on students’ ability to complete their education. The National Statistical Office and ICF (2017) also highlight the role of early marriage and pregnancy in exacerbating dropout rates, particularly among older girls.
The EFA and CFA confirmed that perceived autonomy, competence, and relatedness are distinct and significant constructs influencing dropout rates. These findings are consistent with SDT, which posits that fulfilling these psychological needs fosters intrinsic motivation and engagement. In the Malawian context, Chimombo et al. similarly found that the lack of supportive teacher-student relationships and rigid school structures diminish these psychological constructs, leading to higher dropout rates (Chimombo et al. 2014). Deci and Ryan reported that enhancing perceived autonomy and competence can significantly reduce dropout rates (Deci and Ryan 2012), corroborating our findings that higher perceived autonomy (β = −0.35, p < 0.01) and competence (β = −0.38, p < 0.01) are associated with lower dropout risks. The significance of relatedness (β = −0.32, p < 0.01) further underscores the importance of supportive social environments, aligning with Vallerand et al. (Vallerand, Fbrtier, and Guay 1997), who highlighted the role of relatedness in fostering academic persistence.
The HLM results [distance to school (β = 0.22, p < 0.05) and school resources (β = −0.26, p < 0.05)] revealed significant multi-level influences on dropout rates, including individual psychological factors and school-level variables such as distance to school and school resources, corroborating relevant extant studies.
Consistent with local findings by Gondwe (Gondwe 2016), our LRA identified perceived autonomy, competence, relatedness, household responsibilities, and economic status as significant predictors of dropout. Gondwe also highlighted that economic hardship and domestic duties are significant barriers to girls’ education in Malawi. Globally, UNICEF reported that socioeconomic barriers are primary determinants of school dropout among girls. The strong predictive power of household responsibilities (OR = 1.55, p < 0.01) and economic status (OR = 0.75, p < 0.05) in our study underscores the need for interventions that address these socioeconomic challenges, corroborating extant global studies (Hanushek and Woessmann 2021; UNICEF 2020, 2023; World Bank 2023).
PA revealed that perceived competence directly and indirectly affected dropout rates through academic performance. Our findings corroborate Fortier et al., who noted that academic self-efficacy influences students’ persistence via its impact on academic achievement (Fortier, Vallerand, and Guay 1995). The link between self-perception of competence and academic performance in Malawi has been well-documented. The indirect effect of perceived competence (β = −0.15, p < 0.05) through academic performance further supports the importance of fostering academic self-efficacy to enhance retention.
DA showed that perceived autonomy, competence, relatedness, household responsibilities, and economic status effectively discriminated between dropouts and non-dropouts. This finding is consistent with Deci and Ryan (Deci and Ryan 2012), who emphasized the role of psychological factors and socioeconomic conditions in educational outcomes. Chimombo et al. identified similar discriminating factors in Malawi, highlighting the critical need to address psychological and contextual educational barriers (Chimombo et al. 2014).
CA and LCA identified distinct profiles and latent classes based on students’ survey responses. These analyses revealed groups with varying levels of perceived autonomy, competence, and relatedness corresponding to different dropout risks. This approach is supported by Bowers and Sprott, who demonstrated that identifying student profiles can help tailor interventions to specific needs (Bowers and Sprott 2012). In Malawi, understanding these profiles is essential for developing effective strategies to address the unique challenges faced by different groups of students.
6.1 Policy Recommendations
This study highlights the need for targeted policies to address socioeconomic barriers causing girls’ dropout rates in rural Malawi, using intersectionality and Self-Determination Theory (SDT) to boost motivation. Policymakers can create a supportive educational environment by enhancing girls’ autonomy, competence, and relatedness and considering local cultural contexts. Recommended policies include increasing girls’ decision-making in schools, providing academic support, fostering an inclusive school climate, improving rural school infrastructure, offering financial aid to low-income families, promoting gender-equitable household responsibilities, implementing after-school programs, and establishing monitoring frameworks. These measures aim to support girls’ education, autonomy, and gender equality.
6.2 Sustainable Academic Support Programs in Resource-Constrained Settings: Challenges and Community-Based Solutions in the Global South
The recommendations address the psychological and socioeconomic barriers to girls’ education in rural Malawi through strategies to enhance autonomy, competence, and relatedness. Success depends on careful planning, sustainability, and addressing challenges. Fostering autonomy through student councils and curriculum involvement can be met with resistance, which targeted staff training can mitigate. After-school tutoring and peer mentoring improve competence and resilience but face challenges in resource-constrained settings, requiring partnerships with NGOs and universities to effectively utilize community spaces and resources.
Community engagement is critical for infrastructure development and program sustainability. However, misunderstandings about participation in low-literacy areas can lead to resistance. Awareness campaigns emphasizing shared responsibility and long-term benefits, supported by culturally relevant communication, are essential to foster ownership.
Financial aid is vital for reducing dropout rates but risks fostering dependency or misuse. Tying aid to measurable outcomes, providing financial literacy training, and using robust accountability mechanisms, including digital tracking and community monitoring, ensure effective use of resources and empower recipients. Lessons from Zambia and Kenya show the need for strong governance to prevent corruption and dependency.
Malawi’s cultural diversity requires tailored solutions for patriarchal and matrilineal communities. In patriarchal areas, campaigns should engage male leaders to advocate for equitable household responsibilities. In matrilineal systems, programs can empower women to support girls’ education while addressing domestic burdens through collaborative household solutions and after-school programs.
Robust monitoring frameworks using mobile data collection and partnerships with international organizations can track intervention outcomes and adaptability, ensuring effectiveness in diverse cultural contexts. These strategies collectively foster a more inclusive and equitable educational environment while addressing structural and cultural barriers.
7 Conclusions
This study advances understanding of gender and education in the Global South by highlighting the intersectional factors shaping girls’ educational outcomes in rural Malawi. Integrating SDT and Intersectionality Theory reveals how autonomy, competence, and relatedness interact with socioeconomic and cultural barriers to influence dropout risks. The findings show how poverty and entrenched gender norms undermine psychological needs, emphasizing the need for tailored, intersectional interventions addressing individual and systemic barriers (Vogt, Gardner, and Haeffele 2012).
HLM and logistic regression highlight the roles of psychological and structural factors, such as school distance and inadequate resources, in predicting dropout, stressing the need for both empowerment and infrastructural improvements. PA demonstrates competence’s direct and indirect effects on dropout via academic performance, underscoring the importance of fostering self-efficacy. Discriminant and cluster analyses reveal distinct student profiles, supporting the necessity of context-sensitive, targeted interventions over-generalized approaches.
These recommendations require careful planning, sufficient resources, and sustained monitoring. Insights from this study offer valuable guidance for advancing gender equity and reducing educational disparities, not only in Malawi but across similar Global South contexts.
8 Limitations and Future Research Directions
The study’s reliance on secondary data limits its relevance to current contexts, underscoring the need for future primary data collection tailored to the research questions. Its geographical focus on one district restricts generalizability, highlighting the importance of including multiple regions for broader insights. The cross-sectional design prevents causal inferences, suggesting future longitudinal studies to capture evolving factors and intervention impacts. While profiles of autonomy, competence, and relatedness were identified, the lack of tested interventions for these groups limits practical application. Addressing these limitations through targeted, longitudinal research will enhance strategies to improve girls’ educational outcomes in Malawi.
-
Research ethics: This study utilized secondary data collected as part of a previous thesis research project that was approved by the relevant department under the institution’s research ethics committee. The original data collection followed strict ethical guidelines to ensure the confidentiality, anonymity, and informed consent of all participants. As the current study re-analyzed this pre-existing data, no new data collection was undertaken, and the ethical standards of the original study were maintained throughout this research.
-
Informed consent: Not applicable.
-
Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.
-
Use of Large Language Models, AI and Machine Learning Tools: None declared.
-
Conflict of interest: The authors declare no conflicts of interest related to this study.
-
Research funding: This research received no specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
-
Data availability: Not applicable.
References
A4ID. 2022. SDG 5: Gender Equality. A Legal Guide. www.a4id.org.Search in Google Scholar
Ang, J. W. J., Y. N. Ng, L. H. W. Lee, and J. Y. Yong. 2024. “Exploring Students’ Learning Experience and Engagement in Asynchronous Learning Using the Community of Inquiry Framework through Educational Design Research.” Education Sciences 14 (3). https://doi.org/10.3390/educsci14030215.Search in Google Scholar
Banda, L. O.L., C. V. Banda, J. T. Banda, and T. Singini. 2024a. “Preserving Cultural Heritage: A Community-Centric Approach to Safeguarding the Khulubvi Traditional Temple Malawi.” Heliyon: e37610. https://doi.org/10.1016/j.heliyon.2024.e37610.Search in Google Scholar
Banda, L. O. L., C. V. Banda, J. T. Banda, T. T. Hlaing, and E. Mwaene. 2024b. “Assessing Farmers’ Knowledge of Environmental Policy along the Ayeyarwady River: Strides towards the Indian Ocean Marine Life Safety.” Heliyon 10 (16). https://doi.org/10.1016/j.heliyon.2024.e35503.Search in Google Scholar
Banda, L. O. L., C. V. Banda, J. T. Banda, E. Mwaene, G. N. C. Munthali, T. T. Hlaing, and B. Chiwosi. 2024c. “Unraveling Agricultural Water Pollution Despite an Ecological Policy in the Ayeyarwady Basin.” BMC Public Health 24 (1). https://doi.org/10.1186/s12889-024-19084-7.Search in Google Scholar
Banda, L. O. L., J. T. Banda, C. V. Banda, E. Mwaene, and C. H. Msiska. 2024d. “Unraveling Substance Abuse Among Malawian Street Children: A Qualitative Exploration.” PLoS One 19 (5): e0304353. https://doi.org/10.1371/journal.pone.0304353.Search in Google Scholar
Banda, L. O. L., J. Liu, W. Zhou, and J. T. Banda. 2024e. “Is Face-to-Face Scrambled Teaching Practice Supervision Effective Amidst Natural Disasters and Pandemics? The Teaching Practice Students’ Perspectives.” Sage Open 14 (2). https://doi.org/10.1177/21582440241241368.Search in Google Scholar
Boris, E. 2008. “Book Review: The Politics of Disgust: The Public Identity of the Welfare Queen.” Gender & Society 22 (4): 527–9. https://doi.org/10.1177/0891243208316864.Search in Google Scholar
Bowers, A. J., and R. Sprott. 2012. “Why Tenth Graders Fail to Finish High School: A Dropout Typology Latent Class Analysis.” Journal of Education for Students Placed at Risk 17 (3): 129–48. https://doi.org/10.1080/10824669.2012.692071.Search in Google Scholar
Bryman, A. 2006. “Integrating Quantitative and Qualitative Research: How Is it Done?” Qualitative Research 6 (1): 97–113. https://doi.org/10.1177/1468794106058877.Search in Google Scholar
Calleja, P. 2009. “Foundations of Mixed Methods Research: Integrating Quantitative and Qualitative Approaches in the Social and Behavioral Sciences.” Australasian Emergency Nursing Journal 12 (4): 145. https://doi.org/10.1016/j.aenj.2009.07.004.Search in Google Scholar
Cameron, C., C. Chiiya, M. Chonta, N. Israelite, S. Nixon, D. Njelesani, J. Njelesani, P. Parnes, E. Swai, J. Walker, and A. Figue, et al.., ActionAid. 2020. The Bedrock of Inclusion: Why Investing in the Education Workforce Is Critical to Delivering SDG4. Johannesburg: ActionAid.Search in Google Scholar
CAMFED. 2023. Annual review. www.CAMFED.org.Search in Google Scholar
Chaitanya, G., P. Tevari, and D. Hanumanthappa. 2024. “Path Analysis: An Overview and its Application in Social Sciences.” International Journal of Agriculture Extension and Social Development 7 (4): 299–303. https://doi.org/10.33545/26180723.2024.v7.i4d.556.Search in Google Scholar
Chimombo, J., M. Chibwanna, C. Dzimadzi, E. Kadzamira, E. Kunkwenzu, D. Kunje, and D. Namphota. 2014. “Classroom, School, and Home Factors that Negatively Affect Girls’ Education in Malawi.” International Journal of Educational Development 24 (3): 231–43.Search in Google Scholar
Collins, P. H. 1991. Black Feminist Thought: Knowledge, Consciousness, and the Politics of Empowerment, 11th ed. Oxfordshire: Routledge.Search in Google Scholar
Collins, P. H. 2015. “Intersectionality’s Definitional Dilemmas.” In Annual Review of Sociology, 41, 1–20. San Mateo: Annual Reviews Inc.10.1146/annurev-soc-073014-112142Search in Google Scholar
Crenshaw, K. 1989. “Demarginalizing the Intersection of Race and Sex: A Black Feminist Critique of Antidiscrimination Doctrine, Feminist Theory and Antiracist Politics.” University of Chicago Legal Forum 1: 139–67.Search in Google Scholar
Creswell, J. W., and V. L. Plano Clark. 2017. Designing and Conducting Mixed Methods Research, 3rd ed. Washington, DC: SAGE Publications.Search in Google Scholar
Deci, E. L., and R. M. Ryan. 2012. “Self-determination Theory.” In Handbook of Theories of Social Psychology, edited by P. A. M. Van Lange Arie, W. Kruglanski, H. Tory, and V. Lange, 1st ed. Washington, DC: SAGE.Search in Google Scholar
DeMaris, A. 1995. “A Tutorial in Logistic Regression.” Wiley 57 (4): 956. https://doi.org/10.2307/353415.Search in Google Scholar
Eardi, L., W. Zhang, S. R. Levendovszky, M. W. Weiner, P. Aisen, M. W. Weiner, P. Aisen, et al.. 2024. “Interpretable Discriminant Analysis for Functional Data Supported on Random Nonlinear Domains with an Application to Alzheimer’s Disease.” Journal of the Royal Statistical Society - Series B: Statistical Methodology 86. https://doi.org/10.1093/jrsssb/qkae023.Search in Google Scholar
Edgerton, E., and J. McKechnie. 2023. “The Relationship between Student’s Perceptions of Their School Environment and Academic Achievement.” Frontiers in Psychology 13. https://doi.org/10.3389/fpsyg.2022.959259.Search in Google Scholar
Fitzmaurice, G. M., and N. M. Laird. 2015. Binary Response Models and Logistic Regression, 587–95. Elsevier BV.10.1016/B978-0-08-097086-8.42060-XSearch in Google Scholar
Fortier, M. S., R. J. Vallerand, and F. Guay. 1995. “Academic Motivation and School Performance: Toward a Structural Model.” Contemporary Educational Psychology 20 (3): 257–74. https://doi.org/10.1006/ceps.1995.1017.Search in Google Scholar
Gershenson, S., C. M. D. Hart, J. Hyman, C. Lindsay, and N. W. Papageorge. 2021. The Long-Run Impacts of Same-Race Teachers, 25254. National Bureau of Economic Research. http://www.nber.org/papers/w25254.Search in Google Scholar
Ghavami, N., and L. A. Peplau. 2018. “Urban Middle School Students’ Stereotypes at the Intersection of Sexual Orientation, Ethnicity, and Gender.” Child Development 89 (3): 881–96. https://doi.org/10.1111/cdev.12763.Search in Google Scholar
Gondwe, G. C. 2016. “Factors Influencing Rural Female Pupils Drop Out from Primary Schools, in Nkhata-Bay South District, Malawi.” Journal of Education and Practice 7 (35): 83–9.Search in Google Scholar
Good, C., A. Rattan, and C. S. Dweck. 2012. “Why Do Women Opt Out? Sense of Belonging and Women’s Representation in Mathematics.” Journal of Personality and Social Psychology 102 (4): 700–17. https://doi.org/10.1037/a0026659.Search in Google Scholar
Goodenow, C., and K. E. Grady. 1993. “The Relationship of School Belonging and Friends’ Values to Academic Motivation Among Urban Adolescent Students.” The Journal of Experimental Education 62: 60–71. https://doi.org/10.1080/00220973.1993.9943831.Search in Google Scholar
Graham, S. H. 2018. “Race/Ethnicity and Social Adjustment of Adolescents: How (Not if) School Diversity Matters.” Educational Psychologist 53: 64–77. https://doi.org/10.1080/00461520.2018.1428805.Search in Google Scholar
Greene, J. C. 2007. Mixed Methods in Social Inquiry. Hoboken, NJ: John Wiley & Sons.Search in Google Scholar
Hancock, A. M. 2004. The Politics of Disgust: The Public Identity of the Welfare Queen. New York, NY: New York University Press.Search in Google Scholar
Hanushek, E. A., and L. Woessmann. 2021. “Education and Economic Growth.” In Oxford Research Encyclopedia of Economics and Finance. Oxford: Oxford University Press.10.1093/acrefore/9780190625979.013.651Search in Google Scholar
Hosmer, D. W., S. Lemeshow, and R. X. Sturdivant. 2013. Introduction to the Logistic Regression Model, 1–33. Hoboken, NJ: Wiley.10.1002/9781118548387.ch1Search in Google Scholar
Kumar, M. K. M. S., and H. V. Reddy. 2020. “Comprehensive Models towards for Feature Extraction and Recognition in Machine Learning.” The International Journal of Recent Technology and Engineering (IJRTE) 8 (6): 3638–41. https://doi.org/10.35940/ijrte.f7997.038620.Search in Google Scholar
Lloyd, R., and D. Mertens. 2018. “Expecting More Out of Expectancy Theory: History Urges Inclusion of the Social Context.” International Management Review 14 (1): 24–37.Search in Google Scholar
Malawi Ministry of Education, Science, and Technology. 2019. Education Sector Performance Report 2019. Li: Ministry of Education.Search in Google Scholar
Mamie Martin Fund. 2016. Girls’ Education in Malawi. https://mamiemartin.org/.Search in Google Scholar
McCall, L. 2005. “The Complexity of Intersectionality.” Signs 30 (3): 1771–800. https://doi.org/10.1086/426800.Search in Google Scholar
Meissner, H. I., J. W. Creswell, A. C. Klassen, V. L. P. Clark, and K. C. Smith. 2018. “Mixed Methods Procedure.” In Research Design: Qualitative, Quantitative, and Mixed Methods Approaches, 213–46. Los Angeles: SAGE.Search in Google Scholar
Meteyard, L., and R. Davies. 2020. “Best Practice Guidance for Linear Mixed-Effects Models in Psychological Science.” Journal of Memory and Language 112: 104092. https://doi.org/10.1016/j.jml.2020.104092.Search in Google Scholar
National Statistical Office (NSO) & ICF. 2017. Malawi Demographic and Health Survey 2015–16. Zomba, Malawi, and Rockville. Maryland, USA: NSO and ICF.Search in Google Scholar
OECD. 2019. PISA 2018 Results, 1. Paris: OECD.Search in Google Scholar
Patton, M. Q. 2015. Qualitative Research & Evaluation Methods: Integrating Theory and Practice, 4th ed.. Washington, DC: Sage Publications.Search in Google Scholar
Stoltzfus, J. 2011. “Logistic Regression: A Brief Primer.” Wiley 18 (10): 1099–104. https://doi.org/10.1111/j.1553-2712.2011.01185.x.Search in Google Scholar
Tashakkori, A., and C. Teddlie. 2010. SAGE Handbook of Mixed Methods in Social & Behavioral Research, 2nd ed. Washington, DC: SAGE Publications.10.4135/9781506335193Search in Google Scholar
The Government of Malawi. 2018. National Strategy for Adolescent Girls and Young Women-2018–2022. Lilongwe: Government of Malawi.Search in Google Scholar
Tripathy, P. J. 2013. “Secondary Data Analysis: Ethical Issues and Challenges.” Iranian Journal of Public Health 42 (12): 1478–9.Search in Google Scholar
UNESCO Education Sector 2030. 2017. Unpacking Sustainable Development Goal 4 Education 2030 GUIDE. Paris: UNESCO.Search in Google Scholar
UNICEF. 2018. Malawi: Education Factsheet. UNICEF Malawi. Retrieved from: https://www.unicef.org/malawi.Search in Google Scholar
UNICEF. 2020. Addressing-the-learning-crisis-advocacy-brief-2020.Search in Google Scholar
UNICEF. 2023. For Every Child UNICEF Annual Report. www.unicef.org.Search in Google Scholar
United Nations Development Programme (UNDP). 2022. Education and Gender in Malawi. New York City: UNDP. https://www.undp.org.Search in Google Scholar
Vallerand, R. J., M. S. Fbrtier, and F. Guay. 1997. “Self-Determination and Persistence in a Real-Life Setting toward a Motivational Model of High School Dropout.” Journal of Personality and Social Psychology 72 (5). https://doi.org/10.1037//0022-3514.72.5.1161.Search in Google Scholar
Vogt, W. P., D. C. Gardner, and L. M. Haeffele. 2012. When to Use what Research Design. New York, NY: Guilford Press.Search in Google Scholar
World Bank. 2023. Development A New Era Contents.Search in Google Scholar
© 2025 the author(s), published by De Gruyter on behalf of the Zhejiang Normal University, China
This work is licensed under the Creative Commons Attribution 4.0 International License.