The relationship between women’s occupation and fertility outcomes in The Gambia, 2019–2020

Abstract

Background

Women’s occupation and labour force participation are assumed to lower fertility levels as they compete with women’s household obligations. While research has demonstrated this to be true in industrialized nations, developing countries, such as The Gambia, require further investigation.

Methods

Secondary data extracted from the Gambia Demographic Health Survey (GDHS) 2019–2020 were used. A weighted sample of 11,865 women of reproductive age (15–49 years) was included in the study. At the bivariate level, one-way ANOVA and t-tests were used to examine the relationship between the dependent variable, children ever born, and the independent variable, women’s occupation, as well as each control variable. Multivariable analysis, comprising Poisson regression models, was used to assess the impact of the independent variable on the dependent variable while controlling for confounders.

Results

The analysis revealed that women’s occupational type was significantly associated with fertility in The Gambia. After adjustment for socio-demographic characteristics, women in professional, technical, managerial, and clerical occupations had approximately 12.8% fewer children than non-working women (IRR = 0.872, p < 0.01), while women in agricultural occupations had approximately 6.3% more children (IRR = 1.063, p < 0.01). Age, education, wealth status, contraceptive use, religion, and marital status were also significantly associated with children ever born.

Conclusion

The findings show that measures aimed at lowering fertility should be multifaceted, focusing on variables that were found to be significant in this study. These could be effective tools for empowering Gambian women and lowering fertility. There is also a need for further primary data-based research into the relationship.

Introduction

Women’s empowerment has long been recognized as playing an important role in reducing fertility and stimulating economic growth. Since the International Conference on Population and Development (ICPD), held in Cairo in 1994, many developing countries have implemented reproductive-health and women’s empowerment programmes, including gender-mainstreaming initiatives and efforts to increase women’s participation in non-traditional occupations such as science and technology, the military, business, and finance [1]. These interventions are often intended to improve women’s autonomy, expand economic opportunities, and promote broader socioeconomic development, and they may also shape fertility preferences and outcomes.

Despite this policy focus, the relationship between women’s employment, occupation, and fertility remains contested [1]. It has often been assumed that female employment lowers fertility levels, as it competes with their family obligations of childbearing responsibilities [2, 3]. However, evidence differs on whether there is a direct association between women’s occupation and fertility, the direction of this association, and the mechanisms through which it operates [2].

In the developed world, several studies utilizing cross-sectional data documented a very substantial inverse association between female employment and fertility [2, 4–6]. More recent studies, however, suggest that policies designed to reconcile work and family responsibilities have contributed to a reversal of these patterns in some industrialized countries in recent decades [2, 7].

In developing countries, the empirical picture is more heterogeneous. Several studies have shown that the correlation between women’s occupation and fertility is weak or absent in some contexts [6]. For example, a study in Ghana found that fertility was lower among women in professional, technical, administrative, and clerical occupations than among women in other occupations, while women in agriculture had the highest fertility [8]. A comparative study of employment and fertility across global regions similarly suggests that although women’s employment has generally increased while fertility has declined, this pattern is most evident in developed regions and less consistent in Africa and parts of Asia [6]. One explanation for this pattern in Africa is the persistence of relatively high fertility levels alongside substantial variation in women’s employment opportunities [6].

In sub-Saharan Africa (SSA), evidence suggests that the association between women’s occupation and fertility may be weak or positive in some settings [9, 10]. Several structural and cultural factors common in many developing contexts may explain this pattern. This includes the availability of inexpensive domestic labour, extended-family systems that provide childcare, limited opportunities for women’s paid employment, and customs that restrict women’s mobility and participation in the labour market [10]. Furthermore, the direction and strength of the association often depend on country-specific gender norms and role expectations [3]. Using repeated, nationally representative, and comparable Demographic and Health Surveys (DHS) from 21 SSA countries, Finlay [11] found that women’s employment was associated with lower fertility in some countries, whereas in others, no clear relationship emerged, reinforcing the conclusion that context matters.

In The Gambia, empirical evidence on the association between women’s occupation and fertility remains scarce. This gap is notable given the country’s current demographic and policy context. The Gambia has a youthful age structure, characteristic of an early stage of the demographic transition, and harnessing the demographic dividend is a key objective of the National Population Policy (2023) and the National Development Plan (2023 to 2027). At the same time, increasing women’s participation in the labour market, particularly beyond subsistence agriculture into formal and semi-formal sectors, has been identified as a critical pathway to women’s empowerment, poverty reduction, and broader socioeconomic development.

Without a nuanced understanding of how different types of occupations are associated with fertility behaviour, policies aimed at promoting women’s employment may have uneven or unintended demographic implications. Examining this association in the Gambian context is therefore important for designing policies that support women’s empowerment, inform fertility-related interventions, and contribute to national development priorities. Based on the foregoing discussion, this study addresses the following research questions:

1. Is there an association between women’s occupation and fertility in The Gambia?

2. What other socio-demographic and socio-economic factors are associated with the number of children ever born among women in The Gambia?

Theoretical framework and hypotheses

To understand the work–fertility relationship, the maternal role incompatibility theory, and the societal response hypothesis, theoretical frameworks have been widely used in the socio-demographic literature [12]. The societal response hypothesis has mainly been applied in developed-country contexts to explain the shift from a negative to a positive relationship between fertility and female employment after the 1980s. This shift has been attributed to the expansion of family-friendly labour market policies, institutional childcare, and gender-equitable work-family arrangements [7, 13].

In contrast, this study is grounded in Maternal Role Incompatibility Theory, which has been widely applied in developing countries and SSA contexts using cross-sectional datasets [11, 14, 15]. This theoretical framework posits that fertility outcomes are influenced by the degree to which women’s employment responsibilities are compatible with childbearing and childcare. Fertility tends to be lower when work is time-intensive, inflexible, and spatially separated from the household, and higher when productivity and reproductive rules can be combined [3, 13].

In the Gambian context, women’s employment is highly segmented, with a substantial portion engaged in agriculture, formal trade, and manual labour, alongside a small share employed in formal and semi-formal occupations. At the same time, institutional childcare options are limited, and caring responsibilities remain predominantly borne by women. These structural conditions shape the degree to which occupational roles are compatible with childbearing responsibilities.

Guided by this framework, the study conceptualized women’s occupation as a key determinant of fertility operating through the degree of compatibility between work and childbearing. In addition to occupation characteristics, several sociodemographic and socioeconomic factors are included as control variables because they influence both women’s labour market participation and fertility behaviour. In this study, occupational categories are interpreted in terms of their relative compatibility with reproductive responsibilities. For example, agriculture and manual labour are considered highly compatible occupations, where childcare and productive activities may be combined. However, professional and clerical occupations are conceptualized as low compatibility occupations given their time rigidity and spatial separation from the household. Consistent with Maternal Role Incompatibility Theory and the empirical literature reviewed above, the study advances the following numbered hypotheses:

1. Women engaged in traditional occupations, particularly agriculture, will have higher fertility than women engaged in professional, technical, managerial, and clerical occupations.

2. Women in traditional occupations, particularly agriculture, will have more children than women with no occupation/women not working.

3. Women who use modern contraceptives will have fewer children than women who do not use contraceptives.

Methods and materials

Source of data and sampling design

This study employed secondary data, extracted from the Gambia Demographic Health Survey (GDHS) 2019-20. The GDHS is a nationally representative survey conducted by the Gambia Bureau of Statistics (GBoS) in collaboration with its partners. The 2019–2020 GDHS is the second in the Gambia DHS series, the first being conducted in 2013. The 2019–2020 GDHS employed five sets of questionnaires: The Household Questionnaire, the Women’s Questionnaire, the Men’s Questionnaire, the Biomarker Questionnaire, and the Fieldworker Questionnaire [16]. However, only the women’s questionnaire was used for this study as it contains data on the variables of interest. The sampling design employed a two-stage multistage cluster sampling technique. The first stage involved selecting enumeration areas (EAs) based on a probability proportionate to their size, yielding 281 EAs in total. The second stage involved systematically sampling households, then from the lists of households, a fixed number of 25 households per cluster were systematically selected, resulting in a total sample size of 7,025 households.

From the survey, a total of 11,865 (weighted) women of reproductive age (15–49 years) were interviewed. This forms the sample size for this study. This sample includes estimates for The Gambia as a whole, for both urban and rural areas, as well as each of the eight local government areas (LGAs). This represents a response rate of 95% of women. The 2019–2020 GDHS protocol was submitted for review with the ICF Institutional Review Board. Other ethical considerations include obtaining informed consent from all participants to ensure their voluntary participation. The data were also de-identified upon receipt. The sampling weights provided in the GDHS dataset were applied across all analyses to adjust for unequal selection probabilities and ensure nationally representative estimates.

Variables and their measurements

Independent variable

The main independent variable is women’s occupation, defined as the main type of work women were engaged in before the survey. In the GDHS, occupation was measured among women who were employed or had worked in the 12 months preceding the survey and was categorized as “not working”, “professional/technical/managerial”, “clerical”, “sales and services”, “skilled manual”, “unskilled manual”, “agriculture”, and “other”. For this study, the original occupational categories were regrouped into six categories to simplify analysis and interpretation: (1) managerial/clerical/professional/technical, (2) sales and services, (3) manual labour, (4) agriculture, (5) not working, and (6) other.

Skilled and unskilled manual occupations were combined into the manual-labour category, while the professional, technical, and managerial category was combined with the clerical occupation since they broadly represent formal or skilled employment. The “not working” category refers to women who were not employed during the relevant survey reference period, while “other” refers to occupational responses classified in the DHS as “other” because they did not fall into the specified occupational groups. This regrouping was also informed by previous studies conducted in sub-Saharan Africa that used broad occupational categories to examine the relationship between women’s work and fertility outcomes [6, 8–10].

Dependent variable

The dependent variable is fertility, measured as the total number of children ever born (CEB) to a woman. In the survey, participants were asked the total number of children they have ever given birth to who were born alive. Thus, CEB represents the cumulative number of live births reported by each woman up to the time of the survey. The variable is treated as discrete count data ranging from 0 to 14 in the dataset. Because the sample includes women aged 15–49 years, the measure captures both completed and ongoing fertility. Age was therefore included as an important control variable to account for the mechanical increase in cumulative births with increasing exposure to reproductive years.

Control variables

Nine covariates were included in the study. Some variables were regrouped from their original categories in the dataset to facilitate easier and more understandable analysis and interpretation. These control variables were included in the analysis based on their hypothesized relationship with the number of CEB and their importance in previous fertility research [13, 17, 18]. The variables are listed below and are measured as follows.

• ➢ Age: The age groups were maintained as the seven five-year age groups, namely, “15–19”, “20–24”, “25–29”, “30–34”, “35–39”, “40–44”, “45–49”.

• ➢ Education: The highest educational level attained for women was recoded into four groups: “No education”, “Primary”, “Secondary”, and “More than Secondary”.

• ➢ Religion: The religious affiliations of women were recoded into two categories, namely, “Islam” and “Christianity/Others”. The category “others” included those with no religion and other religions.

• ➢ Place of residence: The two categories measuring the respondent’s place of residence were maintained as “Rural” and “Urban”.

• ➢ Marital status: The previously five categories of women’s marital status were recoded into three: “Never married”, “Currently in union” (included women who were married and living together), and “Formerly in union” (included women who were divorced, separated, and widowed).

• ➢ Ethnicity: This variable included six ethnic groups, namely, “Mandinka/Jahanka”, “Wollof”, “Jola/Karoninka”, “Fula/Tukulur/Lorobo”, “Sarahule”, and “Others”. The category “Others” includes non-Gambian women.

• ➢ Wealth index quintiles: The household wealth status of women was maintained as five categories, “Poorest”, “Poorer”, “Middle”, “Richer”, and “Richest”. The DHS wealth index is constructed using principal component analysis based on household ownership of selected consumer goods (e.g., radio, television, car), utilities (e.g., electricity, water source), and housing characteristics (e.g., floor, wall, and roof material), to mention a few. The first component is extracted and becomes the index, which is then ranked and divided into five quintiles.

• ➢ Contraceptive use: The contraceptive use variable was recoded into three main groupings, namely, “no method”, “folkloric/traditional methods”, and “modern methods”.

• ➢ Employment Status: Women’s employment status was maintained as the two categories, “unemployed” and “employed”. Employment status captures labour-force participation, while occupation describes the type of work among those employed. The two variables, therefore, reflect distinct dimensions of women’s economic activity and are retained as separate covariates.

Methods of analysis

The study employed three major stages of analysis. All analyses were performed using IBM SPSS v25 and accounted for the complex survey design of the 2019–2020 Gambia Demographic and Health Survey (GDHS) by applying the provided sampling weights. However, due to software limitations, cluster and specification were not fully accounted for, which may result in slightly underestimated standard errors. In the first stage, descriptive univariate analyses (using frequencies, percentages, and means) were used to estimate fertility and statistically describe variables used in the study.

The second stage involved bivariate analyses to test the association between the independent and control variables and the dependent variable. Since the dependent variable, CEB, is a count variable, one-way analysis of variance (ANOVA) and t-tests were used to examine differences in the mean number of CEB across occupational categories and other covariates. Before conducting the one-way ANOVA, key assumptions were assessed. Homogeneity of variances across occupational groups was evaluated using Levene’s test, which indicated that this assumption was satisfied. The distribution of CEB within occupational groups was also inspected and, given the large sample size, the ANOVA procedure was considered robust to minor deviations from normality.

Finally, the Poisson regression was used to assess how independent variables influence the dependent variable while accounting for other confounders. Poisson regression was chosen because CEB is a non-negative integer count with a highly right-skewed distribution [7, 13, 19]. Additionally, before model estimation, overdispersion was assessed by comparing the mean and variance of the dependent variable. The results indicated that the dispersion was within an acceptable range, supporting the use of the Poisson model rather than a negative binomial model.

The Poisson model was specified as

Where: μ_i=E(CEB_i∣X_i) represents the expected number of children ever born for woman i. CEB_i denotes the number of children ever born, α is the intercept, β represents the coefficients of the explanatory variables, and X_i represents the independent variables included in the model. The exponentiated coefficients were interpreted as incidence rate ratios (IRRs).

Results

Women’s background characteristics

Table 1 presents the percentage distributions of women according to selected socio-demographic and socio-economic characteristics. For the independent variable, type of occupation, the results show that 40.1% of respondents were not working at the time of the survey. Among those who were working, the largest proportion were engaged in sales and services, accounting for 33.0% of respondents. In contrast, only 5.7% were employed in professional, technical, managerial, or clerical occupations, indicating that formal and skilled employment accounted for a relatively small share of women’s occupations in the sample.

Table 1.

Percentage distribution of respondents by socio-demographic and socio-economic characteristics

Variable	Number	Percentage (%)
Type of occupation
Not working	4,757	40.1
Professional, technical, managerial, clerical	676	5.7
Sales/services	3,916	33.0
Agriculture	1,958	16.5
Manual labour	510	4.3
Other	48	0.4
Place of residence
Urban	8,745	73.7
Rural	3,120	26.3
Age
15–19	2,633	22.2
20–24	2,181	18.4
25–29	2,248	18.9
30–34	1,619	13.6
35–39	1,438	12.1
40–44	1,028	8.7
45–49	718	6.0
Educational level
No education	4,119	34.7
Primary	1,854	15.6
Secondary	5,021	42.3
Higher	871	7.3
Religion
Islam	1,1443	96.4
Christianity/Others	422	3.6
Ethnic group
Mandinka/Jahanka	3,961	33.4
Wollof	1,487	12.5
Jola/Karoninka	1,311	11.1
Fula/Tukulur/Lorobo	2,156	18.2
Sarahule	868	7.3
Other	2,081	17.5
Marital status
Never in union	3,704	31.2
Currently in union	7,526	63.4
Formerly in union	635	5.4
Wealth quintile
Poorest	1,998	16.8
Poorer	2,135	18.0
Middle	2,292	19.3
Richer	2,591	21.8
Richest	2,849	24.0
Employment status
Unemployed	5,876	49.5
Employed	5,989	50.5
Contraceptive use
No method	10,273	86.6
Folkloric and Traditional method	141	1.2
Modern	1,451	12.2

Source: Authors’ analysis of the 2019–2020 Gambia Demographic and Health Survey

Total weighted sample = 11,865

Approximately three-quarters of the women interviewed (73.7%) were urban residents, while 26.3% resided in rural areas. This indicates that the study sample was predominantly urban. With respect to age, the proportion of respondents generally declined with increasing age. The highest proportion (22.2%) were aged 15 to 19 years, while the lowest proportion (6.0%) were aged 45 to 49 years. This pattern reflects the relatively youthful age structure of the female population in the sample.

In terms of educational attainment, 42.3% of respondents had secondary education, representing the largest educational category. Women with no formal education accounted for 34.7%, while 15.6% had primary education. Only 7.3% had attained higher education, indicating that post-secondary education remained relatively uncommon among women in the sample.

Regarding religion, the majority of respondents were Muslim, accounting for 96.4% of the sample. Women who identified as Christian or belonged to other religious groups accounted for 3.6%. The ethnic composition of the sample shows that Mandinka/Jahanka women constituted the largest group, representing 33.4% of respondents. This was followed by Fula/Tukulur/Lorobo women at 18.2%, women classified under “Other” ethnic groups at 17.5%, Wollof women at 12.5%, Jola/Karoninka women at 11.1%, and Sarahule women at 7.3%.

For marital status, the majority of respondents were currently in union, comprising 63.4% of the sample. Women who had never been in union accounted for 31.2%, while 5.4% were formerly in union, including those who were divorced, separated, or widowed.

With respect to household wealth status, the largest proportion of respondents belonged to the richest wealth quintile, 24.0%, followed by the richer quintile, 21.8%. The middle, poorer, and poorest quintiles accounted for 19.3%, 18.0%, and 16.8%, respectively. Employment status was almost evenly distributed. Slightly more than half of the women, 50.5%, were employed, while 49.5% were unemployed.

Finally, contraceptive use was low among respondents. The majority of women (86.6%) reported using no contraceptive method. Modern contraceptive use was reported by 12.2%, while only 1.2% reported using folkloric or traditional methods.

Children ever born

Table 2 presents the mean number of children ever born among women in the study sample. The results show that the average number of children ever born was 2.45, with a standard deviation of 2.67. This indicates variation in fertility experience among women of reproductive age in the sample.

Table 2.

Mean number of children ever born among respondents

Variable	Number	Mean ± SD
Children ever born	11,865	2.45 ± 2.67

Bivariate associations between the type of occupation and fertility

Table 3 below shows the bivariate association between independent and dependent variables. The results of the one-way ANOVA indicate statistically significant differences in the mean number of children ever born across occupational groups (F = 331.390, p < 0.001). Women in technical/clerical/managerial/professional occupations had fewer children than those who work in agriculture. Women with no occupation, as well as women in the “Other” occupational category, had the lowest fertility. As expected, women working in the agricultural sector had the highest fertility rates. Post-hoc comparisons using the Tukey-HSD post-hoc tests revealed significant differences between several occupational categories, particularly between women engaged in agricultural and manual labour occupations and those in professional and clerical occupations. These findings suggest that fertility levels vary systematically by occupational category.

Table 3.

Bivariate associations between fertility (CEB) and types of occupation

Variable	Mean	SD	F-Score	Sign
Type of Occupation			331.390	< 0.001
Not working	1.45	2.11
Professional, Technical, Managerial, Clerical	1.58	1.78
Sales/Services	3.16	2.70
Agriculture	3.67	3.06
Manual Labour	2.81	2.68
Other	1.07	1.49

Bivariate relationships between control variables and fertility

Table 4 below shows the bivariate associations between the control variables and the dependent variable. These bivariate associations are descriptive and do not account for potential confounders. They should also not be interpreted as causal effects. All the variables included in the analysis were significantly associated with fertility. Age had a strong positive association with the number of children ever born to women. The result indicates that as a woman’s age increases, so does the number of children she has ever had, as shown in Table 4 below. Women aged 15–19 had a mean fertility of 0.12, while those aged 45–49 had a mean fertility of 6.05. Bivariate findings also indicate that women who have never been in a union had the lowest mean fertility of 0.14, while women who are currently in a union (married/living together) at the time of the survey had the highest mean fertility (3.54). For the education level, the result suggests that the higher the educational attainment, the lower the mean fertility. Women with a higher level of education had a mean fertility of 1.09, while those without formal education had a mean of 4.04. For the wealth quintile, the results indicate that the mean number of children ever born decreased with increasing wealth status from the poorest to the richest. With regards to employment status, employed women have a slightly higher fertility compared to their unemployed counterparts.

Table 4.

Mean number of children ever born by background characteristics of women

Variable	Mean	SD	F-Score	Sign
Current Age			2771.999	< 0.001
15–19	0.12	0.37
20–24	0.77	1.02
25–29	2.08	1.64
30–34	3.35	1.95
35–39	4.70	2.33
40–44	5.67	2.67
45–49	6.05	2.69
Educational Level			1121.900	< 0.001
No education	4.04	2.86
Primary	2.61	2.55
Secondary	1.31	1.89
Higher	1.09	1.50
Religion			25.556	< 0.001
Islam	2.47	2.68
Christianity/Others	1.80	2.04
Employment status			306.379	< 0.001
Unemployed	1.72	2.326
Employed	3.16	2.784
Ethnicity			7.258	< 0.001
Mandinka/Jahanka	2.40	2.72
Wollof	2.55	2.73
Jola/Karoninka	2.13	2.58
Fula/Tukulur/Lorobo	2.66	2.79
Sarahule	2.47	2.55
Other	2.43	2.48
Marital status			3104.622	< 0.001
Never in union	0.14	0.54
Currently in union	3.54	2.60
Formerly in union	2.92	2.33
Residence			270.191	< 0.001
Urban	2.18	2.49
Rural	3.19	3.00
Wealth quintile			169.172	< 0.001
Poorest	3.42	3.10
Poorer	2.87	2.81
Middle	2.59	2.63
Richer	2.11	2.47
Richest	1.63	2.07
Contraceptive use			465.679	< 0.001
No method	2.16	2.58
Folkloric and Traditional method	3.62	2.48
Modern	4.33	2.49

Considering religion, the result revealed that Muslim women had a higher number of children ever born (2.47) compared to their Christian counterparts and women of other religious affiliations (1.80). In terms of ethnicity, there was very little difference among the ethnic groups. However, women of the Fula/Tukulur/Lorobo ethnic group had the highest fertility rate, 2.66, while the Jola/Karoninka women recorded the lowest mean (2.13).

Considering the place of residence, results indicate that rural women had higher fertility than their counterparts in urban areas, 3.19 and 2.18, respectively. Finally, in terms of contraceptive use, the result produces an intriguing finding. Fertility rates were found to be higher (4.33) among women currently using modern contraceptives compared to women who never used any form of contraceptives (2.16) and those who used folkloric and/or traditional methods (3.62). The higher mean CEB among women who have used modern contraception likely reflects the fact that women often adopt contraception after achieving their desired family size; this pattern is further examined in the multivariable analysis.

Multivariable analysis

Table 5 presents the results of the Poisson regression models that examine the association between a woman’s occupation and her number of children ever born. In Model 1, which includes occupation only, the results indicate that occupational type is significantly associated with fertility. Using women who are “not working” as the reference category, most occupations showed a positive association with fertility. Women engaged in sales and services, agriculture, and manual labour have incidence rate ratios (IRRs) of 2.175, 2.470, and 1.930, respectively, indicating substantially higher expected numbers of children compared with women who are not working. In contrast, women in the “other” occupational category have a lower expected number of children (IRR = 0.723). Although all occupational groups generally exhibit higher fertility than non-working women, those in professional, technical, managerial, and clerical occupations have comparatively lower fertility (IRR = 1.107) than women working in agriculture or manual labour.

Table 5.

Results of the poisson regression: dependent variable – CEB

Variable (RC)	Model I Exp(β)	Model II Exp(β)
Type of Occupation
Not working (RC)	1.00	1.00
Professional, Technical, Managerial, Clerical	1.107**	0.872**
Sales/Services	2.175***	0.992
Agriculture	2.470***	1.063**
Manual	1.930***	1.043
Others	0.723*	0.647**
Current Age
15–19 (RC)		1.00
20–24		3.451***
25–29		6.668***
30–34		9.606***
35–39		12.318***
40–44		14.549***
45–49		15.707***
Residence
Urban (RC)		1.00
Rural		1.011
Wealth quintile
Poorest (RC)		1.00
Poorer		0.952**
Middle		0.903***
Richer		0.827***
Richest		0.740***
Educational Level
No education (RC)		1.00
Primary		0.952**
Secondary		0.777***
Higher		0.596***
Religion
Islam (RC)
Christianity/Others		0.906**
Ethnic Group
Mandinka/Jahanka (RC)
Wollof		0.999
Jola/Karoninka		1.021
Fula/Tukulur/Lorobo		1.028
Sarahule		0.980
Others		1.000
Marital status
Never in union (RC)		1.00
Currently in union		6.001***
Formally in union		4.683***
Employment status
Unemployed (RC)		1.00
Employed		1.038
Contraceptive use
No method (RC)		1.00
Folkloric and traditional method		1.179***
Modern		1.268***

Source: Computed from GDHS, 2019/2020

RC = Reference Category

^*p < 0.05; ^**p < 0.01; and ^***p < 0.001

After adjusting for socio-demographic characteristics in Model II, the association between occupation and fertility remains statistically significant, although the magnitude of the effects changes. Relative to women who were not working (reference category), women in professional, technical, managerial, and clerical occupations (IRR = 0.872), sales and services (IRR = 0.992), and other occupations (IRR = 0.647) show lower expected numbers of children. For example, women in professional, technical, managerial, and clerical occupations have approximately 12.8% fewer children than women who were not working, holding other factors constant. In contrast, women engaged in agriculture (IRR = 1.063) and manual labour (IRR = 1.043) maintain slightly higher expected numbers of children relative to the reference group, with agricultural workers exhibiting the highest fertility among the occupational groups in the adjusted model.

Several socio-demographic variables were also significantly associated with fertility. As expected, age shows a strong positive relationship with the number of children ever born. Using women aged 15–19 years (reference category), women in older age groups have progressively higher expected numbers of children, with IRRs ranging from 3.451 for women aged 20–24 to 15.707 for those aged 45–49. These results reflect the cumulative nature of fertility, whereby older women have had longer exposure to childbearing.

For wealth status, as anticipated, the result indicates a negative effect on fertility. Compared with women in the poorest wealth quintile (reference category), women in higher wealth categories have progressively lower expected numbers of children. The incidence rate ratios decrease to 0.952, 0.903, 0.827, and 0.740 for women in the poorer, middle, richer, and richest quintiles, respectively. For example, women in the richest households have approximately 26% fewer children than women in the poorest households. Similarly, education showed a strong negative effect. Compared with women with no formal education (reference category), women with primary, secondary, and higher education have expected numbers of children that are 0.952, 0.777, and 0.596 times those of women without education. In other words, women with higher education have approximately 40.4% (1-0.596 = 0.404) fewer children than those with no formal education.

Religion and marital status are also significant predictors of the number of children ever born. Christian women and women of other religious affiliations have slightly lower expected numbers of children (IRR = 0.906), representing about a 9.4% lower fertility rate. Using women who were never in union (reference category), women who are currently in union (married/living together) and those who are formerly in union (divorced, widowed, and separated/no longer living together) have an IRR of 6.001, while women who were formerly in union (divorced, widowed, or separated) have an IRR of 4.683, indicating substantially higher numbers of children compared with women who have never been married.

Finally, contraceptive use shows an unexpected association with fertility. Compared with women who have never used any contraceptive method (reference category), women who report using folkloric or traditional methods and those using modern methods have IRRs of 1.47 and 1.26, respectively, indicating higher expected numbers of children. However, because the data are cross-sectional, the temporal ordering between fertility and contraceptive use cannot be determined. The higher fertility observed among women who have used contraception likely reflects contraceptive adoption after achieving the desired family size, rather than contraceptive use leading to higher fertility. This counterintuitive pattern underscores the importance of timing and selection into contraceptive use and has implications for interpreting contraceptive prevalence indicators in higher fertility settings.

Discussion

Using the 2019/2020 Gambia DHS, we examined the association between women’s occupational categories and the number of children ever born, while simultaneously accounting for a range of socio-demographic determinants through Poisson regression models, appropriate for count outcomes. After adjustment for socio-demographic and socio-economic characteristics, women employed in agriculture had a higher expected number of children than women who were not working, while women in professional, technical, managerial, or clerical occupations had a lower expected number of children than the reference group. This pattern supports our hypothesis that women in agriculture will exhibit higher fertility and women in the formal sector will exhibit lower fertility. The findings corroborate those from similar studies [8, 20] where women in agriculture were found to have higher fertility compared to women in any other type of occupation. It is also consistent with the results of a review of the literature, which indicates that modern sector workers have fewer children than their counterparts who work in the informal sector [6, 21, 22]. This pattern could be attributed to the structural nature of agricultural work in The Gambia, which remains largely unmechanised and labour-intensive. Thus, larger families provide additional farm hands, reducing the opportunity cost of childbearing. In contrast, formal-sector occupations impose stricter schedules and higher opportunity costs, creating a role-incompatibility between intensive career demands and child-rearing, which translates into lower fertility.

Age emerged as a strong positive predictor of CEB, with women aged 45–49 reporting the highest number of children ever born, reflecting the cumulative nature of fertility across the reproductive lifespan. These findings are consistent with findings from other studies [6, 14, 23, 24] where fertility was lowest for women in the age group 15–19 years compared to other age groups.

Educational attainment showed an inverse association with fertility, which corroborates the results from other related studies [21, 25, 26]. This could be explained by research that suggests educated women have greater access to information about contraception and more opportunities in the formal labour market, fostering a trade-off between schooling and childbearing. Similarly, the wealth index also showed a negative relationship with CEB. This is consistent with studies such as [23, 25, 27–29] and the intergenerational-wealth-flows theory by Caldwell (1982), which predicts lower fertility when families can rely on non-child resources for old-age security.

Religion also has a strong relationship with CEB. Muslim women, who constitute the vast majority (= 95%) of the sample, had higher fertility, a pattern that may be driven by the prevalence of polygamy and cultural norms that encourage larger families. This corroborates findings from [30, 31], and a Mozambique study [32] where religion significantly influences fertility [33]. Marital status remained a strong predictor, with married women bearing more children, reflecting the normative expectation that births occur within marriage. These findings are in line with those [5, 29, 34–37]. This could be attributed to religious and other socio-cultural factors. The Gambian society strongly prohibits having children out of wedlock, which cuts across cultures and religions, although the trend is gradually changing.

For contraceptive use, women who used modern contraceptives were found to have higher fertility than those who never use any method, and those who only use folkloric or traditional methods. This finding did not support the hypothesis and was contrary to the findings in a cross-sectional study by [20] and a Senegalese study [34] where women who use modern contraception were found to have the lowest risk of becoming pregnant. However, the findings are consistent with other studies [23, 26, 32]. This finding can be explained in relation to the period during which the women used contraceptives. This study uses cross-sectional data, and it is plausible that the women had their ideal number of children before their current use of contraception. There is also a strong relationship between women with several children and contraceptive use; they are more likely to use it to limit childbearing compared to those with none or few children [30]. The findings further revealed that approximately 87% of women used no method of contraception. This finding is consistent with other Gambian studies [31, 33, 38] that also discovered very low contraceptive usage among women of reproductive age in the Gambia. This could be attributed to the issue of access and misconceptions [39], in addition to socio-cultural as well as religious factors [27].

The results suggest that policies aiming to accelerate The Gambia’s demographic transition should address both structural and behavioural dimensions. Expanding women’s participation in formal, non-traditional occupations can lower fertility by increasing opportunity costs and promoting gender-equitable labour markets. Simultaneously, strengthening family-planning services, particularly outreach that targets women with high parity and addresses cultural misconceptions, can help align contraceptive use with reproductive intentions. Such integrated strategies are essential for achieving Sustainable Development Goals 3 (health and well-being) and 5 (gender equality) in The Gambia.

Finally, because the analysis relies on secondary DHS data, further qualitative and longitudinal quantitative research is needed to elucidate the causal mechanisms linking occupation, cultural norms, and fertility decisions; primary data collection would allow for more precise measurement of occupational characteristics, work-family conflict and the timing of contraceptive adoption, thereby informing more nuanced and effective policy design.

Limitations

This study has some limitations. First, the analysis was based on secondary cross-sectional data from the 2019 to 2020 Gambia Demographic and Health Survey. Although the DHS provides nationally representative data and remains a valuable source for examining population and reproductive health outcomes, the dataset was not originally designed specifically to examine the occupation-fertility relationship. As a result, some variables relevant to this study were unavailable.

The cross-sectional nature of the data also limits the ability to establish temporal ordering between women’s occupation and fertility. The dataset links children ever born to women’s occupation at the time of the survey; however, some women may have had their children before entering their current occupation, since occupational status can change over the life course. Therefore, this study identifies associations between women’s occupation and children ever born but cannot establish causal relationships.

Finally, the GDHS dataset did not include detailed measures of occupational conditions, work intensity, childcare availability, timing of employment, duration in occupation, or work-family conflict. Future longitudinal and qualitative studies are therefore needed to better examine the causal mechanisms linking women’s occupation, household responsibilities, cultural norms, and fertility decisions in The Gambia.

Conclusion

This study found significant associations between women’s occupational categories and fertility outcomes in The Gambia after adjusting for selected socio-demographic and socio-economic characteristics. Women in agricultural occupations had a higher expected number of children than non-working women, while women in professional, technical, managerial, or clerical occupations had a lower expected number of children than the reference group. However, because the study used cross-sectional data, these findings should be interpreted as associations rather than causal effects. The observed pattern is consistent with the maternal role incompatibility framework. Agricultural and other labour-intensive occupations may be more compatible with larger family sizes, particularly where children can contribute to household or farm labour. In contrast, formal-sector occupations may involve stricter schedules, spatial separation from the household, and higher opportunity costs for childbearing and childcare.

Beyond types of occupations, the multivariable models showed that age, education, wealth, contraceptive use, marital status, and religion were significantly associated with fertility. Fertility rises sharply with age, declines with higher educational attainment and wealth, is lower among women who are unmarried or practice Christianity, and is paradoxically higher among women who use modern contraceptives, a cross-sectional artifact reflecting that women tend to adopt contraception after reaching their desired family size.

These findings suggest that policies aimed at supporting The Gambia’s demographic transition should address both structural and behavioural factors. Structural interventions may include expanding women’s access to formal and non-traditional occupations, including professional, technical, managerial, and clerical employment, while also improving productivity and mechanisation in agriculture. Behavioural interventions should focus on strengthening family-planning services, addressing misconceptions about modern contraception, and providing tailored counselling for high-parity women.

Such an integrated approach is consistent with Sustainable Development Goal 5 on gender equality and Sustainable Development Goal 3 on health and well-being. It also aligns with the objectives of The Gambia’s National Development Plan 2023 to 2027, particularly those relating to gender equality and improved reproductive-health outcomes. Further qualitative and longitudinal research is needed to better understand the causal pathways linking occupation, cultural norms, household responsibilities, and fertility decisions in The Gambia.

Abbreviations

ANOVA

Analysis of Variance

ASRM

American Society for Reproductive Medicine

CEB

Children Ever Born

COVID-19

Corona Virus Disease 2019

FLFP

Female labour force participation (FLFP)

GDHS

Gambia Demographic and Health Survey

GBoS

Gambia Bureau of Statistic

ICPD

International Conference on Population and Development

SSA

Sub-Saharan Africa

DHS

Demographic and Health Surveys

LA

Latin America

ILO

International Labour Organization

ILOSTAT

International labour Statistics

LGAs

Local Government Areas

OECD

Organization for Economic Cooperation and Development

OLS

Ordinary Least Squares

SDG

Sustainable Development Goal

TFR

Total Fertility Rate

WFS

World Fertility Survey

UNFPA

United Nations Population Fund

UN

United Nations

Authors’ contributions

BJ, AB and JWJ made substantial contributions to the conception, design, literature search, drafting of the manuscript, and the critical review of subsequent drafts of the manuscript. BJ and JWJ acquired, analyzed and interpreted the data. All authors read and approved the final draft of the manuscript.

Funding

No funding was received for this study.

Data availability

The study used secondary, de-identified data from the DHS Program. The dataset is available upon request and registration through the DHS Program website (www.dhsprogram.com). No personal identifiers were attached to the dataset provided to researchers; all data were de-identified upon receipt by the research team.

Declarations

Ethics approval and consent to participate

This study used de-identified secondary data; however, it was conducted in accordance with the Helsinki Declaration to preserve participant confidentiality, safety, and rights, as well as the protection of their data. The survey protocol was reviewed and approved by the ICF Institutional Review Board and The Gambia Government/Medical Research Council Joint Ethics Committee. Informed consent was obtained during the original DHS data collection process. Since the present study involved analysis of anonymized secondary data, no additional participant consent was required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.