The Effect of Sociodemographic Factors on Female Educational Attainment: An Application of the Multipopulation Curie–Weiss Model
Abstract
In recent times, there have been a lot of advocacies to encourage females of all ages to pursue education to the highest level. The Government of Ghana and its stakeholders have championed this to achieve the Sustainable Development Goal 4 in this regard. However, progress has been on a snail’s pace. In this study, we employed the multipopulation Curie–Weiss model to establish the influence of sociodemographic factors on the educational attainment of all females in Ghana. We utilized data from the Ghana Demographic and Health Survey (2008 and 2014). The parameters of the interacting and noninteracting multipopulation Curie–Weiss models were estimated using the partial least squares and ordinary least squares methods, respectively. Our findings suggest that marital status, place of residence, and wealth status of females in Ghana influence their educational attainment. Therefore, to get more females of all ages to attain higher levels of education, the Government of Ghana and its stakeholders should channel their efforts towards eradicating poverty at all levels, preventing early and teenage marriages, providing better infrastructure, and creating better living conditions for all females especially those in the poorest regions of the country.
1. Introduction
Successive governments over the years have made frantic efforts to improve the lives of Ghanaians and make education accessible to all irrespective of their income status, gender, or location in the country (urban or rural). This has led to the introduction of numerous interventions like health insurance, free maternal care, free education from basic school to senior high school, and increased infrastructure [1–3]. Although, this may have had an enormous impact on people’s lives, it cannot be mathematically or statistically substantiated. Data from the World Bank revealed a significant gender disparity in adult literacy rates in Ghana for individuals aged 15 and above. While the overall literacy rate stands at 69.8%, indicating that roughly seven out of ten people in this age group can proficiently read and write, the rates differ by gender. Specifically, 74.1% of males are literate compared to 65.4% of females, highlighting a need for targeted educational interventions to bridge this gap [4].
Akanbang, Aneleru, and Aziabah [5] demonstrated that following the implementation of intervention by the campaign for female education (CAMFED) in some selected districts in Ghana, the number of female students enrolled in schools and their academic advancement significantly rose. However, despite the behaviour modification component of the intervention system, gender-based attitudes still prevent females from receiving an adequate education. Their study comes to the conclusion that in order to maintain the progress gained thus far, government initiatives are needed to tackle the cultural and socioeconomic barriers impeding females’ education.
The Strategic Approach to Girls’ Education (STAGE) programme was initiated to assist underprivileged out-of-school females in Ghana’s northernmost regions in enrolling in and succeeding in primary school. The STAGE relied on the Complementary Basic Education (CBE) strategy of the Ghanaian government, which encourages an accelerated learning programme that offers literacy (reading) and mathematical lessons to out-of-school females between the ages of 8 and 14 in their mother tongue [6]. There have been similar and numerous educational reforms in Ghana, the recent one being the 2018 reforms targeted at 12 key areas mainly geared towards making education accessible at all levels in all 16 regions irrespective of gender, income level, and residential status [7]. STAGE was able to achieve this aim through working with elementary schools, local officials, and families, as well as by teaching reading and arithmetic to out-of-school females to help them succeed in school [6]. Although a number of reforms have taken shape and are being implemented, a lot of bottlenecks still exist and many disparities are observed in terms of coverage and infrastructure across the country with more than half of Ghana’s population being urban [8–10].
Ghana’s policy design and sustainability drives have been mostly urbanised [11]. Ghana’s population generally has more females than males with more females than males, in urban areas as compared to more males than females in the rural areas [12]. According to data from the Ghana Statistical Service (GSS) [13], between 2010 and 2021, the national urban population in Ghana increased from 50.9% to 56.7%. Nearly half of this growth was concentrated in the Greater Accra and Ashanti regions. Urban areas also exhibited a higher concentration of youth, and the female population there exceeded the male population by 5.0%. Conversely, in rural areas, males slightly outnumber females by 0.3%. These demographic shifts are crucial for understanding regional educational needs and resource allocation.
Gender has turned into a vital determinant for admittance to formal schooling in Ghana. Although Ghana is well noted for having a praiseworthy schooling strategy on paper with free training for everybody, the dropout rates are high. This is especially valid for females below the age of 15 who have higher dropout rates than males below the age of 15 [14]. In Ghana, the educational retention and dropout rates at the elementary level show notable differences between males and females.
As of 2024, the population of Ghana is relatively balanced between males and females, with females comprising approximately 50.13% and males 49.87%. This demonstrates a very slight female majority in the demographic composition of Ghana [5]. According to the GSS [13], about 15% of females and 17% of males in primary school are at risk of dropping out, indicating a slightly higher dropout rate for males than females at this educational level. Efforts to improve gender parity in enrolment have resulted in nearly equal levels for males and females in primary education. However, challenges persist in retaining students through to the completion of their schooling. The overall completion rate for primary education is about 71%, although this figure does not provide a gender-specific breakdown [13, 15]. Many females struggle to access quality education due to long distances to schools, gender bias, and poverty [16].
In a study conducted in Ethiopia [17], the researchers focused on the importance of female’s education and its potential impact on various developmental aspects. They highlighted that female’s education is crucial for achieving faster economic growth, increased life expectancy, reduced population growth, improved quality of life, and higher returns on investment in developing countries. A recent paper by Ebrahimi et al. [18] contributed to expanding field of education by highlighting key dimensions that are crucial for empowering females, particularly in terms of their socioeconomic advancement. According to Zhang and Liang [19], there is a positive influence of education on marital satisfaction. They emphasized the significance of educational development in enhancing marital satisfaction, which is a pivotal factor in the enduring success and stability of marriages. They concluded that the quality of a marital relationship significantly affects the mental and emotional well-being of both partners. Therefore, understanding the variables that influence marital satisfaction is critical for developing interventions aimed at enhancing relationship quality.
The significance of education as a crucial factor in determining individual well-being was studied by [20]. Their findings revealed that factors such as material well-being, identity, and capabilities also play a role in enhancing overall well-being. Interestingly, the study suggested that education has a more pronounced impact on female’s well-being compared to male’s well-being.
Li and Qiu [21] highlighted the substantial influence of family background on children’s academic achievement, even within systems of compulsory education designed for social justice. This influence primarily manifests through economic resources and social capital. Economically advantaged families can afford better educational resources, such as tutoring and technology, which directly support learning. Additionally, social capital enables these families to leverage relationships and networks to provide their children with further educational advantages and opportunities. This perpetuates disparities in academic achievement, despite efforts to equalize educational access. Children from rural and urban impoverished communities, members of linguistic and racial minorities, and those who are displaced have long faced prejudice and marginalization in education in Ghana and other regions of Sub-Saharan Africa [22].
In technical and vocational educational training, [23] revealed that a male enrolment has exceeded that of females by about 51.6% for several academic years. They showed that the high expense of materials and equipment needed for practical sessions in these programs was the main barrier to female enrolment and academic advancement. Female enrolment in these programs was also impacted by the absence of female role models and criticisms of Technical Vocational Education and Training (TVET) made by the general public.
Statistical mechanical models can be used to study the relationships and interactions among these sociodemographic variables. The structure of the statistical mechanical models, specifically the multipopulation Curie-Weiss model, has been proven to be a useful tool for studying sociodemographic factors [24]. It is believed that social interactions or sociodemographic factors may influence educational attainment. This has motivated the application of statistical mechanics to the study of various social interactions and interventions [25].
Phan et al. [26] reviewed the ideas of [24–30] and [31] to propose three classes of models moving continuously from judicious towards versatile assumptions. They discussed the gamble and chances of rendering apparatus, strategies, and ideas from statistical mechanics to financial matters together with the correlation between two models of heterogeneous quirky inclinations, compared to cases with extinguished and toughened jumble in measurable physical and social sciences.
Agliari et al. [32] applied the theory of probability and statistical mechanics to investigate social interactions in Italy using annual data from 2002 to 2010 collected by the Italian National Institute of Statistics. Their models showed stronger interactions in urban areas than those in the countryside. Opoku, Osabutey, and Kwofie [33] utilized a model within the framework of statistical mechanics as a worldview for instructive decisions when the population of interest is apportioned by the characteristics of sex orientation and residential settings. They concentrated on how fulfilment in education is influenced by the property of sex orientation and residential settings for five countries in the World Bank’s developing country setting. Their adapted model was made up of a social and a private motivation part with coefficients estimating the influence that people have on one another vis-à-vis their external influence. Estimation of parameters in their model was done using partial least squares (PLS) and ordinary least squares to separately assess the boundaries of the interacting and noninteracting models.
Osabutey, Opoku, and Gyamfi [34] employed the Curie–Weiss and discrete choice models within a social interaction framework to study energy management from the perspective of energy specialists. They showed that the multipopulation forms of the models are good at incorporating social interactions (i.e., specialist views, economic, and sociopsychological). Societal perspectives of urban sustainable development-related concerns and sociodemographic factors affecting education in Ghana vary [11]. It is believed that social interactions or sociodemographic factors may influence educational attainment [35].
The majority of previous studies on female educational attainment have relied on cross-sectional studies or logistic regression. However, our study seeks to employ the multipopulation Curie–Weiss model, a statistical mechanical model, in order to examine the impact of sociodemographic factors on female educational attainment.
The remaining part of our study is in three sections. The first section discusses the methodology applied and the data used in the study. This is followed by the results and discussions from the analyses of the data. The final section is made up of conclusions and recommendations.
2. Data and Methods
2.1. Data
The study employed secondary data from the national cross-sectional Ghana Demographic and Health Survey (GDHS) carried out in 2008 and 2014. Previous surveys were excluded from this study due to the absence of the wealth status variable. The wealth status of households (HHs) (a group of people, related or unrelated, who live together in the same house) of sampled females is an important factor that affects the educational status of females in the study setting. The data comprise of information on females aged between 15 and 49 years collected within enumeration areas (EAs) across Ghana. The study sample was obtained through a multistage sampling procedure which involved the random selection (stratified representative sampling) of EAs and the subsequent selection of HHs within the selected EAs from which the females were sampled. Further details of the sampling processes used in the demographic and health surveys by the implementing agency, GSS, are discussed in [36]. The final sample used for the study entailed 12,388 females (2992 and 9396 females in 2008 and 2014 GHDSs, respectively).
2.2. Study Variables
The study considered educational attainment of the sampled females as the outcome variable. Educational status was categorized as either educated (coded +1) for females who have had any form of formal education or noneducated (coded as −1) for those without any form of formal education. The covariates considered in this study include the type of place of residence (rural or urban), marital status of females (married or never married), and the wealth status (poor or nonpoor) of HHs where the sampled females live.
2.3. Methods
Now, we will present the model to be examined throughout the paper. Our focus is on understanding how an individual’s decisions or choices are influenced by those made by others in their reference group. Additionally, individuals possess certain characteristics like gender, place of living, educational background, ethnicity, and more, which also impact their decision-making process.
2.3.1. The Curie–Weiss Hamiltonian
In this paper, the Curie–Weiss model for discrete choice with social interaction will be employed as a benchmark model [33]. More specifically, we are interested in how the interactions among females, as well as sociodemographic factors such as marital status, residence, and wealth, impact educational attainment choices of females in Ghana.
2.3.2. Multipopulation Curie–Weiss Model
is the average choice made by the members of the partition a. Individual options have now been replaced with collective choices in our approach. HN(σ) returns the overall population’s total level of satisfaction [33].
Theorem 1. For any choice of the parameters and , the limiting pressure admits the following variational representation:
In this paper, we will look at three sociodemographic factors, that is, marital status, place of residence, and wealth status, each with two alternative values, for a total of eight divisions. In what follows, we put and . Furthermore, we can write the probability that the ith female in group α will choose σi ∈ {−1, +1} to be given by
We can see that the expected value of the ith individual’s choice is given by
As a result of the self-consistency (Equation (19)), for every a = 1, 2, ⋯, d,
Note from Equation (23) that ta is the average decision of group members in a. Furthermore, Ua in Equation (23) is a linear regression model with respect to the parameters Jab and ha. The private incentive component ha is modelled as a linear regression as follows:
2.3.3. Estimation
3. Results and Discussion
The study employed secondary data from two national cross-sectional GDHS carried out in 2008 and 2014. The study entailed 12,388 females (2992 and 9396 females in 2008 and 2014 GHDSs, respectively, involving both rural and urban areas, wealth status (poor and nonpoor), and information on marital status (married and never married). Here, we are looking at females that have been partitioned according to three attributes (, , and ) representing residence, marital, and wealth status, respectively. We would want to investigate how residence, marital, and wealth status influence educational attainment on females in Ghana. Table 1 shows the partition of the population into eight subsections, that is, females in rural and nonpoor, females in rural and poor, females in urban and nonpoor, females in urban and poor, females in rural and married, females in rural and never married, females in urban and married, and females in urban and never married. Tables 2 and 3 show the partitioning of the population on educational attainment with respect to their sociodemographic attributes. We will now compute the values of the attributes , , and representing residence, marital, and wealth status, respectively, depending on the group a female belongs, for the eight cases generated in Table 4. The values assigned to the attributes in each of the eight cases describe the relevance of that attribute to the private incentive part of the Hamiltonian. For instance, in case 1, being female in rural as an attribute does not contribute to the private incentive of a group, while being a female in urban area contributes to the private incentive of a group.
| Attribute | Wealth | Marital status | ||
|---|---|---|---|---|
| Residence | Nonpoor | Poor | Married | Never married |
| Rural | 1777 | 5009 | 4281 | 2505 |
| Urban | 4888 | 714 | 3151 | 2451 |
| Attribute | Wealth | Marital status | ||
|---|---|---|---|---|
| Residence | Nonpoor | Poor | Married | Never married |
| Rural | 1552 | 2671 | 2093 | 2130 |
| Urban | 4325 | 427 | 2444 | 2308 |
| Attribute | Wealth | Marital status | ||
|---|---|---|---|---|
| Residence | Nonpoor | Poor | Married | Never married |
| Rural | 225 | 2338 | 2188 | 375 |
| Urban | 563 | 287 | 707 | 143 |
| Cases | Residence | Wealth | Marital status | |||
|---|---|---|---|---|---|---|
| Rural | Urban | Poor | Nonpoor | Married | Never married | |
| 1 | 0 | 1 | 1 | 0 | 1 | 0 |
| 2 | 1 | 0 | 0 | 1 | 1 | 0 |
| 3 | 0 | 1 | 0 | 1 | 0 | 1 |
| 4 | 1 | 0 | 1 | 0 | 0 | 1 |
| 5 | 0 | 1 | 1 | 0 | 1 | 0 |
| 6 | 1 | 0 | 0 | 1 | 1 | 0 |
| 7 | 0 | 1 | 0 | 1 | 0 | 1 |
| 8 | 1 | 0 | 1 | 0 | 0 | 1 |
3.1. Noninteracting Case
When the βj values representing the private incentives of a partition sum to a positive value, the group will tend to favour education in their decision-making. Conversely, when the sum is negative, the group will lean towards a decision that does not prioritize education. Table 5 presents the estimates for these scenarios.
| Cases | Estimation of parameters |
|---|---|
| Case 1: Females in rural and nonpoor, N1 = 1777 | β1 = −0.0833 |
| β2 = −0.1011 | |
| β3 = −0.0319 | |
| β0 = 0.2222 | |
| Case 2: Females in rural and poor, N2 = 5009 | β1 = 0.0933 |
| β2 = −0.1613 | |
| β3 = −0.1253 | |
| β0 = 0.2062 | |
| Case 3: Females in urban and nonpoor, N3 = 4888 | β1 = −0.0227 |
| β2 = 0.1456 | |
| β3 = −0.0925 | |
| β0 = 0.1075 | |
| Case 4: Female in urban and poor, N4 = 714 | β1 = 0.0535 |
| β2 = 0.0691 | |
| β3 = 0.0578 | |
| β0 = 0.0230 | |
| Case 5: Females in rural and married, N5 = 4281 | β1 = −0.0833 |
| β2 = −0.1011 | |
| β3 = −0.0319 | |
| β0 = 0.2222 | |
| Case 6: Females in rural and never married, N6 = 2505 | β1 = 0.0933 |
| β2 = −0.1613 | |
| β3 = −0.1253 | |
| β0 = 0.2062 | |
| Case 7: Females in urban and married, N7 = 3151 | β1 = −0.0227 |
| β2 = 0.1456 | |
| β3 = −0.0925 | |
| β0 = 0.1075 | |
| Case 8: Females in urban and never married, N8 = 2451 | β1 = 0.0535 |
| β2 = 0.0691 | |
| β3 = 0.0578 | |
| β0 = 0.0230 | |
In the case of urban females who have never been married, all βj values are positive and close to zero. This suggests that factors such as residence, marital status, and wealth have minimal influence on individuals’ personal motives to pursue education when there is no social interaction.
Interestingly, the base private incentive β0 has positive values across all eight cases. This indicates that females in Ghana generally have a preference for some level of education.
Furthermore, it is worth noting that the base private incentive component surpasses the sum of estimates for the private incentives associated with residence, marital status, and wealth status in cases 1, 2, 3, 4, 5, and 6.
3.2. Interacting Case
The utility function of the interactive model incorporates both social and private incentives. Individuals belonging to groups a and b prefer to imitate themselves when the social incentive Jab for these groups is positive. However, if Jab is negative, people in these groups tend to make different choices on average, indicating a lack of conformity. It is important to note that J11 represents the interaction strength of rural, nonpoor females interacting with themselves. According to Table 6, the estimate for J11 is negative, indicating that imitation is not rewarding in this case. On the other hand, J33 represents the interaction strength of urban, nonpoor females interacting with themselves, and the estimate for J33 is positive, suggesting that imitation is rewarded.
| Parameters | Estimates |
|---|---|
| J11 | −0.0161 |
| J12 | −0.0040 |
| J13 | −0.0457 |
| J14 | −0.0017 |
| J15 | 0.0012 |
| J16 | −0.0213 |
| J17 | −0.0211 |
| J18 | −0.0263 |
| J21 | −0.0144 |
| J22 | −0.0036 |
| J23 | −0.0409 |
| J24 | −0.0015 |
| J25 | 0.0010 |
| J26 | −0.0191 |
| J27 | −0.0189 |
| J28 | −0.0235 |
| J31 | 0.0488 |
| J32 | 0.0123 |
| J33 | 0.1384 |
| J34 | 0.0052 |
| J35 | −0.0035 |
| J36 | 0.0646 |
| J37 | 0.0639 |
| J38 | 0.0797 |
| J41 | −0.0058 |
| J42 | −0.0015 |
| J43 | −0.0165 |
| J44 | −0.0006 |
| J45 | 0.0004 |
| J46 | −0.0077 |
| J47 | −0.0076 |
| J48 | −0.0095 |
| J51 | −0.0451 |
| J52 | −0.0113 |
| J53 | −0.1279 |
| J54 | −0.0048 |
| J55 | 0.0032 |
| J56 | −0.0597 |
| J57 | −0.0591 |
| J58 | −0.0736 |
| J61 | −0.0534 |
| J62 | −0.0134 |
| J63 | −0.1514 |
| J64 | −0.0056 |
| J65 | 0.0038 |
| J66 | −0.0706 |
| J67 | −0.0699 |
| J68 | −0.0871 |
| J71 | 0.0424 |
| J72 | 0.0106 |
| J73 | 0.1201 |
| J74 | 0.0045 |
| J75 | −0.0030 |
| J76 | 0.0560 |
| J77 | 0.0554 |
| J78 | 0.0691 |
| J81 | 0.0437 |
| J82 | 0.0110 |
| J83 | 0.1239 |
| J84 | 0.0046 |
| J85 | −0.0031 |
| J86 | 0.0578 |
| J87 | 0.0572 |
| J88 | 0.0713 |
| β1 | 0.7835 |
| β2 | −1.1245 |
| β3 | −1.0542 |
| β0 | 0 |
Furthermore, when the βj’s of the private incentive of a group sum up to a positive value, females in that group tend to choose education to some extent at the individual level. Conversely, if the sum is negative, females in the group tend to choose no education. In this context, the positive value of β1 indicates that the place of residence (rural or urban) has a significant impact on the educational choices made by females in Ghana. The estimates β2 and β3, on the other hand, are negative. The base private incentive β0 is zero, which means that in a situation where social interaction is present, females considered in these cases will rely on their residence or marital status or wealth to make a choice towards educational attainment.
3.3. Model Diagnostics and Validation
In this section, we will examine how well our interacting model fits the data. To estimate the parameters of the interacting model discovered in [39], the PLS method was used. Based on that equation, Ua is our dependent variable and ta is our independent variable. PLS analyses (predicts) a set of dependent variables based on a set of independent variables (predictors). This approach yields orthogonal factors, known as latent vectors, from the independent variables and arranges those factors in a decreasing order of their eigenvalues. The number of latent vectors used is determined by which vectors best explain the covariance between the independent and dependent variables [40]. Matrix laboratory (MATLAB) software version R2016a is used to generate our results and conduct our analyses. Table 7 shows the amount of variance explained by the latent vectors used in the estimation in the independent variables (ta) and the dependent variable (Ua), as well as the root mean square error of prediction (RMSEP). Table 7 shows that the first seven latent vectors explain 97.76% of the variance in the Ua and 96.41% of the variance in the ta. The variances explained by the dependent variable are sufficient for modelling or prediction. The RMSEP decreases as the number of latent vectors increases.
| Latent vectors | Percentage of explained variances forta | Cumulative percentage of explained variances forta | Percentage of explained variances forUa | Cumulative percentage of explained variances forUa | RMSEP |
|---|---|---|---|---|---|
| 1 | 45.38 | 45.38 | 32.53 | 32.53 | 0.9337 |
| 2 | 15.22 | 60.6 | 22.67 | 55.20 | 0.6901 |
| 3 | 10.87 | 71.47 | 18.16 | 73.36 | 0.4529 |
| 4 | 8.01 | 79.48 | 6.65 | 80.01 | 0.2561 |
| 5 | 2.51 | 81.99 | 3.10 | 83.11 | 0.2090 |
| 6 | 7.67 | 89.66 | 5.16 | 88.27 | 0.1162 |
| 7 | 6.75 | 96.41 | 9.49 | 97.76 | 0.0158 |
- 1.
Advantages
- a.
Comprehensive analysis: The multipopulation Curie–Weiss model allows for a detailed analysis of the sociodemographic factors that might influence the educational attainment of females. This is evident as the study uncovers the effects of marital status, place of residence, and wealth status on education.
- b.
Combined estimation techniques: By using both the PLS and ordinary least squares methods for different models, the study ensures robustness and better estimation of parameters.
- a.
- 2.
Disadvantages
- a.
Limited time frame: The study only considers data from the years 2008 and 2014, which might not capture the most recent trends or potential shifts in sociodemographic influences on female education.
- b.
Potential omissions: While the study focuses on marital status, place of residence, and wealth status, there might be other significant sociodemographic factors not accounted for in this model.
- a.
4. Conclusion
The study reinforces the value of employing statistical mechanical models in the realm of sociodemographic research. Such models offer a sophisticated lens through which we can assess the interactions among individuals, thereby unravelling the collective dynamics of sociodemographic systems. What many sociologists and economists often refer to as foundational factors can be traced back to results generated by these statistical mechanical frameworks. Specifically, this research delved into a distinct variant of the probabilistic choice models—frequently recognized as the Luce or multinomial logit models in sociodemographic circles. To investigate the impact of sociodemographic elements on the educational achievements of Ghanaian females, this study adopted the multipopulation Curie–Weiss model. The data pool was sourced from the GDHS 2008 and 2014. In analysing this data, we employed both the PLS and ordinary least squares methods to estimate the parameters for the interacting and noninteracting Curie–Weiss models. Our results highlighted the significant role played by factors such as marital status, geographical location, and wealth in shaping the educational attainment of Ghanaian females. As we strive for a future where females of all ages have greater access to advanced education, it is imperative for the Ghanaian government and its allies to implement key initiatives. These should include addressing the root causes of poverty through economic empowerment programs, preventing early and teenage marriages by enforcing relevant laws and educating communities, and upgrading educational infrastructure in underprivileged areas. Enhancing basic living conditions, such as access to clean water and healthcare, is also crucial. Additionally, supporting female education with scholarships, mentorship programs, and gender-equal materials, as well as engaging local communities to shift societal attitudes towards women’s education, is a vital component of this mission. In subsequent studies, we plan to juxtapose the educational achievements of females in Ghana against those in other West African nations or on a global scale. Such comparisons can underscore particular regional issues or effective strategies from different countries that might be suitable for adaptation.
Ethics Statement
The GDHS adheres to ethical guidelines for data collection. The Ghana Statistical Service (GSS) oversees the survey and ensures compliance with ethical standards.
Consent
Informed consent is obtained from participants before interviews and measurements.
Conflicts of Interest
The authors declare no conflicts of interest.
Author Contributions
All authors contributed equally to this manuscript.
Funding
The authors received no specific funding for this work.
Acknowledgments
The authors thank the staff at the Mathematics and Statistics Department of University of Energy and Natural Resources; the Department of Statistics and Actuarial Science, Kwame Nkrumah University of Science and Technology; and the Department of Mathematics, Kwame Nkrumah University of Science and Technology, for their support and kindness during the period that this paper was written. We also acknowledge the Ghana Demographic and Health Survey Program for the dataset for this study.
Open Research
Data Availability Statement
The dataset used in this study is included in the supporting information (available here).
