Abstract
This study compares the role of siblings in the survival of newborns from 1750 to 1900. Utilizing nearly one million longitudinal life history trajectories from Spain and Portugal, we examined whether cooperative breeding and resource dilution theories explain the observed patterns in southern Europe. Our analysis employs both descriptive statistics and probit regression models. Our findings confirm that in all cases, a high number of older siblings significantly increased the chances of newborn survival (by 10–30 percent). Additionally, there were notable differences based on the gender of the baby, with girls born in rural Spain being more adversely affected.
Introduction
When Jane Austen states in her book Mansfield Park, “What strange creatures brothers are!,” she is referring to the differences in behavior and the expression of emotions between brothers and sisters. Masculinity was associated with the ability to be economically self-sufficient and to control one’s emotions. 1 This did not mean that brothers did not care for or show protective behavior toward their siblings, but rather that they did so primarily within the confines of their assigned gender roles.
There is still a debate on whether siblings have a beneficial, negative, or neutral effect on the survival of their brothers and sisters, with a lack of studies focusing on southern Europe. 2 Conversely, it is widely accepted that mothers played a very positive role in the survival and health of their offspring, especially during the early years of life. 3
The role of fathers is less clear and still debated as to whether they had a positive or neutral effect. 4 Fathers appear to have played a more decisive role during their children’s youth and early adulthood, primarily through economic improvements in living conditions, with male wages being crucial for the household’s sustenance. 5
The absence of a father or mother has a significant impact on the survival chances and quality of life of their young children. 6 The economic, social, and epidemiological context obviously affects the health and longevity of individuals. 7 Additionally, children who have lost a young parent may have different life expectancy perceptions and more self-destructive behaviors. 8 Stepparents might mitigate the psychological and social consequences of a parent’s death. 9 However, we know little about how siblings contributed to the upbringing of their younger siblings at different ages and whether the stronger family ties in southern Europe compared to other regions of the continent could have been a beneficial factor for newborns. 10
The aim of this article is to delve into the effect of siblings (both in total and differentiated by gender and age) on the survival and health of younger siblings. We aim to test the hypothesis that a greater presence of siblings is associated with higher survival rates due to the care they could provide in a family-oriented region such as the Iberian Peninsula and the Azores over the long term (birth cohorts 1750–1900), before the onset of the fertility transition that occurred in the study area at the beginning of the twentieth century. 11
We have chosen this period not only for the availability of high-quality data but also because the onset of fertility control marked a paradigm shift regarding family size. 12 We aim to explore the preceding period, when family size, the number of children/siblings, and their gender distribution were more challenging to pre-determine. This study contributes to the discussion on how early life conditions (in this case, those associated with family size) affected the survival and early life experiences of individuals. 13
For this article, we utilized three of the largest databases containing family reconstructions and life trajectories available for Spain and Portugal. Consequently, our research encompasses data for nearly 300,000 families, totaling over 1 million individuals. Our analysis will be based on two of the most prominent theories regarding the presence of siblings: cooperative breeding and resource dilution (see the following section).
This study is innovative for several reasons. First, it is the first comparative study using microdata on life trajectories from various distant regions of Spain and Portugal to examine the role of siblings. Additionally, thanks to the large volume of data, this study delves into the effects of siblings based on their gender and age, allowing us to provide a much broader perspective on the explanatory factors.
The Influence of the Role of Siblings on the Survival of New Offspring
There is still an international debate on the impact that the presence of a high number of siblings, depending on their gender, has had on the survival and well-being of newborn siblings. 14 Some studies find a positive relationship between the number of siblings and the well-being of the newborn, 15 while others find no such relationship. 16 Furthermore, we must consider that the results of each study may be influenced by the available data, geographic area, period, and methodology used.
As a consequence of the multitude of results, a wide variety of possible explanatory theories have been developed, although none has been fully confirmed. In this article, we will focus on two fundamental theories:
The theory of cooperative breeding focuses on whether older siblings had a positive effect on the survival and well-being of their younger siblings through their contribution to their upbringing.
17
As mentioned, several studies have found positive results of older siblings consistent with the hypothesis of cooperative breeding.
18
The resource dilution hypothesis predicts a negative association between sibling size and outcomes in offspring survival and long-term biological well-being.
19
The logical explanation would be that the limited resources available in a pre-industrial family must be distributed among all offspring. Therefore, with a greater number of children, fewer resources are available for each sibling, which would have a direct negative impact on them. Some studies support the resource dilution theory.
20
There are other studies that do not find associations between offspring size and adult height.
21
Recently, this literature has been enriched by contributions from various and diverse perspectives. Gibbs et al. consider the possibility that, in addition to the resources parents provide to their children, social institutions may also contribute additional resources to families, which may vary over time and place, leading to different outcomes. 22 Quanjer and Kok introduce the concept that the increased family support model has contributed to the height increase of the Dutch. 23 Other authors examine how cultural factors may have influenced different countries in similar ways. 24
De Keyser and Van Rossem study birth weight and length. 25 Their findings are significant as they reveal that these factors are not dependent on post-natal living conditions. Among the recent works emerging under the conditional research model approach, notable studies include those by Paiva et al. for Ribeira Seca in the Azores in the nineteenth century; 26 by Park et al. for Korea in the mid-twentieth century; 27 and the study by Riswick for the Netherlands in the nineteenth century. 28
Area, Data, and Methods
Areas: Rural Aragón (Spain), Semi-Urban Fafe and rural Couto do Mosteiro and São Joaninho (Northern Portugal) and Rural Azores
Let 's describe each of these areas. In the case of the database for rural Aragon (Alfamén and Middle Huerva Database), 29 the study area includes sixteen Aragonese villages in northeastern Spain, 30 situated around the middle valley of the Huerva River, whose municipal terms cover an area of over 500 km² (see Figure 1). The distance to the regional capital, Zaragoza, varies from 19 to 40 km. The study area had a population of 5,300 people in 1800 and 7,765 in 1910. This area is located next to the Ebro Valley, at the foot of the Sierra de Algairén, blending plains with low mountain landscapes.
Location of the study area.
The “Alfamén & Middle Huerva Database,” comprising over 120,000 individuals in total, was constructed using the family reconstruction method devised by Michel Fleury and Louis Henry. 31 Data from baptism, marriage, and death parish records were employed for a period spanning from the last quarter of the fifteenth century to 1950. The database includes all individuals who were born and baptized in the reference parishes or migrated to them and were registered in any of the aforementioned events. Individuals who left the parish and therefore had their date of death unknown are not considered for this analysis.
In rural Aragon, the inhabitants were primarily engaged in agriculture, mainly cultivating their own or others’ cereal and vineyard fields, and to a lesser extent, sheep farming. 32 Before the demographic transition, in rural Aragon, marriages where both spouses exceeded the age of forty-nine had an average of around seven live-born children. The last women to exceed the average of seven children were those born between 1881 and 1885, with 7.29 children. From that point on, there began a decline in parity that led to those born just twenty-five years later (between 1906 and 1910) having much lower levels, with 3.43 children. 33
Regarding the sample for Northern Portugal, we have two areas: 1. The semi-urban Fafe, and 2. The two rural villages of Couto do Mosteiro and São Joaninho (district of Viseu). Fafe (in northern mainland Portugal) 34 encompasses a demographic and genealogical sample of thirty-three complete parishes with rural and semi-urban characteristics between 1700 and 1900. With a total of 160,003 individuals who were part of 53,045 families residing in the municipality of Fafe, located in the Baixo Minho subregion of Portugal. Parish records of births, marriages, and deaths have been the primary documentary sources for the construction of the database, following the parish reconstruction method, similar to family reconstruction but with some differences that facilitate cross-referencing data across multiple generations. 35 Its management allows for the extraction of information regarding vital events related to individuals born, married, or deceased in the parishes over successive generations. Being a semiurban locality, slightly over half of the population was engaged in subsistence agriculture, but the remaining nearly 50 percent was involved in artisanal production, commercial activities, and goods transportation, with 10 percent of the population engaged in tasks requiring advanced knowledge such as teaching or healthcare. 36
Regarding the rural sample from the villages of Couto do Mosteiro and São Joaninho (Viseu district), they were selected due to the high quality of their mortality data in the eighteenth and nineteenth centuries (unlike Fafe, which has issues with infant mortality records until the nineteenth century, as we will discuss later). These are two villages contributing 39,445 individuals to the study sample, which we can consolidate into 12,729 distinct families, following the parish reconstruction method. 37 These villages were agricultural throughout the entire study period, specializing in subsistence farming. 38
Finally, for the case of the Azores islands, consisting of small rural communities, we have studied rural communities from five islands of the Azores archipelago (see Figure 1) based on the analysis of their parish records (baptism, marriage, and death books). The five islands are Corvo, Faial, Flores, Pico, and São Jorge. The island of Corvo has a single parish and 8,218 individuals (records available from 1714). The island of Faial has been studied based on the large parish of Pedro Miguel, with observations for 100,427 individuals. The island of Flores has been studied from the locality of Lajes das Flores, with 15,908 individuals (from 1757 onwards). Farmers in Flores own highly productive agricultural land, and these are communities closely tied to the sea. The island of Pico has been analyzed based on three representative parishes: Santa Luzia (in São Roque do Pico), with 7,813 individuals (from 1666), Prainha (in São Roque do Pico), with 12,583 individuals (from 1664), and Ribeiras (in Lajes do Pico), with 15,275 individuals (from 1717). The localities of Pico were chosen due to differences in resources and their connection to the sea. Finally, the island of São Jorge has been studied based on the parish of Ribeira Seca, for which we have information for 12,535 individuals. Parish data were similarly compiled using the parish reconstruction method. 39
When comparing the population trends of these islands with mainland Portugal, we find that these islands follow similar patterns to the mainland, but with some differences. 40 While the relative population stagnation in Portugal in the early eighteenth century showed an increasing trend around the 1750s, in the studied islands, it was not until well into the second half of the century that the upward trend emerged. 41 Similarly, while the population of Portugal continued to grow in the 1840s, the population of these islands began to decline—and continued to do so for decades—as a result of international emigration (primarily to the United States and Brazil). 42
Unfortunately, the high data quality parameters are not equal across the three areas studied. Data on infant and juvenile mortality clearly suffer from underreporting issues in historical periods in southern Europe. However, while for the studied parishes in the Azores, the two rural villages in the Viseu district (Couto do Mosteiro and São Joaninho), and rural Spain, the underreporting problem had been practically resolved by the mid-eighteenth century (when our study period begins), and their mortality rates are consistent with those found in the following 150 years. In the case of Fafe, underreporting issues persist until the mid-nineteenth century. Therefore, analyses with the Fafe sample are only conducted from 1850 onwards. Hence, in the data from Northern Portugal, rural data from Couto do Mosteiro and São Joaninho prevail until 1850. From that point on, given the greater volume of information for Fafe, data from the semi-urban locality predominate in the analysis. Nevertheless, we consider that the results obtained for both areas are relatively close and coherent. In all cases, we are in a period prior to the onset of fertility transition, with high rates of marital fertility and infant and childhood mortality. According to our data, in rural Spain, the infant mortality rate between 1750 and 1910 was 282, and childhood mortality was 572 (considering only individuals for whom we know the date of birth and death). In the Azores, these rates were 273 and 406, respectively, and in Northern Portugal, 279 and 472, respectively.
Variables
As mentioned earlier, we have obtained genealogical data on all families living in the study area between 1750 and 1900 from parish records of baptisms, marriages, and deaths. Using these individual and familial records, we have been able to extract various variables of interest for our study.
Firstly, we have the date of birth and death for all individuals who recorded these events in the study areas. Consequently, we have the age at death for a significant portion of children, enabling us to utilize this information to ascertain the survival or death of children at different ages such as the first year of life or the first five years of life (which will serve as the dependent variable in the models).
Secondly, we can ascertain the number of live-born brothers and sisters, as well as those who died or survived and were able to assist in the upbringing of subsequent siblings.
In Figure 2, we have analyzed the infant mortality rate in the three study areas based on the number of older brothers and sisters. The results reveal two interesting aspects. Firstly, as previously noted, rural Spain exhibits higher infant mortality, followed by the Azores and, to a lesser extent, northern Portugal. This is likely due to living conditions, as the extreme climate and limited dietary diversity were possibly worse in inland Spain than in areas with a milder climate closer to the coast. Secondly, and even more intriguing for the aim of this study, the infant mortality rate decreases in all areas with an increase in both the number of older brothers and sisters. This serves as a starting point for subsequent regression analysis.
Infant mortality rate (death before the first year of life) by the region of residence and number of older siblings at birth, birth cohorts 1750–1910.
To control for the effect of interbirth intervals, we have also calculated the distance between the birth of the individual under study and that of the previous child (or the marriage in the case of the first child). This may provide insights into family customs and sexual parental patterns and their potential effects. Additionally, we have information on the age of both the mother and the father. On the one hand, advanced ages may be associated with greater life experience, which could favor the baby’s survival. On the other hand, advanced maternal age at the child’s birth may be linked to greater health problems for both the mother and the baby. We have also included a variable indicating whether the individual is a twin, as twins in the preindustrial world typically were born with lower birth weight and weaker health compared to singletons, which had an impact on their survival.
Lastly, we have paid attention to the birth period, dividing our study period into four sub-periods: 1750–1799, 1800–1849, 1850–1879, and 1880–1910. It ' important to note that observations for Fafe begin from 1850 onward. Additionally, we have also differentiated by birth area (between rural Spain, northern Portugal, and the Azores) because, as mentioned, there are geographical, climatic, and mortality differences that could influence the results. In Table 1, we can observe a description of the variables analyzed in this paper for each of the study areas. This provides us with an overview of the distribution of the variables.
Description of the Main Variables, 1750–1910.
Source: Parish and municipal registers.
Note: It should be noted that the reported number of children refers to the total number of registered births. In some cases, these children may have left the study area with their families and had no further recorded events, which could result in their exclusion from the statistical regression analyses, depending on the criteria applied in each sample selection.
Methods
In this study, we will examine how the number of siblings affected the likelihood of survival for newborns. We estimated 8 regression tables, each containing between 4 and 6 probit models.
43
All models can be denoted as follows:
Results
We will utilize probit regression models to assess the probability of newborn mortality as the dependent variable. As independent variables, we will introduce: 1. Total number of brothers born into the family before the birth of the reference individual, 2. Total number of sisters, 45 3. Total number of older brothers alive at the birth of the reference individual who survived beyond 5 years of age (in other words, those who were present in the household at the time of the baby’s birth and were 5 years old or older), 4. Total number of older sisters alive under the same conditions, 46 5. Number of days between the birth of the previous child and the birth date of the individual under study, 6. Mother’s age at the birth of the individual, 7. Father’s age, 8. Whether the individual was born in a multiple birth, 9. Birth period. The aim, in all cases, is to examine the effect of living brothers and sisters on the likelihood of survival for a new sibling, controlling for the total number of brothers and sisters born.
In Table 2, we have outlined the first six regression models. The first three correspond to the dependent variable associated with dying in the first year of life (infant mortality), and the last three correspond to dying in the first five years of life (childhood mortality). In all models, we have introduced all the independent variables mentioned in the previous paragraph (and sex), except for those related to the number of siblings, which we have included according to the model. Thus, in models (1) and (4), we have only introduced the total number of brothers and sisters born before the birth of the reference individual (regardless of whether they were alive or not, considering the high childhood mortality before the demographic transition). In models (2) and (5), we have instead included the total number of living brothers and sisters (older siblings) at the time of the birth of the reference individual. Meanwhile, in models (3) and (6), we have included all variables to assess the combined effect. 47
Probability of Having Died According to the Number of Brothers and Sisters, Birth Cohorts 1750–1910.
Source: Parish and municipal registers.
Notes: se denotes the robust standard error. * Statistical significance at the 10% level, ** at the 5% level. *** at the 1% level.
From Table 2, let ' first analyze the results for the total number of living brothers and sisters. In all cases, what we observe is that the total number of siblings is negatively associated with survival. That is, a higher number of siblings is linked to increased likelihood of mortality, and it increases with the total number. For example, in the case of model (3) on infant mortality, the presence of 0 brothers is the reference, the presence of 1–2 brothers has a change in z compared to the reference category of 0.385, the presence of 3–5 brothers has 0.625, and the presence of 6 or more brothers has 0.988. This is the typical pattern in all cases. That is, a higher number of born brothers leads to higher chances of mortality. The same applies to sisters. This effect is possibly the consequence of families with more children experiencing higher mortality due to less secure behaviors, less hygienic environments, or biological/genetic differences (which gave them the opportunity to have new births more quickly).
If we examine the number of living brothers and sisters, we find the opposite effect. The number of living siblings sharing the household with the newborn had a clearly positive effect on the survival of the new siblings (and was highly significant as well). Moreover, we observe the same increasing effect as their number rises. For instance, in model (6) for childhood mortality, having zero older sisters alive is the reference, having 1–2 sisters presents a change in z compared to the reference category of −0.259, 3–5 sisters −0.817, and 6 or more sisters −1.407. In other words, there was a strong increase with the addition of sisters. Similarly, in the case of brothers, we find an even greater impact (the corresponding values would be −0.417, −0.963, and −1.577). Therefore, the number of brothers may be even more important in the survival of new siblings (which we will discuss in the “Discussion” section). Lastly, the combination of the effect of born siblings and older living siblings does not change anything but reinforces the results, as they all appear to be significant in the same direction.
The results from all these models in Table 2 suggest a very slight positive effect on survival for the distance in days from the birth of the previous child. However, we must consider that the death of the previous child quickly favors the possibility of having a new child in a shorter time frame (shorter intervals) and perhaps greater intensity in care to avoid repeating the problem. The mother’s age at the birth of the child, especially at older ages, proves to be a decisive factor in favor of survival in all cases (we will see this especially in the case of Portugal). Thus, children born to mothers over 32.5 years old, and especially over forty years old, had over a 25 percent higher chance of survival. Perhaps this is due to the experience gained by the mother with previous children. However, the father’s age does not seem to be significant.
In general, being born in a multiple birth is associated with higher chances of mortality, especially in the first few months of life. This is possibly due to the greater vulnerability of twins compared to singletons in a period without modern incubators. Finally, the time period also shows to have an influence on the chances of mortality, although this variable will be studied in greater depth later on.
In Table 3, we have replicated models (3) and (5) from Table 2 (for infant mortality and childhood mortality) based on the newborn’s gender. Models (1) and (2) in Table 3 are for males, and models (3) and (4) are for females. The rest of the variables remain the same as in the previous case. The results again highlight the increasing importance of the number of brothers and sisters alive on the survival of newborns. In all cases, we find similar results to those in the previous table, so we will only focus on the differences according to the child’s gender. When comparing the results for males and females, it stands out that while the presence of brothers does not seem to have a strongly differential effect, the presence of sisters could be linked to higher survival rates of male newborns (showing a 10–20 percent more intense role of sisters). This could be a consequence of the greater vulnerability of male babies. If sisters had a more prominent role in family care, they might have been especially beneficial for male babies due to their greater fragility. The greater fragility of males is clearly observed with the variable “birth period”; over time, there is a greater reduction in the chances of male infants dying compared to females, which would result from socioeconomic and health improvements that were more beneficial for vulnerable individuals. Finally, from Table 3, the variable “being twin” should be highlighted, which reflects a significant increase in the chances of dying in the first year of life (especially in the first month of life) if the newborn is female. This would be related to the former. Girls are stronger at birth, but twins, by sharing resources, tend to have worse health status at birth and during breastfeeding. 48
Probability of Having Died According to Sex of the Child, Birth Cohorts 1750–1910.
Source: Parish and municipal registers.
Notes: se denotes the robust standard error. * Statistical significance at the 10% level, ** at the 5% level. *** at the 1% level.
In Table 4, we have replicated the same model as in Table 2 (models 3 and 6) but with the dependent variable being death at different age intervals: 1. In the first 28 days, 2. Between 29 and 365 days, 3. From the first year to the fourth, 4. From the fifth year to the ninth, and 5. Before the age of ten. The aim is to understand the effect of living brothers and sisters at different ages of the individual. The results follow the previous trends but are especially interesting. We again find that families with many children are associated with higher chances of newborns dying, while the number of older brothers and sisters is associated with increased survival chances. We will focus on discussing the results for older siblings. It is clear that in practically all models, there is a strongly significant positive effect between the number of older siblings and higher survival rates. However, this effect is clearly more intense (even double) the younger the newborn. For example, having 3–5 living sisters is associated with a change in z compared to the reference category of −0.450 in the first 28 days, −0.245 between the first and twelfth month of life, and −0.174 between 1–4 years of age. This is similarly reflected in all categories. Of course, as always, the intensity of the positive effect of siblings increases as their number increases. In almost all observations for children under five years old, we see the beneficial effect of living siblings on survival. However, the results from age five onwards (from 5 to 9 years old) are particularly striking. In almost no case do we find significant results (only a beneficial effect of having three to five living sisters with a 90 percent significance). Therefore, the results show us that available siblings were very important in the early stages of the newborn’s life when the baby requires more care and supervision. Other interesting results from Table 3 include being born a twin (which increases the probability of dying, but only between the first month and the twelfth month of life), the birth period (higher probability of dying in the older periods), and the birth area (with Portuguese areas showing lower probabilities of dying than rural Spain).
Probability of Having Died Before 28 Days, After 28 Days and 1 Year, Between 1 and 4 Years, Between 5 and 9 Years, or Before 10 Years, Birth Cohorts 1750–1910.
Source: Parish and municipal registers.
Notes: se denotes the robust standard error. * Statistical significance at the 10% level, ** at the 5% level. *** at the 1% level.
In Table 5, we have replicated the model from Table 4 for infant mortality and childhood mortality, separating the analysis by the different available samples: rural Spain, northern Portugal, and the Azores. As expected, the results for all areas are similar to those obtained previously (which we will discuss later in the “Discussion” section), so we will focus on the differences between the three regions. In general, the variables tend to exhibit greater intensity (mainly in factors related to increases in mortality) in rural Spain than in the Portuguese samples, which may be related to the higher childhood mortality rate in rural Spain due to drier conditions, more extreme weather, and limited dietary variety (focused on wheat). The beneficial effect of older siblings is particularly notable again in the sample available for rural Spain, where the effect of the presence of brothers are approximately 20–25 percent higher than those for northern Portugal and the Azores. However, this changes for childhood mortality in the case of the presence of three or more brothers (categories 3–5 and 6 or more). In that case, the intensity of the effect is slightly higher in the Azores. Therefore, a high number of brothers had a positive effect in all areas studied, but especially in the Azores. Regarding sisters, the results we find are similar, although the positive effect on survival is slightly lower than in the case of brothers (generally between 10 percent and 20 percent lower). We again find that the beneficial effect is particularly intense in the case of rural Spain, but, as before, the Azores stand out again in the case of a high number of older sisters (three or more sisters) in childhood mortality.
Probability of Having Died According to the Region of Living, Birth Cohorts 1750–1910.
Source: Parish and municipal registers.
Notes: se denotes the robust standard error. * Statistical significance at the 10% level, ** at the 5% level. *** at the 1% level.
In Table 6, we have replicated the same models as in Table 4 (including Area as an independent variable) for the subperiods 1750–1799, 1800–1849, 1850–1879, and 1880–1910, as well as for the entire period (to enable comparisons), but exclusively for infant mortality (childhood mortality has been analyzed in Table 7). The aim of this analysis is to understand if there was a temporal evolution in the importance of older siblings in the study areas. The results are consistent with those mentioned earlier, so we will focus on what the temporal evolution adds, especially to the relationship between the number of older siblings and survival.
Probability of Having Died Before One Year Old According to the Period, Birth Cohorts 1750–1910.
Source: Parish and municipal registers.
Notes: se denotes the robust standard error. * Statistical significance at the 10% level, ** at the 5% level. *** at the 1% level.
Probability of Having Died Before Five Years Old According to the Period, Birth Cohorts 1750–1910.
Source: Parish and municipal registers.
Notes: se denotes the robust standard error. * Statistical significance at the 10% level, ** at the 5% level. *** at the 1% level.
What we can observe in the results is that in all sibling variables, generally, there is greater intensity in the older periods which tends to decrease over time. Of course, a high number of siblings born is related to higher mortality, while a high number of older siblings is associated with greater chances of survival, as we discussed before. For example, the change in z compared to the reference category (significant at 99 percent confidence level in all cases) of having three to five older brothers was −0.787 in 1750–1799, −0.642 in 1800–1849, −0.529 in 1850–1879, and −0.507 in 1880–1910. This example reveals a strong importance of the number of older brothers, but at the same time, a reduction in this importance. This reduction is possibly because other factors related to the improvement of socioeconomic and health levels in these societies came into play. Additionally, this temporal evolution is more pronounced in the case of the number of brothers than in the number of sisters, where we even see the highest intensities in the second period (1800–1849). However, we should interpret these results with caution as both are clearly periods prior to industrialization and demographic transition in the study area. In any case, the positive relationship between the number of older siblings and the significant increase in survival chances is confirmed for all periods.
Finally, it ' worth highlighting the result of the “Area” variable, where we observe negative and significant values in the case of northern Portugal and, especially, the Azores compared to rural Spain. This further confirms that mortality rates were possibly higher in the rural agrarian Spanish area analyzed.
Table 7 replicates the models from Table 6 by periods, but with childhood mortality as the dependent variable (instead of infant mortality). The results found are very similar to those in the previous table, albeit with slightly lower coefficient intensity (generally approximately between 10 percent and 20 percent lower). Therefore, we focus on analyzing the effects of the number of older brothers and sisters on newborn survival.
The results once again find a negative relationship between the probabilities of death and the number of living brothers and sisters for all periods. Again, the effect is especially intense in the oldest period and with a decreasing intensity, especially in the case of brothers. As mentioned, this could be the effect of intense childhood mortality in the pre-industrial period and socioeconomic and health improvements throughout the nineteenth century. Similarly to the previous case, childhood mortality in rural Spain appears to have been more intense than in the other Portuguese areas studied, possibly as a consequence of the region’s harsh climate and limited dietary diversity.
In Table 8, we have focused on older siblings but taken it a step further by classifying them not only by sex but also by their age difference with the newborn. The aim is to determine whether the sibling’s age, in addition to their sex, has a differential effect on newborn survival. Given the reduced sample size, we have opted for three separate age groups for brothers and sisters: from zero to four years (little contribution to caregiving), from five to nine years (may play a significant role in assisting the family), and from ten to nineteen years (almost adult individuals who can almost replace the role of parents). Due to the scarcity of observations in some categories, we have decided to use continuous variables, that is, to examine whether there is a relationship between the increase in siblings in a particular category of sex and age and their effect on the increase (or lack thereof) of newborn survival chances (instead of a categorical variable as in previous tables). As in previous cases, we have differentiated between infant mortality (models (1), (2), and (3)) and childhood mortality (models (4), (5), and (6)). Additionally, in models (1) and (4), we have included all cases, but in models (2) and (5), we have focused on whether the individual only has brothers (regardless of their age), and in models (3) and (6), we have focused on whether the individual only has sisters (regardless of their age). Since these are complex models with fewer observations, we have introduced fewer independent variables to ensure larger sample availability. Specifically, we have included the age of the mother and father at the birth of the analyzed individual and the birth period as independent variables.
Probability of Having Died According to the Number of Brothers and Sisters and their Ages, Birth Cohorts 1750–1910.
Source: Parish and municipal registers.
Notes: se denotes the robust standard error. * Statistical significance at the 10% level, ** at the 5% level. *** at the 1% level.
The results from Table 8 largely align with those obtained previously, emphasizing the importance of older siblings in newborn survival. The models indicate that younger brothers (under ten years old) are particularly significant. Brothers aged ten to nineteen only emerge as significant in the case of childhood mortality for the complete model (model (4)), with a very small change in z compared to the reference category of only −0.021. This could be because brothers aged ten and above had already begun to fully participate in the workforce to support the family. Siblings over the age of ten might not even reside within the family unit. Therefore, during the crucial early years where care is especially important, working brothers had a smaller impact.
In the case of sisters, the results do not vary significantly from those of brothers. Sisters have a positive impact on newborn survival. However, the model suggests that their impact, for those under ten years old, would be slightly lower than that of brothers, which is surprising given the traditional role of women from a young age in caring for family members. Additionally, from the age of ten onwards, we only find significant value (favorable to survival) when analyzing newborns who only had sisters. This could be due to a situation similar to that explained for brothers (workforce participation) or due to imperfections in the analysis. Therefore, in the next table, we have reconsidered the analysis from a complementary perspective. In Table 7, it is also noteworthy the positive effect of increasing the mother’s age at the time of the newborn’s birth in all cases. This result is possibly linked to the positive effect of the mother’s greater experience in caring for her children.
In Table 9, we have replicated Model 4 from Table 8 (solely for childhood mortality) but considering only individuals who have brothers or sisters in a single age group. The aim of this analysis, despite the inherent biases, is to delve into how each group of siblings affects survival. Therefore, in interpreting the results, we will focus on the variables of the number of surviving siblings, which have changed compared to the previous table.
Probability of Having Died According to the Number of Brothers and Sisters and their Ages (Just Some Siblings in a Corresponding Age by Model), Birth Cohorts 1750–1910.
Source: Parish and municipal registers.
Notes: se denotes the robust standard error. * Statistical significance at the 10% level, ** at the 5% level. *** at the 1% level.
In Table 9, we observe that, in all cases, the presence of siblings favors the survival of the newborn, but with varying intensities. While the presence of very young sisters has an almost negligible effect, having older sisters multiplies the survival effect, becoming a decisive factor when sisters are between ten and nineteen years old. They were already mature enough to take care of their younger siblings, complementing or substituting mothers in case of need due to work or medical reasons. The surprising results from Table 8 (the previous one) are nuanced with this new analysis. However, for males, the highest intensity of the effect (−1.553, the highest found in the models of Table 8) occurs when brothers are between 5 and 9 years old. Perhaps this is because brothers between ten and nineteen years old were engaged in exhausting workdays, thus contributing positively to the family’s income but having a less intense involvement in newborn care. Siblings over the age of ten might not even reside within the family unit. All these matters will be discussed in the following section.
Discussion
Our findings confirm that the number of older siblings had a significant influence on the survival probabilities of newborns, especially at younger ages (losing their effect after five years), for those born in the earliest decades of the analysis, for both sexes, and particularly for areas with higher mortality rates (in our case, rural inland Spain). The positive results obtained may be attributed to the fact that brothers and sisters could assist in childcare to varying degrees depending on gender and age. In fact, in Tables 8 and 9, we have reconfirmed the importance of gender and age (with older sisters being crucial for caregiving).
Another alternative hypothesis, which is not fully confirmed for our study area, would be that families with a greater number of living siblings were those where parents implemented more successful hygiene and care measures, so we would actually be observing how the previous success of parents is linked to the survival of the newborn. However, the differences found based on the gender and age of the siblings seem to indicate that the older siblings themselves also played a role in the survival of the newborn.
Based on the explanation provided in the previous paragraph, our results seem to confirm the predominance of cooperative breeding. 49 The number of brothers and sisters clearly has a positive effect on the survival of the newborn, which moreover increases with the total number. However, we cannot dismiss the existence of a resource dilution effect. 50 In fact, the total number of siblings born shows a strong significance in the opposite direction (potentially increasing mortality). Nonetheless, the effect of cooperative breeding could be compensating for the resource dilution effect and even surpassing it, demonstrating that the positive effect on survival outweighs the negative one.
A surprising finding is the difference in effect between the presence of older brothers and older sisters. In almost all cases, the results show that the positive effect of brothers on survival is slightly higher (between 10 percent and 20 percent) than the positive effect of sisters. However, this has been nuanced with the results of siblings according to their age. In any case, this could be because brothers not only contribute to caregiving but also, from a very young age, engage in small paid jobs that could help improve the family budget. In fact, Tables 8 and 9 have highlighted the particularly significant role of brothers before the age of 10. That is, when they have not yet fully entered the labor market and still maintain a balance between their household cooperation and economic contribution. Table 8 has underscored the important role of older sisters. Nonetheless, further research is needed to delve into the differences in the effect of older brothers and sisters. 51
Our results for the prefertility transition period seem to reinforce the idea of the importance of family ties in Southern Europe (in this case, the Iberian Peninsula and the Azores), where both nuclear and extended families appear to have strong connections among them, also in favoring the survival of newborns. 52 Due to the lack of data availability for these three study areas, an analysis of the role of siblings introduced to a greater extent in the twentieth century and spanning the early stages of the fertility transition, when parents are reducing their fertility, remains pending.
It is essential to acknowledge the limitations of this analysis, particularly the potential impact of scarring effects—situations in which the death of older siblings may adversely affect the survival prospects of younger children. In high-mortality contexts, children born into families that have already experienced multiple child losses may face elevated mortality risks. These risks may stem not only from biological or environmental vulnerabilities but also from broader structural disadvantages within the household. 53 Although scarring remains a relevant theoretical and empirical concern, its effect may be attenuated in pre-transitional populations, where fertility patterns tend to be less selective and more uniformly distributed across families. Nonetheless, it represents a significant limitation that must be considered, especially in studies seeking to disentangle the causal relationships between fertility and child mortality.
Another limitation to consider is the exceptional use within the same model of variables that account for the number of siblings, which may include different characteristics such as a specific age or simply the fact of having been born. It is evident that in households with many births, there may also be more surviving children, leading to a correlation between the variables and a potential issue of multicollinearity in some models that must be taken into account. While none of the correlations between the variables in any of the models exceed 0.500, the results from the more complex models provide valuable information but should be interpreted with caution.
Conclusions
The results obtained have confirmed that the presence of older siblings clearly favored the survival and well-being of the new siblings in all three studied areas (rural Spain, semi-urban and rural Portugal, and The Azores). However, the results are not uniform across the three areas, with a greater intensity observed in rural Spain compared to the other two areas (though it should be noted that infant mortality was also higher in rural Spain). The differences in intensity could be attributed to varying standards of living in the studied areas (with levels closer to subsistence in rural Spain) or to differences in sibling behavior based on the customs and traditions of each area. Nevertheless, the strong family ties that bind families in Southern Europe could explain the significant role played by siblings. 54
It has been noted that the survival of newborns is especially influenced by older sisters (especially those aged ten or older), who had more knowledge and could provide better care.
Footnotes
Acknowledgments
This article forms part of the framework of the research project of the Ministry of Science, Innovation and Universities of the Spanish Government PID2022-138886NB-I00, and the Consolidated Research Group S55_23R of the Government of Aragon.
Author Contributions
Conception or design of the work: Norberta, Francisco. Acquisition of the data: Norberta, Milene, Francisco. Analysis and interpretation of data: Milene, Francisco.
Availability of Data and Materials
All materials used are available on request. In the case of Spanish data, they are the property of the corresponding author and are obtained with a justified email (fmarcog@unizar.es). In the case of Portuguese data, they are managed by the Repositório Genealógico Nacional and an email must be sent to them requesting access to the data (geral@csarmento.uminho.pt). Codes can be requested from the corresponding author.
Consent for Publication
All authors and their institutions agree to publication.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethics Approval and Consent to Participate
This article uses historical data with full ethical and data protection permissions.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Ministry of Science, Innovation and Universities of the Spanish Government PID2022-138886NB-I00 and Consolidated Research Group S55_23R of the Government of Aragon.
