Cross-National Patterns of Intergenerational Continuities in Childbearing in Developed Countries
发达国家生育的跨国家代际延续模式

Department of Social Policy, London School of Economics, London, United Kingdom
英国伦敦，伦敦经济学院社会政策系

Fertility patterns of parents and their children tend to be positively correlated. Results comparing the completed fertility of successive generations for both historical and more recent populations in a number of now-developed countries up to the later part of the twentieth century are presented in Murphy (1999). This article updates and considerably broadens the geographic coverage to assess the magnitude of intergenerational continuities in childbearing in a number of developed and middle-income societies and, in particular, to establish whether earlier trends have continued into the most recent period. It also considers the relationship with macro-level factors and how much the relationship changes when socioeconomic controls are included.
父母及其子女的生育模式往往呈正相关。Murphy（1999）对多个发达国家直至20世纪后期的历史和近期人口中连续几代人的完整生育率进行了比较。本文更新并大大扩展了地理覆盖范围，以评估多个发达国家和中等收入社会中生育的代际连续性，特别是确定早期趋势是否延续到最近时期。本文还考虑了与宏观因素的关系，以及当纳入社会经济控制时，这种关系会发生多大变化。

The association between parents’ fertility and their children’s fertility has been a subject of scientific interest since the late nineteenth century. Pearson, Lee, and BramleyMoore (1899) investigated the extent of fertility correlations among branches of the British upper classes, one of the few sources of such data at that time. They found the highest correlation coefficient of 0.21 for the fertility of 1,000 mother-daughter pairs from the peerage and baronetage when both had marriages that had lasted for at least 15 years. The lowest reported correlation coefficient for women was 0.04 for 4,418 pairs of mothers
自19世纪末以来，父母生育能力与子女生育能力之间的关联一直是科学界关注的话题。Pearson、Lee和BramleyMoore（1899年）调查了英国上层阶级各分支之间的生育能力关联程度，这是当时此类数据的少数来源之一。他们发现，在1000对来自贵族和男爵阶层的母女中，如果两人的婚姻都持续了至少15年，则生育能力的相关系数最高，为0.21。据报道，在4418对母亲中，女性的最低相关系数为0.04

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The moral rights of the named author(s) have been asserted.
本文为开放获取文章，根据知识共享署名许可协议（http://creativecommons.org/licenses/by/3.0/）进行分发，允许在适当引用原作的前提下，不受限制地使用、分发和复制到任何媒介。署名作者的精神权利得到保护。

Thanks are due to the United Nations Economic Commission for Europe and national statistical agencies for access to FFS and GGP data, to the GESIS Data Archive for access to ISSP data, and to the ESRC Data Archive for access to the Economic and Social Research Council-funded UNSOC.
感谢联合国欧洲经济委员会和国家统计机构提供FFS和GGP数据，感谢GESIS数据档案馆提供ISSP数据，感谢ESRC数据档案馆提供经济和社会研究理事会资助的UNSOC数据。

Address correspondence to Professor Michael Murphy, Department of Social Policy, London School of Economics, Houghton Street, WC2A 2AE, London, UK. E-mail: m.murphy @lse.ac.uk
请致信伦敦经济学院社会政策系迈克尔·墨菲教授，地址：Houghton Street, WC2A 2AE, London, UK。电子邮件：m.murphy @lse.ac.uk
and daughters drawn from a mixture of upper-class groups, with the only restriction being that the daughter should have been married for 15 years or had a spouse who died before 15 years of marriage.
以及来自上层阶级的女儿，唯一的限制是女儿结婚必须满15年，或者配偶在结婚15年之前去世。

Pearson, Lee, and Bramley-Moore (1899:279) assumed that the mechanism of transmission was biological rather than social and that their study populations undertook no deliberate attempt to limit fertility, although they did suggest that fertility control would tend to depress correlations. Studies of less specialized pretransitional historical populations are rare. Langford and Wilson (1985) analyzed 10,931 English ever-married mother-daughter pairs from the mid-sixteenth to mid-nineteenth centuries. They computed a small intergenerational fertility correlation coefficient of 0.02 and concluded that there was no association between the fertility of mothers and the fertility of daughters. Bocquet-Appel and Jakobi (1993) estimated a similar correlation of 0.015 for fertility of 257 pairs of once-married women in successive generations in a French village in the early part of the nineteenth century. They also rejected the existence of intergenerational transmission of fertility in this pretransitional period. Neel and Schull (1972:345) analyzed a rather later population, but one that did not use birth control-Amish couples who married in the period 1820-1879 in Ohio and Indiana-and they computed slightly higher intergenerational correlations between their fertility and the sibship sizes of wives and husbands of 0.09 and 0.07 , respectively.
Pearson、Lee和Bramley-Moore（1899：279）认为，生育率传递的机制是生物性的，而不是社会性的，他们研究的人群没有刻意限制生育，但他们确实认为生育控制会抑制相关性。对过渡期前不太专业的历史人群的研究很少。Langford和Wilson（1985）分析了10931对16世纪中期至19世纪中期的英国已婚母女。他们计算出代际生育率相关系数为0.02，并得出结论，母亲的生育率和女儿的生育率之间没有关联。Bocquet-Appel和Jakobi（1993）估计，19世纪早期，法国一个村庄中257对曾婚妇女的生育率在连续几代人中的相关系数为0.015。他们还否认了生育率在过渡期前存在代际传递。Neel和Schull（1972：345）分析了相当晚的人口，但没有使用节育措施——1820-1879年间在俄亥俄州和印第安纳州结婚的阿米什夫妇——他们计算出他们的生育率和妻子和丈夫的兄弟姐妹数量之间的代际相关性略高，分别为0.09和0.07。

Recently, biological interest in fertility transmission has revived, in part because of its importance for transmission of rare genetic diseases (Tremblay 1997; Austerlitz and Heyer 1998; Gagnon and Heyer 2001; Helgason et al. 2003). Such studies typically analyze the intergenerational fertility patterns of descendants of an initial population over extended periods of time. Based on analyses of two Québec populations, Gagnon and Heyer (2001) also concluded that family size does not have a tendency to run in families. However, much higher correlation coefficients of $0.31(p<{10}^{-6})$$0.31\left(p<{10}^{-6}\right)$0.31(p < 10^(-6))0.31\left(p<10^{-6}\right) for completed fertility between couples and their sons and of $0.23(p<.001)$$0.23(p<.001)$0.23(p < .001)0.23(p<.001) between couples and their daughters were found for a set of 161 genealogies in the Hutterite archetypal natural fertility population (Pluzhnikov et al. 2007). These results cover an extended time period, and so they were adjusted to control for time trend. They show a high correlation in a well-documented and socioeconomically and behaviorally homogeneous population.
最近，人们对生育力传递的兴趣又重新燃起，部分原因是生育力传递对于罕见遗传疾病的重要性（Tremblay 1997；Austerlitz和Heyer 1998；Gagnon和Heyer 2001；Helgason等人2003）。此类研究通常分析初始群体后代在较长一段时间内的代际生育力模式。根据对魁北克两个群体的分析，Gagnon和Heyer（2001）还得出结论，家庭规模不会在家庭中传承。然而，在Hutterite原型自然生育力群体中，对161个家谱的分析发现，夫妻及其儿子之间的完整生育力相关系数 $0.31(p<{10}^{-6})$$0.31\left(p<{10}^{-6}\right)$0.31(p < 10^(-6))0.31\left(p<10^{-6}\right) 以及夫妻及其女儿之间的相关系数 $0.23(p<.001)$$0.23(p<.001)$0.23(p < .001)0.23(p<.001) 要高得多（Pluzhnikov等人2007）。这些结果涵盖的时间跨度较长，因此进行了调整以控制时间趋势。结果表明，在记录完整且社会经济与行为同质的人群中，相关性很高。

Gagnon and Heyer (2001) found a much stronger association between the effective family sizes of “settled populations” in Québec-that is, populations in which both successive generations remained in the study area and survived to be involved in procreation-a result that they attributed mainly to sociodemographic factors rather than biological transmission (Heyer, Sibert, and Austerlitz 2005). In contemporary developed societies, the difference between numbers of total and living children is small, so we are unable to address this issue, although for developing countries, where mortality levels are higher, the strength of the intergenerational associations is similar (Murphy 2012).
加尼翁和海耶（2001）发现，魁北克“定居人口”（即两代人都在研究区域生活并生育后代的人口）的有效家庭规模之间存在更紧密的联系，他们认为这一结果主要归因于社会人口因素，而非生物遗传（海耶、西伯特和奥斯特利茨，2005）。在当代发达社会中，总生育数和存活子女数之间的差异很小，因此我们无法解决这个问题，但对于死亡率较高的发展中国家，代际关联的强度是相似的（墨菲，2012）。

Studies concerned with biological reproductive performance or disease transmission are primarily interested in those individuals who are engaged in such activity, and many studies including those mentioned earlier are confined to married (sometimes further restricted to individuals in long-term, intact marriages) or parous individuals. The results are usually not representative of the overall population concerned, since, for example, they exclude immigrants who arrive during the analysis period. While findings from these pretransitional populations need to be interpreted carefully because of the selected nature
有关生物繁殖性能或疾病传播的研究主要关注从事此类活动的个体，包括前面提到的许多研究都仅限于已婚（有时进一步限制为长期、完整的婚姻）或有生育经验的个体。研究结果通常不能代表相关总体人口，因为例如它们排除了在分析期间到达的移民。由于这些过渡前人口的选择性，需要谨慎解释研究结果
of samples used, such populations tend to show positive but modest and statistically nonsignificant intergenerational fertility correlations that are usually only slightly larger than the early values of Langford and Wilson (1985) and Bocquet-Appel and Jakobi (1993).
样本的使用表明，这些人群的生育率呈现出积极但温和且在统计学上不显著的代际相关性，通常仅略高于Langford和Wilson（1985）以及Bocquet-Appel和Jakobi（1993）的早期数值。

Anderton et al. (1987) used data from the Utah Population Database to show a positive association between the fertility of women born in the period 1830 to 1870 and the fertility of their daughters, but the researchers did not present correlation coefficients. However, Jennings, Sullivan, and Hacker (2012) used the same data source and showed that the correlation coefficients for the fertility of women born between 1840 and 1899 and the fertility of their mothers and mothers-in-law (confined to those who were still married at age 45 in nonpolygamous unions) were 0.085 and 0.055 , respectively. Studies of later populations often show higher values; Madrigal, Relethford, and Crawford (2003) found a correlation of 0.17 between mothers’ and daughters’ fertility in Ireland around the start of the twentieth century. However, as with many studies, this work was based on the comparison of successive generations of married women, so selection effects could be important. In this population, a substantial fraction of babies did not survive to adulthood, very high proportions of women never married, and emigration was widespread. If children from large families were less likely to have survived to adulthood, to be married, or to remain in Ireland, overall intergenerational correlations would be expected to be lower. Reher, Ortega, and Sanz-Gimeno’s (2008) study of the Spanish town of Aranjuez in the period 1871-1970 was able to look at time trends. The correlation between fertility of mothers and fertility of daughters was 0.15 in the period 1871-1970, 0.19 in the period 1891-1910, and 0.14 in the period 1923-1945. These differences are not statistically significant, suggesting the persistence of similar levels of intergenerational continuities. Bresard (1950) used information on the number of children of the father and both grandfathers to examine the association between respondents’ and their fathers’ and mothers’ number of siblings for a representative sample of 3,000 French men aged 18 to 50 in 1948 in a study that covers a similar childbearing period as that used by Reher, Ortega, and Sanz-Gimeno (2008). For the occupational group of farmers and peasants, who accounted for 34 percent of the sample, the correlation coefficient for the respondent’s number of siblings and the maternal grandfather’s number of children was 0.27 ; the correlation coefficient dropped to 0.24 for paternal grandfathers. These correlations are higher than those reported in earlier studies.
Anderton等人（1987）使用犹他州人口数据库的数据，展示了1830年至1870年出生的女性生育率与其女儿生育率之间的正相关关系，但研究人员并未给出相关系数。然而，Jennings、Sullivan和Hacker（2012）使用相同的数据源，并展示了1840年至1899年出生的女性生育率与其母亲和婆婆（限于45岁时仍保持婚姻关系的非一夫多妻制婚姻中的女性）生育率之间的相关系数分别为0.085和0.055。对后来人口的研究通常显示出更高的数值；Madrigal、Relethford和Crawford（2003）发现，在20世纪初左右，爱尔兰母亲和女儿的生育率之间的相关系数为0.17。然而，与许多研究一样，这项工作是基于对连续几代已婚女性的比较，因此选择效应可能很重要。在这个人口中，相当一部分婴儿没有活到成年，很大比例的女性从未结婚，移民现象也很普遍。如果大家庭的孩子不太可能活到成年、结婚或留在爱尔兰，那么总体的代际相关性预计会降低。Reher、Ortega和Sanz-Gimeno（2008）对1871年至1970年西班牙阿兰胡埃斯镇的研究能够观察到时间趋势。在1871年至1970年期间， Bresard（1950）利用父亲和祖父的子女人数信息，研究了1948年3000名18至50岁法国男性的代表性样本中，受访者与其父亲和母亲兄弟姐妹数量之间的关联，该研究涵盖了与Reher、Ortega和Sanz-Gimeno（2008）所使用的生育期相似的生育期。对于占样本34%的农民和农民职业群体，受访者的兄弟姐妹数量与外祖父的子女人数之间的相关系数为0.27；与外祖父的相关系数下降到0.24。这些相关系数高于早期研究中报告的相关系数。

For a slightly later period of childbearing of the second generation, around the 1920s and 1930s, studies in Britain by Berent (1953) and in the United States by Kantner and Potter (1954) reported intergenerational correlations between mothers and daughters of 0.19 and 0.11 , respectively. Both studies sampled surveys, although neither was completely random; Berent’s (1953) results were based on 1,377 couples who were in their first marriage, with that marriage being of at least 15 years duration. He found no statistically significant differences between social classes and birth controllers/noncontrollers, suggesting similarity between these subgroups.
第二代生育时间稍晚，大约在20世纪20年代和30年代，英国贝伦特（1953年）和美国坎特纳和波特（1954年）的研究分别报告了母亲和女儿之间的代际相关性为0.19和0.11。这两项研究都是抽样调查，但都不是完全随机的；贝伦特（1953年）的研究结果基于1377对初婚夫妇，且婚姻持续时间至少为15年。他发现社会阶层与生育控制者/非控制者之间没有统计学上的显著差异，表明这些亚群之间存在相似性。

Kantner and Potter (1954:295) argued that any intergenerational fertility transmission in modern populations is largely determined by the older generation forming notions or instilling preferences about family formation in the younger generation, but they found that such effects were, in any case, trivial. They also argued that with less variation in fertility levels, such effects would be likely to die out. However, Pullum and Wolf (1991) presented data on correlations between the number of living siblings and the number of children for several countries in a kin-modeling study. In the 1985 Canadian General Social Survey of
坎特纳和波特（1954:295）认为，现代人口中任何代际生育力传递在很大程度上取决于老一辈对年轻一代形成观念或灌输家庭组建偏好，但他们发现，这种影响无论如何都是微不足道的。他们还认为，随着生育水平差异的减少，这种影响很可能会消失。然而，普鲁姆和沃尔夫（1991）在亲属模型研究中提供了几个国家在世兄弟姐妹数量与子女数量之间相关性的数据。1985年加拿大一般社会调查