Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Household and gender in a life-course perspective

David Kertzer
Department of Sociology and Anthropology, Bowdoin Brunswick, Maine, United States of America

One of the major developments in social science research over the past two decades has been the shift away from static approaches towards perspectives and methods that can shed light on the true flux of life. Of course, social change has long been a principal focus of social science. What is new is the increased attention to the interplay between the larger, societal forces of change and the actual experiences of people. Social scientists are moving away from simple models of "before" and "after" from traditional to modern, from pre-industrial to industrial, from rural to urban towards much more subtle formulations of how people's lifes are lived and how the course of their lives may vary as a result of macro-level developments.

The life-course perspective which arose from a confluence of movements in sociology and psychology in the 1960s subsequently came to involve economists, anthropologists, historians, and others. It looks at the distinctive series of roles and experiences through which the individual passes as she or he ages from birth to death and inquires into the impact of various changes on these patterns. Moreover, it asks how people's reactions to change affect larger societal forces which in turn influence the course of social change. In short, the perspective calls attention not only to the ways in which people's lives are influenced by broad economic, political, social, and cultural developments, but also to how the collective impact of individuals' reactions to these changes affects the course of subsequent change at the macro-level.

The Household, Gender, and Age (HGA) project of the United Nations University was designed as an attempt to provide better information on just how the many macro-level changes taking place in the developing countries of the world are affecting individuals in their family environment. The emphasis was on the experience of women, whose lives have often been little understood by the technicians and politicians most closely involved in development policy decisions.

Given these goals, the selection of a life-course approach to provide a common methodological framework for the individual country research projects made much sense. The life-course approach provides a sophisticated framework for empirical research without forcing specific, substantive questions on researchers whose regional and national settings may be very different. Hence, the individual researchers were able to formulate their own specific questions according to their local situation while still sharing a common, broader perspective on the relationship of macro-level change to the domestic situation of women. The life-course approach also encouraged the use of a wide range of data collection techniques and analysis, ranging from traditional life-history and other qualitative approaches to some of the most advanced and newly developed methods of longitudinal data-gathering and statistical analysis.

By assuring a dual research focus - the macro-forces affecting women's lives and the changing ways in which those lives are actually being lived - the life-course perspective provides a firm basis for understanding the human consequences of development. As the HGA researchers have so richly documented, many of these consequences are unintended, and some have been disastrous.

One of the major advantages of a life-course approach is the key it provides to understanding the various consequences of social change at different stages of a person's life. All too often development specialists and social scientists speak of the impact of macro-level changes as though everyone in a community equally were affected regardless of gender, age, or situation. However, the point of the life-course which the individual has reached when a change is introduced may have profound implications for the effect of that change on that person. This is obvious in the case of introducing a new school, for example, as this will presumably directly affect youngsters more than the others in a community. Virtually all types of externally induced changes are like this The result is that reference to positive versus negative changes are often misleading, since some changes may have positive implications for those at a certain point in their life-course and negative implications for those at other points.

A final reason for adopting a life-course perspective in the HGA project was that it provides valuable insights on domestic group processes, the "H" of the HGA. The household is not a monolithic unit, despite the fact that planners and social scientists often speak of it as such. Rather, the household should be understood as the continuously changing product of the interaction of the group of individuals of whom it is comprised. Interests vary: those of the male head of household may differ from his wife's; in a polygynous household, the interests of the various wives may differ from one another; the interests of the oldest child may differ from that of younger siblings. The life-course perspective brings this to light, showing the actual complexity of interacting human lives that lies behind the misleading unity of the household label.

The essence of a life-course approach lies in the continuous interplay between social change and the life-course of individuals. People begin their lives in one historical period distinguished by a characteristic set of cultural norms and perceptions and institutional arrangements, and, as they age, these larger forces change. Individuals begin to develop ideas of normative family behaviour and the proper trajectory of their lives at an early age, but when they come to face life later it may be in a society very different from the one from which these ideas and norms first sprung. For example, they develop their ideas of what the life of a grandparent is like and what the relations between grandparents and their children and grandchildren are like when they arc children, yet when they later become grandparents themselves they may be living in a very different society, where being a grandparent could have a very different significance (Riley, 1985).

A life-course approach implies the notion of a society that is structured by age. That is, roles in a society differ by age, so that one can imagine a society that is structured by age in much the same way as it is structured by class and gender. Each society has its own cultural timetable for the appropriate progress of the individual through the life-course. There are social pressures which encourage individuals to make transitions between roles (for example, from unmarried to married) at points that arc defined as culturally appropriate (Hagestad and Neugarten, 1985).

Scholars have long been interested in the life-cycle, the sequences of roles through which the individual passes from birth to death. However, there has been an unfortunate tendency to look at the range of current life-stages - infancy, childhood, adolescence, young adulthood, and so on - and assume that together they represent the stages of life through which any individual in the society passes. As a life-course perspective stresses, this would be true only in the case of a society that was entirely unchanging. Since no current societies arc known to fall into this category, this means that we cannot generalize from the synchronic description of individual life-stages to characterize the pattern of roles and experiences through which any individual (or any cohort) passes over the life-course. The life-course approach, then, puts a sense of change and history into life-cycle research.

A central concept in life-course analysis is the cohort or group of individuals who enter a given system or a given status at a particular time. A marriage cohort, for example, refers to all those who got married in a certain period. A birth cohort, which is the grouping used in this study, consists of all those who were born in the same time-span. Since different cohorts encounter different segments of time as they proceed through their lives, each is affected differently by historical change. One of the major aims of life-course research is to determine how changes are reflected in the way different cohorts age, and how the characteristics of different cohorts (their size, for example) may bring about historical changes (Riley et al., 1972).

This does not mean, however, that cohorts should be seen as homogeneous units. Cohort analysis involves not only comparison between different cohorts, but also the examination of variation within each cohort. Since members of the same cohort have different characteristics, they may be affected differently by the same historical change. For example, the introduction of mandatory elementary education may have a greater effect on women farmers or small traders, who depend on the labour of their young children, than on elite women, who do not. As this example indicates, changes affecting one cohort (in this case children at the time elementary education becomes compulsory) can have important implications for the life-course pattern of other cohorts (in this case, the mothers of these children). Similarly, wage-labour opportunities for young men may have a major impact on the status of the elderly, whose control over their juniors may be diminished. This linkage between cohorts is a major focus of lie-course research.

Life-course analysis introduces a historical perspective into the study of how people live from birth to death. When looking at the household - one of the primary links of social organization - the life-course approach forces us to see the real dynamism of people's lives and to avoid the reification of academic analytical categories. The constellation of people who compose any household is the result of several different factors, including cultural norms about what is an appropriate place to live, demographic variables such as the survival of parents, number of siblings, children, and grandchildren, and economic factors, like the advantages of housing large numbers of related adults together where the household is also the productive unit. Any individual's co-residential experiences are the result of the interaction of these different forces, which themselves are influenced by larger sources of historical change. Cultural norms regarding polygyny, for example, may change over time; decreasing mortality and changes in fertility and migration affect the demographic component of household composition; and changes in the economy lead to different pressures on households, which has a bearing on the advantages or disadvantages of large, complex family households.

A well-entrenched sociological approach to the study of the domestic living unit focuses on the developmental cycle of households, often identified with the family cycle. A major problem with the family-cycle approach is that in concentrating on how household units change over time, it often loses sight of what happens to individuals. For example, a family-cycle model may focus on the cycle involving the death of the elder generation and the passing on of the household to the eldest son. But what of the daughters and the younger sons? And what of those who had only daughters and no sons, or those whose parents died when they were young children? In short, the family-cycle approach tends to yield a picture of uniformity that is rarely found in practice (Kertzer, 1986).

By focusing on the actual experiences of individuals from birth to death, the life-course approach allows the research to capture the full diversity of their co-residence in households. It provides a better position to examine the impact of various changes such as those introduced by development schemes - on people's domestic lives. This cannot be done by simply following household units: it is necessary to look at the history of every individual over time, examining the choices he or she made and the changing economic, demographic, environmental, and cultural forces that influenced the person's decisions and experiences.

The life-course approach to co-residence is especially useful for studying in developing countries, where the Western notions commonly built into family-cycle models are least appropriate. This is most notable in the many cases where polygyny is a cultural norm, for family-cycle research is based almost entirely on the assumption of a simple marital dyed: a single husband with a single wife. Using a life-course approach to co-residence it is possible to follow the co-residential experiences of each individual in whatever kind of household she or he lives.

A life-course approach provides a broad framework for the study of how women's lives and family relations generally have changed in the third world. By focusing attention on the interrelationship between the individual and the ever-changing society, it avoids the pitfalls of examining only macro-level phenomena (such as economic development per se) on the one hand, or, on the other, the particular experiences of the individual unrelated to the broader societal context. Moreover, unlike a simple life-history approach, it provides a means for aggregating the experiences of individuals to test propositions about the nature of the change that is taking place.

While the life-course approach provided a common framework for HGA researchers working in different countries, it did not constrain them to limit their specific substantive focus. Rather, each team could identify developmental and other issues important in their particular societal context and use a life-course framework to design their study of them. The issues differed widely from country to country, but what unites the various studies of the HGA project is, on the one hand, a common interest in how larger societal changes influence the lives of women in their domestic context, and, on the other, the use of a life-course perspective to provide a common means for determining the nature of these changes and their implications.

The studies are also characterized by the multiple levels of comparison that are built into a life-course approach. These involve comparison between cohorts and between communities. The latter is, of course, a standard method in social science. For example, in the original Chinese report there were two levels of geographically based comparison. The first involved comparison between an area (in Sichuan province) that is less industrially and commercially developed than another (in Jiangsu province). The second involved comparisons within each area between larger, more centrally located communities and those that are smaller and more remote.

The distinctive feature of the comparisons in the HGA studies, though, is the use of cohort comparisons in addition to those geographically based. In order to trace the path of changes in people's lives, each national study involved selecting specified younger and older cohorts and comparing their life-trajectories. Such comparisons show how the life-courses of successive cohorts have varied as they have experienced change at different stages.

The method of cohort comparison employed in the HGA studies can be illustrated by reference to figure 1, which provides a hypothetical example of a comparative study of three birth cohorts. The movement of historical time is represented from left to right, with the passage of years shown at the bottom of the figure. The oldest cohort is cohort 1, the members of which were born around 1920, and whose last members are expected to die around 1995, when they are 75. The middle cohort, cohort 2, represents those born 20 years later, around 1940, while the youngest cohort, cohort 3, was born around 1960. The details of the cohort comparison method are simplified here for ease of explanation. Actual studies involve more broadly defined cohorts, for example, those born over a five- or ten-year period. Moreover, the more recent cohorts might be expected to have a longer life-span than the older cohorts.

Fig. 1.: Cohort comparison.

As the members of each cohort age, they pass from one stage of their life-course to a successive stage. For convenience, four general life-stages have been distinguished: child (0-16), young adult (17-39), mature adult (40-59) and old age (over 60). Of course, life-stages are themselves culturally defined and not biologically fixed; indeed, a common theme in life-course research is the question of how a people divide up the life-course (Keith, 1985; Kertzer and Keith, 1984). The four divisions found in figure 1 are therefore not to be interpreted as cross-culturally valid categories, but simply as a convenient device for explaining the cohort comparison method. At any given date, members of different cohorts are of different ages. Examples of such dates are by vertical lines in figure 1. For instance, by following the vertical line from 1970, it is clear that members of cohort I were mature adults, members of cohort 2 were young adults, and members of cohort 3 were children in that year A historical event affects members of the different cohorts at different stages of their lives and so potentially has a different effect on each of them. For example, a general famine in 1970 might have led to malnutrition and increased disease and mortality for all cohorts. This would be a period effect, affecting all cohorts in the same way. However, such a historical event would also be likely to have different long-term implications for the various cohorts. For instance, malnutrition at an early age may lead to a higher incidence of certain disabilities later in life with the result that the old-age experience of cohort 3 after the year 2000 would be different from the old-age experiences of cohort 1 in the 1980s and 1990s. This would represent a cohort effect.

In life-course studies, the focus is on how cohorts differ from one another as a result of historical change. If there were no change, presumably the experience of cohort I would be identical to that of cohorts 2 and 3. Their childhood and young adult experiences would all be the same, despite the fact that they took place in different periods. It was just such an assumption of an unchanging society that allowed earlier anthropologists and others to write studies of the life-cycle in traditional societies divided into chapters on infancy, childhood, marriage, and old age. However, it is clear that such an assumption can no longer apply to any society. It is not possible to generalize on various life-stages in such a simple way, since the same life-stage differs from one cohort to another.

Moving from this rather general theoretical level to considering how the HGA researchers could put this method into practice shows that what may seem to be a straightforward conceptual framework encounters complexities when it comes to the research itself. Consider, for example, a comparison between a group of women who are in their fifties and a group of women in their thirties. If the findings suggest that the older women were more oriented to family life than the younger women, who spent more time in labour outside the household, it might be tempting to conclude that this represents a life-course pattern and that young women go through a period of involvement in the outside labour force which they leave as they grow older. In fact, many earlier analysts made just such assumptions.

However, there are other possible interpretations of these findings which would acknowledge historical change in life-course patterns and hence different experiences for different cohorts. If, for example, a textile factory favouring young female workers had been established in the community in recent years, it could be that the pattern of extra-domestic labour for young women is new, and that the women in the older cohort never passed through such a stage. The life-course approach is sensitive to these important differences of interpretation and is designed to distinguish between such competing explanations of life-course patterns.

Cohort comparison, then, is comparing segments of the life-course which are experienced by different cohorts in different historical periods. For example, returning to figure 1 to examine the course of women's lives as young adults would involve investigating the experiences of cohort 1 in the period 1937-1960 and comparing it with the experiences of cohort 2 in 1957-1980 and cohort 3 beginning in 1977. This would reveal the results of historical change in this phase of the life-course of all the women concerned.

It should now be evident that a lie-course approach requires longitudinal data, that is, information on individuals over time, beginning at their birth and ideally continuing on until they die. This contrasts with most social research, which has relied heavily on collecting cross-sectional data on a population at a certain point in time. This is the typical approach in most survey research. The need for longitudinal data in life-course research places heavy demands on the researcher.

There are a variety of ways in which life-course data can be gathered. It can be collected through archive-based studies that use governmental, church, family, or other organizational records to reconstruct the lives of individuals. It can be done through prospective or panel studies that begin at a certain date and then continually return to the subject population to follow them as they age. Thirdly, there are retrospective studies that take place at a certain point in time but try to elicit information from individuals on their entire life-course from birth to the date of the interview. In the HGA, the last of these alternatives was adopted, since in many settings sufficient historical records for the first approach were not available, while the second approach would require a commitment to a multi-year (if not multi-decade) study which would take a long time to produce results and which, in any case, would require a massive budget.

The design adopted by the HGA resulted in the use of a common instrument in most of the component projects. This was a life-course matrix that was based on retrospective information on various dimensions of the lives of those interviewed, such as marriage, work, education, co-residence, and fertility. In a work history, for example, the person would be asked when she first went to work and what kind of job it was. She would then be asked the date or her age when she left that job and the reason she did so, followed by questions about her next job. Her work history, then, would consist of the dated series of jobs she held from the time she began work to the present.

The vertical dotted line in figure I represents a date (1985) for the collection of such retrospective life-course data. Note that the data collected for cohort 1 are much more complete than those gathered for cohort 2, which, in turn, are fuller than data gathered for the youngest cohort, cohort 3. In fact, the members of cohort 3 are only about 25 years old at the time of the interview, and so nothing can be known about their mature adult or older years, and it is impossible to compare all three cohorts for this portion of the life-course. It is possible to compare all three cohorts' childhood experiences, analysing historical differences and their causes, and it is also possible to compare the early years of young adulthood of all three cohorts. However, a study of mature adulthood is limited to a two-way comparison of cohorts 1 and 2.

One other methodological note to be kept in mind in conducting retrospective life-history research concerns the problem of selectivity bias. The older the members of the cohort being interviewed, the more likely they arc to be a highly selective subset of the original cohort. This is because survival to old age is not randomly distributed among members of a cohort, but is linked to a variety of factors in earlier life. For example, interviewing all living members of cohort 1 in 1985 means interviewing those who had survived to age 65. This could well be a minority of all those who had been original cohort members in 1920, since most of them might have died. Therefore any picture of the childhood experiences of cohort l based entirely on the retrospective reports of those members who were interviewed in 1985 is likely to reach erroneous conclusions if there is no allowance for selectivity bias.

To give an extreme example, imagine a society with no historical change, with each cohort following the same path and having the same characteristics over the life-course as every other cohort. Imagine, though, that half the population is poor and has high infant and childhood mortality rates. This half also has different kinds of early life-course experiences, for example more time in the labour force, than the richer half. Comparing the older cohort to the younger cohort on the basis of retrospective reports of what the youth of each was like would lead to the conclusion that the two cohorts had very different experiences, even though, in fact, there were no differences at all. Since few of the poorer people would have survived to be interviewed at 65, the picture of the childhood of the older cohort would be based largely on the experiences of the wealthy half of the population. There is no space here to go into the details of how such problems need to be handled. However, the problem of selectivity bias must be carefully dealt with in any analysis of retrospective data.

One of the valuable aspects of the life-course approach in the context of the HGA project is that it called for both qualitative and quantitative data. This, too, is reflected in the different individual country projects. While the life-history matrix described above calls for quantitative data, the interpretation of life-course experiences benefits greatly from supplementary qualitative information. Three kinds of approach - open-ended survey questions, participant observation, and life-histories are particularly useful here.

Open-ended survey questions complement close-ended questions on most surveys. By providing an opportunity for the person to respond in her own terms to a broad area of inquiry, such questions allow the researcher greater insight into how people perceive their own lives. They also suggest causal chains that the researcher might not have previously considered, some of which might subsequently be through more close-ended questioning of a broader sample.

Possibilities for participant observation varied from project to project, but it was possible to some degree in virtually all of them. Sharing the daily life experiences of the people in the study provides a richer understanding of their lives. Direct observations of behaviour lead to questions that might not otherwise arise, and observation also allows the researcher to distinguish between what people claim happens in certain situations and what in fact actually happens. More generally, sharing the daily lives of those concerned leads to greater acceptance by the local people and results in better-quality data, including those obtained through the administration of the more formal research instruments.

All of the HGA projects collected some intensive life-histories. In general, an attempt was made to collect life-histories from individuals representing different cohorts and different subgroups in each country study. Such information not only provides richer detail about the course of people's lives, but also allows for people's own interpretation of their lives and of the historical events that affected them. In the end, the best life-course analysis is often the one that can blend these various instruments to describe people's changing lives, to identify the factors that have led to the changes, and to discover the ways in which these changes are viewed by the people themselves.

Although the individual country studies vary in their emphasis on qualitative versus quantitative data, all of them did collect systematic quantitative data on life-course processes. While these data provide many advantages over simple survey-style cross-sectional data, their analysis does introduce new complexities associated with their longitudinal nature. Fortunately, the statistical analysis of longitudinal data has been one of the areas in the social sciences most actively developed in recent years, especially in relation to sociology and economics. Treatment of the technical issues involved in event-history analysis and other forms of statistical analysis for life-course data is beyond the scope of this chapter. However, the reader is referred to such works as Tuma and Hannan (1984), Allison (1984), and Kalbfleisch and Prentice (1980) for a presentation of many of the relevant statistical methods. Dennis P. Hogan, professor of sociology at Pennsylvania State University, served as principal consultant on analytical methods for the HGA project.

The use of quantitative life-course data also introduces complexities in data management. Event-history data, like that gathered for the HGA projects, entail collecting information on all relevant events like births, marriage and marriage dissolution, new jobs, and unemployment for each individual in the study throughout her life-course, that is, from birth to the time of the interview. This leads to a form of data very different from that found in typical cross-sectional studies. The typical survey, for example, consists of a set of questions, each of which has a list of possible responses. The resulting data set has a regular rectangular form: each variable occupies a certain column or columns, and each person contributes a single record to the database.

Event histories, on the other hand, do not have this simple structure. This is partly due to the fact that the lives of some people are more eventful than others. Person A may have had 20 different jobs while person B has never been in the workforce. At the same time, person A may never have been married, while person B has been married and divorced four times. To deal with an event-history database as if it were a cross-sectional database would require allowing, for each such variable as jobs and marriage, as many occurrences as are found in the life of the person with the most such events (the most jobs, the most marriages, and so on). The result would be an enormous database consisting very largely of blanks. If, for example, just one out of 1,000 respondents had had 40 different jobs, there would have to be space in each of the 1,000 records for information on 40 jobs, even though most people had had no more than three.

The solution adopted by many of the HGA projects was a computer database management system devised by Nancy Karweit, a sociologist at the Centre for Social Organization of Schools at Johns Hopkins University in Baltimore. The nature of this system is described in Karweit and Kertzer (1986). It involves dividing the life-course data into different domains (marriage, fertility, employment, and schooling, for example) and providing a separate record for every single event occurring to a person in each domain. Hence a person who married three times would have three marriage records, while a person who never married would have none. More than a way of simply economizing space in the data file, this system, known as CASA, provides a means for retrieving data to capture life-course processes and to permit statistical analysis. Versions of CASA have been prepared to run on both mainframe and IBM-compatible microcomputers.

With the tremendous changes that have been taking place in the developing countries of the world, interest has inevitably gone beyond broader questions of economic development, demographic change, and political revolution to the question of what these macro-forces mean to the people, to individuals. In looking at how lives have been changed by these historical developments, it is necessary to move beyond conventional sociological models which seek to ferret out relationships between variables (such as occupation and fertility) in a system that is assumed to be unchanging. A life-course perspective concentrates on the link between the larger field of economic, political, demographic, and institutional change on the one hand and the lives of individuals on the other. It avoids analyses which ignore the actual life experience of individuals and also goes well beyond micro-level studies that fail to relate the lives of individuals to the changing constellation of macro-forces that mould individual lives and lead to different patterns of experience for successive cohorts.

Life-course study represents a blending of old and new methods, of quantitative and qualitative instruments. These projects focused on the lives of women, but the life-course approach meant that their lives could only be validly seen in terms of the broader social context - as represented in marriage and kinship relationships, for example - that included men as well as women. While focusing much on how larger social and economic factors have altered the lives of women in their household setting, the life-course approach also reveals how individual women have responded creatively to these external pressures in fashioning satisfactory lives for themselves and their families. These innovations, introduced from below, in turn influence the course that younger women will follow later, thus having an impact on the course of economic development and the path of social change.


Allison, Paul D. 1984. Event History Analysis. Sage, Beverly Hills, Calif.

Hagestad, Gunhild O., and Bernice L. Neugarten. 1985. Age and the Life Course. In: Robert H. Binstock and Ethel Shanas, eds., Handbook of Aging and the Social Sciences. 2nd ed. Van Nostrand Reinhold, New York.

Kalbfleisch, John D., and Ross L. Prentice. 1980. The Statistical Analysis of Failure Time Data. Wiley, New York.

Karweit, Nancy, and David I. Kertzer. 1986. A Life Course Approach to Co-residence for Life Course Family Research. In: David I. Kertzer, ed., Family Relations in Life Course Perspective. Current Perspectives on Aging and the Life Cycle, vol. 2. JAI Press, Greenwich, Conn.

Keith, Jennie. 1985. Age in Anthropological Research. In: Robert H. Binstock and Ethel Shanas, eds., Handbook of Aging and the Social Sciences. 2nd ed. Van Nostrand Reinhold, New York.

Kertzer, David I. 1986. A Life Course Approach to Co-residence. In: David I. Kertzer, ed., Family Relations in Life Course Perspective. Current Perspectives on Aging and the Life Cycle, vol. 1. JAI Press, Greenwich, Conn.

Kertzer, David I., and Jennie Keith, eds. 1984. Age and Anthropological Theory. Cornell University Press, Ithaca, N.Y.

Riley, Matilda White. 1985. Age Strata in Social Systems. In: Robert H. Binstock and Ethel Shanas, eds., Handbook of Aging and the Social Sciences. 2nd ed. Van Nostrand Reinhold, New York.

Riley, Matilda White, Marilyn Johnson, and Anne Foner. 1972. Aging and Society. Sociology of Age Stratification, vol. 3. Russell Sage Foundation, New York.

Tuma, Nancy B., and Michael T. Hannan. 1984. Social Dynamics: Models and Methods. Academic Press, Orlando, Fla.

Contents - Previous - Next