Three eras of young adult home leaving in twentieth-century America (1).

This article divides the history of coming of age in the U.S. in the twentieth century into three distinct time periods: one that ran from 1880 until World War II, one that started in the 1940s and continued until the 1960s, and a third that began by 1970 and was clearly still in evidence in 1990. The story is based on data that recorded whether young people were living with one or both of their parents at the time of each of the decennial censuses of the U.S., from 1880 through 1990. (2) We focus on the ages at which young men and women left home based on census data, and we interpret those levels and trends in terms of the broad social conditions in which they lived. In doing so we see that during the twentieth century, the process by which young people passed from dependence to independence underwent major changes.

The news here is in the history of home-leaving ages up until the time of the second world war. Virtually all earlier studies assert that the age of home-leaving was declining from as early as it could be measured until 1970. Our results say otherwise. From 1880 until 1940 for males and 1950 for females, the age at leaving home did not decline, it rose (see Figure 1). The decline came later. Led by men in 1940, the generation of the Second World War experienced a sharp decline in the age at which they left home, one that continued for everyone until 1960 and for white men and women until 1970. Beginning with 1970, the age of home-leaving rose again, reaching relatively high levels by 1990. These patterns hold generally for white and black Americans, and for males and females, but we will show later that there are significant differences among these four groups. (3) In the 1940s turnaround, for example, all four groups had varying experiences. Among other differences, trends for blacks are less well defined t han for whites, so that the ups and downs are less pronounced.

This research makes use of census data that document relationships within each household. We focus on young people aged 15 to 29, because those are the ages at which most Americans made the transition from living with their parents to living away from their parents. Our approach is shaped by the fact that the census is a static document that records living arrangements on a given date. These data do not capture the transition from living with at least one parent to living away from both parents. Nor do they indicate whether an adult has left home, then returned in later years. Nevertheless, the census allows us to examine the characteristics of persons who are living with a parent or not living with a parent, and then ask whether the characteristics of those who are living at home or away have changed over time, or whether they differ between groups.

The most important of the characteristic differences between young people at home and those living on their own is their age. We say that young people left home "later" in 1940 than in 1920 because the average ages of those still at home in 1940 were greater than those of 1920, after taking into account the overall age distribution of the population. Other important characteristics, in addition to race and sex, are the region in which people lived, whether they were foreign-born or native-born, whether they were single or married, whether they lived on a farm or not, and whether they had migrated out of their region of birth. These are not the only personal attributes that shaped the home-leaving experience, but the reliable census data that exist about these characteristics allow us to tell an important story about the history of home-leaving in the U.S.

The process of home leaving for young adults is closely related to becoming an independent adult, and the motivations that have caused young people through the twentieth century to leave the parental home have varied. Leaving home may occur as a result of parental mortality, increased personal or financial opportunities outside the home, or social expectations or rules requiring that the young person live away from parents. While the wording is over-simplified, we can describe these two groups of processes as either involuntary or voluntary separation of children from their parents' home. Involuntary home-leaving occurs when both parents die or when poverty leads to the disintegration of the family household. In contrast, children may leave their parents' home voluntarily when they choose to attend school or work away from home, join the military, or when they marry or otherwise establish an independent household separate from that of their parents. Over the century, the prevalence of what we describe as inv oluntary home-leaving has decreased, largely due to a decline in the likelihood of becoming an orphan between the ages of 15 to 29. The more complex and socially driven reasons for leaving home, however, are important to the leaving home process throughout the twentieth century.

The history of home-leaving in the U.S. since 1880 largely reflects changes in social conventions, family relationships, and individual characteristics. During the Long First Half of the Twentieth Century, one of the most important factors in the rising age of home-leaving was declining adult mortality, which led to declining rates of orphanhood. High rates of immigration into the U.S. up until the 1920s also influenced the age at which young people left home, because young adults who immigrated by themselves during the peak years were necessarily away from the home of their parents. Finally, social change that led to decreased child labor and increased schooling in the first decades of the century led to later home leaving.

Those trends and experiences changed in the period we call War and Post-War, when the median age of leaving home decreased for all groups that we study here. The driving force behind the male experience of leaving home, we hypothesize, was the military draft. The continued downward trend in the age at leaving home through the 1960 census is probably a result of increased educational opportunities at the college level, particularly as a result of the 0.1. Bill, and of the post-war marriage boom.

Finally, beginning during the 1960s, in the era we call the Baby Boomers and After, the age at leaving home increased. College enrollment continued to be important, in part due to an increasing number of institutions offering more flexible educational plans. Marriage ages rose during this era, and marriage became a less important influence on the home-leaving behavior of women. Young people during this period were both more likely to move away from their parents simply to form a new and independent household, and more likely to return home and live with parents after they had initially left. The result of these social changes was an upward trend in the age of leaving home, one that is more dramatic for females than for males, and one that was dramatic between 1980 and 1990.

Background

The story we tell about the ages of home leaving in the twentieth century U.S. expands upon and retells a story told by other historians. Most of these earlier studies use either individual longitudinal data, such as surveys and matched censuses, or local or national census data. Longitudinal data allow the researcher to trace individuals' transitions, the sequencing of those transitions, and the changes in those individuals' lives over time. (4) Census statistics provide the researcher with information about one time point only. Census data are more limited in their usefulness, because they do not allow the researcher to see the experiences of individuals experiencing the transition from youth to adulthood, but rather only their conditions either before or after the transition. (5)

The most influential work on this subject has been done by Goldscheider and her colleagues. At the national level, they show that for most of the twentieth century, each successive cohort of young people left home at an average age below that of the cohort immediately preceding them. (6) While there has been relatively little research dealing with the period prior to the 1930s, what there has been suggests that the downward trend began even earlier. (7) There is consensus, however, that around 1980 the decline in the age at home-leaving began to reverse. One possible explanation for this change is the increase in the age at marriage since the mid-1960s. (8) A second trend is the progression over time toward a narrower range of ages at which young people left home, reflecting a more homogenized coming-of-age experience.

While there has not been much explicit theorizing--from either a social or economic perspective--about home-leaving, most authors tend to identify four factors that play an important role in determining home leaving, at both the group and individual levels. The first of these factors is marriage timing. Until at least the 1960s, most young women and many young men first left their parents' home at the time they married. Differences in marriage ages between men and women led to the common observation that young women left home at younger ages than did young men. (9) Moreover, as others have shown and as we will show in this article, changes in marriage age through time are strongly aligned with changes in the ages of home leaving (Figure 2). Still, marriage and gender do not always work perfectly together. In his study of late nineteenth century Ireland, Guinnane argues that gender differences often affect the age at which a young person leaves home less than other systematic differences. Men did not always l eave home later, and economic pressures were in many cases more important than gender in predicting the age at leaving home. (10)

Just as marriage and gender do not always work in perfect harmony, nor do trends in marriage and race. The data on marriage in Figure 2 clearly show that the boom in marriages that took place from the later 1940s until the 1970s was a characteristic of the white population much more than of the black population. As recently as 1950, black men and women aged 20 to 24 were more likely to be married than their white counterparts. By 1960 that pattern was reversed, a reversal that has continued despite the decline in percent married that took place after 1970.

By the end of the twentieth century, many young people chose to leave home for pre-marital residential independence. If they can afford to, young adults increasingly prefer to live alone or in non-familial arrangements. (11) Acceptance of the trend, especially for young women, is nonetheless highly variable across social groups. The relatively new phenomenon is generally accepted among highly educated Protestants and Jews, but is more controversial in other communities. (12)

As Guinnane's research suggests, structures of economic opportunity constitute the second major factor determining home leaving ages. Young people stay at home if the household can offer financial opportunities to those who remain. Alternatively, they leave home if the household cannot offer them opportunities, or if better opportunities exist elsewhere. According to Florey and Guest, the strongest predictor that a late-nineteenth century American farm boy would remain home was the likelihood that he would inherit the farm. (13)

Economic opportunity in the community also influenced when a child left home. A young adult who could not find a way to contribute to the family economy while remaining at home might leave to look for work. Florey and Guest suggest that local economic development 'pulled' boys from the farm toward more lucrative work. (14) Although children had been a significant part of the U.S. labor force at the turn of the century, their involvement decreased by the 1940s and i 1950s (Figure 3). Approximately two-thirds to one half as many children aged 15 to 19 were in the labor force in 1940 as had been in 1900. The change was due in part to technological advances in factories and legal reforms requiring children to be in school. New cultural perspectives also contributed to the change. One result was that while middle class parents viewed urban or non-family related work as damaging to young people, they believed that labor involvement in a family farm or small business was morally and physically wholesome. (15)

After the 1950s, the involvement of children in the labor force changed again. Previously, American children had worked as preparation for adult occupations. By the 1960s, however, young workers were increasingly likely to be employed in service jobs, such as pumping gas or serving food, which financed personal consumption but were unrelated to later work. As the U.S. service economy expanded, young adults were more likely to work, but their jobs were less likely to lead to economic self-sufficiency. (6)

The third factor in home leaving is the possibility of alternatives to marriage and employment. Two such alternatives, education and military service, have played important and changing roles in the United States in the twentieth century. Greater access to education has been a national trend in the twentieth century. As more children received primary and secondary educations, entry into the labor force occurred at a later age. The transition to mass college education after the 1940s led to a decreased age at leaving home as financially dependent students lived away from their parents while at school. The increase in college enrollment was due to several factors. First, the number of high school graduates increased due to improved education in the first half of the century. Second, the G.I. Bill allowed World War II and Korean War veterans an affordable undergraduate education. Finally, the increasing number of public junior colleges and two-year institutions in the 1970s allowed more students to attend colle ge, despite financial or scholastic constraints. (17)

After a period of stability until 1900 and from 1910 to 1920 for whites (but not for blacks), the proportion of both males and females aged 15-19 attending school rose steadily, rising by half by 1940-1950, and doubling by 1970 (Figure 4). The lack of change from 1910 to 1920 may be more exaggerated by the data than is appropriate, because the census enumerators instructions for 1910 encouraged them to give an occupation to persons who were employed in family businesses, including farms. (18) The likelihood of people between 15 and 19 attending high school increased dramatically until 1970, with the likelihood that they would live at home.

Participation in voluntary or mandatory military service also determines the age at which a young person leaves home. According to Goldscheider and Goldscheider, the cohort of men who turned 18 during World War II left home two years earlier than the cohort who had turned eighteen during the depression. More than half of the men in the World War II cohort first left home as a direct result of a desire to enter military service. (19) Modell reports that military service was important for the generation that fought in World War II and the Korean War, but after that, participation became less common. (20) Goldscheider and Goldscheider, however, argue that even after World War II, participation in military service was an important determinant of the age at which men left home. (21) Our results (Figure 5) confirm that finding by showing that the impact of military service for males persisted at least until the cohort aged 15-29 in 1970.

The fourth major factor influencing patterns of home leaving is migration, which includes internal migration within the U.S. as well as international immigration. The widespread twentieth century migration from rural settings to cities and suburbs led to homogenization in the age at leaving home. The migration from rural areas to urban areas that occurred in the 1920s was led by young people leaving family farms to find work. (22) Stevens describes this as a process in which "the country moved from an agricultural society where young men from farm families were the last to leave home, to an urban society in which all occupational classes experience this transition at the same time." (23)

International migration has also been instrumental in shaping American society (Figure 6). Many of the young people who arrived in the U.S. in the last century did so without their parents. Immigrants who came to the U.S. in the late nineteenth and early twentieth century brought their own cultural practices and values. As some ethnic differences declined over the twentieth century, however, the age at which children left home became more homogenized for all Americans. (24)0

A final influence on the age at which young people leave home is whether their parents are still living. We define "living at home" as being in the same household as at least one parent. Orphaned young people appear in our results to have left home, although they did not make the choice to do so. The trend in orphanhood is difficult for us to describe because the census sample data that would produce precise results do not specify whether a person is orphaned or not. Instead of making direct computation of orphanhood rates from the census sample data, we have instead used standard U.S. life tables to estimate the likelihood of a both parents of an individual dying by a specified age, given average ages of childbearing for those parents. (25) The estimates in Figure 7 show, for example, that the estimated probability of orphanhood for a twenty-year old child in 1900 was 0.0.37, meaning that slightly fewer than four percent of twenty-year olds were orphans. These probabilities declined to a negligible level by the middle of the twentieth century.

The estimated probability of orphanhood in the United States varies dramatically by age (as would be expected), and declines steadily from 1880 to 1990. In the most extreme case, a person aged 30 had a 14.7% chance of being orphaned in 1880, which fell to twelve percent in 1900, to less than ten percent in 1920, and to less than eight percent in 1920. The same declines are evident for orphanhood at ages 15, 20, and 25, although they start at lower levels. Even in 1880, a 15-year hold had only a one-in-40 chance of being orphaned, which declined by 1940 to less than one percent.

The conclusion that we draw from our estimates of orphanhood is that the steady decline of adult mortality in the U.S. from 1880 to 1990 had reduced the role of orphanhood in determining the ages at which young people have left home, with a decline by half taking place between 1880 and 1940. The rising overall age of home-leaving from 1880 to 1940 is consistent with this evdence. What also needs to be said is that at the relatively young ages at which most people left their parents' homes in the late nineteenth century, the role of orphanhood was a modest factor.

Young adults in the U.S. experienced changing living conditions between 1880 and 1990, a result of evolving patterns of health, economic life, and the community's expectation of them and their families. These changed conditions influenced the length of time that they lived with their parents. Up until the outbreak of the Second Would War in 1941, many of the changes that took place, especially in orphanhood, school attendance, and employment, should have led to later home leaving. The war introduced one significant change, a dramatic increase in military service for men, and indirectly led to two others-more marriage and more higher education-that reversed the earlier trend, and led to earlier home leaving. The war and post-war trends ran their course by the 1970s, and were followed by stability in schooling and work, and a decline in the propensity for young people to marry.

A Brief Introduction to Data and Methods

Before we present our analysis of home-leaving ages, we begin with a brief description of our sources and methods. Those wishing more information should consult the appendixes. The analysis reported here makes use of household-level and individual-level data drawn from the U.S. Census of population from 1880 through 1990, with the exception of 1890 and 1930 (see note 2). These data about individuals and households have been digitized from a sample of records extracted from the original manuscripts of the census, and maintained as part of the IPUMS data collection by the University of Minnesota. (26) In every census used here, the enumeration recorded relationships within each household, and it is usually possible from those relationships to identify the parents of each child, if those parents lived within the household. The censuses changed the questions they asked from time to time, but consistently allow identification of race, sex, place of current residence, birthplace (both with varying levels of detail) , marital status, and whether the individual lived on a farm or not. (27) We use these important characteristics of individuals to categorize the subjects of our analysis. The characteristics that we do not use are also significant. It is not possible to categorize the subjects of our analysis by their parents' socioeconomic status, for example, because we usually cannot know the parental SES of those who have already left home. This is a weakness in our analysis, but it does not diminish the long-term trends and differences between major population sub-groups that we have discovered.

We use the census data in three complementary ways.

The simplest thing we do is tabulate the distribution of persons with some characteristic according to whether they still lived with one parent or not. This information is available directly from the census. In Figure 3, for example, we report the percentage of women aged 15-29 by marital status who live away from their parents. This information, while simple, is comparable across censuses and sub-groups, and is immediately meaningful. If more married women in the south than in the north lived away from home in a given time period, the results are meaningful.

The second way we use the data is to estimate the ages at which young people left home by quartile, in order to produce estimates of the age at which one fourth (the first quartile), one half (the median), and three-fourths (the third quartile) of all young men and women had left home. This information is not available directly from the census, and therefore requires more complex methods, because census data do-not include information about actual transitions living at home to living away, but are limited to cross-sectional information about whether individual persons do or do not live with their parents at a single point in time. As a demographic transition, leaving home is generally analogous to marriage, and we might have borrowed from two methods used by demographers for the study of marriage. (28) Because of the limitations of demographic approaches to estimating home-leaving ages, the results reported here draw on statistical rather than demographic methods. The statistical approach is described in gre ater detail in Appendix A, and involves fitting the recorded proportions having left home to a logit model for each race-sex sub-population, with other covariates (region, farm/non-farm, immigration status, and so on). We can estimate medians and quartiles from the predicted values of those fitted models. These estimated medians and quartiles allow us in Table B. 1, for example, to give decennial estimates of the median age at which white women left home, despite the fact that the census does not record that information itself.

The third way we use the data is an extension of the statistical approach just described, which allows us to gauge the relative importance of the factors that influenced home leaving ages. The logit regressions of the home leaving process produce coefficients that can be transformed into odds ratios, which tell us just how much various factors influenced home leaving. While these regressions cannot always be compared across time periods (unlike the estimated medians and quartiles, and the simple distributions of people by home-leaving status), they allow us to take into account possible interactions, for example if farm-resident black males in the south were different from farm-resident white males in the south. They also allow us to gauge the overall strength of all relationships by looking at measures of goodness-of-fit for the overall model, such as r-squared. We generally use these statistical results to confirm and illuminate the more easily interpreted and compared results produced by the first two met hods we have described.

Results

The key starting points for our research are reported in Figure 1. In this section of our article we display trends in proportions away from home or the median home-leaving age graphically, because of the ease with which they can be interpreted. The tables on which these figures are based are presented in Appendix B. Figure 1 presents the median time spent living with parents for males and females, by race, for the entire U.S. The most important initial finding that we demonstrate here is the difference in the chronological pattern of our results and the patterns reported in the previous literature.

Our results for the entire United States population, for both males and females, suggest that the age at which young people ceased to live with their parents rose in the early twentieth century, before reaching a peak in 1940 for men, and in 1950 for women. Following those peak years, the age declined until 1970 before rising slightly from 1970 to 1990. These findings are in contrast to the consensus view of home leaving in the literature. Several authors assert that the age at leaving home has been declining since the early twentieth century. Our results show more complex changes in the leaving home age until 1970, when the two sets of results become similar.

Although we report only the median age at leaving home in Figure 1, we also consider the first and third quartiles--that is, the ages at which a quarter of the population had left home, and the age at which three-quarters of the population had left home. The interquartile range, which spans the first to the third quartile, indicates the uniformity of the leaving-home experience among all young people. A large interquartile range suggests a variety of choices or experiences that affect the age at which a person leaves home, while a smaller value indicates a narrower range of choices.

The most striking feature of Figure 8 is the extent to which patterns in the interquartile range over time are associated with sex. For both white and black men, the interquartile range remained stable from 1880 until 1940. In 1950 and 1960, the range decreased sharply, indicating that leaving-home behavior became more uniform. We hypothesize that this is a result of increased participation in military service during this period. From 1970 to 1990, the interquartile range for men increased again, as military service became less common and social norms about leaving home may have become more diverse. For both white and black women, the interquartile range was more variable. Diversity in leaving home generally increased from 1920 to 1950, perhaps as a result of new opportunities in education for some women. From 1960 to 1990, the interquartile range for women followed a similar pattern to that of men, but with less variation. Because fewer women than men served in the military in the post-1940 era, some of tha t decrease may have to do with a more prescribed path in higher education. The sharp increase that both men's and women's interquartile ranges showed from 1980 to 1990 may reflect a new social paradigm in which young people leave home for the sole purpose of establishing a new household. For both men and women, in almost all years, whites had a slightly higher interquartile range than blacks, which may result from more cultural or economic choices in home-leaving.

For both sexes, although the phenomenon is more pronounced for women, we find the most important predictor of home-leaving is marital status. Figure 9 shows the extent to which married women were likely to live away from their parents. With the exception of 1950, no fewer than 85 percent of both white and black married women lived away from their parents. In 1950, the percent of married women living away from their parents decreased, especially for blacks, in response to post-World War II housing shortages. The behavior of white and black married women was almost identical from 1880 to 1920. After 1920, however, black women were consistently more likely to live with their parents after marriage.

In Figure 10, we show the percent of married people who lived with at least one parent. From 1880 to 1990, there was little difference between married men and women. Race became more important after 1920, as married blacks were more likely than married whites to live with a parent. Both races experienced a sharp increase in 1950, because housing was scarce in 1940 and 1950 relative to 1960 and later. (29) In 1940 more than twenty percent of the population lived more than one person per room, a percentage that fell to about sixteen percent in 1950, but only 8.5 percent in 1960. Housing shortages in 1940 and 1950 led to a relatively large proportion of married young people living with a parent, a pattern that was sharply reduced by 1960, and had only partly returned in 1980 and 1990. (30) The increase that all groups experienced in living with a parent from 1980 to 1990 may reflect the recent tendency for adults to return home, either for financial reasons or to care for aging parents.

The age at which young men and women left home varied considerably across the regions of the United States. For white men and women, a consistent pattern emerges (Figure 11). Young men left home earliest in the West and latest in the Northeast. Young people in all four regions experienced the same relative changes in home-leaving overtime. One exception is the low age at leaving home for Western young men in 1880, which is a result of westward migration and employment possibilities for the very young. Overall, these results are very similar to the national results for whites (Figure 1), because of the large proportion in all time periods of young people who were white.

The median ages at leaving home by region for black males and females reveal very different results than the figures for whites. The relative ages at which young people left home in each region are more complex: in most years, they left home earliest in the West and latest in the South, but there are many exceptions. By 1980 and 1990, young people living in the Northeast and Midwest left home later than any other group. For every region, the age which black men (and to a lesser extent women) left home varied considerably over time, suggesting that these groups were more sensitive to social and economic changes than were whites.

Farm residence is another important factor that determined the timing of home leaving for young people in the U.S. For all four groups (black and white, male and female), farm residents left home later than non-farm residents until at least 1940 (Figures 12). This could be a function of different social conditions for those who lived on the farm, so that they were slower to leave their parents' homes. Alternatively, it could be an artifact of social change, so that the prevailing pattern of home leaving was one where young people moved from farm to non-farm residence when they left home. Beginning in 1950, the differentials changed. For males, since 1950 those not living on farms had higher median ages at leaving home, something that was especially so for black men. This may represent the final surge of off-farm migration. For females, the pattern appears rather different. Farm females continued to leave home later than non-farm females, and the difference becomes more pronounced beginning in 1950.

The final set of issues we report have to do with migration. It stands to reason that young people who have migrated away from their place of birth were likely to have home-leaving ages younger than those who still lived in their place of birth. We examine this question from two points of view. In Figure 13 we report the differences between native-born and foreign born young white people in terms of home leaving ages (there are relatively few foreign born blacks, which makes their results less interesting). This graph demonstrates that there was a dramatic difference between the U.S.-born and the foreign born in the ways we expected, and that the differences diminished over time, so that there is little difference in home-leaving ages for females by 1970, and not much more difference for males in recent years. In Figure 14 we describe the differences in the median age at leaving home between native-born young people who are still living in their region of birth, and those who are currently living in a differ ent region. These results show that those living in the region of their birth consistently stay with their parents longer than do those who leave, and that the differences for males is larger than for females. It is also interesting to note that the differences are larger for blacks than for whites beginning in 1950.

In the analysis thus far, we have emphasized what we can learn by looking at the trends in median home-leaving ages and proportions away from home for specific groups within the population. We have also combined these analyses and estimated logistic regression models for each year. We estimate the regressions in order to see if there are combinations of independent variables that when used simultaneously change our overall interpretation or give us a different way of seeing the role played by one independent variable or one set of social conditions. We are especially interested in using this analysis--which we explain and report fully in Appendix C--to give us a sense of the relative importance of the independent variables in explaining home leaving, which is not clear from the medians and proportions we have discussed thus far.

For men and women of both races, a few variables clearly have the most impact in influencing whether a person lives at home or away. In these results, age, marital status, and foreign nativity explain most of the variance in living away from home. The importance of age is predictable, because as young people aged they were more likely to be away from their parents.

Despite the logical importance of age, marital status is in fact the strongest predictor of leaving home for all groups, which is a predictable consequence of the high association in U.S. society for married people to establish their own households. The effects are greater for whites than for blacks, and for women than for men. For women, the importance of marriage in determining the likelihood of leaving home declines steadily from the late nineteenth century to the present, with important declines between 1900 and 1910, and after 1960 (for white females) and 1970 (for black females), with a very sharp decline after 1970. The overall decline in the importance of marriage for home-leaving for females is very substantial. We suspect that the decline in importance in marriage as a predictor of home-leaving has two causes. The first is the rising age at marriage. As fewer people were married at each age, their importance in the overall pool of home-leavers diminished. The second is the genuine trend for married people to be living with their parents shown in Figures 3 and 4. Married black women are less likely to live away from their parents than white women, in every time period.

Nativity is the second major predictor of leaving home. The variable is much more likely to be significant among whites than blacks, because only a few blacks were recorded as foreign born until recent years. What is most interesting about the foreign-born and home-leaving is the change in the direction of the relationship that takes place beginning after 1950. The general pattern up through 1950 is that the foreign-born were much more likely to be living away from their parents than the native-born, something we would expect based on a migration pattern in which a large proportion of young people came to the United States separately from their parents. Beginning in 1960 for white males and 1970 for white females (and in 1930 for black males and females), the effect is reversed: the foreign born are less likely to live away from their parents, perhaps a reflection of more family-oriented migration streams that have emerged since the second world war. (31) The variable that identifies people living away from the region of their birth describes a strong and predictable relationship between migration and home-leaving. Consistently over the century, young people who have moved away from their region of birth are twice as likely as those who have not to live away from their parents.

Conclusion

We began this article with the assertion that three time periods define the long-term trend in home leaving ages in the United States since 1880. The evidence of those trends is clear from our first pages, where Figure 1 demonstrates those trends. What we have done in the pages that follow is to show the underlying elements that contributed to the creation of our periodization, as well as to reveal key differences between young Americans by race and sex, region and residence.

In the Long First Half of the Twentieth Century, from 1880 until 1940 for men, and 1950 for women, ages at leaving home rose in the United States. This is the most dramatic finding of our research, because it contradicts assertions by other historians that the age of home leaving was falling during much of this time period, part of a long trend that they assert did not end until the 1960s or later. We show otherwise. The explanation is clear in the supporting data we present. While orphanhood never separated a large proportion of young Americans from their parents, declining mortality gave it less and less of an impact. Much more important than mortality were increases in school attendance and declines in the employment of teenagers. Children attending secondary school but not employed were much more likely to live at home with their parents. Even a lack of new housing in the 1930s contributed to the change.

The Second World War brought dramatic changes to the timing of home-leaving for young Americans. Increased military service and the transformation of economy and society spurred by wartime mobilization drew young people away from their homes. After the war the process was never fully reversed, so that at the time of the 1950 census young men were still more likely than in 1940 to be living apart from their parents, and young women were about to experience the same trend. The draft and near-compulsory military service was only part of the process, even with the continuing military obligations of the United States in Asia and Europe throughout the 1950s and 1960s, and into the 1970s. The end of the Second World War brought with it a sharp increase in marriages for whites, if not for blacks. Through policies like the G.I. bill and state-sponsored post-secondary education, more and more young people attended colleges and universities, and in doing so lived away from home.

The War and Post-War decline in home-leaving ages was short-lived in our view, especially when compared with the assessments of other historians. There was no century-long trend toward declining ages of leaving home. Instead, the pattern was over after 1960 for blacks and 1970 for whites, to be followed by a new set of paradigms for young people. These new paradigms were experienced by the young people we have called the Baby Boomers and After. Reduced obligations for military service led more young people to live at home with their parents. So did more available local higher education in the form of community colleges. Finally, and most difficult to define, changing expectations in terms of consumption and living arrangements made it more likely that some young people would choose to return to living with their parents in order to spend more of their disposable income on cars, food, and consumer goods, and less on rent.

The results we have presented are not simply about trends through time that cut across all groups within the society of the United States. We have identified equally dramatic variations by sex and race. The importance of military service in our explanations means that our findings are more dramatic for men than for women. For women, the temporal trends that we associate with the Second World war and its aftermath are less dramatic than they are for men. On the other hand, the liberating social forces in the second half of the century that allowed women more flexibility in the choices they make about their lives have led to substantial long-term changes in the home-leaving experiences.

Similarly, young people of African descent in America have had a noticeably different set of experiences than those of the white majority. While the gendered impact of military service for men was shared by blacks and whites, racial variations in education and employment have distinguished the two largest racial groups in the U.S. The housing shortage that led to more black than white young married couples living with their parents was probably a consequence of both greater black poverty and slower housing construction in some communities. The boom in higher education that followed the Second World War was more pronounced for white men and women than for blacks, despite 0.1. Bill benefits for all veterans. Employment opportunities have generally been greater for whites than blacks as well. Taken together, these factors led to less dramatic changes for blacks than for whites after World War II, especially in the 1960s when the impact of higher education growth may have been most rapid. In all this, region, res idence, and migration have mattered as well, a sign of their continued importance in the development of American society.

The timing of home-leaving matters because, for much of the twentieth century, home-leaving was the starting point for a range of processes that signaled the transition from youth to adulthood. Most young people left home to marry, complete their education, serve in the military, or to work. With those changes came parenthood and economic independence. The timing of the components of home-leaving has changed, making the whole process occur earlier or later, or even become reversible as people return home. In these transitions we see the outline of change in their lives and in the experiences of American society as a whole.

The wider message in our results is about the close correlation between the timing of home-leaving for young Americans over the past century and longer, and the broadest social and cultural transformations that we have experienced. There should be little surprise in that, but the story told here shows that the big themes of transformation in migration, education, and military service, coupled with the expansion of civil rights and the social emancipation of women and young adults, has had a measurable impact on lives. We have made a cultural and social world for our children, and they have lived in it and shaped it and their own lives at the same time.

[Figure 1 omitted]

[Figure 2 omitted]

[Figure 3 omitted]

[Figure 4 omitted]

[Figure 5 omitted]

[Figure 6 omitted]

[Figure 7 omitted]

[Figure 8 omitted]

[Figure 9 omitted]

[Figure 10 omitted]

[Figure 11 omitted]

[Figure 12 omitted]

[Figure 13 omitted]

[Figure 14 omitted]

ENDNOTES

(1.) This research has been supported by grant HD33554 from the National Institute of Child Health and Human Development. We thank Michele Butler, who got us started on this topic, and on the other members of the Great Plains Population and Environment group in Austin, as well as to Matt Sobek. We presented earlier versions of this paper at the Annual Meetings of the Population Association of America in New York, March, 1999, and the Social Science History Association in Fort Worth, November, 1999, and at seminars at Brown University, at SUNY University at Albany, and the University of Michigan in 2000. We are grateful for all the helpful suggestions we received from those audiences.

(2.) The census enumeration forms for 1890 were destroyed in a fire, making detailed analysis of the 1890 census impossible; no 1930 census data are available for this analysis, because those census schedules will not be available for consultation until 2002, and the preparation of public use data files will then take several years. See the discussion of sources and methods later in this article for more details.

(3.) In order to simplify analysis, we have excluded persons who were not classified as "white" or "black" in the census. In all years, persons of Hispanic origin are classified as "white."

(4.) Frances K. Goldscheider and Calvin Goldscheider, "Leaving and Returning Home in 20th Century America," Population Bulletin 48 (1994): 1-34; Timothy W. Guinnane and Myron P. Gutmann, "Leaving Home in Nineteenth Century Texas" (Unpublished paper presented at the Annual Meeting of the Social Science History Association: 1992); Francesca A. Florey and Avery M. Guest, "Coming of Age among U.S. Farm Boys in the Late 1800s: Occupational and Residential Choices," Journal of Family History 13 (1988): 233-249; John Modell, et al., "Social Change and Transitions to Adulthood in Historical Perspective," Journal of Family History 1 (1976): 7-32.

(5.) For an example of census-based research, see David A. Stevens, "New Evidence on the Timing of Early Life Course Transitions: The United States 1900 to 1980," Journal of Family History 15 (1990): 163-178. While our findings contradict his, we do not emphasize the divergence here. We believe that there are errors in his quantitative analysis, but the article is unclear about methods, and we could not locate the author for clarification.

(6.) Goldscheider and Goldscheider 1994.

(7.) Modell et al.; Stevens.

(8.) Goldscheider and Goldscheider 1994; F. Goldscheider and C. Goldscheider, The Changing Transition to Adulthood: Leaving and Returning Home (Thousand Oaks, 1999).

(9.) C. Horace Hamilton, "The Annual Rate of Departure of Rural Youths from Their Parental Homes," Rural Sociology 1(1936): 164-179

(10.) Timothy W. Guinnane, The Vanishing Irish: Households, Migration, and the Rural Economy in Ireland, 1850-1914 (Princeton, 1997).

(11.) Fred C. Pampel, "Changes in the Propensity to Live Alone: Evidence from Consecutive Cross-Sectional Surveys, 1960-1976," Demography 20 (1983): 433-447.

(12.) Frances K. Goldscheider and Calvin Goldscheider, Leaving Home Before Marriage: Ethnicity, Familism and Generational Relationships (Madison, 1993), 169-186.

(13.) Florey and Guest.

(14.) Ibid, 246-247.

(15.) Elliott West, Growing Up in Twentieth-Century America: A History and Reference Guide (London, 1996), 207-208.

(16.) Ibid, 288-295.

(17.) Claudia Goldin, "A Brief History of Education in the United States." Working Paper Series on Historical Factors in Long Run Growth, Historical Paper 119 (1999): 13-14.

(18.) Matthew Sobek, "A Century of Work: Gender, Labor Force Participation, and Occupational Attainment in the United States, 1880-1990" (University of Minnesota: Unpublished Ph.D. dissertation, 1997).

(19.) Goldscheider and Goldscheider, 1994.

(20.) John Modell, Into One's Own: From Youth to Adulthood in the United States, 1920-1975 (Berkeley, 1989).

(21.) Goldscheider and Goldscheider, 1994.

(22.) Ibid.

(23.) Stevens: 176.

(24.) Stevens.

(25.) For a more complete discussion of our computation of orphanhood rates, see Appendix B.

(26.) Steven Ruggles and Matthew Sobek, Integrated Public Use Microdata Series: Version 2.0 (Minneapolis: Minnesota Historical Census Projects, University of Minnesota, 1997).

(27.) The definition of a farm changes from census to census. See Ruggles and Sobek; Margo J. Anderson, The American Census: A Social History (New Haven, 1988); U.S. Bureau of the Census, Twenty Censuses: Population and Housing Questions, 1790-1980 (Washington, 1979).

(28.) John Hajnal, "Age at Marriage and Proportions Marrying," Populations Studies 7 (1953): 111-36; Ansley J. Coale, "Age Patterns of Marriage," Population Studies 25 (1971): 193-214; Ansley J. Coale and David McNeil, "The Distribution by Age of the Frequency of First Marriage in a Female Cohort," Journal of the American Statistical Association 67 (1972): 743-40.

(29.) There were relatively few housing starts between 1930 and 1945, which led to severe housing shortages during the war, which continued for some time afterward. Kenneth T. Jackson, Crabgrass Frontier: The Suburbanization of the United States (New York, 1985), 232.

(30.) U.S. Bureau of the Census, Census of Housing, 1960, Volume I: States and Small Areas, Part I: United States Summary (Washington, D.C., 1961), xxviii-xxiv.

(31.) The decreasing likelihood of immigrants to leave home in the late twentieth century is a complex issue. One cause of the reversal in that appears in our regression results is the increasing trend for immigrants to enter the United States as a family (West, 1992, 299). A second possibility is that foreign-born young adults were less likely to leave home for non-job related reasons, which include education and financial pressures (Goldscheider and Goldscheider, 1999, 135). We do not control for these variables in this regression. Finally, we know that immigrants were less likely than native-born people to leave home for premarital residential independence in the late twentieth century (Goldscheider and Goldscheider, 1993, 35). The importance of marriage in the leaving-home patterns of the foreign-born may explain the difference in Table 11 and our regression results, as only the regressions control for independent variables such as marriage.

(32.) Ruggles and Sobek.

(33.) John Hajnal, "Age at Marriage and Proportions Marrying," Population Studies 7 (1953): 111-136.

(34.) Coale; Coale and McNeil.

(35.) Michael R. Haines, "Estimated Life Tables for the United States, 1850-1910," Historical Methods 31 (1998): 149-170; Vital Statistics of the United States, 1990, Volume II: Mortality: Part A (Hyattsville, MD, 1994), 6.14.

APPENDIX A

Data and Methods for this Analysis

Sources of Data

For this paper, we use individual-level data from the United States Census of Population. Our data come from the Integrated Public Use Microdata Series (IPUMS) prepared by the University of Minnesota. (32) We examine patterns in home leaving ages for the U.S. from 1880 to 1990. Within that context we look at the differences between males and females, blacks and whites, farm and non-farm residents, regional populations, and married and unmarried. Our analysis is twofold. We calculate medians and quartiles of the age at leaving home by sex and race, using a fitted curve derived from a logit regression model. We also use logit regressions for each sex and race subgroup to test for the influence of contextual factors on the age at leaving home. In this analysis we assume that the IPUMS samples are flat samples, and we have not used weights or attempted to adjust for unrepresentative cases.

The coding of family relationships in the IPUMS data sets gives us a straightforward way to characterize whether young people were still living with their parents or not, and then to estimate the ages at which the transition took place. The IPUMS data sets consistently provide two variables (momloc and poploc) that tell whether a person's mother or father is living in the same household, and then identifies that person in the data. We have used any valid value for these variables to indicate that a person is still living with his or her mother or father; a value of zero indicates that the person has already left his or her household of origin.

We analyze this transition from living with parents, to living away from parents, within a multivariate statistical framework. Such a framework allows us to include several predictors of the transition simultaneously, to estimate the intensity of their effects individually and in the presence of other predictors, to evaluate the statistical significance of these effects, and to produce fitted values.

Estimating the Distribution of Home-Leaving Ages

Our major methodological challenge is that our data do not include information about actual transitions, but are limited to cross-sectional information about whether individual persons do or do not live with their parents at a single point in time. An age distribution at the time of a census must be synthesized into the behavior over time of an artificial or synthetic cohort, in the same sense that a life table is a synthesis of cross-sectional data on mortality and survivorship. A life table is typically summarized with the mean age at death, or expectation of life, but it could also be summarized with the median age at death, and perhaps the first and third quartiles of the age at death. Here we would like to estimate the quartiles of the distribution of age at leaving home, within categories of covariates.

As a demographic transition, leaving home is much more comparable to marriage than to death. Most of these transitions occur within a fairly narrow range of ages, about fifteen years, and not everyone makes the transition. Persons who have made the transition remain in the population and can be observed in a census or survey.

We will briefly describe two methods, well established in the demographic literature, for describing the distribution of age at first marriage. The first, due to John Hajnal, is called the singulate mean age at marriage (SMAM). (33) It uses the proportions of women at each age who are still unmarried, typically from a census. After removing an estimate of the proportion of women who will not marry by age 55, the SMAM calculates the mean number of years lived in the unmarried state, which is interpreted as a synthetic cohort estimate of the mean age at marriage (for those who marry by age 55).

A second method to describe age at marriage is due to Coale and McNeil. (34) They observed a widespread international regularity in the distribution of age at marriage, which they were able to fit very well with a specific function having three parameters. The risk or hazard that an unmarried person of age a will marry in the interval a to a + da was found by Coale and McNeil to be closely approximated with a double exponential function. The three parameters can be interpreted as an initial age when marriages begin to be observed, the mean age at marriage, and the proportion who never marry.

These two approaches can be modified to produce estimates of the mean age at leaving home or the quartiles of the age at leaving home, with cross-sectional data on the proportions still at home, rather than the proportions still unmarried. However, neither approach has any of the desirable properties of a multivariate statistical model. Estimates must be re-calculated within separate categories or combinations of predictors, there is no easy way to hold other covariates constant, and there are no estimates of standard errors. Therefore, we prefer to draw on statistical rather than demographic methods.

For most populations, when cross-sectional data on the proportion of persons who have left home is graphed against age, the curve has the clear shape of an ogive. For the youngest ages, the proportion is close to zero; it typically increases gradually in the early teens, very steeply in the late teens and early twenties, and then levels off at a high level in the late twenties. It does not appear to have an asymptote of one.

This kind of curve can be fitted with hazard models, but it can also be fitted very well, and more simply, with a logit model that includes age, age squared, and covariates. A quadratic term for age is needed because the deceleration in the rate of leaving home in the later ages is typically less than the acceleration in the earlier ages.

After considerable testing of alternatives, we ran such logit regressions on the individual-level files for persons aged 15--29 for each census year, with a binary dependent variable that was 1 if the person was not living with parents and 0 if still living with parents. Because sex and race frequently had significant interactions with the other four covariates--marital status, farm/nonfarm residence, nativity, and region-we did separate analyses within each of the four combinations of sex and race. We first did logit regressions with no covariates; with each of the four covariates; and then with all four, in which each of these covariates would be adjusted for the effect of the other three. The significance of a covariate, before or after adjustment for the others, is indicated by the standard error for the coefficients in these logit regressions or by the improvement in fit when the covariate is included.

The median age (or second quartile) at leaving home was found by determining the value of age for which the fitted logit was log(.50/.50) = 0. The first quartile of the distribution is the value of age for which the fitted logit is log(.25/.75) = -log(3) and the third quartile is the value of age for which the fitted logit is log(.75/.25) = log(3). Standard errors of the estimated quartiles were also estimated but are not included in the present paper, in order to avoid excessive complexity.

Calculating Estimates of Orphanhood

Haines gives plausible decennial life tables for 1850 through 1900, based on the Coale and Demeny model West life tables. After 1900, the U.S. Department of Health and Human Services gives decennial life tables based on vital statistics data. (35) A complete series of life tables is available from these sources for the total population of the United States, separately for males and females but not for other subpopulations.

We assumed that persons were born when both parents were exactly age 30 and then estimated the probability that the mother and father (separately) would have survived to exact ages 45, 50, 55, and 60, at which time the child would be aged 15, 20, 25, and 30, respectively. For example, to estimate the probability that the father of a 20-year old in 1900 was still alive, we calculated the probabilities of surviving from 30 to 35, 35 to 40, 40 to 45, and 45 to 50. These four survival probabilities were drawn from life tables for 1885, 1890, 1895, and 1900, and multiplied together. (The probability of surviving from 30 to 35 in 1885, for example, was obtained as the geometric mean of the probability of surviving from 30 to 35 in 1880 and 1890.) Finally, the probability that this father was not still alive in 1900 was multiplied by the probability that the mother was not still alive, giving the probability that both had died and the child was an orphan. This step assumes that the survival of the mother and the su rvival of the father are independent. It is likely that they are not independent, but are positively correlated.

APPENDIX B

Source Tables for Graphical Data Presented in the Text.

Table B.1

Estimated Quartiles for Number of Years Lived with One or Both Parents,
United States, 1880-1990

                        1880  1900  1910  1920  1940  1950  1960  1970

White Males

 1st Quartile           18.5  18.8  18.9  20.0  20.9  20.6  18.5  18.6
 2nd Quartile (Median)  22.3  22.6  22.0  23.3  24.3  23.4  20.5  20.4
 3rd Quartile           26.2  27.0  26.5  28.0  28.9  27.5  23.3  22.9

White Females

 1st Quartile           15.2  16.6  17.4  17.9  18.4  19.5  17.8  18.0
 2nd Quartile (Median)  19.4  20.1  20.2  20.5  22.2  23.3  20.7  20.7
 3rd Quartile           22.9  23.7  23.7  24.1  26.5  27.9  24.6  24.4

Black Males

 1st Quartile           17.2  17.6  17.9  18.5  18.8  18.8  17.8  18.1
 2nd Quartile (Median)  20.3  20.8  20.9  21.4  21.6  21.2  19.4  19.7
 3rd Quartile           24.6  25.2  25.4  26.1  26.2  24.8  21.5  21.8

Black Females

 1st Quartile           15.9  16.2  16.6  16.9  17.1  18.0  17.2  17.8
 2nd Quartile (Median)  18.3  19.8  19.1  19.3  20.0  21.3  19.7  20.2
 3rd Quartile           21.3  22.8  22.6  22.6  24.3  27.1  23.1  23.3

                        1980  1990

White Males

 1st Quartile           18.8  19.1
 2nd Quartile (Median)  20.9  21.7
 3rd Quartile           23.6  25.3

White Females

 1st Quartile           18.7  18.5
 2nd Quartile (Median)  21.7  22.6
 3rd Quartile           25.9   *

Black Males

 1st Quartile           18.2  18.5
 2nd Quartile (Median)  19.9  20.6
 3rd Quartile           22.1  23.4

Black Females

 1st Quartile           18.4  18.5
 2nd Quartile (Median)  21.0  21.8
 3rd Quartile           24.6  27.2

(*)Estimate unavailable due to limitations in the data. As the median
age at which white women leave home increases, it approaches the upper
age limit of our data sample. This, combined with the slope of the
change over time, produced clearly erroneous results.

Source: Estimated from the Integrated Public Use Microdata Series
(IPUMS) data for each year.
Table B.2

Percent Ever Married, by Age, United States, 1880-1980

               1880  1900  1910  1920  1940  1950  1960  1970  1980

White Males
 Aged 20-24    20.5  19.5  22.3  27.2  26.6  37.6  32.0  45.4  33.7
White Females
 Aged 15-19    11.0  11.2  11.1  12.0  10.8  13.7  14.0  12.1   9.6
White Females
 Aged 20-24    49.4  45.9  48.6  52.9  51.2  65.3  41.2  65.3  51.8
Black Males
 Aged 20-24    39.0  34.1  41.0  44.6  40.8  43.9  30.6  40.9  22.2
Black Females
 Aged 15-19    18.4  16.3  18.4  22.9  18.4  18.8  13.9  11.5   4.8
Black Females
 Aged 20-24    65.3  59.9  66.2  69.2  63.5  65.6  38.8  56.7  32.4

               1990

White Males
 Aged 20-24    23.8
White Females
 Aged 15-19     6.0
White Females
 Aged 20-24    40.4
Black Males
 Aged 20-24    14.8
Black Females
 Aged 15-19     3.1
Black Females
 Aged 20-24    20.5

Source: Tabulated from Integrated Public Use Microdata Series (IPUMS)
data for each year.
Table B.3

Percent of Persons aged 15-19 in Labor Force, United States, 1880-1980

               1880  1900  1910  1920  1940  1950  1960  1970  (*) 1980

White Males    70.0  67.5  70.1  57.6  34.3  43.6  44.4  64.1    61.2
White Females  23.7  27.4  32.6  34.7  18.2  26.1  28.7  49.5    56.2
Black Males    86.6  81.3  84.9  69.4  49.8  52.3  36.6  49.1    42.4
Black Females  54.1  48.0  60.3  40.5  24.4  21.4  19.8  36.9    37.0

               1990

White Males    54.5
White Females  51.4
Black Males    39.1
Black Females  38.1

Source: Tabulated from Integrated Public Use Microdata Series (IPUMS)
data for each year.

Figures for 1980 are for persons aged 16-19.
Table B.4

Percent of Persons Aged 15-19 Attending School, United States, 1880-1980

               1880  1900  1910  1920  1940  1950  1960  1970  1980

White Males    35.2  31.5  39.5  39.6  57.7  64.0  72.0  81.1  75.3
White Females  29.5  32.2  40.5  41.5  56.3  59.7  68.0  76.7  75.0
Black Males    14.0  17.6  25.6  31.8  43.5  52.1  64.0  74.0  74.6
Black Females  14.9  21.7  32.5  36.8  45.9  53.7  62.6  72.3  76.6

               1990

White Males    80.5
White Females  81.1
Black Males    77.7
Black Females  79.8

Source: Tabulated from Integrated Public Use Microdata Series (IPUMS)
data for each year.
Table B.5

Percent of Persons Who Have Served in Military by Cohort Aged 15-19

               1930  1940  1950  1960  1970  1980

White Males
 Veteran        6.4  58.6  73.3  47.1  37.0  15.0
 No Service    17.6  39.5  24.9  51.5  63.0  85.0
 N/A           75.6   1.9   1.8   1.4   0.0   0.0

Black Males
 Veteran        4.5  44.1  58.5  33.4  30.7  19.9
 No Service    18.1  54.8  39.8  64.8  69.3  80.1
 N/A           77.4   1.1   1.7   1.8   0.0   0.0

White Females
 Veteran        0.0   0.0   0.0   1.1   1.2   1.5
 No Service     0.0   0.0   0.0  98.8  98.8  98.5
 N/A            0.0   0.0   0.0   0.0   0.0   0.0

Black Females
 Veteran        0.0   0.0   0.0   1.2   1.8   2.7
 No Service     0.0   0.0   0.0  98.7  98.2  97.3
 N/A            0.0   0.0   0.0   0.1   0.0   0.0

Source: Tabulated from Integrated Public Use Microdata Series (IPUMS)
data for each year. For explanation of tabulations, see caption for
Figure 5.
Table B.6

Population Aged 15-19 by Nativity, Percentages

               1880  1900  1910  1920  1940  1950  1960  1970  1980

White Males
 Native Born   86.7  85.8  75.0  87.6  96.7  97.5  96.8  96.0  94.0
 Foreign Born  13.3  14.2  25.0  12.4   3.3   2.5   3.2   4.0   6.0

Black Males
 Native Born   99.6  99.4  96.7  98.6  98.9  98.9  98.9  98.2  95.8
 Foreign Born   0.4   0.6   3.3   1.4   1.1   1.1   1.1   1.8   4.2

White Females
 Native Born   87.5  86.1  81.0  88.5  96.5  97.1  96.3  95.5  94.4
 Foreign Born  12.5  13.9  19.0  11.5   3.5   2.9   3.7   4.5   5.6

Black Females
 Native Born   99.6  99.9  97.5  99.0  99.0  99.1  99.2  98.1  96.0
 Foreign Born   0.4   0.1   2.5   1.0   1.0   0.9   0.8   1.9   4.0

               1990

White Males
 Native Born   90.6
 Foreign Born   9.4

Black Males
 Native Born   94.0
 Foreign Born   6.0

White Females
 Native Born   92.1
 Foreign Born   7.9

Black Females
 Native Born   94.1
 Foreign Born   5.9

Source: Tabulated from Integrated Public Use Microdata Series (IPUMS)
data for each year.
Table B.7

Estimated Probability of Orphanhood (both parents dead) by time period
and child's age, United States, 1880-1990

     1880   1890   1900   1910   1920   1930   1940   1950   1960   1970

15  0.025  0.024  0.018  0.016  0.012  0.010  0.006  0.003  0.002  0.001
20  0.047  0.047  0.037  0.033  0.027  0.023  0.017  0.010  0.005  0.004
25  0.086  0.083  0.071  0.061  0.052  0.046  0.038  0.025  0.015  0.011
30  0.147  0.140  0.122  0.113  0.095  0.086  0.076  0.056  0.038  0.028

     1980   1990

15  0.001  0.001
20  0.003  0.002
25  0.008  0.006
30  0.021  0.015

Source: Estimated from life tables presented in Historical Methods and
Vital Statistics of the United States. For explanation of method, see
Appendix A.
Table B.8

Estimated Inter-Quartile Ranges in Age at Leaving Home, 1880-1990

               1880  1900  1910  1920  1940  1950  1960  1970  1980

White Males     7.8   8.2   7.6   8.0   8.0   6.9   4.8   4.3   4.8
White Females   7.7   7.1   6.3   6.2   8.1   8.4   6.8   6.4   7.2
Black Males     7.4   7.7   7.5   7.6   7.4   6.0   3.7   3.7   3.9
Black Females   5.4   6.6   6.0   5.7   7.2   9.1   5.9   5.5   6.2

               1990

White Males     6.2
White Females     *
Black Males     4.9
Black Females   8.7

Source: Estimated from the Integrated Public Use Microdata Series
(IPUMS) data for each year.

For 1990 the third quartile figure is unavailable. (see Table 1).
Table B.9.

Percent of Persons Aged 15-29 who live with Parents, by Marital Status,
Race and Sex

                         1880  1900  1910  1920  1940  1950  1960  1970

Married White Males
 Live with Parents        6.8   6.3   9.4  11.0  12.4  14.2   5.9   4.8
 Live away from Parents  93.2  93.7  90.6  89.0  87.6  85.8  94.1  95.2

Unmarried White Males
 Live with Parents       65.2  67.4  66.1  76.8  84.8  87.8  74.2  73.3
 Live away from Parents  34.8  32.6  33.9  23.2  15.3  12.2  25.8  26.7

Married Black Males
 Live with Parents        4.3   4.4   8.1  10.5  14.8  22.9  11.6   9.6
 Live away from Parents  95.7  95.6  91.9  89.5  85.2  77.1  88.4  90.4

Unmarried Black Males
 Live with Parents       51.2  55.6  59.2  63.9  73.0  80.8  68.7  71.1
 Live away from Parents  48.8  44.4  40.8  36.1  27.1  19.2  31.3  28.9

Married White Females
 Live with Parents        4.9   5.8   9.6  10.7  11.6  15.5   6.8   5.9
 Live away from Parents  95.1  94.2  90.4  89.3  88.4  84.5  93.2  94.1

Unmarried White Females
 Live with Parents       72.3  73.6  79.8  83.0  84.3  89.1  79.8  76.8
 Live away from Parents  27.7  26.4  20.2  17.0  15.7  10.9  20.2  23.2

Married Black Females
 Live with Parents        4.6   6.6   9.6  10.8  14.9  24.1  13.4  11.3
 Live away from Parents  95.4  93.4  90.4  89.2  85.1  75.9  86.6  88.7

Unmarried Black Females
 Live with Parents       61.0  65.6  68.4  73.6  74.1  82.9  72.2  73.1
 Live away from Parents  39.0  34.5  31.6  26.4  25.9  17.2  27.8  26.9

                         1980  1990

Married White Males
 Live with Parents        5.4   7.8
 Live away from Parents  94.6  92.2

Unmarried White Males
 Live with Parents       66.2  65.3
 Live away from Parents  33.9  34.7

Married Black Males
 Live with Parents       10.6  14.3
 Live away from Parents  89.4  85.7

Unmarried Black Males
 Live with Parents       66.2  64.7
 Live away from Parents  33.8  35.3

Married White Females
 Live with Parents        5.1   6.7
 Live away from Parents  94.9  93.3

Unmarried White Females
 Live with Parents       66.7  65.1
 Live away from Parents  33.3  35.0

Married Black Females
 Live with Parents       11.6  14.4
 Live away from Parents  88.4  85.6

Unmarried Black Females
 Live with Parents       64.1  61.0
 Live away from Parents  36.0  39.0

Source: Tabulated from Integrated Public Use Microdata Series (IPUMS)
data for each year.
Table B.10

Median Age at Leaving Home by Region, Sex, and Race

               1880  1900  1910  1920  1940  1950  1960  1970  1980

White Males
 Northeast     22.9  22.7  22.3  24.0  25.8  24.8  21.5  21.4  22.1
 Midwest       22.2  22.6  22.5  23.6  24.4  23.3  20.7  20.5  21.0
 South         22.1  22.6  22.2  22.7  23.4  22.9  20.1  20.0  20.5
 West          19.8  21.7  20.8  22.3  22.9  22.2  19.6  20.0  20.2

Black Males
 Northeast     18.8  17.9  19.7  19.5  22.6  22.9  20.5  21.0  22.4
 Midwest       20.2  18.6  20.8  19.8  22.4  22.6  20.6  20.7  22.2
 South         19.4  20.3  20.2  20.7  22.2  23.6  21.1  20.8  21.7
 West          22.1   *    17.4  18.2  20.9  21.6  18.7  19.3  20.3

White Females
 Northeast     20.8  20.8  21.5  22.3  23.1  22.8  20.2  20.5  20.9
 Midwest       20.0  20.8  21.3  21.5  21.4  20.9  19.3  19.7  19.9
 South         20.2  20.8  20.5  20.7  20.7  20.5  19.2  19.4  19.6
 West          19.3  20.6  20.1  20.8  20.7  20.3  18.9  19.4  19.5

Black Females
 Northeast     18.2  16.6  18.8  18.8  20.1  21.3  19.3  20.0  21.3
 Midwest       18.1  19.2  19.4  19.3  20.1  20.7  19.1  19.9  20.9
 South         18.3  19.2  19.2  19.3  20.0  21.5  20.0  20.4  21.1
 West          17.4   *    17.5  19.1  19.0  19.6  18.4  19.4  20.5

               1990

White Males

 Northeast     23.0
 Midwest       21.8
 South         21.4
 West          20.9

Black Males

 Northeast     23.0
 Midwest       23.0
 South         22.0
 West          20.0

White Females

 Northeast     21.6
 Midwest       20.5
 South         20.3
 West          20.2

Black Females

 Northeast     22.5
 Midwest       21.8
 South         21.7
 West          20.8

(*)There are so few Blacks in the west in the 1900 IPUMS sample that the
computation is unreliable.

Source: Estimated from the Integrated Public Use Microdata Series
(IPUMS) data for each year.
Table B.11

Median Age at Leaving Home by Sex and Farm Residence, 1880-1990

                            1880  1900  1910  1920  1940  1950  1960

White Males on Farms        22.7  24.2  23.8  24.8  25.3  25.3  22.9
Black Males on Farms        20.7  21.7  21.6  21.6  21.7  21.4  20.2
White Males not on Farms    19.6  21.8  21.3  21.7  23.0  25.5  23.6
Black Males not on Farms    18.9  20.2  19.9  19.8  20.5  22.4  21.7
White Females on Farms      21.9  21.8  21.5  22.9  24.1  23.1  20.3
Black Females on Farms      20.7  20.4  20.7  21.4  21.6  21.1  19.4
White Females not on Farms  19.2  18.6  19.5  19.7  21.8  22.6  20.5
Black Females not on Farms  18.0  18.3  18.6  18.8  19.8  20.9  19.5

                            1970  1980  1990

White Males on Farms        23.8  23.8  25.9
Black Males on Farms        21.4  21.4  22.8
White Males not on Farms    23.3  25.5  25.8
Black Males not on Farms    22.1  24.9  24.6
White Females on Farms      20.3  20.8  21.6
Black Females on Farms      19.7  19.9  20.6
White Females not on Farms  20.7  21.7  22.6
Black Females not on Farms  20.1  21.0  21.8

Source: Estimated from the Integrated Public Use Microdata Series
(IPUMS) data for each year.
Table B.12

Median Age of Leaving Home by Nativity, Sex, and Race

               1880  1900  1910  1920  1940  1950  1960  1970  1980

White Males
 Foreign       19.7  19.8  19.2  21.0  24.3  21.7  19.9  20.3  20.2
 Native Born   22.6  23.1  23.2  23.7  24.3  23.5  20.5  20.4  20.9

Black Males
 Foreign       18.8  15.0  18.2  16.7  22.7  21.0  19.0  19.6  20.6
 Native Born   19.4  20.1  20.2  20.6  22.2  23.3  20.8  20.7  21.8

White Females
 Foreign       17.9  18.1  18.8  19.5  21.6  19.9  18.9  19.7  19.8
 Native Born   20.6  21.2  21.4  21.7  21.6  21.2  19.4  19.7  19.9

Black Females
 Foreign       18.7   1.4  18.6  17.9  20.0  19.6  18.6  19.2  20.5
 Native Born   18.3  19.0  19.1  19.3  20.0  21.3  19.7  20.2  21.1

               1990

White Males
 Foreign       20.3
 Native Born   21.8

Black Males
 Foreign       21.0
 Native Born   22.0

White Females
 Foreign       20.2
 Native Born   20.7

Black Females
 Foreign       21.1
 Native Born   21.8

Source: Estimated from the Integrated Public Use Microdata Series
(IPUMS) data for each year.
Table B.13

Median Age of Leaving Home by Migration Status, Sex, and Race

                         1880  1900  1910  1920  1940  1950  1960  1970

White Males
 Left Birth Region       20.0  20.2  19.5  21.0  22.4  21.1  18.8  19.2
 Stayed in Birth Region  22.9  23.3  23.6  24.1  24.7  23.8  21.0  20.9

Black Males
 Left Birth Region       18.6  14.1  18.1  18.2  21.3  20.7  19.0  19.3
 Stayed in Birth Region  19.4  20.3  20.3  20.8  22.4  23.9  21.4  21.3

White Females
 Left Birth Region       18.4  18.6  19.1  19.7  20.4  19.6  18.5  19.1
 Stayed in Birth Region  20.7  21.3  21.6  21.9  21.7  21.4  19.6  19.9

Black Females
 Left Birth Region       17.6  15.4  18.2  18.1  19.0  19.2  18.3  19.1
 Stayed in Birth Region  18.3  19.1  19.2  19.4  20.2  21.7  20.1  20.5

                         1980  1990

White Males
 Left Birth Region       19.6  20.1
 Stayed in Birth Region  21.3  22.2

Black Males
 Left Birth Region       19.8  20.0
 Stayed in Birth Region  22.3  23.0

White Females
 Left Birth Region       19.2  19.8
 Stayed in Birth Region  20.1  20.9

Black Females
 Left Birth Region       19.9  20.3
 Stayed in Birth Region  21.3  22.1

Source: Estimated from the Integrated Public Use Microdata Series
(IPUMS) data for each year.

APPENDIX C

Logistic Regressions

In addition to the median age at leaving home and the first and third quartile ages, we have undertaken logistic regressions that demonstrate the influence of each contextual variable on the age at leaving home.

For these regressions, our dependent variable indicates whether a given person lived away from both parents. Our independent variables represent a variety of factors that we believe may explain the variance in living with or away from parents, including age and farm residence, which we have already described by use of the estimated quartiles and medians. We provide separate regressions for the four sex and race groups as a recognition of the differences between these groups, and in order to avoid including a large number of interaction terms in each of the models. We assume that there would be interactions between sex and race and most of the independent variables in any combined model. These are the independent variables used in the analysis:

Age and Age Squared. Age is the dominant factor in determining whether a young person has left their parental home or not, in the sense that there is a steady increase by age in the number of persons who are no longer living in their parents' home, no matter what their other characteristics. We also include age squared because the relationship between age and home-leaving is not linear. These variables serve more as controls that define the overall process, and their odds ratios are nor necessarily easy to interpret.

Marital Status. While age is the most important factor overall in determining the time at which a young person leaves home, we will see that marital status is the single most important factor overall that determines whether he or she lives with parents or not. Simply put, married people are much less likely to live with their parents than the unmarried, although the strength of the effect varies through time and across sex and race groups.

Region of Residence. We divide the United States into four regions following standard U.S. census divisions: Northeast, Midwest, South, and West. We hypothesize that living conditions will be different in different regions and that these patterns will be expressed in different home leaving ages. We recognize that the regional differences may reflect different proportions urban, different racial and ethnic compositions, and different prevailing categories of socio-economic status, not all of which we can control for in this analysis.

Stayed in Region. This variable represents the residence status of the individual relative to her or his region of birth. We hypothesize that a person who has stayed in the region of birth is less likely to live away from his or her parents.

Foreign-born. We hypothesize that persons who were born in a foreign country (and therefore who have migrated to the U.S.) have a different status relative to living with their parents than do the native born. The analysis permits us to recognize difference in the social conditions of migration for the foreign born in different time periods. In some eras it is possible that migrants were more likely to live with their parents than others, and that they were more likely to live with their parents than the native born, holding age constant.

Farm Residence. This is the most consistently available variable in the census samples that gives an indication of residence and economic role. Following the results reported for the estimated quartiles, we hypothesize that farm residence is associated with later home-leaving, but we recognize that there is a complication if a large proportion of home leaving involves a move from farm to non-farm residence.

Percent aged 15-19 in the state attending school. We hypothesize that high levels of involvement in public education in each state are associated with later ages at leaving home. One reason for this outcome is that formal education for this age group is conversely related to participation in the labor force. Young people who are attending high school are more likely to be still living with their parents, partly because they are not under the financial pressures that apply to people in this age group who are in the labor force.

We report regression results in Tables C.1 through C.4. The tables present odds ratios rather than regression coefficients because of their ease of interpretation. For categorical variables, the odds ratio represents the relative strength of the effect for the category measured relative to the reference category, so that in Table C.1 in 1880, for example, foreign-born white males were 1.41 times as likely as native-born white males to live away from their parents.

We begin by comparing a model that simply includes age and age squared with one that includes the full complement of independent variables. The results of these comparisons show two things. First, the full models have higher values for r-squared, which confirms that the addition of variables other than age helps explain the variation in home leaving age in a significant way. Second, the r-squared values for the models for white males and females generally increase from 1880 to 1970, before falling back in 1980 and 1990. The trend of r-squared values for the black male and female models is not so clear from 1880 to 1970, but has the same reduction in 1980 and 1990 that the models for whites have. This reflects the same reversal in patterns of convergence in home-leaving that we saw in the trend for inter-quartile ranges. In 1980 and 1990 there is clear evidence that the trend toward more similarities in the living experiences of young people had turned around.

Table C.1

Logistic Regression Results, Age at Leaving Home, 1880-1990: Odds Ratios
for White Males (*)

Model 1: Age and Age Squared only

                       1880    1900    1910     1920     1940     1950

Intercept             0.001   0.000   0.000    0.000    0.000    0.000
Age                   1.398   1.607   2.463    2.423    2.216    3.121
Age Squared           0.999   0.996   0.987    0.987    0.990    0.983

R squared             0.232   0.230   0.250    0.265    0.264    0.323

                        1960     1970     1980     1990

Intercept              0.000    0.000    0.000    0.000
Age                    5.824    6.710    4.897    3.664
Age Squared            0.969    0.967    0.974    0.979

R squared              0.357    0.384    0.357    0.307

Model 2: Full Model

                       1880    1900    1910     1920     1940     1950

Intercept             0.002   0.001   0.000    0.000    0.000    0.000
Age                   1.654   1.671   2.271    2.121    1.842    2.216
Age Squared           0.993   0.992   0.986    0.988    0.990    0.987
Marital Status
 Not Married (ref)    1.000   1.000   1.000    1.000    1.000    1.000
 Married             13.855  18.091  11.400   15.716   19.942   19.709
Place of Birth
 U.S.-born (ref)      1.000   1.000   1.000    1.000    1.000    1.000
 Foreign-born         1.405   1.576   1.741    1.344    0.544    1.288
Since Birth
 Yes (ref)            1.000   1.000   1.000    1.000    1.000    1.000
 Reg. Same as Birth   0.485   0.526   0.376    0.369    0.368    0.337
Region
 Northeast            0.437   0.686   0.689    0.686    0.583    0.603
 Midwest              0.632   0.923   0.876    0.943    0.818    0.990
 South                0.556   0.970   0.738    1.238    1.036    1.001
 West (reference)     1.000   1.000   1.000    1.000    1.000    1.000
Farm Residence
 Non-farm (ref)       1.000   1.000   1.000    1.000    1.000    1.000
 Farm                 0.781   0.437   0.640    0.475    0.627    0.500
% in School 15-19     0.986   1.006   1.005    1.008    1.005    0.999

R squared             0.340   0.359   0.381    0.414    0.427    0.477

Number of cases      62,026  12,286  61,862  121,635  159,270  184,511

                        1960     1970     1980     1990

Intercept              0.000    0.000    0.000    0.000
Age                    5.043    6.600    5.235    4.034
Age Squared            0.969    0.964    0.970    0.975
Marital Status
 Not Married (ref)     1.000    1.000    1.000    1.000
 Married              22.661   23.408   13.363    9.618
Place of Birth
 U.S.-born (ref)       1.000    1.000    1.000    1.000
 Foreign-born          0.667    0.521    0.648    0.833
Since Birth
 Yes (ref)             1.000    1.000    1.000    1.000
 Reg. Same as Birth    0.250    0.309    0.383    0.410
Region
 Northeast             0.597    0.627    0.447    0.509
 Midwest               0.847    0.904    0.722    0.731
 South                 0.951    0.937    0.764    0.732
 West (reference)      1.000    1.000    1.000    1.000
Farm Residence
 Non-farm (ref)        1.000    1.000    1.000    1.000
 Farm                  0.325    0.229    0.273    0.283
% in School 15-19      0.987    0.989    1.024    1.023

R squared              0.492    0.499    0.449    0.397

Number of cases      152,515  208,694  252,654  231,733

(*)Coefficients in boldface are significant at alpha=0.05 or less
Independent variables are described in Appendix C.
Table C.2

Logistic Regression Results, Age at Leaving Home, 1880-1990: Odds Ratios
for Black Males (*)

Model 1: Age and Age Squared only

                     1880     1900     1910    1920    1940    1950

Intercept           0.036      0.001  0.000   0.000   0.000   0.000
Age                 1.053      1.445  2.457   2.735   1.559   1.741
Age Squared         1.006      0.999  0.987   0.985   0.996   0.994

R squared           0.235      0.244  0.267   0.273   0.236   0.241

                      1960    1970    1980    1990

Intercept            0.000   0.000   0.000   0.000
Age                  2.464   2.734   2.497   2.124
Age Squared          0.986   0.984   0.986   0.988

R squared            0.276   0.288   0.266   0.175

Model 2: Full Model

                     1880     1900     1910    1920    1940    1950

Intercept           0.022  2,365.867  0.000   0.000   0.011   0.007
Age                 1.168      1.422  2.098   2.039   1.201   1.295
Age Squared         1.001      0.996  0.988   0.988   0.999   0.997
Marital Status
Not Married         1.000      1.000  1.000   1.000   1.000   1.000
Married             9.439     15.782  7.859   8.044   8.005   6.821
Place of Birth
U.S.-born (ref)     1.000      1.000  1.000   1.000   1.000   1.000
Foreign-born        0.925      0.705  0.771   2.495   0.633   0.948
Mig. Since Birth
Yes (ref)           1.000      1.000  1.000   1.000   1.000   1.000
Reg. Same as Birth  0.487      0.000  0.315   0.329   0.485   0.324
Region
Northeast           2.459      0.000  0.505   0.608   0.818   1.230
Midwest             1.468      0.000  0.321   0.604   0.765   1.168
South               2.278      0.000  0.651   0.951   1.491   1.881
West (ref)          1.000      1.000  1.000   1.000   1.000   1.000
Farm Residence
Non-farm (ref)      1.000      1.000  1.000   1.000   1.000   1.000
Farm                0.951      0.329  0.453   0.406   0.609   0.510
% in School 15-19   1.006      1.003  1.006   1.005   1.013   1.001

R squared           0.294      0.365  0.347   0.370   0.333   0.339

Number of cases     8,807      1,708  6,183  14,383  18,081  27,402

                      1960    1970    1980    1990

Intercept            0.000   0.000   0.000   0.000
Age                  1.930   2.397   2.749   2.212
Age Squared          0.989   0.985   0.982   0.986
Marital Status
Not Married          1.000   1.000   1.000   1.000
Married              7.229  10.354   6.768   5.719
Place of Birth
U.S.-born (ref)      1.000   1.000   1.000   1.000
Foreign-born         1.290   1.105   0.774   0.657
Mig. Since Birth
Yes (ref)            1.000   1.000   1.000   1.000
Reg. Same as Birth   0.379   0.425   0.383   0.419
Region
Northeast            0.757   0.586   0.791   0.701
Midwest              0.624   0.625   0.751   0.768
South                0.965   0.876   0.873   0.780
West (ref)           1.000   1.000   1.000   1.000
Farm Residence
Non-farm (ref)       1.000   1.000   1.000   1.000
Farm                 0.422   0.465   0.395   0.569
% in School 15-19    0.994   1.023   0.968   0.982

R squared            0.365   0.385   0.337   0.246

Number of cases     18,294  26,150  37,469  31,191

(*)Coefficients in boldface are significant at alpha=0.05 or less
Independent variables are described in Appendix C.
Table C.3

Logistic Regression Results, Age at Leaving Home, 1880-1990: Odds Ratios
for White Females (*)

Model 1: Age and Age Squared only

                       1880    1900    1910     1920     1940     1950

Intercept             0.000   0.000   0.000    0.000    0.000    0.000
Age                   1.971   2.177   2.821    3.028    3.629    4.773
Age2                  0.991   0.989   0.983    0.982    0.978    0.973

R squared             0.211   0.218   0.231    0.247    0.266    0.310

                        1960     1970     1980     1990

Intercept              0.000    0.000    0.000    0.000
Age                    8.801    8.468    6.518    4.582
Age2                   0.961    0.962    0.968    0.975

R squared              0.377    0.399    0.375    0.345

Model 2: Full Model

                       1880    1900    1910     1920     1940     1950

Intercept             0.004   0.002   0.000    0.000    0.000    0.000
Age                   1.447   1.434   1.777    1.831    2.282    2.374
Age Squared           0.994   0.994   0.990    0.989    0.985    0.984
Marital Status
 Not Married (ref)    1.000   1.000   1.000    1.000    1.000    1.000
 Married             39.662  37.089  25.984   25.206   23.684   21.392
Place of Birth
 U.S.-born (ref)      1.000   1.000   1.000    1.000    1.000    1.000
 Foreign-born         2.232   1.809   1.547    1.455    0.832    1.458
Mig. Since Birth
 Yes (ref)            1.000   1.000   1.000    1.000    1.000    1.000
 Reg. Same as Birth   0.681   0.554   0.560    0.575    0.590    0.527
Region
 Northeast            1.119   1.541   1.227    1.091    0.659    0.591
 Midwest              1.212   1.318   1.235    1.298    1.042    1.050
 South                1.124   1.267   1.015    1.429    1.070    1.073
 West (reference)     1.000   1.000   1.000    1.000    1.000    1.000
Farm Residence
 No-farm (ref)        1.000   1.000   1.000    1.000    1.000    1.000
 Farm                 0.625   0.476   0.697    0.656    0.592    0.742
% in School 15-19     0.998   1.011   1.016    1.019    1.000    1.005

R squared             0.440   0.451   0.454    0.467    0.469    0.477

Number of cases      60,752  12,343  57,506  122,258  161,003  190,989

                        1960     1970     1980     1990

Intercept              0.000    0.000    0.000    0.000
Age                    4.940    5.977    6.019    4.409
Age Squared            0.969    0.966    0.967    0.973
Marital Status
 Not Married (ref)     1.000    1.000    1,000    1.000
 Married              22.075   19.856   14.824   10.299
Place of Birth
 U.S.-born (ref)       1.000    1.000    1.000    1.000
 Foreign-born          1.062    0.647    0.575    0.646
Mig. Since Birth
 Yes (ref)             1.000    1.000    1.000    1.000
 Reg. Same as Birth    0.484    0.519    0.515    0.517
Region
 Northeast             0.743    0.674    0.459    0.590
 Midwest               1.146    1.027    0.759    0.824
 South                 1.132    0.984    0.817    0.801
 West (reference)      1.000    1.000    1.000    1.000
Farm Residence
 No-farm (ref)         1.000    1.000    1.000    1.000
 Farm                  0.469    0.299    0.315    0.297
% in School 15-19      1.025    1.017    1.042    1.031

R squared              0.504    0.510    0.470    0.429

Number of cases      154,992  213,530  249,252  225,118

(*)Coefficients in boldface are significant at alpha = 0.05 or less
Independent variables are described in Appendix C.
Table C.4

Logistic Regression Results, Age at Leaving Home, 1880-1990: Odds Ratios
for Black Females (*)

Model 1: Age and Age Squared only

                       1880        1900    1910    1920    1940    1950

Intercept             0.000       0.000   0.000   0.000   0.000   0.000
Age                   2.795       2.202   3.195   3.414   2.772   2.664
Age2                  0.983       0.988   0.980   0.979   0.983   0.984

R squared             0.222       0.211   0.220   0.242   0.216   0.215

                       1960    1970    1980    1990

Intercept             0.000   0.000   0.000   0.000
Age                   3.631   3.743   3.045   2.460
Age2                  0.978   0.978   0.983   0.986

R squared             0.261   0.308   0.295   0.222

Model 2: Full Model

                       1880        1900    1910    1920    1940    1950

Intercept             0.003  17,602.567   0.001   0.000   0.001   0.001
Age                   1.561       1.264   1.806   1.793   1.588   1.514
Age Squared           0.993       0.998   0.989   0.990   0.992   0.993
Marital Status
 Not Married (ref)    1.000       1.000   1.000   1.000   1.000   1.000
 Married             19.726      20.562  13.436  13.675  10.052   8.695
Place of Birth
 U.S.-born (ref)      1.000       1.000   1.000   1.000   1.000   1.000
 Foreign-born         0.722     999.000   0.665   1.385   0.648   1.041
Mig. Since Birth
 Yes (ref)            1.000       1.000   1.000   1.000   1.000   1.000
 Reg. Same as Birth   0.712       0.332   0.567   0.430   0.450   0.373
Region
 Northeast            1.080       0.000   1.214   1.800   0.925   1.120
 Midwest              0.962       0.000   0.679   1.099   0.727   1.010
 South                1.000       1.000   1.000   1.000   1.000   1.000
 West (reference)     1.000       1.000   1.000   1.000   1.000   1.000
Farm Residence
 Non-farm (ref)       1.000       1.000   1.000   1.000   1.000   1.000
 Farm                 0.584       0.426   0.499   0.573   0.648   0.736
% in School 15-19     1.002       0.994   1.006   1.008   1.008   1.003

R squared             0.373       0.398   0.368   0.393   0.353   0.341

Number of cases       9,438       1,790   6,830  16,076  20,309  31,227

                       1960    1970    1980    1990

Intercept             0.000   0.000   0.000   0.000
Age                   2.215   2.611   2.922   2.505
Age Squared           0.986   0.983   0.982   0.984
Marital Status
 Not Married (ref)    1.000   1.000   1.000   1.000
 Married              8.413   9.553   5.597   4.505
Place of Birth
 U.S.-born (ref)      1.000   1.000   1.000   1.000
 Foreign-born         1.140   1.135   0.697   0.740
Mig. Since Birth
 Yes (ref)            1.000   1.000   1.000   1.000
 Reg. Same as Birth   0.447   0.536   0.518   0.550
Region
 Northeast            1.144   0.906   0.942   0.811
 Midwest              1.000   0.894   1.018   0.997
 South                1.000   1.000   1.000   1.000
 West (reference)     1.000   1.000   1.000   1.000
Farm Residence
 Non-farm (ref)       1.000   1.000   1.000   1.000
 Farm                 0.522   0.550   0.332   0.476
% in School 15-19     1.004   1.035   0.996   0.982

R squared             0.370   0.410   0.352   0.273

Number of cases      20,384  28,795  40,992  34,020

(*)Coefficients in boldface are significant at alpha=0.05 or less
Independent variables are described in Appendix C.