This study identified and investigated the impact of several potential predictors of TCEB that have not been previously explored or thoroughly examined by past researchers. These new predictors were combined with existing global predictors of TCEB to develop a new model for women's fertility rates in Nigeria. The base model (Model 1) includes predictors already established in the literature, while 16 additional models assess the marginal effect of each newly identified predictor. Based on their AICc values, the variables "Residence type" and "Anaemia level," as well as "Decision maker for using contraception," were found not to improve the base model and were thus excluded from the final model (Model 17). The results of Model 17 are presented under fixed effects, non-linear effects, and spatial effects as follows:
4.1 Interpretation of fixed effects estimation results
Table 6 presents the posterior mean estimates of predictors modeled parametrically, along with their 95% credible intervals. Additionally, it includes estimates for four metrical predictors (Age at first birth, Age at first cohabitation, Age at first sex, and Number of Co-wives), which were initially assumed to have a non-linear effect on the response variable but were modeled linearly. The findings indicate that the TCEB for women in the North-East, North-West, and South-East regions are 12.12%, 16.95%, and 3.53% higher, respectively, than those in the North-Central region. In contrast, women in the South-West have a 7.71% lower TCEB than those in the North-Central. The posterior mean estimates for the North-East, North-West, and South-West are significant at the 5% level.
Furthermore, women with no education, primary education, and secondary education have 12.55%, 11.91%, and 5.31% higher TCEB, respectively, compared to those with higher education. These estimates are all significant at the 5% level. In terms of wealth index, women in the poorest category have 2.16% more children than those in the middle category, although this estimate is not significant at the 5% level. Women using female sterilization, injections, male condoms, or no contraceptive have 10.29%, 4.63%, 9.3%, and 5.09% fewer children, respectively, than those using emergency contraception.
Additionally, women living with their partner, married women, and widowed women have 16.13%, 14.67%, and 6.04% more children, respectively, than divorced women. Women who have never been in a union have 26.37% fewer children than divorced women. Only the estimate for widowed women is not significant at the 5% level. Women whose healthcare decisions are made solely by their husband/partner have 33.56% more children than those whose healthcare decisions are made by others. Women who share healthcare decision-making with their husband have 1.62% fewer children than those whose husband alone decides. Both estimates are not significant at the 5% level.
Women who used the internet in the last 12 months have a TCEB that is 10.97% lower, significant at the 5% level, compared to those who never used the internet. Women who have had a terminated pregnancy have 3.82% fewer children, significant at the 5% level, compared to those who never had. Women whose husbands previously lived in Bauchi, Imo, Jigawa, Katsina, and Niger States have 5.53%, 7.76%, 5.51%, 6.34%, and 6.89% higher TCEB, respectively, than those whose husbands previously lived in Abia State. Conversely, those whose husbands previously lived in Enugu State have 6.53% significantly lower TCEB compared to Abia State, at the 5% significance level. Finally, increases in age at first birth, age at first cohabitation, age at first sex, and number of co-wives increased the TCEB by 0.9544, 1.0017, 0.9983, and 0.9984, respectively. Only the estimate for age at first birth is significant at the 5% level.
4.2 Interpretation of smooth terms results
Figure 2 displays plots illustrating the relationship between various factors and TCEB: respondent age, age of partner, duration of amenorrhea, years since first cohabitation, months of abstinence, preceding birth interval, and succeeding birth interval. The TCEB increases with the age of the respondents. Similarly, the TCEB increases with the age of the partner up to 40 years, after which it declines. For women with amenorrhea, the TCEB fluctuates for periods shorter than 25 months before stabilizing.
The relationship between "years since first cohabitation" and TCEB is ∩-shaped, with an increase in the number of children born up to 15 years of cohabitation, followed by a decline. The TCEB also fluctuates as the months of abstinence increase up to 60 months, after which a continuous increase is observed. Increases in the preceding birth interval and succeeding birth interval are associated with a decrease in TCEB.
4.3 Spatial Effects
Figure 3 illustrates the spatial effects on TCEB. The reference region is depicted in grey. Regions coloured red indicate states in Nigeria with higher TCEB compared to the reference region, while regions coloured blue indicate states with lower TCEB. The states with the highest number of children born are Adamawa, Taraba, Borno, Plateau, Nasarawa, Benue, and Enugu. Conversely, the states with the lowest birth rates are Bayelsa, Niger, Kebbi, Kwara, and Rivers.