Homogamy and heterogamy in sub-Saharan Africa: Implications for couples’ fertility behaviour

ABSTRACT

This study examines patterns of homogamy and heterogamy and their implications for couple’s fertility behaviour. Data were obtained from demographic and health surveys conducted in Ghana, Nigeria and Zambia between 2001 and 2014. The study population were couples in a monogamous union where wives were aged 35 years and over: 1785 in Ghana, 3185 in Nigeria and 3175 in Zambia. Results of the descriptive analysis show that educational, religious and ethnic homogamy and occupational heterogamy were predominant among the couples and mean children ever born varied according to couples’ characteristics. The significant predictors of fertility as revealed in the multivariate analysis included educational homogamy and heterogamy, occupational homogamy, age difference and religious and ethnic homogamy. To expedite fertility transition in the three countries and other sub-Saharan African countries, programmes and policies aimed at reducing fertility in the region should incorporate strategies that focus on the characteristics of couples in union.

KEYWORDS:

• Homogamy
• heterogamy
• couples sub-Saharan Africa
• fertility

1. Introduction

Although fertility in sub-Saharan Africa has been declining, the region still has the highest total fertility rate (TFR) (5.1) in the world (Bongaarts 2017; Mbacké 2017). Fertility is of concern in sub-Saharan Africa given its implications for population development. The persistent high fertility levels are driving population growth, and the imbalance between rates of population growth and development presents challenges to the achievement of development goals in the region (Shapiro & Gebreselassie 2013; Bongaarts 2017). Also, the occurrence of stalled fertility transition in some countries within this region implies that the achievement of sustainable replacement fertility levels will be delayed further (Bongaarts & Casterline 2013; Shapiro & Gebreselassie 2013).

Over the years, the social, cultural and economic contexts and the individual-level factors associated with fertility behaviour and transitions in many sub-Saharan African countries have been documented (Bongaarts et al. 1984; Isiugo-Abanihe 1998; Shapiro & Tambashe 2002; Garenne 2008; Bongaarts & Casterline 2013). However, the relationship of couples’ characteristics to their fertility behaviour has not received much attention even though marriage is an important determinant of fertility in the region. Most of the existing studies have been based on analysis of samples of individual women, and of women’s characteristics, thereby neglecting the important role that characteristics of a couple play in influencing fertility behaviour.

Relating couples’ characteristics to their fertility behaviour is of importance because marital fertility contributes substantially to the high TFR in sub-Saharan Africa (Garenne 2008). Marriage is nearly universal, childbearing is closely associated with marriage, and regulation of birth is minimal (Bongaarts et al. 1984; Bledsoe 1990). Also, in patriarchal African societies, husbands tend to make decisions on behalf of their wives and influence their wives’ fertility decisions and behaviour (Ezeh 1993).

Furthermore, evidence from existing studies show a strong association between couples’ characteristics and various reproductive outcomes. For instance, studies have shown that fertility outcomes for homogamous couples, meaning those couples who have similar achieved characteristics such as the same level of education or have similar ascribed characteristics such as age, ethnicity and religion, tend to differ from heterogamous couples, whose characteristics are different (Casterline et al. 1986; Gyimah et al. 2008; Huber & Fieder 2011; Tsou et al. 2011).

Tsou et al.’s (2011) study in Taiwan has shown that women who are married to husbands with equal education or higher education have more children than women married to a husband with lower education. A study by Huber and Fieder (2011) in America has shown that educational homogamy among couples is associated with lower odds of reproductive failure; for instance, women who were married to husbands of the same educational level were less likely to remain childlessness compared with heterogamous couples. A study in Ghana that examined denominational affiliation and fertility behaviour showed that religious homogamy among couples was associated with fertility: couples with the same religious affiliation had higher levels of fertility than couples with different religious affiliation (Gyimah et al. 2008).

This study therefore examines the relationship between couples’ characteristics and fertility behaviour, as indicated by number of children ever born (CEB), in three sub-Saharan African countries—Ghana, Nigeria and Zambia—where TFR has shown signs of decline. Of the three countries, TFR has declined fastest in Ghana, to a current level of 4.2, whereas Nigeria and Zambia retain a high fertility regime of 5.5 and 5.3, respectively (Population Reference Bureau 2016). Specifically, our study looks at the level of homogamy (having similar characteristics) and heterogamy (having different characteristics) in achieved and ascribed characteristics among couples, the mean CEB across homogamous and heterogamous categories and the relationship between homogamy and hetrogamy in achieved and ascribed characteristics of couples and their fertility behaviour.

2. Method and measures

2.1. Data source and study population

Data for this study were obtained from demographic and health surveys (DHS): 2003, 2008 and 2014 for Ghana, 2003, 2008 and 2013 for Nigeria and 2001–2, 2007 and 2013–14 for Zambia. The three countries were purposively selected to represent moderately low and high fertility regimes in the region. To increase the sample size and examine patterns of homogamy and heterogamy, datasets for the three survey years were pooled for each country. Patterns of homogamy are usually studied with newly formed or all existing marital union (Hamplova 2009). Thus, the study population was a weighted sample of couples in monogamous union (married and living together with partner) where wives were aged 35 years and over: 1785 in Ghana, 3185 in Nigeria and 3175 in Zambia. Given that the female median age at first marriage in the three countries is below 20 years, and the closeness of first birth to age at first union, the age of wife was reduced to 35–49 to eliminate younger women who are not near completion of childbearing.

The outcome variable was fertility behaviour, indicated by CEB reported by wives. Wives’ CEB was used for all the levels of analysis because the mean CEB of wives is closer to the TFR in the three countries. In Zambia, wives’ mean CEB was 6.3 whereas husbands’ was 7.0 and Zambia’s TFR is 5.3. In Ghana, CEB for wives was 4.8 and 5.3 for their husbands, and the country’s TFR is 4.2; in Nigeria, wives’ CEB was 5.7 whereas their husbands’ CEB was 6.1 and the country’s TFR is 5.5. Total CEB of a woman is a commonly used measure of completed fertility. Thus, we considered it a more accurate indicator of fertility behaviour than other measures provided in DHS such as number of living children.

2.1.1. Variables and measures

Explanatory variables were achieved and ascribed characteristics of the couples. Achieved characteristics usually refer to educational attainment and occupation (Hamplova 2009). Thus, highest educational level and occupation were the indicators of achieved characteristics. Education was categorised into: both have none, both have secondary and higher, both have primary (educational homogamy), husband more educated, and wife more educated (educational heterogamy). Occupation was categorised as: not same occupation (occupational heterogamy) and both whitecollar (professional, managerial, technical and clerical), sales and services, agriculture (employers and employees), and manual and others. In Nigeria, cases where both had manual and other occupations were included in the same category as agriculture because they were few. Ascribed characteristics were age, religion and ethnic origin. Age was categorised into: same age (homogamy), wife older, wife 1–4 years younger, wife 5–9 years younger, and wife younger by 10 years and over (heterogamy). Religion was grouped into: both Catholics, other Christian, Muslim, traditionalist (homogamy), wife Catholic/husband other Christian, and wife Christian/husband Muslim, traditionalist and other (heterogamy). Ethnic origin was categorised into: same ethnic origin (homogamy) and inter-ethnic (heterogamy). Some factors, such as place of residence, household wealth, fertility preferences, marital status, that are associated with homogamy and fertility behaviour in previous studies were included as control variables (Casterline et al. 1986; Schoen & Weinick 1993; Dodoo & Seal 1994; Oyediran & Isiugo-Abanihe 2002; Hamplova 2009). However, type of union was excluded as a control variable in the multivariate analysis because couples who were not in a formalised union (living together with partner) in Zambia were few. Place of residence and household wealth are similar for husband and wife, fertility preference was re-coded to reflect a couple’s preference, but marital status was that reported by wives. Year of survey was also added as a control.

2.1.2. Analytical strategy

Analysis was conducted using Stata 12 for Windows in three levels: univariate, bivariate and multivariate. The univariate analysis consisted of a description of the study population by selected couple’s characteristics using weighted percentages. The bivariate analysis presented the mean CEB and standard deviation for each category of the selected explanatory variables. At the multivariate level, Poisson regression was used to model fertility behaviour of couples as a function of their characteristics while controlling for potential confounders. Poisson regression was used because of the count nature of CEB. Goodness-of-fit test was insignificant in each country analysis, confirming the suitability of the Poisson model for the multivariate analysis. To obtain robust standard errors, vce (robust) command in stata was used to avoid mild violation of any of the Poisson regression assumptions. The result is presented as the incidence rate ratio (IRR) and 95% confidence interval. Statistical significance was set at 0.05 but 0.10 was considered a marginal significant level. Because of correlation of 0.67 and 0.69 between place of residence and household wealth index in Zambia and Ghana, respectively, different models were estimated with each of the variables. However, the results of these different models were only slightly different from the model that contained both variables, therefore the model with the two variables is presented.

3. Results

Results of descriptive analysis showing the profile of the study population by selected couples’ characteristics is shown in Table 1, and mean CEB with standard deviation by survey years and couples’ characteristics is presented in Table 2. Results of the Poisson regression estimating the IRR for fertility behaviour are presented in Table 3.

Table 1. Percent distribution of the study population by selected characteristics of couples.

Characteristic	Nigeria (n = 3 185)	Zambia (n = 3 175)	Ghana (n = 1785)
CEB
0–4	34.7	23.2	46.3
5+	65.3	76.8	53.7
Education
Both have none	23.5	2.7	15.4
Husband more	26.4	38.2	37.9
Wife more	13.7	8.3	7.4
Both secondary/higher	23.9	14.3	36.3
Both primary	12.5	36.5	3.0
Occupation
Not same	67.2	61.2	56.1
Whitecollar	5.7	2.2	2.9
Sales and services	13.2	6.4	6.8
Agriculture	–	28.3	31.0
Manual and others	–	1.9	3.2
Agriculture and others	3.9	–	–
Age
Same	3.2	3.3	4.1
Wife older	4.5	5.4	8.5
Wife 1–4 years younger	25.2	32.9	30.9
Wife 5–9 years younger	42.2	40.2	33.5
Wife 10+ years younger	24.9	18.2	23.0
Religion
Both Catholic	8.8	17.6	9.1
Both other Christian	44.7	75.3	64.6
Both Muslim	36.7	–	10.9
Both traditionalist	4.0	–	3.7
Wife Catholic/husband other Christian	2.1	4.2	2.9
Wife Christian/husband Muslim, traditionalist and other	3.7	2.9	8.8
Ethnicity
Inter-ethnic	11.6	45.0	18.4
Same ethnic group	88.4	55.0	81.6
Place of residence
Urban	42.0	38.4	44.9
Rural	58.0	61.6	55.1
Household wealth
Poorest	18.6	18.2	16.3
Poorer	15.4	19.2	20.1
Middle	16.7	20.4	17.7
Richer	19.0	20.4	19.2
Richest	30.3	21.8	26.7
Fertility preference
Both no more	26.1	55.6	50.9
Both want or undecided	45.4	16.3	24.3
Wife wants or undecided/husband no more	9.9	10.4	11.1
Wife no more/husband wants or undecided	18.6	17.7	13.7
Marital status
Married	98.6	99.5	91.5
Living together with partner	1.4	0.5	8.5

Table 2. Mean CEB by selected couples’ characteristics.

Characteristic	Nigeria	Zambia	Ghana
	Mean CEB (standard deviation)
Mean CEB
2001–2	–	6.8 (2.8)	–
2003	6.6 (2.9)	–	5.3 (2.3)
2007	–	6.4 (2.6)	–
2008	5.9 (2.8)	–	4.6 (2.2)
2013	5.3 (2.5)	–	–
2014	–	–	4.5 (2.1
2013–14	–	6.2 (2.5)	–
Pooled data	5.7 (2.7)	6.3 (2.6)	4.8 (2.2)
Education
Both have none	7.4 (2.9)	6.5 (3.1)	6.0 (2.0)
Husband more	5.9 (2.5)	6.4 (2.6)	5.1 (2.3)
Wife more	5.0 (2.5)	5.9 (2.3)	4.9 (2.0)
Both secondary/higher	4.1 (2.0)	4.8 (2.3)	4.1 (1.9)
Both primary	5.9 (2.4)	7.0 (2.4)	6.3 (2.1)
Occupation
Not same	5.9 (2.7)	6.2 (2.5)	4.5 (2.1)
Whitecollar	3.8 (2.0)	3.7 (2.0)	3.5 (1.1)
Sales and services	5.3 (2.7)	5.4 (2.4)	4.0 (2.1)
Agriculture	–	7.2 (2.5)	6.0 (2.1)
Manual and others	–	5.8 (2.5)	4.1 (2.0)
Agriculture and others	6.3 (2.6)	–	–
Age
Same	5.2 (2.7)	5.8 (2.6)	4.5 (2.6)
Wife older	5.5 (3.0)	5.2 (2.5)	4.8 (2.3)
Wife 1–4 years younger	5.4 (2.7)	6.3 (2.6	4.7 (2.2)
Wife 5–9 years younger	5.8 (2.7)	6.7 (2.5)	5.2 (2.1)
Wife 10 years younger	6.2 (2.8)	6.3 (2.6)	5.1 (2.3)
Religion
Both Catholic	5.2 (2.7)	6.7 (2.5)	5.1 (2.2)
Both other Christian	5.0 (2.3)	6.4 (2.5)	4.7 (2.2)
Both Muslim	6.7 (2.9)	–	5.3 (2.1)
Both traditionalist	7.2 (3.0)	–	6.2 (2.3)
Wife Catholic/husband other Christian	5.4 (2.2)	6.2 (2.7)	4.8 (2.0)
Wife Christian/husband Muslim, traditional or other	5.5 (2.7)	5.8 (2.4)	5.4 (2.1)
Ethnicity
Inter-ethnic	5.6 (2.7)	5.9 (2.6)	4.7 (2.3)
Same ethnic group	5.7 (2.7)	6.7 (25)	5.1 (2.2)

Table 3. Poisson regression models estimating the IRR for CEB and homogamous and heterogamous characteristics of couples in Nigeria, Zambia and Ghana.

Characteristic	IRR (95% confidence interval)
	Nigeria	Zambia	Ghana
Education
Both have none (RC)	1.00	1.00	1.00
Husband more	0.95 (0.90–1.00)^†	1.07 (0.96–1.21)	0.93 (0.88–0.98)*
Wife more	0.86 (0.81–0.92)***	1.02 (0.90–1.15)	0.92 (0.85–0.98)*
Both secondary/higher	0.80 (0.75–0.85)***	0.95 (0.84–1.08)	0.83 (0.79–0.88)***
Both primary	0.93 (0.88–0.99)*	1.11 (0.98–1.24)^†	1.05 (0.96–1.14)
Occupation
Not same (RC)	1.00	1.00	1.00
Whitecollar	0.86 (0.80–0.92)***	0.78 (0.68–0.88)***	0.92 (0.84–1.01)^†
Sales and services	1.01 (0.96–1.06)	0.95 (0.89–1.02)	1.01 (0.93–1.10)
Agriculture	–	1.03 (1.00–1.06)*	1.07 (1.03–1.12)**
Manual and others	–	1.00 (0.89–1.11)	1.00 (0.90–1.11)
Agriculture and others	0.98 (0.94–1.03)	–	–
Age
Same (RC)	1.00	1,00	1.00
Wife older	0.98 (0.88–1.10)	0.92 (0.83–1.03)	1.06 (0.95–1.19)
Wife 1–4 years younger	0.98 (0.91–1.07)	1.02 (0.94–1.12)	1.05 (0.96–1.16)
Wife 5–9 years younger	1.01 (0.93–1.09)	1.07 (0.98–1.17)	1.12 (1.02–1.23)*
Wife 10+ years younger	1.00 (0.92–1.09)	1.00 (0.91–1.10)	1.09 (0.99–1.20)^†
Religion
Both Catholic (RC)	1.00	1.00	1.00
Both other Christian	1.01 (0.95–1.07)	0.99 (0.96–1.02)	0.97 (0.92–1.03)
Both Muslim	1.20 (1.12–1.28)***	–	1.13 (1.05–1.21)**
Both traditionalist	1.08 (0.89–1.32)	–	1.06 (0.97–1.17)
Wife Catholic/husband other Christian	1.06 (0.96–1.17)	0.95 (0.88–1.02)	0.95 (0.86–1.05)
Wife Christian/husband Muslim, traditional or other	1.09 (0.99–1.21)^†	0.92 (0.84–1.00)^†	0.99 (0.92–1.07)
Ethnicity
Inter-ethnic (RC)	1.00	1.00	1.00
Same ethnic group	1.02 (0.97–1.07)	1.07 (1.04–1.10)***	1.02 (0.97–1.07)
Fertility preference
Both no more (RC)	1.00	1.00	1.00
Both want or undecided	0.72 (0.69–0.74)***	0.67 (0.64–0.70)***	0.62 (0.58–0.65)***
Wife wants or undecided/husband no more	0.81 (0.76–0.86)***	0.76 (0.72–0.80)***	0.78 (0.74–0.83)***
Wife no more/husband wants of undecided	0.90 (0.86–0.94)***	0.87 (0.83–0.90)***	0.81 (0.77–0.85)***
Place of residence
Urban (RC)	1.00	1.00	1.00
Rural	1.05 (1.01–1.10)**	1.01 (0.98–1.06)	1.01 (0.96–1.06)
Household wealth
Poorest (RC)	1.00	1.00	1.00
Poorer	0.94 (0.89–0.99)*	0.97 (0.93–1.01)	0.96 (0.91–1.06)
Middle	0.91 (0.86–0.96)**	0.95 (0.92–0.99)*	0.95 (0.89–1.01)
Richer	0.84 (0.80–0.89)***	0.86 (0.81–0.90)***	0.78 (0.72–0.84)***
Richest	0.68 (0.64–0.73)***	0.75 (0.70–0.80)***	0.64 (0.59–0.70)***

***p < 0.001.

**p < 0.01.

*p < 0.05.

^†p < 0.10.

The majority of wives in the three countries had given birth to five or more children. With respect to education, the distribution varied across the three countries, Many couples were in unions where the husband was more educated than the wife. Unions where the wife had attained more education than husband were more prevalent in Nigeria than Zambia and Ghana. Couples where both partners had secondary or higher education were many in Zambia (36.3%) and the highest percentage of couples without education was in Nigeria. The majority of the couples were occupationally heterogamous, but partners who worked in similar occupations were predominant in agriculture in Zambia and Ghana and sales and services in Nigeria. The majority of couples in all three countries were in unions where husbands were older. Homogamy in age was similar in Nigeria and Zambia and slightly higher in Ghana. Couples where the wife is older were more prevalent in Ghana than in other countries. Most couples in the three countries were religiously homogamous, but inter-faith marriage was more prevalent in Ghana than Nigeria and Zambia. More than 80% of couples in Nigeria and Ghana were ethnically homogamous, whereas slightly more than 45% of couples in Zambia were in an inter-ethnic union.

Results in Table 2 show variations in CEB across the survey years and the different categories of couples by their homogamous and heterogamous characteristics. In the most current DHS, the lowest mean CEB was in Ghana (4.5), and the largest was in Zambia (6.2). Mean CEB declined over the years in each country. In the pooled data, mean CEB was 5.7 in Nigeria, 6.3 in Zambia and 4.8 in Ghana. Unions where the couple had secondary/higher education had the lowest mean CEB in the three countries. With respect to heterogamy in all the countries, couples where the wife was more education had fewer CEB than where the husband was more educated. In the three countries, couples who worked in whitecollar occupations were more likely to have had fewer children whereas those in agriculture had the highest number of CEB.

The distribution of CEB by spousal age difference revealed that couples who were of the same age had the least mean CEB in Nigeria and Ghana, but in Zambia the least mean CEB was among couples where wife was older than her husband. The highest mean CEB in the three countries was prevalent in unions where the husband was older than the wife, especially when the gap was five years or more.

A similar pattern of CEB was observed for Nigeria and Ghana in religious homogamy and heterogamy. The least CEB in both countries was among couples affiliated to other Christian denominations whereas the highest was among couples where both were traditionalists. In Zambia, the highest CEB was among couples who were both Catholics and the least was among couples in mixed-faith unions. In general, couples who were of the same faith or denomination were likely to have higher CEB than those in a religiously heterogamous union. Couples of similar ethnic origin had higher mean CEB than those in an inter-ethnic union across the three countries, although the difference in Nigeria was small.

Poisson regression results revealed no difference in the direction of association for both educational homogamy and hetrogamy in Nigeria. Compared with couples who had no education, the incidence of CEB was significantly lower among couples who had attained any level of education. For instance, the incidence rate for CEB among couples where both secondary/higher education was 0.80 times lower than for couples where neither was educated (p < 0.001). In Zambia, the incidence rate for CEB was higher for couples where both had primary education than for those who had no education but the statistical significance was marginal (IRR 1.11, p < 0.10). In Ghana, the incidence of CEB was significantly lower in all the categories of homogamy and heterogamy except for couples who had primary education. Occupational homogamy in whitecollar jobs decreased the incidence of CEB by 14% in Nigeria (IRR 0.86, p < 0.001), 22% in Zambia (IRR 0.78, p < 0.01) and 8% in Ghana (IRR 0.92, p < 0.10). Although couples where both worked in agriculture had the highest mean CEB in the three countries (Table 2), the observed difference was only statistically significant and in the expected positive direction in Zambia and Ghana. The incidence of CEB was higher among couples who both worked in agriculture in Zambia (IRR 1.03, p < 0.05) and in Ghana (IRR, 1.07 p < 0.01).

Age difference was only of statistical significance in Ghana, where the incidence rate for CEB was higher among couples where the wife was younger by 5 or more years compared with couples who were of the same age. Religious affiliation of couples significantly predicted higher incidence of CEB in Nigeria among Muslim couples (IRR 1.20, p < 0.001) and where the wife was a Christian but the husband of another religion (IRR 1.09, p < 0.10) compared with Catholic couples. In Zambia inter-faith union marginally decreased the incidence of CEB by 8% (IRR 0.92, p < 0.10) and in Ghana Muslim couples had a higher incidence of CEB than their counterparts who were both Catholics (IRR 1.13, p < 0.01). The incidence rate ratio for CEB was significantly higher in ethnically homogamous unions than in inter-ethnic unions in Zambia (IRR 1.07, p < 0.001). With respect to the control variables, the incidence of CEB was lower when both or one of the couples still wanted or was undecided about having another child than when both wanted no more. Couples residing in rural areas in Nigeria had a significantly higher rate of CEB (IRR 1.05, p < 0.01). Household wealth was inversely associated with CEB in all the countries. In Nigeria, compared with the poorest, all the other categories had significantly lower incidence of CEB. The same pattern was revealed in Zambia and Ghana but the poorer category was not significant in Zambia and in Ghana only the richer and richest categories were significant.

4. Discussion

This study has examined the level of homogamy and heterogamy in achieved characteristics (education and occupation) and ascribed characteristics (age, religion and ethnicity) of couples in Nigeria, Zambia and Ghana and the association with their fertility behaviour. Most couples in the three countries were educationally homogamous, the highest proportion being among those who had attained secondary and higher education although almost an equal proportion in Nigeria were couples in which both partners were uneducated. Past studies show that there is an increasing tendency for people to marry homogamously and that educational homogamy is becoming increasingly more important in mate selection than ascribed characteristics (Kalmijn 1991; Halpin & Chan 2003), especially in first marriages (Schwartz & Mare 2010).

With respect to educational heterogamy, the most prevalent pattern was unions where the husband was more educated. The highest proportion of couples where wives were more educated than their husbands was in Nigeria. This may be an indication of more openness in Nigerian society with regard to status homogamy in mate selection. It may also be an indication of the high value placed on marriage, particularly for women, in Nigerian society; an educated woman would rather ‘marry down’ than remain unmarried (Isiugo-Abanihe 2000; Ntoimo & Isiugo-Abanihe 2014). Although occupational heterogamy was predominant in the three countries, homogamy, especially in agricultural occupations, was common in Zambia and Ghana. Consistent with past studies, in all the ascribed characteristics all three countries remained conservative, the majority of couples having older husbands, same religion and same ethnic origin (Ortega et al. 1988; Gyimah et al. 2008; Ibisomi 2014) However, Zambia had a large number of inter-ethnic unions. Homogamy and endogamy are indications of how open a society is and of transitions in social inequality (Kalmijn 1991, 1998). Thus, the high level of inter-ethnic marriage in Zambia may be an indication of higher inter-ethnic tolerance in that country than in Nigeria and Ghana.

Educational homogamy and heterogamy predicted the fertility behaviour of couples in Nigeria and Ghana. However, there was no difference in the direction of association between the significant homogamy and heterogamy variables and CEB in the two countries. This indicates the significance of education in lowering fertility even when only one partner has some education. The findings in educational homogamy and heterogamy are consistent with a study conducted in Taiwan by Tsou et al. (2011). Also, occupational homogamy, particularly in whitecollar occupations, predicted lower fertility in the three countries. Of note is that the highest CEB in the three countries was among couples who were homogamous in agricultural occupations and it predicted higher CEB in Zambia and Ghana. Previous studies in sub-Saharan Africa document the use of children for labour in agriculture as one of the determinants of high fertility in Africa (Caldwell & Caldwell 1987). It is likely that this practice is still highly prevalent in these countries. These findings suggest that encouraging post-primary education and the creation of more employment opportunities in the formal sector are significant drivers of fertility transition in the three countries.

Relating homogamy and heterogamy in age to CEB showed that lower mean CEB was observed across the three countries when the couples were of the same age, or the wife was older or younger by 1–4 years, than when the husband was older with a larger age gap. However, the difference was statistically significant only in Ghana. According to Casterline et al. (1986), age difference affects relations between the spouses, which impacts on marital satisfaction, family size preferences, and contraceptive use, all of which directly influence fertility. Also, studies in sub-Saharan Africa show the dominant role of men’s fertility desire in determining family size, particularly when the wife is younger and less educated (Dodoo & Seal 1994; Oyediran & Isiugo-Abanihe 2002). A notable finding in religious homogamy in Nigeria and Ghana was that couples who were both traditionalist had more children than Muslim couples, corroborating a study in Ghana (Gyimah et al. 2008). This finding reveals that couples who are affiliated to traditional religions should be targeted for programmes and interventions to lower fertility in these countries. Couples affiliated to other Christian denominations had fewer children than Catholics and Muslims in Ghana and Nigeria. In general, ethnic and religious homogamy presented higher CEB than inter-faith and inter-ethnic unions, indicating the impact of similar cultural orientations in the fertility behaviour of couples. When couples share similar values and beliefs, it affects their reproductive choices and behaviour (Yeatman & Trinitapoli 2008). Couples of the same religion are more likely to be similarly committed to religious norms about large family size and opposition to the use of contraceptives. Also, religious homogamy increases marital satisfaction (Ortega et al. 1988), and according to Casterline et al. (1986) marital satisfaction influences fertility behaviour.

5. Limitations of the study

Mate selection (homogamy and hetrogamy) is influenced by availability of preferred or suitable mates in the marriage market and by sex ratios. The data source for this study provides no information on these factors, so the effect of sex ratio could not be taken into account in the analysis. Also, DHS provide information on current educational status, which may have improved since marriage, and decisions about number of children ever born may have taken place before the upgrade of education. However, using cross-sectional data provides insight into status homogamy in prevailing marriages in the population at a given time, which has been shown to provide a better picture of assortative mating principles (Hamplova 2009). In spite of the limitations of cross-sectional data, using pooled data in this study enhanced the advantage of applying the same measures to comparable samples across generations.

6. Implication for family demography and post-2015 development agenda in Africa

The findings are relevant to understanding changes in the marriage market in sub-Saharan Africa with regard to mate selection preferences. Patterns of partnering with regard to couples’ characteristics are vital to understanding transformations in family formation, structure and breakdown. They also have consequences for demographic outcomes that are directly or indirectly linked to the family, such as longevity, health, fertility preferences and fertility behaviour. The current study shows that homogamy and heterogamy in marital unions are associated with the couple’s fertility behaviour. For instance, homogamy does not necessarily promote smaller family size, particularly if couples are of the same ethnic and religious orientation. Targeting particular ethnic and religious groups in fertility decline programmes will expedite the pace of fertility decline and promote women’s reproductive health and child health in the region.

People have a tendency to marry within their social groups or to marry people who share similar characteristics to them (Kalmijn 1998). Thus, homogamy and endogamy are indications of how open a society is and how much mobility individuals have in the social hierarchy. The results of the current study indicate that, to a large extent, the three countries studied are still rather closed with respect to social mobility through marital union, but there are signs of positive change.

7. Conclusion

The results of this study indicate that most couples in the three countries were homogamous in educational attainment, religion and ethnic origin. Occupational heterogamy was predominant in all the countries, but homogamy in agricultural occupations was high in Zambia and Ghana. Unions in which the husband was older than the wife by five or more years were highly prevalent. There was a declining trend in fertility among the couples. Educational homogamy and heterogamy in Nigeria and Ghana and occupational homogamy especially in whitecollar employment in the three countries were significant predictors of lower fertility. On the other hand, homogamy in agricultural occupation in Zambia and Ghana, age gap of five years and over in Ghana, religious homogamy for Muslims in Nigeria and Ghana and ethnic homogamy in Zambia were associated with high fertility among couples. These results highlight the importance of couples’ characteristics in determining marital fertility. Thus, programmes and policies aiming to reduce fertility in the region should incorporate strategies that focus on the characteristics of couples in union. More studies are recommended on trends, patterns, transition and consequences of homogamy and heterogamy in sub-Saharan Africa.

Acknowledgement

The authors acknowledge the support of the Department of Science and Technology and the National Research Foundation (DST-NRF) Centre of Excellence in Human Development, South Africa. Opinions expressed and conclusions arrived at are those of the authors and not to be attributed to the Centre of Excellence in Human Development.

Disclosure statement

No potential conflict of interest was reported by the authors.