Fertility refers to the biological capacity to reproduce and have children. It is a key aspect of reproductive health influenced by various factors. Fertility rates reflect the average number of children born to a woman over her lifetime, providing key insights into population dynamics.
Globally fertility rates have been declining over the past few decades, with the global average fertility rate currently estimated at around 2.5 children per woman. In developed nations, fertility has generally declined in recent decades, often falling below the replacement level of 2.1 children per woman. In Africa, fertility rates are usually higher than the global average, with some countries having rates well above 4 children per woman. The total fertility rate in sub-Saharan Africa was 4.6 children per woman.
Particularly in Ethiopia, the fertility rate is relatively high compared to the global average and even the African average. The total fertility rate in Ethiopia is estimated to be around 4.1 births per woman; Crude birth rate is 24.2 and crude death rate is 5.7 per 1000 population (CSA projection).
The key determinants of fertility rates include socioeconomic factors, cultural/religious factors, demographic factors, biological/health factors, and government policies. The complex interplay of these multifaceted factors shapes fertility patterns globally. Fertility rates in Ethiopia have persistently fallen over the previous few decades, although the underlying causes of this trend remain unknown.
This study aims to pinpoint the key determinants influencing fertility levels in the country. Data from the 2019 mini Ethiopian Demographic Health Survey were utilized, encompassing a sample of 8,885 reproductive women selected through stratified random sampling.
Read also: Chad Outfit Guide
The study was done in Ethiopia, with data from the 2019 Ethiopian Demographic and Health Survey. This dataset contains information on a variety of fertility-related variables, socioeconomic characteristics, and other relevant factors.
The study population for our research comprised all women of reproductive age (15-49 years) who participated in the 2019 EDHS, which included both urban and rural locations. A two-stage cluster sampling approach was employed.
Key Variables and Methods
Outcome variable
The outcome variable for this study was number of children ever born.
Independent variables
The independent variables are: the age of the mother, place of residence, region, Mother Education level, household members, Wealth index, religion, Contraceptive use, and current pregnancy.
We selected a negative binomial regression model after thoroughly evaluating overdispersion, goodness-of-fit, alternative specifications, excess zeros, and the dispersion parameter. This careful model selection process allowed us to identify the negative binomial approach best suited to the count-based data. We selected predictor variables from the EDHS data set based on past research and subject knowledge process.
Read also: South Africa Eclipse: Find out more
The study used both descriptive and inferential statistics. The inferential analysis was conducted using negative binomial regression, with the IRR and p-value utilized to examine associations and their statistical significance. In addition, a Decomposition analysis based on place of residence was carried out to evaluate and quantify the effects of numerous factors on overall fertility levels among Ethiopian women. Furthermore, the data was analyzed with Stata version 17.
The model equation for the negative binomial distribution that results is as follows:
$$\eqalign{& \Pr \left( {Y = yi|{\mu _i},\alpha } \right) = \cr & {{{\rm{\Gamma }}\left( {{y_i} + {{\rm{\alpha }}^{ - 1}}} \right)} \over {{\rm{\Gamma }}\left( {{y_i} + 1} \right){\rm{\Gamma }}({{\rm{\alpha }}^{ - 1}})}}{\left( {{{{\alpha ^{ - 1}}} \over {{\alpha ^{ - 1}} + {\mu _i}}}} \right)^{{\alpha ^{ - 1}}}}{\left( {{{{\mu _i}} \over {{\alpha ^{ - 1}} + {\mu _i}}}} \right)^{yi}} \cr}$$The parameter µ is the mean incidence rate of y per unit of exposure.
This study utilized publicly available survey data, with strict adherence to protocols to safeguard respondent privacy and anonymity. The research approach and potential implications were evaluated to ensure the ethical use of demographic information in support of evidence-based policies and programs benefiting all Ethiopians.
Fertility in Ethiopia (1950 - 2022)
Descriptive Analysis
Table 1 shows that in 2019, an average of 273 infants were born per 100 mothers. According to the survey, the youngest mother was 15 and the oldest was 49, with an average responder age of 27.56.
Read also: Witnessing the 2027 Solar Eclipse
| Variable | Mean | Standard Deviation | Minimum | Maximum |
|---|---|---|---|---|
| Age of Mother | 27.56 | 7.71 | 15 | 49 |
| Number of Children Born per 100 mothers | 273 | - | - | - |
Table 2 shows descriptive statistics for the categorical variables. Regarding their place of residence, 6024 respondents (67.8%) live in rural areas, while 2861 (32.2%) live in cities. Religiously, the majority of respondents (41.47%) identify as Orthodox, followed by Muslims (29.48%) and other religious groups (29.05%). In terms of mother’s educational level, the majority have no education (40.4%); household members have five or fewer family members (50.90%); and marital status, the majority are married (64.64%), with the remaining (33.56%) unmarried, divorced, or widowed. The wealth index showed 46.84% as rich, 34.35% as poor, and 18.81% as middling wealth.
| Variable | Category | Frequency | Percentage |
|---|---|---|---|
| Place of Residence | Rural | 6024 | 67.8% |
| Urban | 2861 | 32.2% | |
| Religion | Orthodox | - | 41.47% |
| Muslims | - | 29.48% | |
| Other religious groups | - | 29.05% | |
| Mother’s Education Level | No education | - | 40.4% |
| Household Members | Five or fewer | - | 50.90% |
| Marital Status | Married | - | 64.64% |
| Unmarried | - | 33.56% | |
| Divorced or Widowed | - | 33.56% | |
| Wealth Index | Rich | - | 46.84% |
| Poor | - | 34.35% | |
| Middling wealth | - | 18.81% |
In terms of obstetric considerations, Table 3 reveals that 92.25% of women were not pregnant at the time of the study, whereas 7.78% were. The majority (71.24%) did not use any form of contraception, while 28.76% did. Furthermore, the majority of respondents (42.04%) do not have any children aged five or younger. 33.84% of respondents have a single child, 19.59% have two children, and 4.53% have three or more.
| Variable | Category | Percentage |
|---|---|---|
| Current Pregnancy | Not Pregnant | 92.25% |
| Pregnant | 7.78% | |
| Contraceptive Use | No Contraception | 71.24% |
| Contraception | 28.76% | |
| Children under Five | None | 42.04% |
| One Child | 33.84% | |
| Two Children | 19.59% | |
| Three or More | 4.53% |
Checking the presence of over-dispersion
The Poisson regression model assumes that the mean and variance are equal. However, if the variance exceeds the mean, this implies over-dispersion. This was corroborated by the over-dispersion test results in Table 4, which were comparable to those in Table 1. As a result, we reject the ‘equidisperssion’ hypothesis, implying that the Poisson distribution is not an optimal fit for our data. To address this issue, we should consider using a negative binomial regression model because the variance is larger than the mean (Table 1).
| Test | Value | P-value |
|---|---|---|
| Over-dispersion test | Comparable to Table 1 | - |
The goodness of fit test
Hypothesis tests assume an adequate fit (null hypothesis), whereas the alternative hypothesis implies a lack of fit. According to the findings in Table 5, both the deviance and Pearson goodness of fit tests are non-significant, indicating that the model fits the data well.
| Test | Result |
|---|---|
| Deviance | Non-significant |
| Pearson | Non-significant |
The factors such as the age of the mother, place of residence, religion, mother’s education level, household members, wealth index, current pregnancy, use of contraceptive children under five, and marital status significantly affect number of children ever born.
Map of Ethiopia showing regional divisions.
Negative binomial regression analysis
Table 6 shows that many circumstances have a major impact on the overall number of children ever born. To begin, for every additional year of maternal age, the number of children born increases by 8%. Somali and Gambela regions have higher fertility than Tigray, but Addis Ababa has lower fertility than Tigray. Furthermore, living in a rural region compared to an urban increases the number of children born by 9%. Being Muslim or belonging to ‘other religions’, as compared with ‘orthodox’, increases the number of children born by 13% and 16%, respectively.
Furthermore, mothers with a primary education level have fewer children than those with no education, by 16%. Moreover, mothers with a secondary education level or above have even fewer children than those with no education by 39%. Having six to nine or more than nine household members is associated with higher values compared to having fewer than six household members. Being in the ‘rich’ category is associated with a lower impact on the number of children compared to being ‘poor’.
Moreover, mothers who are currently pregnant have 8% more children than those who are non-pregnant (IRR = 1.08). In addition, mothers who use contraceptives have 13% more number of children ever born than non-users (IRR = 1.13). Having children under the age of five, specifically one, two, or more than two children under five, is associated with higher impacts than having no children under five. Being single, divorced, or separated is associated with a lower impact compared to being married.
| Variable | IRR | P-value |
|---|---|---|
| Maternal Age | 1.08 | 0.00 |
| Rural Residence | 1.09 | 0.00 |
| Muslim | 1.13 | 0.00 |
| Other Religious Groups | 1.16 | 0.00 |
| 6-9 Household Members | 1.24 | 0.00 |
| >9 Household Members | 1.14 | 0.04 |
| One Child Under 5 | 1.35 | 0.00 |
| Two Children Under 5 | 1.77 | 0.00 |
| >2 Children Under 5 | 1.99 | 0.00 |
| Currently Pregnant | 1.08 | 0.00 |
| Contraceptive Use | 1.13 | 0.00 |
| Primary Education | 0.84 | 0.00 |
| Secondary Education | 0.61 | 0.00 |
| Richest Household | 0.94 | 0.00 |
| Single/Divorced/Widowed | 0.49 | 0.00 |
Decomposition analysis
Table 7 shows that examining various variables impacting the overall number of children born in rural and urban areas yields intriguing results. A comparison between maternal age and the number of children born in both rural and urban areas reveals a strong relationship with the number of children born. Maternal age has an incidence rate ratio of 1.08 in both rural and urban environments, meaning that for every one-year increase in maternal age, the risk of fertility rises by 8% in both settings. This link is stable across rural and urban locations, with comparable IRRs and confidence intervals, implying a consistent relationship between maternal age and number of children born regardless of residency.
Religion’s impact on the number of children born varies between rural and urban locations, with ‘Muslim’ and ‘other religions’ having greater IRR values in urban areas than ‘Orthodox’. The IRR for being Muslim is 1.12 in rural regions and 1.21 in urban areas, but the IRR for ‘others’ is 1.20 in rural areas and 1.33 in urban areas, demonstrating a substantial relationship with fertility. The IRRs for Muslims and other religions show that they have a higher risk of fertility than the “orthodox” in both rural and urban environments.
A comparison of education levels (Primary, Secondary, and above) and number of children born demonstrates considerable disparities between rural and urban locations. In rural areas, mothers with primary education have an IRR of 0.85, indicating a negative relationship with number of children born, whereas those with secondary or higher education have an IRR of 0.67, indicating a substantial negative association. In urban settings, moms with primary education have an IRR of 0.81, indicating a reduced likelihood of number of children born, while those with secondary or higher education have an IRR of 0.57, indicating the lowest possibility.
There is a noticeable difference in the association between the mother’s educational level and the number of children born in rural and urban settings, especially for those with Secondary and above Education where urban areas show lower IRRs. The IRRs for Primary, Secondary, and above Education levels compared to no education are lower in both rural and urban areas, indicating a protective effect against outcome.
Both rural and urban areas had positive IRRs for the number of household members, implying a moderate positive relationship with the number of children born. The influence of household members on the overall number of children born is comparable between rural and urban locations, with rural areas having somewhat higher IRR values in the ‘six up to nine’ category. The IRR for having ‘six up to nine’ household members is 1.24 in rural regions and 1.28 in urban areas, but for having more than nine household members, it is 1.17 in rural areas and 1.15 in urban areas, indicating a substantial relationship with the result.
The IRRs for the number of household members show a slight increase in the risk of fertility because the number of household members is small in both rural and urban areas. In the analysis of wealth index categories, a woman from the richest wealth index is 23% less likely to have more children than a woman from the poorest wealth index in the case of urban residence, but there is no significant relationship with rural residents.
The analysis of currently pregnant found that being currently pregnant has a high positive correlation with the number of children in comparison to ‘nonpregnant’ in rural residents, whereas being currently pregnant does not affect the number of children ever born in urban areas.
The examination of contraceptive use shows that it has a strong positive relationship with several children when compared to ‘nonusers’ in both rural and urban areas.
Trends in contraceptive use and total fertility rate in Ethiopia (1990-2019).
Popular articles:
tags: #Ethiopia
