A New Approach to Explaining Fertility Patterns: Preference Theory

Author(s): Catherine Hakim

Source: Population and Development Review, Vol. 29, No. 3 (Sep., 2003), pp. 349-374
Published by: Population Council
Stable URL: http://www.jstor.org/stable/3115278 .
Accessed: 10/06/2011 04:50

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=popcouncil. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Population Council is collaborating with JSTOR to digitize, preserve and extend access to Population and
Development Review.

http://www.jstor.org
A New Approach to
Explaining Fertility
Patterns: Preference
Theory

CATHERINE HAKIM

PATTERNSof family life and declining fertility in modern societies

CHANGING
have attracted substantial research attention and policy debate, yet we seem
no nearer to a full understanding of current trends, nor to explanations
that are adequate enough to provide the basis for predictions. Preference
theory was elaborated by the author in Work-Lifestyle Choicesin the 21st Cen-
tury (2000). This article shows how it provides a new theoretical frame-
work for understanding current changes in modern societies and for pre-
dicting future developments; constitutes a qualitative break from economic
theories of fertility change; and provides an alternative basis for the devel-
opment of family policy. It presents findings from a national survey in Brit-
ain designed to test preference theory's predictions regarding fertility and
employment, with positive results.
One notable feature of current research in demography and the soci-
ology of the family is the absence of any central guiding theory on the rela-
tive importance of childbearing in women's lives. Some argue there is con-
vergence on the nuclear (or conjugal) family, while others claim that
urbanization, industrialization, and changes in women's economic position
are fragmenting the family. Economists insist that economic factors are the
engine driving social change, while some sociologists and demographerssug-
gest that ideological factors are now crucial. One common theme is rising
female employment and changes in the status of women, but this is vari-
ously treated as both cause and effect. Some studies posit that women's
economic independence is the cause of declining marriage rates (and hence
lower fertility), while others show that marriage rates are higher for col-
lege-educated women with high-grade occupations and good earnings
(Oppenheimer 1977, 1988, 1994, 1997; Oppenheimer and Lew 1995; Ma-
son and Jensen 1995; Brewster and Rindfuss 2000; McDonald 2000).

POPULATION AND DEVELOPMENT REVIEW 29(3):349-374 (SEPTEMBER 2003) 349

350 EXPLAINING FERTILITY PATTERNS

Two coterminous trends-rising female employment rates and falling

fertility-are linked in economic theories (Easterlin 1976; Becker 1981;
Lesthaeghe 1998; Lesthaeghe and Willems 1999) and lead to the plausible
idea that more family-friendlyemployment policies could solve the problem
of decliningfertility(Chesnais1996;McDonald2000; Castles2003). The search
for the social, economic, and institutional determinants of fertility has led to
research on the macro-level correlatesof (declining) fertilityin cross-national
comparisons (OECD2001; Kohler, Billari, and Ortega 2002; Castles 2003).'
There are few longitudinal studies, and few qualitative studies of micro-level
processes-Gerson's (1985) brilliant study of women's choices between ca-
reer and family life being a rare exception.
Some social scientists are now emphasizing the role of ideological fac-
tors in family change and fertility trends. The results of the World Fertility
Survey led Cleland to conclude that an ideational theory of change was
more appropriate than a structural theory of change in family formation
and use of birth control (Cleland 1985: 243)-a conclusion that was criti-
cized on the ground that it ignored the massive structuraltransformationac-
companying these changes (Demeny 1987: 345-346). However, Lesthaeghe
and Meekers (1986) as well as Lesthaeghe (1995) and other contributors to
Mason and Jensen's (1995) collection also focus on changing attitudes and
values as the driving force in contemporary developments. At least one so-
ciologist explains low fertility in Spain by attitude change rather than the
usual economic factors of high unemployment and housing costs (Dfez-
Nicolas 2001). This theoretical perspective has also led Mason and Jensen
(1995: 7-9) to criticize researcherswho treat entire nations as a meaningful
unit of analysis instead of adopting a micro-level focus on individual actors
and their goals.
In contrast with much recent research on explaining fertility patterns,
preference theory emphasizes personal values and decisionmaking at the mi-
cro-level. However, it also specifies for the first time in the theoretical litera-
ture the particularsocial, economic, and institutional contexts within which
preferences become the primary determinant of women's choices. The key
break with other theories is that women are understood to be heterogeneous
in lifestyle preferences, whereas economic theories always assume (often im-
plicitly) that women are homogeneous within countries and across Western
societies, so that one-size-fits-all theories and policies work well. Even theo-
ries that emphasize ideational factors usually assume the relative homoge-
neity of values, or claim that preferences cannot be measured.
Preference theory breaks with current perspectives in several other
ways as well. Where fertility rates have fallen in the past, this has been
largely a consequence of male decisionmaking. After the contraceptive revo-
lution of the 1960s, female control of fertilityreplacesmale control. Women's
views, perspectives, and goals become the key to understanding current and
CATHERINE HAKIM 351

future changes in women's position in society and changing patterns of fer-

tility. This fundamental change requires new types of research as well as
new theories.
There are three striking features of demographic research and debates
on fertility decline. First, they are all variable-centered. For example, writ-
ers discuss whether delayed age of marriage or longer years of education
are a cause of lower fertility rates, with almost no reference to the social
processes and the motivations of the women and men behind these statisti-
cal measures. In contrast, preference theory relies on person-centered analy-
sis. A key feature of this approach is that it reveals how variables may have
a hugely different impact, depending on the context (social group or situa-
tion) in which they occur. Person-centered analysis recognizes the hetero-
geneity of respondents; denies that people are homogeneous in their re-
sponses to social and economic influences and experiences; and often focuses
on extreme cases, which can amount to 20 percent at either end of a distri-
bution (Magnusson and Bergman 1988; Cairns,Bergman, and Kagan 1998;
Magnusson 1998).
Second, all variables, including time, are treated as continuous, un-
marked by historical time, and equivalent across countries. This feature fol-
lows from variable-centeredanalysis. Substantivedifferencesbetween coun-
tries or between time periods are reduced to variationson a series of variables
that measure fragments of social reality and social change. In contrast, pref-
erence theory insists on case-by-case analysis of countries, to identify
whether they have achieved the new scenario for women (as defined in
Table 2).
Third, with rare exceptions, almost no attention is paid to women's
intentions, values, and motivations, and to how these differ from those of
men. The emphasis is usually on social institutions and public policy, macro-
level factors that policymakers believe they can influence (McDonald 2000
is one example). In contrast, preference theory places attitudes and values
at the center of causal explanations.
Unfortunately, most large studies continue to focus on process vari-
ables rather than causal factors. Just one example is the series of Fertility
and Family Surveys (FFS) organized in the 1990s by the United Nations
Economic Commission for Europe. A brilliant research design and model
questionnaire were ignored by most of the participating countries, which
collected process variables in exhaustive detail, but generally omitted the
"unessential" attitudinal questions that would have enabled them to ex-
plain their results. Spain was one of the few countries to implement the
model questionnaire unaltered, and one of the few to offer any explana-
tion for declining fertility (Delgado and Castro Martin 1998). Most FFS na-
tional reports are essentially descriptive statistical compendia. This adher-
ence to routine demographic statistics is all the more remarkablegiven that
352 EXPLAINING FERTILITY PATTERNS

a solid base of empirical evidence on the long-term impact of values and

life goals is already available.

The long-term impact of values and

personal goals
From the 1960s onward, the National Longitudinal Surveys (NLS) in the
United States provided longitudinal data that allowed the long-term influ-
ence, or insignificance, of aspirations and life goals to be measured rigor-
ously-in this case preferences for a life centered on marriage and children
or an employment careet. Of particular interest is the cohort of young
women aged 14-24 in 1968, who were interviewed almost every year up
to 1983 when aged 29-39 years. This cohort was asked in 1968, and again
at every subsequent interview, what they would like to be doing when they
were 3 5 years old, whether they planned to be working at age 35, or whether
they planned to marry, keep house, and raise a family at age 35.2 Com-
pared to the length and complexity of work-commitment questions included
in some surveys (Bielby and Bielby 1984), the question is crude and conflates
preferences and plans. But because it asked about women's personal plans,
rather than generalized approval/disapprovalattitudes on sex-role ideology
(Hakim 2003b, 2003c), the question turned out to have strong analytical
and predictive power, and it was used again in the second United States
youth cohort study initiated in 1979.
There are a number of independent analyses of the extent to which
early work plans were fulfilled by age 35 in the United States in the 1980s.
They all show that women achieved their objectives for the most part, re-
sulting in dramatic "mark-ups"to career planners in terms of occupational
grade and earnings (Mott 1982; Rexroatand Shehan 1984; Shaw and Shapiro
1987). Furthermore, career planners were more likely to choose typically
male jobs, had lower job satisfaction than other women (like men gener-
ally),3 and adapted their fertility behavior to their work plans (Waite and
Stolzenberg 1976; Stolzenberg and Waite 1977; Spitze and Waite 1980).
Work plans were a significant independent predictor of actual work behav-
ior. After controlling for other factors affecting labor force participation, a
woman who consistently planned to work had a probabilityof working that
was 30 percentage points higher than did a woman who consistently planned
not to work. Of the women who held consistently to their work plans, four-
fifths were actually working in 1980, at age 35, compared to only half of
the women who consistently intended to devote themselves exclusively to
homemaker activities. Women who planned to be working at age 35 were
likely to do so unless they had large families or a preschool child. Women
who had planned a "marriagecareer"nevertheless were obliged to work by
economic factors in half the cases: their husband's low income, divorce, or
CATHERINE HAKIM 353

the opportunity cost of not working led half to be at work despite aiming
for a full-time homemaker role.
Planning to work yielded a significant wage advantage. Women who
had consistently planned to work had wages 30 percent higher than those
of women who never planned to work. Those women who had planned to
work in the occupation they actually held at age 35 had even higher wages
than women whose occupationalplans were not realized.Women who made
realistic plans and acquired necessary skills fared best in the labor market.
Those who fared worst were women who planned an exclusive homemak-
ing career but ended up working for economic reasons. However, career
planners were only one-quarter of the cohort of young women; the vast
majority of the cohort had unplanned careers (Table 1), as did women in
the older NLS cohort aged 30-44 years at the start of the study in the late
1960s (Mott 1978, 1982; Shaw 1983). One-quarter of women planned, or
aspired to, a marriage career,4but as usual this group was not analyzed in
any detail.
As far back as the 1980s, the NLS longitudinal data overturned the
results of cross-sectional studies suggesting that women's work behavior is
heavily determined by the number and ages of any children, rather than
the other way round. Those who work only if their family responsibilities
permit them to do so are in effect fulfilling a prior choice of emphasis on
the homemaker career. Fertility expectations have only a small negative
effect on young women's work plans, whereas work plans exert a powerful
negative effect on young women's childbearingplans (Waite and Stolzenberg
1976; Stolzenberg and Waite 1977; Sproat, Churchill, and Sheets 1985: 78).
Factors that have long been held to determine women's labor force partici-
pation, such as other family income, educational qualifications, marital sta-

TABLE 1 Young women's work plans and outcomes in the United States
Distribution Percent working
of sample at age 35
Homemakercareer: 28 49
consistentlyindicatesno plans for work;aim is
marriage,family, and homemakingactivities
Driftersand unplanned careers: 47 64
(a) highly variableresponsesover time,
no clearpatternin plans for age 35 35
(b) switch to having future work expectations
at some point in their 20s 12
Careerplanners: 25 82
consistentlyanticipateworkingat age 35
throughout their 20s
SOURCE:Derived from Tables 2 and 3, reporting National Longitudinal Surveys data for the cohort of young
women first interviewed in 1968, when aged 14-24 years, in Shaw and Shapiro (1987: 8-9).
354 EXPLAINING FERTILITY PATTERNS

tus, and age of youngest child, were revealed as being most important in
relation to women with little or no work commitment, who have so far
been in the majority. The minority of women with definite career plans
manifested a rather inelastic labor supply, similar to that of men (Shaw and
Shapiro 1987).5
Overall the NLS results have repeatedly shown the importance of moti-
vations, values, and attitudes as key determinants of employment patterns,
occupational status, and even earnings, an influence that is independent of
conventional human capital factors and frequently exceeds the influence of
behavioral factors (Pares 1975; Andrisani 1978; Mott 1982; Sproat, Churchill,
and Sheets 1985). These "psychological" variables are usually omitted from
sociological and economic research, so their importance has been overlooked,6
and they are never studied for women who pursue the marriage career.
Similar results emerge from other longitudinal studies, on the rare oc-
casions when researchers address the long-term impact of values and life
goals. Attitudes have a specially strong impact on women's behavior today,
because women have gained genuine choices regarding employment ver-
sus homemaking. But attitudes and values have also been shown to have a
major impact among men as well. For example, Szekelyi and Tardos ana-
lyzed 20 years of Panel Study of Income Dynamics (PSID) microdata for
1968-88 to show that people who plan ahead and express confidence and
optimism about their plans subsequently earn significantly higher incomes
than those who do not, after controls. The long-term effects of attitudes
were stronger than short-term effects. Attitudes affected the earnings of both
male heads of households and their wives (Szekelyi and Tardos 1993).
Similar results are reported by Duncan and Dunifon (1998) from an-
other analysis of 24 years of PSID data, this time covering men only. Motiva-
tion (as measured in the respondents' early 20s) had a large impact on long-
term success (as measured by hourly earnings 16-20 years later), and the
effect remained after controlling for other factors. Only a small part of the
impact of motivation worked through its effect on greater investment in train-
ing and education; a substantial part remained after this control. The study
showed that values commonly found among women, such as religiosity (as
measured by church attendance) and a preference for affiliation (as measured
by a preference for friendly and sociable work settings rather than challeng-
ing work settings) both had a negative effect on earnings. Work orientations
that emphasized challenge rather than affiliation and a clear sense of per-
sonal efficacy boosted earnings in the early 40s.
In sum, there is already substantial evidence that attitudes, values, and
life goals have an important impact on outcomes in adult life, for men as well
as women, in modern societies. However, there has so far been no attempt to
integrate this new knowledge into social scientific theory, and empirical studies
routinely ignore these substantive findings.7 Preference theory builds on these
CATHERINE HAKIM 355

results to provide a new model of women's life goals and priorities that helps
to predict employment patterns and fertility patterns.

Preference theory
Preference theory is a new approach to explaining and predicting women's
choices between market work and family work, a theory that is historically
informed, empirically based, multidisciplinary, prospective rather than ret-
rospective in orientation, and applicable in all rich modern societies (Hakim
2000). Lifestyle preferences are defined as causal factors that need to be
monitored in modern societies. In contrast, other social attitudes (such as
patriarchal values and sex-role ideology) are either unimportant as predic-
tors of behavior or else have only a very small impact in creating a particu-
lar climate of public opinion on women's roles (Hakim 2003b, 2003c).
Preference theory specifies the historical context in which core values
become important predictors of behavior. It notes that five historical changes
collectively produce a qualitatively new scenario for women in rich mod-
ern societies in the twenty-first century, giving them options that were not
previously available (Table 2). Small elites of women born into wealthy fami-
lies or into prosperous families with liberal ideas sometimes had real choices
in the past, just as their brothers did. Today, genuine choices are open to
women in the sense that the vast majority of women have choices, not
only particular subgroups in the population. The five social and economic
changes started in the late twentieth century and are now producing a quali-
tatively different scenario of options and opportunities for women in the
twenty-first century.
These changes are historically specific developments in any society.
They are not automatic and do not necessarily occur in all modern societ-
ies. They may not occur together, at a single point in time in a country. The
timing of the five changes varies greatly between countries. The effects of
the five changes are cumulative. The two revolutions-the contraceptive
revolution and the equal opportunities revolution-are essential and con-
stitute the core of the social revolution for women. The five changes collec-
tively are necessary to create a new scenario in which women have genu-
ine choices and female heterogeneity is revealed to its full extent.
With rare exceptions (Cleland 1985; Murphy 1993; Castles 2002), male
demographers have generally overlooked the social and psychological sig-
nificance for women of what Westoff and Ryder (1977) term the contracep-
tive revolution. Demographers discuss the use of contraception without dis-
tinguishing between the methods controlled by men and those controlled
by women. Modern forms of contraception (the pill, IUD, and female steril-
ization) are thus defined primarily by their greater reliability, overlooking
the crucial fact that they transfer control over reproduction from men to
356 EXPLAINING FERTILITY PATTERNS

TABLE 2 The central tenets of preference theory

1. Five separate historical changes in society and in the labor market that started in the
late twentieth century are producing a qualitatively different and new scenario of
options and opportunities for women. The five changes do not necessarily occur in
all modern societies and do not always occur together. Their effects are cumulative.
The five causes of a new scenario are:
-the contraceptive revolution, which, from about 1965 onward, gave sexually
active women reliable control over their own fertility for the first time in history;
-the equal opportunities revolution, which ensured that for the first time in history
women had equal access to all positions, occupations, and careers in the labor
market. In some countries, legislation prohibiting sex discrimination went further,
to give women equal access to housing, financial services, public services, and public
posts;
-the expansion of white-collar occupations, which are far more attractive to
women than most blue-collar occupations;
-the creation of jobs for secondary earners, people who do not want to give
priority to paid work at the expense of other life interests; and
-the increasing importance of attitudes, values, and personal preferences in the
lifestyle choices of affluent modern societies.
2. Women are heterogeneous in their preferences and priorities vis-a-vis the conflict
between family and employment. In the new scenario they are therefore heteroge-
neous also in their employment patterns and work histories. These preferences are
set out, as ideal types, in Table 3. The size of the three groups varies in rich modern
societies because public policies usually favor one or another group.

3. The heterogeneity of women's preferences and priorities creates conflicting interests

between groups of women: sometimes between home-centered women and work-
centered women, sometimes between the middle group of adaptive women and
women who have one firm priority (whether for family work or employment). The
conflicting interests of women have given a great advantage to men, whose interests
are comparatively homogeneous; this is one cause of patriarchy and its dispropor-
tionate success.
4. Women's heterogeneity is the main cause of their variable responses to social
engineering policies in the new scenario of modern societies. This variability of
response has been less evident in the past, but it has still impeded attempts to predict
women's fertility and employment patterns. Policy research and future predictions
of women's choices will be more successful if they adopt the preference theory
perspective and first establish the distribution of preferences between family work
and employment in each society.

SOURCE:Hakim (2000).

women. Control over their fertility produces a change of perspective among

women, even a psychological change, creating a sense of autonomy, respon-
sibility, and personal freedom that is not achieved with contraception con-
trolled by men. The contraceptive revolution is thus an essential precondi-
CATHERINE HAKIM 357

tion in order for the equal opportunities revolution and other changes to
have any substantial effect on women's lives. Qualitative studies of contra-
ceptive practice using the old methods show clearly that women did not feel
they had any control over their childbearing and had fatalistic rather than
calculating attitudes (Fisher 2000). It follows that fertility decline prior to
the contraceptive revolution of the 1960s is attributableprimarilyto changes
in male values and priorities, inter alia, rather than to female values. As
Greene and Biddlecom (2000) point out, it is odd that demographers have
focused on women in fertility research, even in periods and countries where
men were the crucial decisionmakers responsible for contraceptive practice.
In Western Europe, North America, and other modern societies, these
five changes only took place from the 1960s onward. The timing and pace of
change have varied, even between countries in Europe. However, the strong
social, cultural, economic, and political links between modern countries sug-
gest that no country will lag behind on any of the changes indefinitely. All
five changes were completed early in the United States and Britain, so that
the new scenario was well established by the last two decades of the twenti-
eth century in these two countries. Thus they provide the main illustration
of the consequences of the new scenario for women. Reviews of recent re-
search evidence (Hakim 1996, 2000) show that once genuine choices are
open to them, women choose among three differentlifestyles:adaptive,work-
centered, or home-centered (Table 3). These divergent lifestyle preferences
are found at all levels of education and in all social classes.
Adaptivewomenprefer to combine employment and family work with-
out giving a fixed priority to either. They want to enjoy the best of both
worlds. Adaptive women are generally the largest group and will be found
in substantial numbers in most occupations. Certain occupations, such as
schoolteaching, are attractive to women because they facilitate a more even
work-family balance. The great majority of women who transfer to part-
time work after they have children are adaptive women, who seek to de-
vote as much time and effort to their family work as to their paid jobs. In
some countries (such as the United States and southern European coun-
tries) and in certain occupations, part-time jobs are still rare, so other types
of job are chosen. For example, seasonal jobs, temporary work, or school-
term-time jobs all offer a better work-family balance than the typical full-
time job, especially if commuting is also involved. Fautede mieux, adaptive
women sometimes take ordinary full-time jobs.
Work-centered women are in a minority, despite the massive influx of
women into higher education and into professional and managerial occu-
pations in the last three decades. Work-centered people (men and women)
are focused on competitive activities in the public sphere-in careers, sports,
politics, or the arts. Family life is fitted around their work, and many of
these women remain childless, even when married. Qualificationsand train-
 358 EXPLAINING FERTILITY PATTERNS

TABLE 3 Classification of women's work and lifestyle preferences in the

twenty-first century
Home-centered Adaptive Work-centered
20% of women 60% of women 20% of women
varies 10%-30% varies 40%-80% varies 10%-30%

Family life and This group is most diverse Childless women are
children are the and includes women who concentrated here.
main priorities want to combine work and Main priority in life
throughout life. family, plus drifters and those is employment or
with unplanned careers. equivalent activities
in the public arena:
politics, sports, arts, etc.
Prefer not to work. Want to work, but not totally Committed to work or
committed to work career. equivalent activities.
Qualifications Qualifications obtained with Large investment in
obtained as the intention of working. qualifications/training
cultural capital. for employment or other
activities.

Number of children This group is very responsive Responsive to

is affected by to government social policy, economic opportunity,
government employment policy, political opportunity,
social policy, equal opportunities policy, artistic opportunity,
family wealth, economic cycle/recession/ etc.
etc. growth, etc. Not responsive to
Not responsive Such as: social/family policy.
to employment income tax and social welfare
policy. benefits
educational policies
school timetables
childcare services
public attitude toward working
women
legislation promoting female
employment
trade union attitudes toward
working women
availability of part-time work and
similar work flexibility
economic growth and prosperity
and institutional factors generally.
SOURCE:Hakim (2000).

ing are obtained as a career investment rather than as an insurance policy,

as in the adaptive group. The majority of men are work-centered, com-
pared to only a minority of women, even women in professional occupa-
CATHERINE HAKIM 359

tions (Hakim 1998: 221-234, 2003a: 183-184). Preference theory predicts

that men will retain their dominance in the labor market, politics, and other
competitive activities, because only a minority of women are prepared to
prioritize their jobs (or other activities in the public sphere) in the same
way as men. This is unwelcome news to many feminists, who have assumed
that women would be just as likely as men to be work-centered once op-
portunities were opened to them, and that sex discrimination alone has so
far held women back from the top jobs in any society.
The third group, home-centered or family-centered women, is also a mi-
nority, and a relatively invisible one in the Western world, given the cur-
rent political and media focus on working women and high achievers.
Home-centered women prefer to give priority to private life and family life
after they marry. They are most inclined to have larger families, and these
women avoid paid work after marriage unless the family is experiencing
financial problems. They do not necessarily invest less in acquiring qualifi-
cations, because the educational system functions as a marriage market as
well as a training institution. Despite the elimination of the sex differential
in educational attainment, an increasing proportion of wives in the United
States and Europe are now marrying men with substantially better qualifi-
cations, and the likelihood of marrying a graduate spouse is greatly increased
if the woman herself has obtained a degree (Hakim 2000: 193-222). This
may be why women remain less likely to choose vocational courses with a
direct economic value, and are more likely to take courses in the arts, hu-
manities, or languages, which provide cultural capital but have lower earn-
ings potential.
The three preference groups are set out, as sociological ideal-types, in
Table 3, with estimates of the relative sizes of the three groups in societies,
such as Britain and the United States, where public policy does not bias the
distribution. In this case, the distribution of women across the three groups
corresponds to a "normal" distribution of responses to the family-work con-
flict. In practice, in most societies, public policy favors one group or another,
by accident or by design, so that the percentages vary between modern soci-
eties, with a bias toward work-centered women or toward home-centered
women.
Each of the three lifestyle preference groups has a substantively differ-
ent value system, as well as differing life goals. These differences sometimes
bring women into conflict with each other-for example, on whether pub-
lic childcare services are necessary or not, whether positive discrimination
in favor of women for promotion to top jobs is a good thing or not. In a
sense, there is no single, representative group of women in modern society,
but three contrasting, even conflicting groups with sharply differentiated
work and lifestyle preferences. In the United States, the conflict between
work-centered and home-centered women has been expressed through the
360 EXPLAINING FERTILITY PATTERNS

two women's movements: the feminist women's liberation movement and

the maternalist movement, with conflict often focused on the issues of abor-
tion and the proposed Equal Rights Amendment.
The United States, Britain, and probably also the Netherlands currently
provide the prime examples of societies that have achieved the new sce-
nario for women. This does not mean that sex discrimination has been en-
tirely eliminated in these countries. As definitions of sex discrimination keep
expanding, from direct discrimination to increasingly arcane forms of indi-
rect discrimination, this battle is arguably never won. However, these coun-
tries have trenchant equal opportunities legislation, backed up and enforced
by a system of tribunals, equal opportunities commissions, and other tan-
gible public and political support for converting the letter of the law into
reality. Most European countries still have little or nothing to actively en-
force equal opportunities legislation, so that little has changed in practice.
For example, in Greece, Italy, and Spain, there is evidence of informal bar-
riers to women's access to the labor market: female unemployment rates
are more than double those of males, and there is some evidence of the
disparity widening over time rather than falling. Within the European Union,
only Britain, Ireland, and the Netherlands have a public body responsible
for enforcing equal pay and equal opportunities laws. Equally important,
Britain and the United States both have large and diverse populations, en-
suring that cultural diversity and differences in values become accepted and
even welcomed. Some European countries (notably the Scandinavian coun-
tries) have not yet come to terms with the ethnic, religious, and cultural
diversity that generally ensues from decades of immigration, and they have
low acceptance of diversity in values and lifestyles.
Identifying Britain and the United States as two countries that have
achieved the new scenario for women does not mean we expect conver-
gence in employment rates and lifestyle choices in these countries. Even
the most liberal society and laissez faire polity still has institutions, laws,
customs, national policies, and cultures that shape and structure behavior.
Choices are not made in a vacuum. Social and economic factors still matter,
and will produce national variations in employment patterns and lifestyle
choices. In addition, the choices people make are molded by an unpredict-
able constellation of events: economic recessions and booms, wars, changes
of government, as well as events in private lives, individual ability, acci-
dents or ill health, "disastrous" marriages, and "brilliant" marriages. For ex-
ample, Britain and the United States differ in the size and character of their
part-time workforce. Universal and free access to health care in Britain means
that people are free to choose their job and working hours, and even whether
to work at all, without regard to any health benefits offered by employers.
In the United States, health insurance benefits are a key feature of full-time
jobs that biases choices away from part-time work, or non-work. As a re-
CATHERINE HAKIM 361

suit of these differences, some adaptive women in Britain will choose per-
manent part-time jobs or decide not to be employed at all, while some of
their counterparts in the United States will choose full-time jobs. Jobless
single mothers and their children receive exactly the same health care as
everyone else in Britain. A variety of social and economic factors guarantee
there will always be differences between the new scenario countries in pat-
terns of fertility and employment.
In sum, lifestyle preferences determine:
-women's employment pattern over the life cycle: choices between
careers and jobs, full-time and part-time work, and associated job values;
-women's fertility: the incidence of childlessness and, for the major-
ity who do have children, family sizes;
-women's responsiveness to public policies, employer policies, eco-
nomic and social circumstances, including pronatalist policies.
Preferences do not predict outcomes with complete certainty, even
when women have genuine choices, because of variations in individual abili-
ties and factors in the social and economic environment. However, in pros-
perous modern societies, preferences become a much more important de-
terminant, maybe even the primary determinant of women's behavior.
The three lifestyle preference groups differ in values, goals, and aspi-
rations. That is, they are defined by their contrasting lifestyle preferences
rather than by behavioral outcomes. The three groups also differ in consis-
tencyof aspirationsand values, not by strong versus weak preferences. People
who argue that women's choices are always shaped by external events and
the situation around them are describing adaptive women, who are in the
majority. The distinctive feature of the two polar groups of women (and
equivalent men) is that they do not waver in their goals, even when they
fail to achieve them. Work-centered people are defined by prioritizingmar-
ket work (or other competitive activities in the public sphere) over family
work and family life, not by exceptional success in the public sphere.

The 1999 British survey

Preference theory is empirically based, in that it was built up from a review
and synthesis of hundreds of social science studies in several disciplines us-
ing a variety of research methods (Hakim 1996, 2000). To test the impact of
lifestyle preferences on employment and fertility, a new survey was carried
out in 1999. The aim was to select the smallest possible number of survey
questions and indicators appropriate to a structured interview survey that
could be used to identify the three lifestyle preference groups among women.
This had previously been done most effectively by qualitative studies based
on in-depth interviews, as illustrated by Gerson's landmark study of how
women decide about motherhood and careers (Gerson 1985: Table C22;
362 EXPLAINING FERTILITY PATTERNS

Hakim 2000: 149-154). Our aim was to identify classificatoryquestions and

variables that might be included in any large survey.
The survey was carried out as one of 27 projects selected for an Eco-
nomic and Social Research Council (ESRC) Research Programme on the
Future of Work running over five years 1998-2003 in Britain. The inter-
view survey was carried out for the author by the Office of National Statis-
tics (ONS) in Britain in January and February 1999.
The survey was based on a probability random sample of households
and face-to-face interviews with one person aged 16 and older chosen ran-
domly within each household. The proportion of households in which the
selected informant was the head of household or spouse was 81 percent in
our sample. From a sample of 5,388 eligible addresses, an overall response
rate of 68 percent was achieved, producing data for a nationally represen-
tative sample of 3,651 persons aged 16 and older in Britain. The final sample
included 1,691 men and 1,960 women, with a substantial proportion (20
percent) aged 65 and older. Excluding the pensioners reduces the sample
for the population of working age to 2,900, including 2,345 married and
cohabiting couples.
The survey was used to operationalize the identification of lifestyle
preferences in the context of a large-scale structured interview survey, to
test the classification against women's lifestyle choices and behavior, and to
explore further applications of the taxonomy in sociological research on
women's employment.8
The tables presented in the following sections are from the 1999 Brit-
ish survey.9

Preferences and lifestyle choices

Three questions were used to operationalize lifestyle preferences. Two ques-
tions were taken from the Eurobarometer series.10The third, a question on
work commitment, has been widely used, in slightly different versions, in
research on work orientations in the United States and Britain. All three
questions produced results in line with those obtained in Eurobarometer
and other surveys.
A question on ideal family models identifies home-centered women:
women who prefer to focus their time and energy on home and family work,
and thus seek a marriage with complete role segregation. Just under one-
fifth of the sample fell into this category.
Two questions on work orientations identify people for whom market
work is central to their identities and lifestyle. A question on work commit-
ment identifies people who claim they would continue with paid work (not
necessarily in the same job) in the absence of economic necessity. The in-
troduction of a national lottery in Britain in the 1990s made this hypotheti-
cal situation more realistic than previously. Primaryand secondary earners
CATHERINE HAKIM 363

were identified by a question asking about the main income-earner(s) in

the household. People who classified themselves as sole or joint main
earner(s) were classified as primary earners; all others were classified as
secondary earners. The question was treated as an opinion question, and
analyses of responses show clearly that that is what it is. For example, mar-
ried men adopt the identity of primary (co)earner irrespective of income
level and even when they are not in employment. In contrast, women who
regard themselves as primary earners when single switch immediately to
the secondary earner identity after marriage, almost irrespective of their
income level. Work centrality is defined as a combination of adopting a pri-
mary earner identity and having nonfinancial commitment to one's paid
work. For married women, this means in practice those who regard them-
selves as joint main earners as well as being committed to their employ-
ment activities. Less than one-fifth of married women passed this test, and
overall only one-quarter of women (compared to half of men) were classi-
fied as work-centered. The residual group of women with more complex,
or contradictory, values was classified as adaptive.
The distribution of lifestyle preferences among women of working age
(Table 4) and wives of working age (Table 5) is close to that suggested by
preference theory (Table 3). The distributionvaries slightly according to the
population base. For example, among wives and cohabitants aged 20-59
years, the distributionbecomes 13 percent home-centered, 77 percent adap-
tive, and only 10 percent work-centered.

TABLE 4 Characteristics of women in the three lifestyle preference groups

Home- Work-
centered Adaptive centered
% employed
full-time 40 35 63
part-time 16 37 15
% not in employment 44 28 22
% married/cohabiting 71 80 45
average number of children
aged <16 at home 1.28 1.02 .61
% left full-time education
by age 16 54 55 42
17-20 years 28 28 32
age 21+ 18 17 26

Base=100% 171 870 194

National distribution
of the three groups (%) 14 70 16

NOTES: Women aged 20-59 who have completed their full-time education. The fertility indicator is shown for
marriedand cohabitingwomen aged20-54 years.
364 EXPLAINING FERTILITY PATTERNS

In line with preference theory, Table 4 shows that lifestyle choices dif-
fer substantially between the three preference groups. Two-thirds of work-
centered women are in full-time employment. In contrast, two-thirds of
adaptive women work part-time or not at all. Almost half of the home-
centered women are not in employment, and a small minority have never
had a job. A relatively high 40 percent of home-centered women have full-
time jobs. The reasons for this unexpected result are explored in the full
report, and show that in certain circumstances economic necessity can over-
ride personal preferences (Hakim 2003a: Chapters 5 and 8).
Home-centered and adaptive women are most likely to marry or co-
habit and to stay married. This is not surprising, as their preferred lifestyle
is heavily dependent on having a breadwinner spouse who is in regular
employment. Work-centered women are least likely to marry and most likely
to be separated or divorced. Women who regard themselves as financially
independent anyway have less motive to marry and to stay married. Most
important, home-centered women have twice as many children as work-
centered women, many of whom seem to be childless. The fertility mea-
sure here is the "own child" measure: the average number of children be-
low age 16 years living at home per woman aged 20-54 years. It does not
include older children (who may no longer live at home anyway), so it
understates total fertility. The measure shows clearly that fertility levels vary
markedly between the three preference groups, along with marriage rates
and employment patterns.
Educational standards differ between the three preference groups, but
not widely. Work-centered women are slightly more likely to have higher
education: 26 percent compared to about 18 percent in each of the other
two groups. The difference is small enough to be explained by differential
self-selection into higher education. As predicted by preference theory,
lifestyle preferences cut across education groups as well as socioeconomic
and income groups. Overall, the key features of the three lifestyle prefer-
ence groups are in line with preference theory. In broad terms, preferences
predict outcomes. Further analysis (Hakim 2003a: 134-137, 153-160) shows
that attitudes predict behavior, but that behavior does not predict attitudes.
That is, attitudes are not a post hoc rationalization for decisions already taken.

Careers and fertility

Analysis of the British survey shows that lifestyle preferences are even more
important than educational qualifications in shaping women's choice be-
tween market work and a life centered on family and children. The theses
that women's fertility declines with increasing social status and that higher
education invariably leads women to become career-oriented are not sup-
ported by the evidence.
CATHERINE HAKIM 365

TABLE 5 The relative importance of lifestyle preferences and

education: Full-time employment rates
Percent working full time Distribution
Highly Other Highly Other
qualified women qualified women

Lifestyle preferences:
home-centered (28)a 34 11 14
adaptive 48 26 71 78
work-centered 93 65 18 8
All wives 20-54 54 30 100 100
Base=100% 119 589 119 589

a( ) indicatessmallbasenumbers.
NOTES:Marriedwomen aged 20-54 who have completedtheir full-time education.In the absenceof
informationon educationalqualifications,the highly qualifiedare definedas those completingtheir full-time
educationat age 21 and later,becausein Britainfirstdegreesare normallycompletedby age 21. People
completingfull-timeeducationat age 20 or earlierare assumedto have qualificationsbelow tertiary
educationlevel.

The analysis in Tables 5 and 6 is restricted to wives because women's

choices only become sharply defined, and can only be implemented, after
marriage to a breadwinner spouse.1 The analysis is also restricted to wives
aged 20-54 years because women aged 55 and older rarely have children
below age 16 at home, and because many women (and men) quit the labor
market from age 55 onward. We focus on the choice between a career, as
indicated by full-time employment (Table 5), and a life centered on family
work, as indicated by the fertility measure (Table 6).
The possession of higher education qualifications is a good general in-
dicator of women's social status. It is an indicator of women's self-confi-

TABLE 6 The relative importance of lifestyle preferences and

education: Fertility
Average number of Percent with no
children <16 years at home children at home
Highly Other All Highly Other All
qualified women women qualified women women

Lifestyle preferences:
home-centered (2.0)a 1.18 1.29 (8)a 42 37
adaptive 1.00 1.09 1.08 39 42 42
work-centered .71 .62 .65 68 64 65
All wives 20-54 1.08 1.08 1.08 41 44 44
Base=100% 119 589 708 119 589 708

a( ) indicatessmallbase numbers.
NOTES:Marriedwomen aged20-54 who have completedtheirfull-timeeducation.See notes to Table5.
366 EXPLAINING FERTILITY PATTERNS

dence and self-assertiveness; of their potential earnings if they choose to

work; and a rough indicator of their socioeconomic status, either through
their own job or through their husband's status. Tables 5 and 6 show virtu-
ally no effect of education or social status on wives' career orientation and
fertility levels.
Education does have an effect on employment: full-time employment
rates are 24 percentage points higher among highly qualified women. How-
ever, lifestyle preferences are far more important-as a determinant of both
employment and fertilitylevels. Work-centeredwives have much higher full-
time employment rates than home-centered (or adaptive) wives, whether
they are highly educated or not. Small base numbers mean the results for the
home-centered highly qualifiedwomen are not entirely reliable,but the pat-
tern is consistent and strong in both groups of women.
Fertilityamong home-centered women is double the level among work-
centered women. Again the differences are even largeramong highly quali-
fied women, with our index of fertility almost tripled compared with work-
centered wives. Overall, lifestyle preferences are more important than the
variables more commonly measured in surveys, such as education or social
status. It appears that lifestyle preferences are the hidden, unmeasured fac-
tor that determines women's behavior to a very large extent.
The bifurcated results in Tables 5 and 6 are supported by other studies
displaying the polarization of women's lifestyle preferences and activities
(Hakim 1996, 2000: 84-156). Most recently, a study of childbearing pat-
terns among young women with and without higher education qualifica-
tions found a polarization of childbearingbehavior that is most pronounced
among highly qualified women, who generally began motherhood some
five years later than others. Graduate women were most likely to remain
childless: 23 percent compared to 15 percent of the nongraduate women.
However, graduate women were also more likely than nongraduate women
to have a second and third child, and to do so quickly (Rendall and Small-
wood 2003).

Policy implications
Preferencetheory offersa new approachto policy development, one that takes
account of the diversity of lifestyle preferences instead of adopting the usual
one-size-fits-all approach.Policiesthat treat women as a homogeneous group
are bound to fail or to work poorly. In contrast, policies that are designed to
be neutral between the three preference groups, offering each of them a flex-
ible benefit, are likely to be highly successful in terms of take-up rates and
political popularity. The policy implications of preference theory are set out
more fully elsewhere (Hakim2000: 223-253). Here, we consider those poli-
cies that might slow or reverse declining fertilityrates.
CATHERINE HAKIM 367

At present, social policy and family policy generally focus on the work-
ing mother and ignore home-centered women. It is often argued that ma-
ternity leave (unpaid or paid) helps women to combine paid work with
having children. However, a preference theory perspective clarifies that it is
mainly work-centered women (and to a lesser extent adaptive women) who
benefit from maternity leave and related job rights-that is, women who
have the lowest fertility and the lowest probability of increasing it.
Governments that are serious about raising fertility rates (and few are,
as Demeny [19871 points out) should focus instead on policies to support
home-centered women, who have the highest fertility rates and can most
easily be persuaded to increase their family size. Such policies would also
benefit those adaptive women who lean toward the family, rather than to-
ward market work, as their main priority. In practice, the focus of social
and family policy has swung so far toward the working mother that there is
now a systematic policy bias against nonworking mothers in most modern
societies-most obviously in relation to single mothers.
Until recently, policymakers accepted that it was in children's best in-
terest for the sole parent to be a full-time parent, even if this meant long-
term dependence on welfare, social housing, and other benefits. Policy has
now swung against full-time mothers. Single mothers in the United States
are pressed to accept low-paid and unrewarding jobs in Welfare-to-Work
schemes. In Britain, similar pressures push single mothers into jobs on the
grounds that they are "better off" psychologically and financially. Publicity
for such schemes underlines the low public esteem accorded to full-time
mothers and parenting work generally, and reinforces the idea that full-
rights citizenship is dependent on gainful employment, however low-status
and low-paid. Crittenden (2001) maintains that in the United States full-
time parenting tends to be equated with "doing nothing."
Governments find it difficult to accord reproductive work the same
status, dignity, and value as productive work. This is probably because gov-
ernments and public policy are still male-dominated, even in modern soci-
eties, and men insist on treating women's unpaid reproductive work as
taken-for-granted, "natural"women's work that does not merit the same
valuation and rewards as male-style productive work. The bottom line in
public policy is that women should provide reproductive services for free.
Unfortunately many women have absorbed this phallocentric thesis, lead-
ing to policies that deny professional fees to surrogate mothers, denigrate
sex workers, deride couples who pay to adopt children, and disdain other
activities involved in the industrializationof sexual and reproductivework.'2
Similarly, many governments have fiscal policies that discriminate against
single-earner families, including Britain and Sweden.
Preference theory exposes the bias against motherhood in current fis-
cal, social, and family policies. It also helps us to identify policies that are
368 EXPLAINING FERTILITY PATTERNS

neutral between the three preference groups. One example is the homecare
allowance introduced in various forms in the 1990s in Finland, Norway,
and France (Ilmakunnas 1997; Hakim 2000: 232-235). The homecare al-
lowance is a salary paid to the mother (or any parent) who stays at home to
care for children without using public daycare nurseries. It is separate from
financial benefits paid for dependent children, which are intended to help
parents with the extra costs of children. The homecare allowance can be
regarded as a wage for childcare at home-as a partialreplacement for earn-
ings forgone-or it can be used as a subsidy for purchased childcare services
that enable the parent to return to work, whether full-time or part-time.
The scheme has been hugely popular wherever it is introduced, with take-
up rates close to 100 percent even in the early years, unlike the much-
publicized parental leave rights. The value of the homecare allowance var-
ies between countries and schemes, but is never nominal. For example, in
Finland the homecare allowance for one child amounts to 40 percent of the
average monthly earnings of female employees. The allowance is a public
statement of the social value accorded to full-time parenting and the dig-
nity of motherhood. By raising the social status of motherhood as com-
pared with paid jobs, it redresses the bias against motherhood as an activity
and can affect fertility rates.

Conclusions
There is no shortage of empirical research on current trends in family for-
mation and family arrangements, fertility rates, and women's employment.
What has been lacking is a theoretical framework that helps us to under-
stand social processes and foresee future developments, including responses
to changes in public policy. Preference theory does all this. The initial for-
mulation (Hakim 2000) is open to further revision and development in the
light of new research (Hakim 2003a).
Preference theory breaks with contemporary demographic theory in
two ways. Its first distinctive feature is the recognition that the contracep-
tive revolution of the 1960s, and several other recent social and economic
changes, create a new scenario of opportunities and options for women.
This is a fundamental and radical change in women's position in society
and the lifestyle choices open to them. Male demographers (and many other
social scientists) have tended to assume that motherhood is a natural, even
biologically determined choice for women, and that the high levels of fertil-
ity seen in the past were "normal."They have failed to recognize that sexu-
ally active heterosexual women had no direct control over their fertility,
and thus had little choice about the shape of their lives, prior to the intro-
duction of modern methods of contraception. The contraceptive revolution
gave women independent control of their fertility, if necessary without the
agreement or cooperation of male partners, for the first time in history.
CATHERINE HAKIM 369

The current emphasis on the first and second demographic transitions

overlooks the fundamental importance of the recent sea change in fertility
control. When women control their own fertility, it is their preferences and
values that shape responses to public policy. And public policy has not, in
practice, paid much attention to women's wishes. All the evidence is that
most women in modern societies want two, or at most three, children
(Chesnais 1996: 736; Fahey 2001). The large families and high fertility lev-
els of the past were unavoidable rather than chosen, or were chosen by
men, and are unlikely to recur naturally. New theories are needed for this
new situation. Romantics like to believe that couples decide jointly, but in
practice one partner has always had the overriding vote. Women have now
become the deciding factor, and replacement fertility levels will only be
achieved if women are satisfied with the terms and conditions offered for
their reproductive and childrearingrole-an activity that spans 20-30 years
rather than the 3-4 years that are the narrow focus of preschool childcare
policy debates. Sharpfalls in fertilityeven in familisticsocieties such as Spain,
Italy, and Greece demonstrate that in many countries public policy has not
yet produced the right terms and conditions for motherhood, once women
can choose to avoid it.
The second distinctivefeature of preference theory is the recognition of
female heterogeneity in preferences for a life centered, like that of men, on
employment or other competitive activities in the public sphere or else a life
centered on the noncompetitive activitiesof privatelife. It is this heterogeneity
of lifestyle preferencesthat impedes attempts to predictfertilityafter the con-
traceptiverevolutionand the equal opportunitiesrevolutionhave given women
genuine choices over the shape of their lives for the firsttime in history.
The appearance of voluntary childlessness after the contraceptive revo-
lution, raising childlessness to around 20 percent in most modern societies,
has generally been ignored by demographers. It disproves the unstated as-
sumption that women will always want to have and rear children. Some do
not. Similarly there are women who want to pursue careers in the same
way as men, but again they are a minority rather than typical. The typical
woman is adaptive and is highly responsive to social pressures and policy
signals. At the moment, all the signals point to paid employment as the
favored activity. The equal opportunities revolution has hugely improved
the rewards, terms, and conditions in the labor market. In contrast, the so-
cial status and rewards of motherhood have stood still (at best) or, more
often, declined. The social stigma of solo motherhood has vanished, only to
be replaced by the social and economic stigma of female joblessness. In this
context, work-centered women often choose childlessness, adaptive women
are not encouraged to have children, and home-centered women feel them-
selves to be ignored or belittled.
It is possible to reverse the decline in fertility in modern societies, but
policymakers will have to start with a revaluation of motherhood and re-
370 EXPLAINING FERTILITY PATTERNS

productive work, and change the emphasis in public policy. The bias to-
ward support for working mothers needs to be balanced by new measures
to support home-centered women. The emphasis on family-friendly em-
ployment policies should be replaced by a diversity of policies supporting
all groups of women.

Notes
1 Unfortunately cross-national compari- women enteringthe labormarketin the 1980s
sons can lead to the ecological fallacy,as illus- had stronger work expectations and work
tratedby Castles's(2002, 2003) conclusionthat commitment than did previous cohorts. In
high fertility and high female employment 1979, young women were only half as likely
rates are causallylinked at the individuallevel as young women in 1968 to say they expected
because they appear to be linked in analyses to be housewives not in the paid labor force at
of national, aggregatedatasetsthat include the age 35, with only one-quarter planning to be
Scandinaviancountries. Macro-level analyses housewives (Sproat, Churchill, and Sheets
are inappropriatefor social policies that will 1985: 76-78, 318, 335-336).
operate at the micro-level;such analyses need 5 An even largerAmericanstudy, the Na-
direct information on social processesand de- tional LongitudinalStudy of the High School
cisionmakingat the individuallevel. Class of 1972 (NLS72), has produced results
2 Two versions of the question have been that corroboratethose of the NLS. It showed,
used in the NLS.In the initial 1968 survey, re- for example, that young women who subse-
spondentswere asked"NowI would like to talk quently became mothers before the age of 25
to you about your future plans. What would differedsignificantlyfromthose who remained
you like to be doing when you are 35 years childless:they were less work-oriented, more
old?" From 1969 onward the question was likely to plan to be homemakers at age 30, less
modified to read "Now I would like to talk to likely to plan a professional career, and held
you about your futurejob plans. What kind of more traditionalsex-role attitudesand aspira-
work would you like to be doing when you are tions before they gave birth. Parenthood
35 years old?"In both versions,keeping house strengthenspreexistingtraditionalattitudesin
or raisinga familywas a possibleresponse.This young white men and women alike (Waite,
particularquestion was omitted when the NLS Haggstrom,and Kanouse 1986; Morgan and
was administeredto women aged 35 yearsand Waite 1987). None of these resultsis acknowl-
older. edged or reflectedin recent analyses of NLS72,
3 In all modern societies, studies invari- which compare men's and women's occupa-
ably show higher levels of job satisfaction tional aspirationswhile ignoringsex-role atti-
among women than among men, even when tudes, to the point that women hoping to be
women are in jobs that have lower status and full-time homemakersat age 30 are simply ex-
lower pay than men's jobs. Preferencetheory cluded from analyses (e.g., in Rindfuss,Cook-
explains these resultsby the qualitativediffer- sey, and Sutterlin 1999).
ence between men's and women's work and 6 Some studies have shown work com-
lifestyle priorities(Hakim 1996:70, 100-101). mitment to have a much bigger effect among
It is thus not surprisingthat the minority of married women than other social structural
women who adopt the typicallymale careerist factors,especiallywhen the husband'sattitudes
approachto employment have lowerlevels of are also taken into account. Geerkenand Gove
job satisfaction, like men generally, even show that these two factors produce a 50-70
though they are generally in higher grade and percentage point increase in economic activ-
higher paidjobs than are typicalfor women. ity ratesof wives in the United States (Geerken
4 The large new youth cohort survey ini- and Gove 1983: 66). A study of Canadian
tiatedin 1979 showed that the cohortof young working wives also found strong associations
CATHERINE HAKIM 371

between work commitment, higher-status substantiveimportanceof their results, which

jobs, and husband'ssupportfor his wife's em- may be minimal.
ployment (Chappell 1980). 10 The Eurobarometerseries of surveys
7 For example, recent NLS analyses con- are run by the European Commission to in-
tinue to treat childcareresponsibilitiesas a key form European Union policymaking. They
determinantof women's laborforce participa- cover all EU member states and focus on so-
tion, as illustratedby Charleset al. (2001). cial and politicalattitudes.
8 A full reporton the 1999 Britishsurvey 11 The analysis was repeated for women
and an equivalent 1999 Spanish survey is who were neither marriednor cohabiting. In
given in Hakim (2003a). This report includes this subgroup, there is little variation in full-
data for men and analyses of sex differences. time employment rates and fertility between
9 Tests of statistical significance are not preferencegroups.Only in the context of mar-
reported. These tests indicate whether the riagedo the three preferencegroups show dif-
ferentiatedbehavior patterns.
sample size was large enough for a small per-
centage difference to be reliable. We know 12 One reason women resistthe industri-
that our survey was large enough to produce alizationand commercializationof sexual and
reliableresults.More to the point, we are only reproductivework is that such steps introduce
interested in large and substantively impor- competitive values into an area of social life
tant differencesbetween the three preference currently dominated by cooperative and car-
groups, not in small but statistically signifi- ing values. These contrastingvalue systems are
cant differences, which can be ignored. Too features of the market economy and the do-
often, as noted by Morrison and Henkel mestic economy respectivelyand not, as some
(1970), researchersuse tests of statisticalsig- sociologists have assumed, characteristicsof
nificance as a substitute for addressing the men and women per se.

References
Andrisani, Paul J. 1978. WorkAttitudesand LaborMarketExperience: Evidencefrom the National
LongitudinalSurveys.New York and London: Praeger.
Becker, Gary S. 1981, 1991. A Treatiseon theFamily.Cambridge,MA: HarvardUniversity Press.
Bielby, Denise D. V. and William T. Bielby. 1984. "Work commitment, sex-role attitudes
and women's employment," AmericanSociologicalReview49: 234-247.
Brewster, Karin L. and Ronald R. Rindfuss. 2000. "Fertilityand women's employment in
industrialized nations," Annual Reviewof Sociology26: 271-296.
Cairns, Robert B., LarsR. Bergman, and Jerome Kagan (eds.). 1998. Methodsand Modelsfor
StudyingtheIndividual.London and Thousand Oaks: Sage.
Castles, Frank G. 2002. "Three facts about fertility: Cross-national lessons for the current
debate," FamilyMatters63: 22-27.
. 2003. "The world turned upside down: Below replacement fertility, changing pref-
erences and family-friendly public policy in 21 OECDcountries," Journalof European
SocialPolicy,forthcoming.
Chappell, N. L. 1980. "Paid labor: Confirming a conceptual distinction between commit-
ment and identification," Workand Occupations 7: 81-116.
Charles, Mariaet al. 2001. "Thecontext of women's market careers:A cross-national study,"
Workand Occupations 28: 371-396.
Chesnais, Jean-Claude. 1996. "Fertility,family, and social policy in contemporary Western
Europe," Populationand DevelopmentReview22: 729-739.
Cleland, John. 1985. "Marital fertility decline in developing countries: Theories and the
evidence," in J. Cleland and J. Hobcraft (eds.), ReproductiveChangein DevelopingCoun-
tries:Insightsfrom the WorldFertilitySurvey.Oxford: Oxford University Press, pp. 223-
252.
372 EXPLAINING FERTILITY PATTERNS

Crittenden, Ann. 2001. ThePriceof Motherhood: WhytheMostImportantJob in the Worldis Still

the LeastValued.New York:Henry Holt and Company.
Delgado, Margarita and Teresa Castro Martin. 1998. Encuestade Fecundidady Familia 1995
(FFS),Opiniones y Actitudes No. 20. Madrid: Centro de Investigaciones Sociol6gicas.
Demeny, Paul. 1987. "Pronatalistpolicies in low-fertility countries: Patterns, performance,
and prospects," in Kingsley Davis et al. (eds.), Below-Replacement Fertilityin Industrial
Societies.Cambridge:CambridgeUniversity Press, pp. 335-358.
Diez-Nicolas,Juan. 2001. CausasyConsequencias del RecienteDescensode la Fecundidaden Espaha,
paper presented to the Seventh Congress of the Spanish Sociological Federation,
Salamanca, September.
Duncan, Gregg J. and R. Dunifon. 1998. "Soft skills and long run labor market success,"
Researchin LaborEconomics17: 123-149.
Easterlin, Richard A. 1976. "The conflict between aspirations and resources," Population
and DevelopmentReview2: 417-42 5.
Fahey, Tony. 2001. "Trendsin Irish fertility rates in comparative perspective," Economicand
SocialReview32: 153-180.
Fisher, Kate. 2000. "Uncertainaims and tacit negotiation: Birth control practices in Britain,
1925-50," Populationand Development Review26: 295-317.
Geerken, Michael and Walter R. Gove. 1983. At Homeand At Work:TheFamily'sAllocationof
Labor.Beverly Hills, CA: Sage.
Gerson, Kathleen. 1985. HardChoices:How WomenDecideabout Work,Careerand Motherhood.
Berkeley and Los Angeles: University of CaliforniaPress.
Greene, Margaret E. and Ann E. Biddlecom. 2000. "Absent and problematic men: Demo-
graphic accounts of male reproductive roles," Populationand DevelopmentReview26:
81-115.
Hakim, Catherine. 1996. KeyIssuesin Women'sWork:FemaleHeterogeneity and the Polarisation
of Women'sEmployment.London: Continuum Press.
1998. SocialChangeand Innovationin the LabourMarket.Oxford: Oxford University
Press.
.2000. Work-Lifestyle Choicesin the 21st Century:PreferenceTheory.Oxford: Oxford Uni-
versity Press.
2003a. Modelsof the Familyin ModemSocieties:Idealsand Realities.Aldershot: Ashgate.
. 2003b. "Lifestyle preferences and patriarchal values: Causal and non-causal atti-
tudes," in Janet Z. Giele and Elke Holst (eds.), ChangingLifePatternsin WesternIndus-
trial Societies.New York:Elsevier.
. 2003c. "Public morality versus personal choice: The failure of social attitude sur-
veys," BritishJournalof Sociology54: 339-345.
Ilmakunnas, Seija. 1997. "Publicpolicy and childcare choice," in I. Persson and C. Jonung
(eds.), TheEconomicsof the Familyand FamilyPolicies.London: Routledge, pp. 178-193.
Kohler, Hans-Peter, Francesco C. Billari, and Jos6 Antonio Ortega. 2002. "The emergence
of lowest-low fertility in Europe during the 1990s," Populationand DevelopmentReview
28:641-680.
Lesthaeghe, Ron. 1995. "The second demographic transition in Western countries: An in-
terpretation," in Karen Oppenheim Mason and An-Magritt Jensen (eds.), Genderand
FamilyChangein IndustrializedCountries.Oxford: Clarendon Press.
. 1998. "On theory development: Applications to the study of family formation,"
Populationand DevelopmentReview24: 1-14.
Lesthaeghe, Ron and Dominique Meekers. 1986. "Value changes and the dimensions of
familism in the European Community," EuropeanJournalof Population2: 225-268.
Lesthaeghe, Ron and Paul Willems. 1999. "Islow fertility a temporary phenomenon in the
European Union?," Populationand DevelopmentReview25: 211-228.
Magnusson, David. 1998. "Thelogic and implications of a person-oriented approach,"in R.
B. Cairns, L. R. Bergman, and J. Kagan (eds.), Methodsand Modelsfor Studyingthe Indi-
vidual. London and Thousand Oaks: Sage, pp. 33-64.
CATHERINE HAKIM 373

Magnusson, David and Lars R. Bergman. 1988. "Individualand variable-based approaches

to longitudinal research on early risk factors,"in M. Rutter (ed.), Studiesof Psychosocial
Risk:ThePowerof LongitudinalData. Cambridge and New York: Cambridge University
Press, pp. 45-61.
Mason, Karen Oppenheim and An-Magritt Jensen (eds.). 1995. Genderand FamilyChangein
IndustrializedCountries.Oxford: Clarendon Press.
McDonald, Peter. 2000. "Gender equity in theories of fertility transition," Populationand
DevelopmentReview26: 427-439.
Morgan, S. Philip and Linda J. Waite. 1987. "Parenthoodand the attitudes of young adults,"
AmericanSociologicalReview52: 541-547.
Morrison, Denton E. and Ramon E. Henkel. 1970. TheSignificanceTestControversy: A Reader.
Chicago: Aldine.
Mott, Frank L. (ed.). 1978. Women,Workand Family.Lexington, MA: D.C. Heath.
.1982. The EmploymentRevolution:YoungAmericanWomenof the 1970s. Cambridge,
MA: MITPress.
Murphy, Michael . 1993. "The contraceptive pill and women's employment as factors in
fertility change in Britain 1963-1980: A challenge to the conventional view," Popula-
tion Studies47: 221-243.
OECD. 2001. "Balancing work and family life: Helping parents into paid employment," in
EmploymentOutlook.Paris:OECD,pp. 129-166.
Oppenheimer, Valerie Kincade. 1977. "Thesociology of women's economic role in the fam-
ily," AmericanSociologicalReview42: 387-406.
1988. "Atheory of marriage timing," AmericanJournalof Sociology94: 563-591.
. 1994. "Women's rising employment and the future of the family in industrial soci-
eties," Populationand DevelopmentReview20: 293-342.
. 1997. "Women's employment and the gain to marriage:The specialization and trad-
ing model," Annual Reviewof Sociology23: 431-453.
Oppenheimer, Valerie Kincade and Vivian Lew. 1995. "American marriage formation in
the 1980s: How important was women's economic independence?," in Karen
Oppenheim Mason and An-Magritt Jensen (eds.), Genderand FamilyChangein Indus-
trializedCountries.Oxford: Clarendon Press, pp. 105-138.
Parnes, Herbert S. 1975. "The National Longitudinal Surveys: New vistas for labor market
research," AmericanEconomicReview65: 244-249.
Rendall, Michael S. and Steve Smallwood. 2003. "Higher qualifications, first-birth timing,
and further childbearing in England and Wales," PopulationTrends111: 18-26.
Rexroat, C. and C. Shehan. 1984. "Expected versus actual work roles of women," American
SociologicalReview49: 349-358.
Rindfuss, Ronald R., E. C. Cooksey, and R. L. Sutterlin. 1999. "Young adult occupational
achievement: Early expectations versus behavioral reality," Workand Occupations26:
220-263.
Shaw, Lois B. (ed.). 1983. UnplannedCareers:The WorkingLivesof Middle-AgedWomen.Lex-
ington, MA: D.C. Heath.
Shaw, Lois B. and D. Shapiro. 1987. "Women's work plans: Contrasting expectations and
actual work experience," MonthlyLaborReview110/11: 7-13.
Spitze, Gary D. and Linda J. Waite. 1980. "Laborforce and work attitudes," Workand Occu-
pations7: 3-32.
Sproat, Kezia V., Helene Churchill, and Carol Sheets (eds.). 1985. TheNationalLongitudinal
Surveysof LaborMarketExperience: An AnnotatedBibliographyof Research.Lexington, MA
and Toronto: Lexington Books.
Stolzenberg, R. M. and Linda J. Waite. 1977. "Age, fertility expectations and plans for em-
ployment," AmericanSociologicalReview42: 769-783.
Szekelyi, M. and R. Tardos. 1993. "Attitudesthat make a difference: Expectancies and eco-
nomic progress," Discussion Papers of the Institute for Research on Poverty, Univer-
sity of Wisconsin.
374 EXPLAINING FERTILITY PATTERNS

Waite, Linda J. and R. M. Stolzenberg. 1976. "Intended childbearing and labor force par-
ticipation of young women: Insights from nonrecursive models," American Sociological
Review 41: 235-252.
Waite, Linda J., G. W. Haggstrom, and D. Kanouse. 1986. "The consequences of parent-
hood for the marital stability of young adults," American Sociological Review 50: 850-
857.
Westoff, Charles F. and Norman B. Ryder. 1977. The ContraceptiveRevolution. Princeton, NJ:
Princeton University Press.