Total fertility and marital fertility rates of 5.3 and 7.8 (shown elsewhere) children per woman, respectively, obtained in this study was similar to the finding in Gondar, North West Ethiopia done in 2007 . The fertility level is still one of the highest. This could be attributed to the credence of the wider community to large family size norm as children assisted households in subsistence farming and petty trade. Though disparities were observed across major regions of the world, children were considered as assets to their parents when they get older as shown in a study using the Demographic and Health Survey in 43 countries . This posit was further supported by the statistically-significant finding of higher fertility among women who were members of larger households compared to those who belonged to smaller sized households (less or equal to 4 members) in this study. In this study, household constituted individuals regardless of their blood relations that live in one or more houses with the same cooking arrangement. Most members of the household were nuclear family members. The fact that women of parity 4 had more than 79 percent chance to have the 5th parity augmented the deep rooted culture of larger family size that might have been supported in the study community in the foreseeable future. Thus, the total-fertility-rate goal stipulated in the Ethiopian population policy is far from reach .
Women that married in their teens had a significantly higher fertility compared to those married after they celebrated their 20th birthday. Moreover, this study revealed that current contraceptive prevalence among women in the reproductive age group and married women were 15 and 25 percent, respectively (figure not shown). Besides women who marry early in life may have an increased risk of having many children, in particular, if they started childbirth before the age of 20 years. On the contrary, several studies [16, 17] have shown that postponement of first childbirth to later ages leads to fertility reductions since women would have fewer years of reproduction window which may introduce parity specific controls even after the initiation of child birth.
On the other hand, women who had many years of education had significantly lower fertility as compared to those who had never been enrolled into any formal education system. This corroborates with similar studies [13, 17–19] and may be attributable to the postponement of childbirth due to longer schooling. Educated women might be more worried to have many children if their area of usual residence had been stricken by frequent food shortage . Nonetheless, a study among Sidamas in Southern Ethiopia indicated that fertility was higher among women with primary level of education compared to those who never attended any formal education. Higher educational level of women gives an opportunity of social and economic empowerments. Thus, able women might feel that they could take care of many children and opted for large family size. This is consistent with the claim by some researchers that increased family income leads to increased fertility when family planning use is low . Education might also have impacts to bring about change in the knowledge and attitude towards low fertility. By the same token, women who did not know the time at which they could be pregnant had higher fertility as compared to those who knew it.
Disparities in level of fertility between urban and rural communities in this study population were similar to the finding in Gondar  that could be attributed to differences in contraceptive prevalence and age at first marriages between urbanites and rural residents in Ethiopia since women in urban areas had better access to media, general knowledge and services. It was however difficult to document reasons for the change in the direction of associations between residence ecology type and fertility when other factors were controlled. Meanwhile, informed urbanites might have a positive attitude towards smaller family size besides having better access to family planning services. For instance the urban fertility level of 3.3 children per woman found in this study (result not shown) was similar to the level found for Awassa town which is the capital of the study region . In a similar context, fertility was significantly lower among women whose major livelihoods are based on trade or services compared to those households who got their main incomes from subsistence farming. Most of the households who relied on trade or services as main source of household income could be influenced by urban culture and practice which showed a low fertility norm.
Although food security was widely misunderstood, researchers considered it as shortages in the quality and quantity of food at any time while others defined it as shortage or absence of edible food . In this study, encounter to household food scarcity in the past calendar year was collected. This variable might not precisely measure the household food security which could be cited as a limitation. However, it indirectly showed the household economic status and risk to vulnerability. Fertility was significantly higher among women whose households suffered from food shortages. There could be an egg-chicken dilemma in this claim since the food shortage might be caused by the large family size which in turn could mainly be triggered by higher fertility. In a resource constrained environment such as the study area, people had to share the scarce food and could probably be exposed to shortages particularly in rainy seasons, the time at which grain foods of farming households would be depleted . Fertility was also found to be lower among malnourished mothers compared to nourished ones in a similar study done for Sidamas in Southern Ethiopia .
An in-migrant to the DSA is a person who went there to live or stayed in it for 6 or more months if did not have such intention. No significant association has been shown between migration status of women and fertility in this study. This could be due to the fact that the duration of residence of in-migrants may not be long enough to influence fertility and reproductive health behavior in the study area as migration was categorized broadly. It may also that in-migrant women could have come from rural villages in the same or nearby districts of the country which had similar socio-cultural and behavioral characteristics. Nevertheless, a study in China had indicated that rural-urban return migrants from middle and large cities had adapted positive fertility behavior towards small family size norm compared to their rural-rural return migrant counterparts . The same study had also revealed that areas with higher prevalence of rural-urban return migrants particularly from middle and mega cities would have a positive effect to adapt small family size.
Studies done on fertility responses to childhood mortality have focused on insurance and replacement effects . Couples in high-mortality settings anticipate the death of some of their children that might dictate them to change their reproductive preferences and behavior. Childhood mortality could also have the combined biological and volitional replacement effects to reduce the time to subsequent conception if the death occurs within a given interval. The time to conception could also be reduced if a childhood death occurs during a prior birth interval. Accordingly, this study had revealed child mortality as a strong and significant predictor of fertility as documented for the same study population a decade ago and elsewhere [17–19]. Epidemic and frequent episodes of drought in the study community, which might have claimed the lives of their children, provoked mothers to replace the lost ones .
Contrary to the findings of many studies, fertility was significantly higher among women who had no preference to the sex of their children. Conversely, studies done in Ethiopia and Asia had revealed more preference towards sons compared to daughters because males inherit properties from their ancestors in a patrilineal society. A similar study in USA indicated a stable marital union or request for the custody of their child by fathers if the marriage was dissolved when couples had sons instead of daughters [22, 25, 26]. Religiosity among women in the study community had probably dimmed their decision.
The use of Butajira DSS database to recruit study participants and administration of standard maternity history questionnaire by clinical nurses could be mentioned as strengths for the current study. The demographic surveillance staffs knew study participants for a long time and this built trust on the participants side to provide reliable information. The study was also conducted in a peak harvesting season which reduced the participation of certain community groups that could be away from their home by virtue of their work.