Two Is Best? The Persistence
of a Two-Child Family Ideal
in Europe
Tomáš Sobotka
Éva Beaujouan
Does ideal family size in low-fertility countries differ from actual fertility
trends? Surveys conducted in the last several decades suggest that intended
and ideal family size in Europe and many low-fertility settings outside Europe
might have stabilized around the replacement level. Hagewen and Morgan’s
(2005: 510) analysis of data for the United States concluded that “there is a
remarkably pervasive desire (and supporting norms) for a family size of two
children.” This stability has contributed to a general decline in demographers’
interest in the concept of ideal family size and has prompted some researchers to question the usefulness of studying fertility ideals and preferences,
especially when making inferences about likely future changes in family size
(Simons 1978; Demeny 2003).
Nevertheless, there seems to be no intrinsic reason why individuals deciding about their family size should embrace a replacement-level
ideal of having two children (Bongaarts 2002). Very low fertility might
be unlikely to persist over the long term without being accompanied by a
corresponding change in family-size ideals (Coleman 1998; Testa and Grilli
2006; Goldstein, Lutz and Testa 2003). This mechanism of adjustment is part
of the hypothesized “low-fertility trap” (Lutz, Skirbekk, and Testa 2006),
which suggests that declining fertility rates lead to a spiral of falling number
of births, rapid population aging, diminishing economic opportunities for
younger generations, and, eventually, fewer encounters with children and
declining family-size ideals. Cultural changes may also contribute to the
eventual erosion of a two-child family ideal: as societal norms related to
family, living arrangements, and gender roles become more permissive over
time and across cohorts (Inglehart 1990; Lesthaeghe 1995), women and
men should feel free to express less-conventional family-size ideals. Furthermore, increasing family instability, complexity of living arrangements,
POPULATION AND DEVELOPMENT REVIEW 40(3): 391–419 (SEPTEMBER 2014)
©2014 The Population Council, Inc.
391
392
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
and longer periods of single living are likely to put downward pressure on
fertility ideals, especially among the cohorts experiencing family instability
in childhood and young adulthood.
We revisit data on ideal family size in Europe and provide a systematic
analysis of their trends since the late 1970s, focusing especially on the possible erosion of two-child family ideals. We address two questions: 1) How
persistent and universal has the two-child family ideal been in Europe during
the last three decades? and 2) Is there evidence of an emerging preference
for small family size (zero or one child) in European countries with a long
history of subreplacement fertility?
To address these questions, we analyzed 168 surveys conducted in
1979–2012 in 37 European countries (data for eastern and western Germany
are analyzed separately).1 This extensive dataset allowed us to go beyond the
existing analyses and, despite relying on slightly different questions on ideal
family size, yielded a long and remarkably stable time series of ideal family
size in a large sample of countries.
We focus on the responses of women of reproductive age (15–49) regarding the ideal number of children for a family. First we discuss the motivation for studying trends in ideal family size in Europe, the recent literature,
and available data. After analyzing time trends in ideal family size across all
surveys, we discuss developments for broad European regions and the evidence for selected countries. We assess whether observed declines in mean
ideal family size tend to slow down or reverse when they reach a low level.
Following the low-fertility trap hypothesis, we discuss whether the countries
experiencing a fall in fertility to very low levels subsequently saw a fall in ideal
family size to subreplacement levels. The concluding discussion looks at the
reasons why a two-child ideal continues to dominate in all parts of Europe.
The motivation for studying fertility ideals
Questions on ideal family size were first asked in a 1936 Gallup poll in
the United States (Blake 1966). Since then they have appeared in different forms in hundreds of surveys globally. Such surveys have repeatedly
attracted criticism for their perceived shortcomings, including the bewildering variety of questions on the ideal number of children and possible
interpretations of these questions, encompassing the distinct concepts of
personal/individual, societal, group-specific, and situational ideals (see discussions in Ware 1974; Philipov and Bernardi 2011). In addition, the term
“ideal family size” may not include the choice of childlessness in the minds
of many respondents, preventing them from mentioning zero as an ideal
(Blake 1974). Some researchers have even questioned the validity of the
concept of reproductive ideals. Hauser (1967: 404) considered the question
meaningless, suggesting that since fertility ideals are closely correlated with
T o m á š S o b o t k a / É va B e a u j o u a n 393
completed family size, the “ideal tends to be what has actually occurred.”
Ryder and Westoff (1969, quoted in Ware 1974: 7) concluded their comparison of intended, expected, desired, and ideal family size in the United
States with a clearly negative assessment for the concept of ideal family size,
seeing it as “the least profitable variable to explore further” and positing
that it “lacks face validity, is relatively unreliable, and it has small variance.”
Over time, studies have highlighted the failure of questions on ideal family size to predict actual fertility behavior. In the European context of low
fertility, attention has repeatedly been paid to the perceived “gap” between
ideal and actual family size (e.g., Testa 2012), which suggests either that
respondents might be unrealistic about their fertility preferences or that
ideals are too abstract and removed from real decisionmaking. As Demeny
(2003) speculated, fertility preferences may turn out to resemble the title
of a Günter Grass novel, Kopfgeburten—births that increasingly occur only
in the minds of would-be parents.
Some of these criticisms are clearly valid. The concept of ideal family
size remains ambiguous and open to various interpretations (Blake 1966):
there is no consensus among demographers about the preferred way(s)
of asking about fertility ideals (Philipov and Bernardi 2011). But some of
these perceived shortcomings are subjective and might even be interpreted
as strengths of the measure. Ideals, as distinct from intentions that are
conditioned by structural constraints, can be seen as a reflection of societal
norms (Trent 1980; Philipov and Bernardi 2011). Their interchangeable use
with intentions or desires, found in some of the past research, is misleading. Ideals are also clearly not well suited for predicting fertility levels. Past
research shows that ideals may evolve quickly in times of rapid cultural
change and thus can give useful signals about shifting family norms and
images about “desirable” family size. Such a shift was documented by Blake
(1974) for the United States between the mid-1960s and 1972, when ideal
family size fell especially among those below age 25 and a two-child ideal
became most common.
Reproductive ideals, preferences, desires, and intentions are relatively
“soft” concepts since they are open to uncertainty, doubt, and revision (e.g.,
Quesnel-Vallée and Morgan 2003), but nevertheless form a basic link in the
chain leading to the decision to have a child. They are often considered as key
determinants of reproductive decisionmaking (Philipov and Bernardi 2011)
and behavior (Schoen et al. 1999). Ideals reflect more general societal pronatalist or antinatalist norms (Trent 1980). Family size ideals, especially personal
ones, can also be seen as reflecting the number of children people would like
to have if they lived under “ideal conditions” and faced no constraints such
as poor health, infertility, not having a suitable partner, a demanding job,
precarious economic circumstances, or inefficient contraception (Philipov and
Bernardi 2011). Under these assumptions, change over time in ideals could be
394
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
seen as reflecting the air du temps, in particular the changing attitudes toward
family, childbearing, and the value of children in society.
Trends in fertility ideals: Shifting to
subreplacement levels?
While an extensive literature in the last two decades has been devoted to
studying short-term fertility intentions and their determinants and realization in low-fertility countries, less attention has been paid to aggregate-level
fertility preferences. In particular, research on fertility ideals has fallen out of
fashion and the question has been removed from a number of recent international surveys (Philipov and Bernardi 2011). This may be the result of the
widespread notion that family-size ideals are stable and do not matter much
for reproductive decisions. Consequently, little is known about ongoing trends
in fertility ideals in different parts of the world.
Tentative evidence suggests that ideal family size has fallen below the
replacement level in several low-fertility settings. Goldstein, Lutz, and Testa
(2003) suggested that Austria and Germany might be the first countries
experiencing such a shift. Their analysis of Eurobarometer data from 2001
indicated that the mean ideal family size in these two countries has dipped
to 1.6–1.7 children per woman. They expect that a similar decline is likely to
take place in other low-fertility countries. Testa and Grilli (2006) reported a
close link between actual family size among older generations (women aged
40–60) and ideal family size among younger women aged 20–39 in Europe.
Outside Europe, a shift toward one-child preferences has been reported for
younger and better-educated women in urban China (Ding and Hesketh
2006), where mean ideal family size fell well below two children (Merli and
Morgan 2011; Basten and Gu 2013).
Data
We assembled data from cross-sectional surveys conducted since 1979, especially drawing from the World Values Survey (WVS), the European Values
Study (EVS), the International Social Survey (ISSP), and the Eurobarometer
(EB) surveys. We analyzed complementary datasets for selected countries and
years, specifically data from several surveys conducted in France, the Swiss
Household Panel Survey (SHP), and the Generations and Gender Survey
(GGS) in Austria,2 as well as tabulated data from the Demographic and Health
Survey (DHS) in Albania and Ukraine.
Data sources and questions asked are listed in Appendix 1; supplementary online Appendixes 2 and 4 (http://videurrep.oeaw.ac.at/two-is-bestpdr/) list all individual surveys considered and selected. We focus on general
T o m á š S o b o t k a / É va B e a u j o u a n 395
family-size ideals and do not study personal ideal family size, which has been
included only in the Eurobarometer surveys.3
The questions on ideal family size were identical in the WVS and EVS
and were very similar in the ISSP and EB surveys. The first two included an
additional qualifying statement, “if any,” following the general question on
“how many children” the ideal family should have (see Appendix 1). The
question in DHS surveys clearly differed from the other surveys considered.
The DHS results are, however, closely aligned with the other surveys analyzed
for these two countries.4
Does adding “if any” to the question on family-size ideals make a difference? While our use of multiple surveys with somewhat different questions on ideal family size can make our results more difficult to interpret, we
found remarkable consistency in responses to different surveys. Overall, these
responses seem to be very little affected by slight differences in questions on
reproductive ideals (see online Appendix 3 at http://videurrep.oeaw.ac.at/
two-is-best-pdr/ for further discussion). This consistency is also illustrated in
the reconstructed time trends in fertility ideals for selected countries, which
are derived as a combination of different surveys.
We carefully considered the criteria of data selection. We were particularly concerned with the small number of respondents in some surveys
and with missing responses to the questions on ideal family size. To obtain a
reasonable sample size, we included all women of reproductive age, broadly
defined as 15–49 years. We might have selected a younger and narrower
group of respondents, but that would have limited the sample size in most
surveys to very small numbers.5
In the questionnaires, the “don’t know” option was not explicitly included in the ISSP surveys, and it cannot be inferred from the questionnaires
whether it was offered by the interviewers in the WVS and EVS. The Eurobarometer surveys additionally included in 2001–11 the response option “there
is no ideal number, it depends.” Again it is unclear whether EB respondents
were explicitly presented with these options or whether they were included
for the interviewers’ coding only. In the latest EB survey in 2011 the interviewers were instructed to keep the question open-ended (see Appendix 1).
In addition, the distinction between missing responses and “don’t know”
answers is often not available in the datasets. In the 1979 Eurobarometer
survey, nonresponse in the questionnaires was coded “0,” the same way as
an ideal of having no children. After investigating the subsequent surveys
from the 1980s, we adjusted these data by considering all these responses as
either missing or “do not know” answers—that is, assuming no respondents
were choosing a zero-child ideal family size. Indeed in the 1980s, the proportion of respondents in different surveys choosing an ideal family size of zero
was negligible, reaching highest values of only 1.1–1.3 percent in Austria,
Belgium, and the Netherlands in 1988 (ISSP survey).
396
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
In all the surveys and for each country, we identified the proportion of
respondents who did not give a numerical answer to the question on ideals or
whose responses were missing. Eurobarometer surveys overall had a higher
proportion of these answers, exceeding 25 percent in four surveys. After
inspecting the proportion of unspecified values in all the surveys, with high
levels signaling potential data quality problems, we chose a threshold of 18
percent, above which the survey was not included for further analysis (online
Appendix 3 gives more detail on the selection of this threshold).6
Indicators such as mean ideal number of children and ideal family-size
distribution for each survey were computed from individual data, applying
individual survey weights. We computed two alternative indicators of ideal
family-size distribution for each country and survey: 1) the proportion of
women by their ideal family size, retaining nonresponse and non-numerical
responses as a distinct category, and 2) the adjusted distribution of respondents with a specific ideal family size computed only for women indicating
their ideal number of children. This adjustment has no influence on the computation of mean ideal family size, which is derived from numerical responses
only and cannot capture uncertain or missing responses. Main results for each
survey by country are presented in online Appendix 4.
Finally, we excluded datasets for countries with a population below 1
million (Cyprus, Iceland, Luxembourg, Malta, and Montenegro) and Northern Ireland. Except for the two surveys available for Iceland and the 2001
Montenegro WVS, these surveys had low sample size (82 to 194 female
respondents aged 15–49; see also online Appendix 3). Altogether 168 surveys for 37 European countries and regions, out of the total of 202 datasets
considered, met our selection criteria and were selected for our analyses (see
the list of all considered, included, and excluded surveys in online Appendix
2 and, with more details, in online Appendix 4).
The analysis of time series of reproductive ideals constituted a good
check of the accuracy of results for different countries, which were typically
based on small samples. Partly to compensate for the small sample size in
most surveys, we conducted additional analyses for broader European regions
that also reflect major differences in fertility patterns across Europe (Sobotka
2013). We used the following regional groupings:
Western and Northern Europe: Belgium, Denmark, Finland, France,
­Netherlands, Norway, Sweden, United Kingdom.7
Southern Europe: Greece, Italy, Portugal, Spain.
German-speaking countries: Austria, eastern and western Germany,
­Switzerland.
Central Europe: Croatia, Czech Republic, Estonia, Hungary, Latvia,
­Lithuania, Poland, Slovakia, Slovenia.
Eastern and South-eastern Europe: Albania, Belarus, Bulgaria, Macedonia,
Moldova, Romania, Russia, Serbia, Ukraine.
T o m á š S o b o t k a / É va B e a u j o u a n 397
Changes in ideal family size in Europe since 1979
Trends over time: All surveys combined
We first inspect changes across Europe in ideal family size and in the proportion of women expressing a two-child ideal, computing selected key indicators
for all analyzed surveys in the period 1979–2012. To improve comparability
of data over time and obtain a reasonable number of surveys for each period,
we grouped the surveys into six periods that do not cover equal time intervals,
but rather are centered on the years of the important multicountry surveys
analyzed here. Potentially, this analysis can be biased by the changing composition of countries and surveys analyzed in different periods, including the
repeated presence of some countries in one period.
Despite uneven country and survey coverage, Table 1 gives a concise
picture of the changes over time. First, it shows a gradual decline in the
average value of mean ideal family size (MIFS)8 by about 0.1 per decade:
the average ideal family size in the analyzed surveys dropped from 2.53 in
1979–83 to 2.21 in 2008–12. The median ideal family size across all surveys
declined more slowly and from a lower initial value of 2.38 in 1979–83 to
2.19 in 2008–12. At the same time, the orientation toward a two-child family
ideal has slightly strengthened. In the surveys conducted since 1993, around
60 percent of respondents expressed a two-child ideal when the data were
adjusted for nonresponse and non-numerical answers, up from 55 percent
in the 1980s. More important, the proportion of surveys where over half of
respondents express a two-child ideal has risen over time, from 58 percent
in 1979–83 to 76–78 percent in the 1990s and early 2000s, and 85 percent
in 2008–12. In only two of 109 surveys conducted since 1993 did the most
frequent ideal differ from two: in Norway in 1996 (WVS survey) and in
Finland in 2006 (EB survey), the proportion of women stating a three-child
ideal slightly surpassed that with a two-child ideal. While a two-child ideal
firmly prevails today throughout Europe, the gradual decline in mean ideal
family size was fuelled by a declining proportion of women with an ideal of
three or more children, accompanied by a slower increase in the proportion
of women expressing a subreplacement ideal of zero or one child. Across all
surveys, this proportion has more than doubled from 5 percent in 1979–83
to 11 percent in 2008–12 when adjusting for nonresponse.
Figures 1 and 2 offer a more detailed look at the change in ideal family
size. The decline in the variance of mean ideal family size in different surveys
and countries (Table 1) has been driven especially by a diminishing number of
countries with high ideal family size. This fact is also illustrated by the rapidly
declining level of the highest reported MIFS in each analyzed period, from
3.88 in 1979–83 (Ireland in EVS in 1981) to 2.54 in 2008–12 (Albania in DHS,
2008–09).9 By contrast, the minimum reported MIFS suggests that there is
TABLE 1 Summary results for European countries, all selected surveys combined, by period, women aged 15–49
Mean ideal family size
No. of
No. of
Standard
Period
surveys countries Average Median deviation
Surveys
with >50%
stating
2-child
ideal (%)
Average %
of women
with
2-child
ideal
Adjusted for nonresponse
Average % Average % Average %
of women of women of women
with ideal with ideal with ideal
0 or 1
2
3+
1979–83 19
112.53
2.38
0.44
1987–91 38
272.45
2.42
0.26
1993–97 36
272.35
2.38
0.23
1998–200225
192.34
2.36
0.20
2005–07 23
232.25
2.22
0.19
2008–12 27
242.21
2.19
0.19
5851 55540
6354 55638
7858 76033
7655 75933
7857 96229
8556 116128
FIGURE 1 Mean ideal family size in Europe, summary results for
all selected surveys by period, women aged 15–49
4.0
3.5
Mean ideal family size
3.0
Maximum
75% (upper quartile)
2.5
Mean
2.0
25% (lower quartile)
1.5
Minimum
1.0
0.5
2
–1
–0
08
20
19
20
05
–2
98
7
2
00
7
93
–9
1
19
–9
87
19
19
79
–8
3
0.0
FIGURE 2 Percent of respondents with an ideal of two
children, summary results for all selected surveys by period,
women aged 15–49
80
Maximum
70
Mean
50
25% (lower quartile)
Minimum
40
30
20
10
8–
12
20
0
5–
07
20
0
02
20
19
98
–
3–
97
19
9
7–
91
19
8
9–
83
0
19
7
Percent with two-child ideal
75% (upper quartile)
60
400
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
a lower floor to the decline in ideal family size, as it was remarkably stable
and close to 2.0 in all the analyzed surveys and periods except in 1998–2002
(Figure 1). Since the mid-1990s a majority of surveys show a MIFS in the
range of 2.0–2.5, with a slightly declining tendency.
There was a clear trend toward greater uniformity in a two-child ideal
across surveys, as illustrated also by a sharp rise in the minimum proportion
of respondents with a two-child ideal from 17 percent in the earliest period
(Ireland in 1981) to 43 percent in the latest (Finland in 2011). A majority of
the surveys conducted in 2008–12 fall into the narrow range of 50–60 percent
of respondents expressing a two-child ideal (Figure 2), suggesting a considerable convergence across Europe.
Comparison of major European regions
Did the trends in ideal family size, evidenced in the analysis of all surveys
combined, progress similarly in different parts of Europe? Our analysis focuses
on broad time periods since the late 1980s—except for Western and Northern
Europe—in order to increase the number of surveys in each region and make
the regional data more comparable over time. Subtle shifts and regional differences can be seen in trends in mean ideal family size and in the ideals of
larger family size (three or more children) and smaller family size (none or
one child). Gradual declines in MIFS can be observed in Central, Southern,
South-eastern, and Eastern Europe, with only Western and Northern Europe
and German-speaking countries showing general stability (Figure 3a). The
FIGURE 3 Mean ideal family size (a) and percent of respondents with an ideal
of two children in broader European regions (b), women aged 15–49, average
values across all surveys in a given period
a)
Mean ideal family size
2.5
F
T
T
F
T
F
2.0
1.5
F
T
Western Europe
Austria, Germany, Switzerland
Southern Europe
Central Europe
Eastern Europe
1.0
Percent with two-child ideal
80
b)
70
60
T
F
F
T
F
T
50
40
30
1979–83 1987–94 1995–2002 2005–12
NOTE: Figure 3b adjusted for nonresponse and non-numerical responses.
1979–83 1987–94 1995–2002 2005–12
T o m á š S o b o t k a / É va B e a u j o u a n 401
average value for Eastern Europe in 2005–12 reaches a low of 2.0, followed
by Southern Europe and German-speaking countries with an average of 2.1.
The data depict considerable stability in the ideal of having two children
(Figure 3b). By the late 1980s, and probably earlier, this ideal had become
dominant in all parts of Europe. None of the regions studied shows any sign
of an erosion of the two-child ideal between the late 1980s and 2005–12. In
the most recent period, the adjusted average proportion of respondents expressing a two-child ideal surpasses 60 percent in all parts of Europe except
Western and Northern Europe, where the proportion has been slightly lower
during the analyzed period.
Trends in selected countries
To gain a finer-grained picture of changes in reproductive ideals, we analyze
trends in ideal family size in selected countries with a longer time series of
data (Figure 4; data for more countries are analyzed in Figure 5 in Sobotka
and Beaujouan 2014). We focus especially on the largest European countries
and on providing contrasting examples for each region. Data for the two
German-speaking countries plus predominantly German-speaking Switzerland are discussed separately. We do not analyze trends in countries in
Eastern and South-eastern Europe, for which we lack longer time series of
comparable data.
France represents the higher end of the European range of ideal family
sizes. The slightly prevailing ideal of two children coexists with the high proportion of women expressing an ideal of three children, whereas the ideal of
one or no child remains rare. Corresponding to that distribution, mean ideal
family size approaches 2.5 children. Similar results were obtained for the
Nordic countries (country trends not shown here). In France, where we have
compiled a longer time series beginning in 1955, the two-child ideal started
to prevail continuously over the “three or more” option only in the 1990s.
In Belgium, the ideal of two dominates more strongly, and a slight increase
can be observed in the proportion of women with an ideal of zero or one.
Spain, alongside other countries in Southern Europe, depicts a trend
toward a low MIFS around 2.0, a sharp decline in the proportion of women
with a larger family ideal, and a concurrent rise in the preference for a small
family size. In Portugal and Italy (country trends not shown for the latter),
the proportion of women with a zero or one-child ideal has already surpassed
the proportion with an ideal of three or more children. The sharp decline in
family-size ideals between 2006 and 2011 in Portugal and Greece (country
trends not shown for the latter) suggests the possible negative influence of
the protracted economic recession (Testa 2012; Testa and Basten 2013).
In Central Europe, the Czech Republic already had a very strong orientation toward a two-child ideal in 1990, with only about 20 percent of women
FIGURE 4 Changes in ideal family size in selected European countries,
women aged 15–49 (data adjusted for nonresponse)
a) Western and Northern Europe
Ideal of 3+ H
children (%)
40
H
H
H
H
Ideal of 1 or no children (%)
B
B
2.0
1.5
H
20
10
70
2.5
Mean ideal family size
50
30
3.0
B
B
B
B
0
1.0
0.5
60
10
80
3.0
70
30
H
H
20
10
B
B
B
B
H
B
B
B
1.0
0.5
1.5
1.0
0.5
B
B
B B
B
B
B B B
B
B B
Portugal
0.0
3.5
3.0
60
2.5
50
2.0
40
1.5
H
30
H
H
20
1.0
B
H
10
2011 EB
2006 EB
2001 EB
2000 WVS
1995 EVS
1994 ISSP
H
B
B
B
1990
EVS
2001
EB
2006
EB
0.5
0
0.0
1981 EVS
0
B
H
2.0
H H H
1955EC
1967 EC
1976 EC
1979 EB
1981 EVS
1982 EC
1987 EC
1990 EVS
1998 EIF
2001 EB
2006 EB
2011 EB
1.5
H
H
Percent with a given ideal
H
H
20
3.5
2.0
40
2.5
H H
H
H
30
0
2.5
50
H
40
0.0
Mean ideal family size
60
H
1990 EVS
Percent with a given ideal
70
3.0
H
b) Southern Europe
Spain
3.5
H
50
1979 1981 1990 2001 2006 2011
EB EVS EVS EB EB EB
80
France
Mean ideal family size
60
80
Mean ideal family size
Ideal of
2 children (%)
3.5
Percent with a given ideal
70
Belgium
Mean ideal family size
Percent with a given ideal
80
0.0
2011
EB
c) Central Europe
H
1.0
20
0.5
10
B
0
1.5
B
B
B
B
1992 1994 1995 2006 2011
EVS ISSP WVS EB
EB
0.0
1.5
30
20
H
H
H
H
B
10
0
B
B
B
H
B
B
H
1.0
0.5
0.0
Mean ideal family size
H
2.0
40
2011 EB
H
H
2.5
50
2006 EB
H
3.0
1998 WVS
2.0
40
3.5
60
1994 ISSP
2.5
50
30
3.0
70
Czech Republic
1991 EVS
60
80
1990 WVS
70
3.5
Percent with a given ideal
Slovenia
Mean ideal family size
Percent with a given ideal
80
T o m á š S o b o t k a / É va B e a u j o u a n 403
at that time having an ideal of three or more children. By 2011, this latter
proportion has further eroded, and the one-child ideal has become more
prevalent, pushing the MIFS to the lowest level in Europe (1.93). Slovenia,
despite having low fertility similar to that in the Czech Republic, shows a
strong stability in ideal family size and a very low proportion of women with
an ideal of zero or one, which does not increase over time.
Ideal family size in Austria, Germany, and Switzerland
The three predominantly German-speaking countries of Europe share a long
history of low fertility (Sobotka 2011). With the exception of two surveys
that do not align well with longer time trends—the Austrian 1994 ISSP
survey and the 2001 EB survey in eastern Germany—the time series for
Austria and Germany show remarkable stability in fertility ideals, especially
in western Germany (Figure 5). This is best illustrated by comparing the
earliest and latest surveys, conducted in 1979 and 2011. The two surveys
show identical mean ideal family size for western Germany of 2.10 and an
identical proportion of respondents with a two-child ideal of 67 percent
when adjusted for nonresponse and non-numerical responses. There are
no noticeable shifts in the distribution of large or small family-size ideals.
Research indicates that women and men in western Germany had adopted
a two-child family ideal at least two decades earlier: a 1957 survey of married women and men aged 15–45 showed that 45 percent had an ideal of
two children, with a MIFS of 2.5810 (Freedman, Baumert, and Bolte 1959,
Table 2). A comparison of the eastern German 1990 EVS with the 2011 EB
reveals striking continuity, with the MIFS increasing marginally from 1.96
to 1.98. The data for Austria suggest a slight decline in MIFS going hand in
hand with a gradual increase in the proportion of women having an ideal of
one or no children. In Switzerland we observe a trend to a lower family-size
ideal, with a notable decline in the ideal of having three or more children.
Unlike in Austria and Germany, the ideal of having one or no children remains marginal in Switzerland.
Our results do not support the widely discussed findings of Goldstein,
Lutz, and Testa (2003) that younger cohorts of women in Austria and Germany, after experiencing decades of low fertility, have begun holding fertility
ideals well below the replacement threshold. What is the main reason for the
lack of supporting evidence in our study? First, we analyze general ideal family size, whereas Goldstein, Lutz, and Testa looked at personal ideals, which in
Austria have been slightly lower than the general ideals. Second, we excluded
the 2001 EB surveys for Austria and western Germany, which did not conform to our selection criteria because of the high proportion of nonresponse
and of respondents who stated that “there is no ideal family size,” altogether
19 percent in western Germany and 23 percent in Austria. These surveys
reported very low ideals that were not replicated in subsequent surveys in
404
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
FIGURE 5 Changes in ideal family size in Austria, Switzerland, and western
and eastern Germany, women aged 15–49 (data adjusted for nonresponse)
50
70
2.5
2.0
40
H
H
Ideal of 3+ children (%)
H
20
H
10
80
B
B
B
1.5
H
H
B
B
B
Ideal of 1 or no
children (%)
1.0
0.5
Western Germany
70
60
20
10
H
H
B
H
H
H
H
H
H
B
B
B
B
B
B
B
0
1.0
0.5
2011 EB
2006 EB
1997 WVS
1994 ISSP
1990 EVS
1988 ISSP
1981 EVS
0.0
1979 EB
30
1.5
H
H
1.0
H
20
0.5
10
1989
WVS
70
1.5
2.0
H
40
3.0
2.0
2.5
50
80
40
30
60
3.5
2.5
50
3.0
0
0.0
1988 1990 1994 2008 2011 2012–13
ISSP EVS ISSP GGS EB GGS
Percent with a given ideal
30
3.5
B
B
1996
WVS
2002
SHP
B
0.0
2005
SHP
Eastern Germany
3.5
3.0
60
2.5
50
2.0
40
30
20
10
Mean ideal family size
Mean ideal family size
3.0
Switzerland
1.5
B
1.0
B
H
B
H
B
H
H
H
H
B
B
0
Mean ideal family size
60
80
Percent with a given ideal
Ideal of
2 children (%)
3.5
Mean ideal family size
70
0
Percent with a given ideal
Austria
Mean ideal family size
Percent with a given ideal
80
0.5
0.0
1990 1994 1997 2001 2006 2011
EVS ISSP WVS EB EB EB
NOTE: The figure includes only the surveys selected for analysis in this study (see the list of all surveys in online
Appendixes 2 and 4).
these countries, as shown in our analyses and in additional surveys analyzed
by Philipov and Bernardi (2011: 503, Table 1). One of the surveys analyzed
by Goldstein, Lutz, and Testa, the 2001 EB in eastern Germany, is included
in Figure 5, but it constitutes an interruption in an otherwise relatively stable
time series. In our whole dataset, this survey showed the lowest value of
MIFS at 1.78, well below the next-lowest value of 1.93 reached in Romania
in 2006 and in the Czech Republic in 2011, which adds to our suspicion that
the 2001 EB survey in eastern Germany might suffer from methodological
problems, such as having a sample selection not well representative of the
eastern German population.11
405
T o m á š S o b o t k a / É va B e a u j o u a n Does a decline in mean ideal family size slow
down when it reaches low levels?
Our analyses suggest that a mean ideal family size around two constitutes a
low threshold that so far has not been consistently crossed in Europe, except
for a few one-off surveys and some variation around this value. This finding is remarkable because the prolonged experience of very low fertility in
a number of countries could be expected to eventually bring about a corresponding fall in ideal family size (Lutz, Skirbekk, and Testa 2006). There is
no pertinent reason to believe that ideals should follow the homeostatic logic
of the “replacement level of fertility” model and stop falling when reaching
levels around two.
To investigate this topic in a more systematic fashion, we analyzed
changes in mean ideal family size depending on the initial level of MIFS.
Specifically, we selected for each country all pairs of consecutive surveys
that occurred within a period of up to ten years and analyzed the changes in
MIFS from one survey to the next. To standardize the results, we computed
the annual change in MIFS between the year of the initial survey and the
year of the subsequent survey. The results displayed in Figure 6 are by defi-
FIGURE 6 Average annual change in the mean ideal family size (MIFS)
between two consecutive surveys, conditional on the initial MIFS value;
127 pairs of surveys in Europe, 1979–2011
0.10
Annual absolute change in MIFS
0.08
0.06
0.04
0.02
0.00
–0.02
–0.04
–0.06
–0.08
–0.10
3.0
2.8
2.6
2.4
2.2
2.0
1.8
1.6
Initial level of MIFS
NOTES: For the initial MIFS level only the data with an initial MIFS at or below 3.0 were selected (this
resulted in the omission of four data pairs). The trendline is a second-order polynomial curve supplied by
MS Excel package.
406
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
nition affected by the fact that we combined different surveys with not fully
comparable questions, sample sizes, nonresponse options, and other features.
This should increase the noise in our time series as well as instability in the
analyzed results. This concern aside, our analysis suggests that there is a
relationship between the initial MIFS and its subsequent change: the lower
the initial MIFS, the weaker its subsequent decline. A crossover even occurs
when the MIFS falls below 2.3 children: when a survey shows a MIFS below
this threshold, it is more likely to increase rather than to decline in the next
period. Although this effect is not very strong, it nevertheless suggests a broad
stabilization of mean ideal family size once it reaches a level close to replacement. We found a similar relationship for the two-child family ideal with
the stabilizing level at around 60 percent. Above that level, the proportion
of respondents with a two-child ideal is more likely to fall than to rise in the
next survey (see Figure 7b in Sobotka and Beaujouan 2014).
The links between subreplacement fertility and
ideal family size
If the changes in ideal family size tend to follow, at least partly, the changes
in fertility behavior as hypothesized by Lutz, Skirbekk, and Testa (2006), two
predictions related to its possible decline to levels significantly below replacement can be made: 1) the earlier a country experiences a decline in fertility to
very low levels, the more likely it is to see a shift to MIFS below replacement;
and 2) the lower fertility falls in a country, the more likely that the country
will experience a shift to below-replacement MIFS.
We addressed these predictions by ranking European countries by the
timing and intensity of their fertility declines to low levels and analyzing their
most recent MIFS levels. To avoid tempo and parity composition distortions
present in the most conventional measure of fertility, the period total fertility
rate, we used the completed cohort fertility rate (CFR). There appears to be no
systematic relation between the early spread of low fertility and recent ideal
family size. Among the four countries where completed fertility fell below
1.8 children per woman in the cohorts born before 1960—western Germany,
Austria, Switzerland, and Italy—the first three countries retain comparatively
higher MIFS at 2.1–2.3 than many countries with later cohort fertility declines, including eastern Germany and Spain (Sobotka and Beaujouan 2014).
When we rank countries with the lowest fertility levels, as measured
by their estimated cohort fertility rate among women born in mid-1970s, a
tentative link emerges between the magnitude of cohort fertility decline and
MIFS level (Table 2). Specifically, all countries except Greece with a CFR below 1.6 children per woman have reached a low MIFS in the narrow range
of 1.95–2.11.12 Even in these countries, however, ideal family size remains
very close to the replacement threshold.
407
T o m á š S o b o t k a / É va B e a u j o u a n TABLE 2 European countries with the lowest completed cohort
fertility among women born in 1974–75 and most recent level of MIFS
CFR (cohort 1974–75)
Spain
1.37–1.41
Italy
1.42–45
Ukraine1.51a
Germany, eastern
1.55
Romania
1.55
Bulgaria
1.56
Germany, western
1.57
Greece
1.55–58
Portugal
1.57–58
Belarus1.58a
Poland
1.61–62
Russia
1.62
Austria
1.63–64
Switzerland
1.63–65
Slovenia
1.66–67
Most recent MIFS
Year
Level
20112.07
20111.95
20071.97
2011
2.10
20111.97
20112.03
2011
1.98
20112.29
20112.00
19962.11
20112.15
19952.15
20122.20
20052.18
20112.38
Data pertain to the 1972 cohort.
SOURCES: MIFS data are based on the most recent surveys listed in online Appendixes 2 and 4. Data on
completed cohort fertility are partly projected and for most countries based on the computations by Prioux and
Barbieri (2012: 537, Table A7) for the 1974–75 cohort (trend projection). Data for Russia and for eastern and
western Germany are for the 1975 cohort and were computed by Myrskylä, Goldstein, and Cheng (2013: Table
2). Data for Belarus and Ukraine pertain to the 1972 cohort and are based on the European Demographic Data
Sheet 2014 (VID 2014; computations by Kryštof Zeman).
a
Discussion
Main findings
In the last several decades a two-child ideal has become nearly universal
among women in Europe, with countries that used to display higher ideal
family size converging over time toward a two-child norm. Six out of ten
women in Europe consider two children ideal, and this proportion is very
similar in different regions, irrespective of their fertility patterns and levels.
Similarly, the mean ideal family size has become relatively closely clustered
around 2.2 in most countries. Gradual shifts can be documented toward more
women expressing an ideal of one child (and, quite rarely, no children), which
goes hand in hand with the decline in an ideal of three or more children.
These trends are regionally differentiated and have been more pronounced
in Southern Europe (possibly in part due to the recent economic recession)
and in Eastern and South-eastern Europe. However, there are no signs that
women in any of the European countries we analyzed are abandoning a twochild ideal in favor of an ideal of having only one child, as has been evidenced
in urban China (Basten and Gu 2013).
408
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
An increasing number of European countries, especially those that have
reached a low level of completed fertility around 1.6 or lower, saw their mean
ideal family size fall to a relatively low level around 1.95–2.15. But with the
exception of one survey for eastern Germany and a few additional surveys
not included here because of high nonresponse, none of the analyzed surveys
suggests a decline in mean ideal family size to levels considerably below replacement—that is, below 1.9 children per woman. Indeed, once a relatively
low ideal family size below 2.3 is achieved, the next survey is more likely to
show a slight increase rather than a further decline in MIFS. Evidence for the
stability of reproductive ideals around replacement level is most surprising for
Austria, western Germany, and Switzerland, which share an early fertility decline to levels well below replacement. Women in western Germany born in
the mid-1950s reached low completed fertility around 1.6, yet their daughters
born in the 1980s continue to express considerably higher ideals, averaging
around 2.1. This evidence is at odds with earlier reports on fertility ideals in
Austria and Germany falling to the very low levels of 1.6–1.7 (Goldstein, Lutz,
and Testa 2003), which might have been partly affected by high nonresponse,
low sample size, or other data-related problems in the 2001 Eurobarometer
survey (see also Philipov and Bernardi 2011: 502, fn 4).
Our conclusions are based on the largest collection of data on ideals analyzed to date. The use of multiple surveys reinforced our results and allowed
us to reconstruct country-specific trends over time. In the absence of detailed
information on uncertainty, nonresponse, and non-numerical and missing
responses in different surveys, we excluded datasets with high proportions
of these responses. The “don’t know” answers often have substantive meaning inasmuch as uncertainty is inherent to fertility decisionmaking (Morgan
1981; Ní Bhrolcháin and Beaujouan 2011). However, sensitivity analysis
showed that our results on the whole are stable and robust to different survey
selection criteria (see more details in online Appendix 3). While we have not
analyzed fertility ideals for men, available studies indicate that they come very
close to those of women in most countries (see Testa 2012: 57, Table A.2.1,
for the 2011 Eurobarometer survey).
The two-child family ideal is not unique to Europe. The evidence presented here is similar to that for non-European countries with low or very
low fertility, including Canada, the United States, Australia, Japan, South
Korea, and Taiwan. In those countries, having two children became a dominant ideal decades ago, and it remains widely adhered to (see discussion in
Sobotka and Beaujouan 2014, Section 5.7). Furthermore, an analysis based
on reproductive preferences in the Demographic and Health Surveys and
Reproductive Health Surveys suggests a spread of a two-child orientation in
many middle-income countries in Asia and Latin America where the fertility
transition is still in progress (Sobotka and Beaujouan 2014).
Our study shows a divergence between near-replacement reproductive
ideals and below-replacement fertility in European societies. The difference
T o m á š S o b o t k a / É va B e a u j o u a n 409
between completed fertility in the EU, namely 1.7 children per woman born
in 1972 (VID 2014), and the most recent data on family size ideals, averaging
around 2.2, is on the order of 0.5 children. Several factors contribute to this
mismatch between preferences and outcomes. They include competing career
and leisure preferences, inability to find the right partner, marriage and partnership disruption, disagreement between partners, and the fact that many
women and men face unstable employment conditions, economic difficulties,
problems with combining work and childrearing, health problems, or infertility (Bongaarts 2002; Adsera 2006a; Morgan and Rackin 2010; Iacovou and
Tavares 2011; Testa, Cavalli, and Rosina 2011).13 Considering that couples in
most European countries use reliable contraception and women often have
access to abortion or emergency contraception, one might find it surprising
that the observed differences in Europe are not even larger.
Why an ideal of two?
The spread and subsequent persistence of a two-child family ideal in Europe
are remarkable. Neither long-term experience of below-replacement fertility nor major social and economic upheavals or cultural and technological
changes seem to have altered the widespread perception that having two
children is ideal, both personally and for society. The factors identified in
earlier studies as influencing fertility ideals and preferences, such as religiosity
(Adsera 2006a), unemployment and labor market position (Adsera 2006b),
experience of single living (Waite, Goldscheider, and Witsberger 1986), number of siblings and siblings’ fertility (Axinn, Clarkberg, and Thornton 1994),
as well as rising family instability, might be expected to erode the two-child
family ideal as societies progressively become more secular, actual family size
shrinks, more women enter the labor market, employment becomes more
precarious, and more children grow up in unstable family environments.
It can be argued that these forces have contributed to the observed gradual
increase in the prevalence of the one-child ideal, but so far have not made
any significant dent in the widespread adherence to a two-child family norm.
Moreover, it is plausible that these developments have had a stronger influence on reducing fertility intentions, which in contrast with ideals imply a
degree of commitment to act (Bachrach and Morgan 2013), or on the extent
to which these intentions are eventually realized (Spéder and Kapitány 2014).
The notion that fertility ideals and norms might not remain permanently
centered on two or more children has a long history. Coleman (1998: 20, 21)
argued that the basic question is “whether there is any imaginable reason why
the average should be two.… It is difficult to think of a mechanism, outside
the realm of psychological ‘instincts’, or of biology, how such norms could be
permanently protected.” But childlessness is often viewed as an “extreme”
choice and an “illogical” ideal that would imply a population collapse in one
generation (Hagewen and Morgan 2005). From the perspective of evolution-
410
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
ary biology and psychology, Foster (2000: 211) argued that humans, and in
particular women, are biologically predisposed to nurturing behavior, and this
“need to nurture” is strong enough to ensure that “women will, other things
being equal, want to bear at least one child, despite the substantial costs of so
doing.” Research on the value of children to parents has identified a broad
array of reasons why people want to have children, ranging from economic
motivation to “stimulation and fun” (e.g. Hoffman and Hoffman 1973; Hoffman and Manis 1979). Since parenthood is perceived by most people as a
unique, valuable, and desired experience, childless couples may attain the
status of parents and satisfy their “baby longing” by having just one child
(Rotkirch et al. 2011).
There are many pertinent arguments why having only one child might
be best from an individual point of view. Parents can allocate more time and
resources to interacting with their only child while minimizing the negative
effect children have on their leisure time, employment, career prospects, and
income. Their resources and attention can be concentrated on one child, who
does not have to experience conflict and competition with siblings (Mancillas
2006). From the perspective of parents, having one child is “the ‘cheapest’ option for those wishing to experience parenthood” (Jefferies 2001: 4). Wilson
(2013: 1383) suggested that if the demographic transition “is hypothesised as
a change from quantity to quality in children, then the logical endpoint is one
child, not two.” Striessnig and Lutz (2013: 411) argued that “at the individual
level it is sufficient to have one child (under low child mortality conditions)
if the primary goal is to pass on one’s genes and continue to live on in the
next generation.” Blake (1981), reviewing the contemporary evidence on
only children, found that despite widespread negative stereotypes, they are
intellectually superior, record better educational achievement, tend to count
themselves as happy, and suffer no obvious character or personality defects
or disruptive behavior when compared with children from larger families. A
more recent review by Mancillas (2006) broadly corroborated these findings,
citing high achievement and intelligence as “the most consistent and strongest advantages of only children” (pp. 270–71). Finally, the first child may
substantially increase parents’ happiness, while the second child increases
happiness only slightly or not at all (Kohler, Behrman, and Skytthe 2005;
Myrskylä and Margolis 2012).
We have identified five broader factors, outlined below, that motivate
people in contemporary low-fertility settings to view having more than one
child as an ideal. With the exception of the last one (fitting the social norm),
they do not necessarily provide a rationale for preferring two children rather
than, say, three or four, but they do provide a clear case for preferring more
than one. Given that parents face high opportunity costs of having children,
two children provide an “ideal strategy” to conform to the factors discussed
below with the least effort.
T o m á š S o b o t k a / É va B e a u j o u a n 411
Having one child of each sex provides a strong motivation for preferring two
children. Interacting with a boy and with a girl may be perceived as qualitatively distinct experiences linked with different kinds of practical and psychological benefits, enjoyment, stimulation, communication, and play. Having
two children is then a minimum needed to experience an interaction with
one child of each sex. Indeed, research on sex preferences and actual fertility
behavior documents considerable motivation among couples to have at least
one child of each sex. This is most clearly illustrated in the studies of fertility
behavior among the parents of two or more children, where “a distinct and
stable preference for at least one child of each sex can be observed as a common pattern … across many different social, economic and cultural contexts,
whether they be located in developing countries or in highly industrialized
nations” (Hank 2007: 765).
A single child is viewed as spoiled. Even in contemporary low-fertility societies there are enduring negative stereotypes about only children, who are often
perceived as spoiled, self-centered, domineering, and quarrelsome (Mancillas
2006; Hagewen and Morgan 2005; Blake 1981; Almodovar 1973). Mancillas
(2006: 268) traces these stereotypes to the psychologist G. Stanley Hall, who
claimed in his 1898 study that “being an only child is a disease in itself.” The
“onlies” are also expected to suffer from their status, experiencing loneliness,
anxiety, and deprivation. Although most of these stereotypes are unfounded
(Blake 1981; Mancillas 2006), they create a negative perception of parents
who have only one child as selfish.
A companion for the first child. Closely related to the negative views of only
children, many parents consider it desirable to provide a companion, a brother
or a sister, to their first child—someone with whom the first child can play,
interact, quarrel, and mutually develop (Jefferies 2001). This sibling interaction might provide unique emotional rewards and evolve into a relationship
with a high level of trust, care, and attention matched only in parent–child
and in couple relations. In line with this, earlier studies found that providing
companionship for a sibling is the most prominent value cited in support of
desiring a second child (e.g., Bulatao 1981).
Two children as an insurance strategy. In high-mortality societies, individuals had a strong motivation to “produce” many offspring to achieve a high
likelihood that at least one survived to adulthood and would be able to reproduce. This motivation for having more than one child has sharply diminished
as almost all children now survive to middle age or beyond. But the possibility
that the only child may die or be seriously harmed has not entirely vanished,
and this thought may still affect the decisionmaking of more anxious parents
as well as those who know a family with such an experience. Furthermore, a
more subtle “insurance strategy” to have more than one child may still be at
work. If one child does not meet parental expectations because it is troublesome and, later in life, drops out of school, experiences poor health, abuses
412
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
drugs, turns violent, or severs ties with parents, having another child greatly
improves the chances that at least some of the parents’ expectations and
desires related to their children will be met. In many societies, children still
serve as the main providers of care for their ailing parents, and having two
children strongly increases the likelihood that parents will have someone to
care for them when they get old.
Fitting the social norm. In Europe and many countries outside Europe,
having two children has become a norm widely shared across generations
and social groups. Because this norm is widespread, most people know about
it. Individuals uncertain about their preferences may state, for convenience,
that their ideal family size is two since they know this conforms to general
expectations. Moreover, many people may receive cues from their parents,
other relatives, friends, peers, and colleagues, perpetuating the notion that
two children are ideal. People responding to the question on the “societal
ideal” may also answer it having in mind a rational (and broadly correct)
perception that in order for the population to replace itself, each couple needs
to have at least two children.
Finally, some respondents expressing a two-child ideal may be influenced
by their current family size and the size of their family of origin. An online
survey focused on fertility ideals revealed that respondents with children had
difficulties stating their ideals “retrospectively” without being affected by their
current family status (Hin et al. 2011: 147, fn 9). At younger ages, respondents
may base their preferences on the size of their family of origin (Régnier-Loilier
2006; Hayford 2009), which now frequently means two children.
Clearly, these are pertinent explanations for the persistence of two-child
ideals. These ideals, even more so than intentions, seem to be sustained by
enduring underlying schemas about the image of a family and the affective
value of family life, discussed for intentions by Bachrach and Morgan (2013).
But since these ideals are partly based on misconceptions (negative stereotypes about only children) or on perceived social norms that may eventually
erode, the current near-universal dominance of the two-child ideal in Europe
may not prevail in all parts of the continent. Although the earlier predictions
of the imminent spread of subreplacement ideals have not materialized,
subtle shifts toward a higher proportion of women with one-child ideals are
observed in many countries. They provide a signal about the potential for a
future further spread of one-child ideals in Europe. Whether this will indeed
take place is an open question.
T o m á š S o b o t k a / É va B e a u j o u a n 413
Appendix 1: Surveys and questions analyzed
All surveys considered for inclusion and those selected for the analysis are listed in
online Appendix 2; online data Appendix 4 provides a more detailed list with selected
key indicators for each survey. Discussion of the selection criteria and of the robustness
of the presented results is provided in online Appendix 3 (all supplementary online
materials are available at http://videurrep.oeaw.ac.at/two-is-best-pdr/).
Multicountry surveys
—Eurobarometer (EB, 1979, 2001, 2006, 2011)
Data source: European Commission, Brussels: Eurobarometer 11, 56.2, 65.1; TNS OPINION
& SOCIAL, Brussels (Producer); GESIS, Cologne (Publisher): ZA1036, ZA3627, ZA4505,
ZA5564.
Questions and predefined response options:
EB 1979: “In (country name) today, what do you think is the ideal number of children
for a family?”
Coded response options: 0 none; DK; NA; 1 one; 2 two; 3 three; 4 four; 5 five; 6 six; 7
seven; 8 eight; 9 nine or more.
EB 2001, 2006, and 2011: “Generally speaking, what do you think is the ideal number of
children for a family?”
Coded response options (partly survey- and country-specific), including 9 there is no ideal
number, it depends; 10 don’t know.
It is not clear whether the interviewers were expected to explicitly list all the response
options. Only in the 2011 survey were the interviewers explicitly instructed to allow openended responses and not to read the pre-coded options.
—European Values Study (EVS, 1981–2008)
Data source: European Values Study 1981–2008, Longitudinal Data File ZA4804. Spanish
data files ZA4451 and ZA4487. GESIS Data Archive, Cologne.
Question: “What do you think is the ideal size of a family—how many children, if any?”
Coded response options (in master questionnaire):
EVS 1981: open question.
EVS 1990 (it is not clear whether the pre-codes were read out):
0 none; 1 one child … continuing up to 9 nine children; 10 ten or more children; 99 don’t
know.
EVS 1999: Asked only in Sweden, coding information not available.
—International Social Survey (ISSP, 1988 and 1994)
Data source: ISSP Research Group, International Social Survey Programme (ISSP): 1988
ZA1700, 1994 ZA620. Distributor: GESIS, Cologne
Question: “All in all, what do you think is the ideal number of children for a family to
have?”
Coded response options: no predefined codes; open question (write number in box).
414
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
—World Values Survey (WVS, 1981–2008)
Data source: World Values Survey 1981–2008 Official aggregate v.20090901, 2009. World
Values Survey Association (www.worldvaluessurvey.org). Aggregate File Producer: ASEP/
JDS, Madrid.
Question: “What do you think is the ideal size of a family—how many children, if any?”
Coded response options (in master questionnaire):
WVS 1981: open question.
WVS 1990 (no clear interviewer instruction is available on this question, so it was not
necessarily read out): 0 none; 1 one child … continuing up to 9 nine children; 10 ten or
more children; 99 don’t know.
WVS 1995, WVS 2000 (again, no clear interviewer instruction): 0 none; 1 one child …
continuing up to 7 seven children; 8 eight or more children; 9 don’t know, no answer.
—Demographic and Health Survey (DHS)
Data source: ICF International, 2012. MEASURE DHS STATcompiler—http://www.statcompiler.com— data downloaded April 17, 2013.
Question: (a) Has living children: “If you could go back to the time you did not have any
children and could choose exactly the number of children to have in your whole life, how
many would that be?”
(a) No living children: “If you could choose exactly the number of children to have in your
whole life, how many would that be?”
Coded response options (phases 5 and 6): None 00; Number … (specified); Other (specify)
96.
Individual country surveys
—INED and INED-INSERM surveys in France, 1955–2010
Data sources:
• Les attitudes et la conjoncture démographique: natalité, structure familiale et limites de la vie
active. 1re partie, 1955, INED. Enquête conjoncture (1955 EC).
• La conjoncture démographique: régulation des naissances, famille et natalité, 1967, INED.
Enquête conjoncture (1967 EC).
• Attitudes des Français sur la conjoncture démographique, la natalité et la politique familiale à
la fin de 1976, INED. Enquête conjoncture (1976 EC).
• Une enquête d’opinion sur la conjoncture démographique, 1982, INED. Enquête conjoncture
(1982 EC).
• L’opinion sur la politique démographique, la nuptialité et les nouvelles techniques de procréation
en mai 1987, INED. Enquête conjoncture (1987 EC).
• Enquête permanente sur les conditions de vie des ménages (EPCV), partie mobile: intentions
de fécondité - mai 1998, INSEE. Enquête sur les intentions de fécondités (1998 EIF).
• Enquête Fecondité-Contraception-Dysfonctions sexuelles FECOND – juin 2010–janvier 2011,
INED/INSERM.
Question (identical in all surveys) : “D’après vous, quel est le nombre idéal d’enfants dans
une famille?” (In your opinion, what is the ideal number of children in a family?).
Coded response options: open-ended question, only in 1998 was it possible to give a range
(we selected the minimum value in our computations).
415
T o m á š S o b o t k a / É va B e a u j o u a n —Swiss Household Panel Surveys (SHP), waves 2002 and 2005.
Data source: Swiss Household Panel Surveys, Swiss Centre of Expertise in the Social Sciences
(FORS), data accessed in April 2014
Question: “What would you say is the ideal number of children for a family living in
Switzerland?”
Coded response options: open-ended question.
—Generation and Gender Surveys (GGS) Austria, wave 1 (2009) and
wave 2 (2012–13)
Data source: Austrian Generation and Gender Surveys, Statistics Austria/Vienna Institute of
Demography.
Question: „Was glauben Sie, ist die ideale Zahl von Kindern für eine Familie in Österreich?“
(What do you think is the ideal number of children for a family in Austria?)
Coded response options: open-ended question.
Notes
This research was funded by the European Research Council under the European
Union’s Seventh Framework Programme
(FP7/2007-2013) / ERC Grant agreement
no. 284238 (EURREP project). Results of
the study have been presented at the 2014
Population Association of America Annual
Meeting in Boston and at the 2014 European Population Conference in Budapest.
An earlier expanded version is available as
a working paper (Sobotka and Beaujouan
2014). More details concerning data, data
quality, and data selection criteria, as well as
a sensitivity analysis of the presented results,
are available in supplementary online Appendixes 2–4 (http://videurrep.oeaw.ac.at/
two-is-best-pdr/). Data for Switzerland are
derived from the Swiss Household Panel (SHP
surveys), which is based at the Swiss Centre
of Expertise in the Social Sciences FORS
(project financed by the Swiss National Science Foundation). We thank Kryštof Zeman,
who provided help with the analysis of the
SHP surveys, and to Dimiter Philipov and
Anna Matysiak, who provided comments on
an earlier draft of this study.
1 This separation of datasets for eastern
and western Germany is based on the fact
that they formed two separate countries until
1990, but even more so on the distinctiveness
of family and fertility patterns in these two
areas that has persisted after reunification
(Kreyenfeld 2003; Goldstein and Kreyenfeld
2011; Sobotka 2011).
2 The question of ideal family size was not
included in GGS conducted in other countries.
3 The EB surveys of 2001, 2006, and 2011
asked respondents about both general and
personal ideal family size. (The question was:
“And for you personally, what would be the
ideal number of children you would like to
have or would have liked to have had?“) Testa
(2012: 9) found that in the 2011 EB there was
very close agreement in responses to these
two questions. The slightly higher personal
ideal family size could be partly explained by
women at higher parities reporting higher
personal than general ideal family size, since
they presumably wanted to avoid reporting a
discrepancy between their personal ideals and
actual behavior.
4 Our inclusion of these DHS datasets was
motivated by the effort to expand the available
datasets for Central and Eastern Europe, where
only a few relevant recent surveys exist.
5 It might be argued that by selecting a
wide age range for our analyses we would
miss the opportunity to detect shifts in ideal
family size among young adults, who could be
early adopters of low family-size ideals. However, the available literature does not suggest
any important shift among young respondents. A detailed analysis of the data from the
416
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
2006 and 2011 Eurobarometer survey by Testa
(2007, 2012) shows that mean general ideal
family size is relatively little differentiated by
age. In a large majority of countries, women
in the youngest age group, 15–24, showed
somewhat higher mean ideal family size (with
an EU-wide average of 2.27) than the next
two age groups considered, 25–39 and 40–54
(EU average of 2.11 and 2.13, respectively).
Further analysis (not presented here) showed
that these differences are too small to cause
shifts in the overall ideal family size for the
broad age group 15–49 when the population
progressively “ages” over time. In other words,
our results are robust to age-group selection
and would not change if we considered narrower age groups.
6 The high proportion of respondents
choosing the answer “There is no ideal number” in some of the Eurobarometer surveys
provides indication of the possible erosion
of the notion of ideal family size. This issue
is worth analyzing more fully, but since that
option is not available in the other surveys,
we did not analyze these responses and treat
them merely as “statistical noise” that may
affect the aggregate results. Moreover, in
many countries there is little consistency
in the share of missing and non-numerical
responses across different Eurobarometer
surveys (see online Appendixes 3 and 4),
which makes it difficult to pursue a substantive interpretation in terms of uncertainty
being systematically higher or lower in specific countries.
7 We did not include data for Ireland in
this regional comparison. The surveys consistently displayed considerably higher fertility
ideals in Ireland than in the other countries
in this region but were only available for the
earlier period until 1994. This affected the regional comparison of trends after 1994 when
no survey was available for Ireland (the EB
2001–11 surveys for Ireland showed a high
proportion of non-numerical responses be-
tween 23 and 30 percent and, in accordance
with our selection rule, were not included in
the analyses).
8 These averages refer to the simple average of the aggregate country-level results
across all surveys considered in a given period;
these data were not weighted by population
size of the analyzed countries.
9 The contrast between the earlier and
later period is partly affected by excluding the
2001–11 Eurobarometer surveys for Ireland
because of the high proportion of nonresponse
and non-numerical responses. If we included
these data, the maximum MIFS value would
remain higher, amounting to 2.70 in 2008–11.
10 If the survey also included unmarried men and women—as has been common
practice in surveys since the 1990s—the mean
ideal family size would probably be lower
since some of those who were unmarried
conceivably formed a select group with lower
family-size ideals.
11 The study by Philipov and Bernardi
(2011: 503, Table 2) reported a similarly low
mean ideal family size of 1.70 in the 1992
Fertility and Family Survey (FFS) for eastern
Germany.
12 Despite some correspondence between
fertility level and ideal family size among the
countries with the lowest cohort fertility in
Europe, relatively low levels of MIFS are also
found in some countries with higher cohort
fertility. For instance, the country with the
lowest MIFS in the 2011 Eurobarometer survey, 1.93 in the Czech Republic, has comparatively high CFR estimated at 1.77–1.78 for the
1974–75 cohorts (Prioux and Barbieri 2012).
13 In addition, as discussed above, ideals
may be seen as abstract general statements
that are often disconnected from the actual
fertility plans of individual men and women.
From this perspective, it is preferable to compare reproductive intentions, not ideals, with
realized fertility (Sobotka and Lutz 2011).
T o m á š S o b o t k a / É va B e a u j o u a n 417
References
Adsera, Alicia. 2006a. “An economic analysis of the gap between desired and actual fertility:
The case of Spain,” Review of Economics of the Household 4(1): 75–95.
———. 2006b. “Religion and changes in family-size norms in developed countries,” Review of
Religious Research 47(3): 271–286
Almodovar, Jean-Pierre. 1973. “Existe-t-il un ‘syndrome de l’enfant unique’?” Enfance 26(3):
235–249.
Axinn, William G., Marin E. Clarkberg, and Arland Thornton. 1994. “Family influences on
family size preferences,” Demography 31(1): 65–79.
Bachrach, Christine A. and S. Philip Morgan. 2013. “A cognitive–social model of fertility intentions,” Population and Development Review 39(3): 459–485.
Basten, Stuart and Baochang Gu. 2013. “National and regional trends in ideal family size in
China,” paper presented at the IUSSP International Population Conference, Busan, Korea,
26–31 August 2013. Accessed November 2013 at http://www.iussp.org/sites/default/files/
event_call_for_papers/GU%20and%20Basten%20Long%20Abstract.pdf.
Blake, Judith. 1966. “Ideal family size among white Americans: A quarter of a century’s evidence,” Demography 3(1): 154–173.
———. 1974. “Can we believe recent data on birth expectations in the United States?”
­Demography 11(1): 25–44.
———. 1981. “The only child in America: Prejudice versus performance,” Population and
­Development Review 7(1): 43–54.
Bongaarts, John. 2002. “The end of the fertility transition in the developed world,” Population
and ­Development Review 28 (3): 419–443.
Bulatao, Rodolfo A. 1981. “Values and disvalues of children in successive childbearing decisions,” Demography 18(1): 1–25.
Coleman, David A. 1998. “Reproduction and survival in an unknown world,” NIDI Hofstee
Lecture Series 5, NIDI, The Hague.
Demeny, Paul. 2003. “Population policy dilemmas in Europe at the dawn of the twenty-first
century,” Population and Development Review 29(1): 1–28.
Ding, Qu Jian and Therese Hesketh. 2006. “Family size, fertility preferences, and sex ratio in
China in the era of the one child family policy: Results from national family planning
and reproductive health survey,” BMJ, 333(7564): 371–373.
Foster, Caroline. 2000. “The limits to low fertility: A biosocial approach,” Population and Development Review 26 (2): 209–234.
Freedman, Ronald, Gerhard Baumert, and Martin Bolte. 1959. “Expected family size and family
size values in West Germany,” Population Studies 13(2): 136–150.
Goldstein, Joshua R. and Michaela Kreyenfeld. 2011. “Has East Germany overtaken West
Germany? Recent trends in order-specific fertility,” Population and Development Review
37(3): 453–472.
Goldstein, Joshua R., Wolfgang Lutz, and Maria Rita Testa. 2003. “The emergence of sub­
replacement family size ideals in Europe,” Population Research and Policy Review 22(5-6):
479–496.
Hagewen, Kellie J. and S. Philip Morgan. 2005. “Intended and ideal family size in the United
States,” Population and Development Review 31(3): 507–527.
Hank, Karsten. 2007. “Parental gender preferences and reproductive behaviour: A review of
the recent literature,” Journal of Biosocial Science 39(5): 759–767.
Hauser, Philip M. 1967. “Family planning and population programs” (book review article),
Demography 4(1): 397–414.
Hayford, Sarah R. 2009. “The evolution of fertility expectations over the life course,” Demography 46(4): 765–783.
Hin, Saskia, Anne Gauthier, Joshua R. Goldstein, and Christoph Bühler. 2011. “Fertility preferences: What measuring second choices teaches us,” Vienna Yearbook of Population Research
2011(9): 131–156.
418
P e r s i s t e n c e o f a T w o - C h i l d F a m i ly I d e a l i n E u r o p e
Hoffman, Lois Wladis and Martin L. Hoffman. 1973. “The value of children to parents,” in J.T.
Fewcett (ed.), Psychological Perspectives on Population. New York: Basic Books, pp. 19–76.
Hoffman, Lois Wladis and Jean Denby Manis. 1979. “The value of children in the United
States: A new approach to the study of fertility,” Journal of Marriage and the Family 41(3):
583–596.
Iacovou, Maria and Lara Patrício Tavares. 2011. “Yearning, learning, and conceding: Reasons
men and women change their childbearing intentions,” Population and Development Review
37(1): 89–123.
ICF International. 2012. MEASURE DHS STATcompiler. Accessed 17 April 2013 at http://www.
statcompiler.com.
Inglehart, Ronald. 1990. Culture Shift in Advanced Industrial Society. Princeton University Press.
Jefferies, Julie. 2001. “A reluctance to embrace the one-child family in Britain?” paper presented at EURESCO conference “The Second Demographic Transition in Europe,” Bad
Herrenalb, Germany, 23–28 June. Available at http://www.demogr.mpg.de/Papers/
workshops/010623_paper05.pdf.
Kohler, Hans-Peter, Jere R. Behrman, and Axel Skytthe. 2005. “Partner + children = happiness? The effects of partnerships and fertility on well-being,” Population and Development
Review 31(3): 407–445.
Kreyenfeld, Michaela. 2003. “Crisis or adaptation reconsidered: A comparison of East and
West German fertility patterns in the first six years after the ‘Wende,’” European Journal
of Population 19(3): 303–329.
Lesthaeghe, Ron. 1995. “The second demographic transition in Western countries: An interpretation,” in K.O. Mason and A.-M. Jensen (eds.), Gender and Family Change in Industrialized
Countries. Oxford, Clarendon Press, pp. 17–62.
Lutz, Wolfgang, Vegard Skirbekk, and Maria Rita Testa. 2006. “The low fertility trap hypothesis: Forces that may lead to further postponement and fewer births in Europe,” Vienna
Yearbook of Population Research 2006: 167–192.
Mancillas, Adriean. 2006. “Challenging the stereotypes about only children: A review of the literature and implications for practice,” Journal of Counseling & Development 84(3): 268–275.
Merli, M. Giovanna and S. Philip Morgan. 2011. “Below replacement fertility preferences in
Shanghai,” Population-E 66(3): 519–542.
Morgan, S. Philip. 1981. “Intention and uncertainty at later stages of childbearing: The United
States 1965 and 1970,” Demography 18(3): 267–285.
Morgan, S. Philip and Heather Rackin. 2010. “The correspondence between fertility intentions and behavior in the United States,” Population and Development Review 36(1):
91–118.
Myrskylä, Mikko, Joshua R. Goldstein, and Yen-Hsin Alice Cheng. 2013. “New cohort fertility
forecasts for the developed world: Rises, falls, and reversals,” Population and Development
Review 39(1): 31–56.
Myrskylä, Mikko and Rachel Margolis. 2012. “Happiness: Before and after the kids,” MPIDR
Working Paper WP-2012–013, Max Planck Institute for Demographic Research, Rostock.
Ní Bhrolcháin, Máire and Éva Beaujouan. 2011. “Uncertainty in fertility intentions in Britain,
1979–2007,” Vienna Yearbook of Population Research 2011(9): 99–129.
Quesnel-Vallée, Amélie and S. Philip Morgan. 2003. “Missing the target? Correspondence
of fertility intentions and behavior in the U.S.,” Population Research and Policy Review 22
(5–6): 497–525.
Philipov, Dimiter and Laura Bernardi. 2011. “Reproductive decisions: Concepts and measurement in Austria, Germany and Switzerland,” Comparative Population Studies 36(2–3):
495–530.
Prioux, France and Magali Barbieri. 2012. “Recent demographic developments in France:
Relatively low mortality at advanced ages,” Population (English edition) 67(4): 493–550.
Régnier-Loilier, Arnaud. 2006. “Influence of own sibship size on the number of children desired
at various times of life: The case of France,” Population (English Edition) 61(3): 165–194.
T o m á š S o b o t k a / É va B e a u j o u a n 419
Rotkirch, Anna, Stuart Basten, Heini Väisänen, and Markus Jokela. 2011. “Baby longing and
men’s reproductive motivation,” Vienna Yearbook of Population Research 2011(9): 283–306.
Ryder, Norman B. and Charles F. Westoff. 1969. “Relationships among intended, expected,
desired, and ideal family size: United States 1965,” Population Research, March.
Schoen, Robert, Nan Marie Astone, Young J. Kim, and Constance A. Nathanson. 1999. “Do
fertility intentions affect fertility behavior?” Journal of Marriage and the Family 61(3):
790–799.
Simons, John. 1978. “Illusions about attitudes,” in Population Decline in Europe: Implications of a
Declining Stationary Population. Edward Arnold Publishers, Council of Europe, pp. 197–214.
Sobotka, Tomáš. 2011. “Fertility in Austria, Germany and Switzerland: Is there a common
pattern? “ Comparative Population Studies 36(2-3): 263–304.
———. 2013. “Pathways to low fertility: European perspectives,“ Expert Paper No. 2013/8,
United Nations Department of Economic and Social Affairs, Population Division. http://
www.un.org/en/development/desa/population/publications/pdf/expert/2013–8_­
Sobotka_Expert-Paper.pdf.
Sobotka, Tomáš and Éva Beaujouan. 2014. “Two is best? The persistence of a two-child family
ideal in Europe,” Vienna Institute of Demography Working Papers 3/2014.
Sobotka, Tomáš and Wolfgang Lutz. 2011. “Misleading policy messages derived from the period
TFR: Should we stop using it?“ Comparative Population Studies 35(3): 637–664.
Spéder, Zsolt and Balázs Kapitány. 2014. “Failure to realize fertility intentions: A key aspect
of the post-communist fertility transition,” Population Research and Policy Review 33(3):
393–418.
Striessnig, Erich and Wolfgang Lutz. 2013. “Can below-replacement fertility be desirable?”
Empirica 40(3): 409–425.
Testa, Maria Rita. 2007 “Childbearing preferences and family issues in Europe: Evidence from
the Eurobarometer 2006 survey,” Vienna Yearbook of Population Research 2007(5): 357–379.
———. 2012. “Family sizes in Europe: Evidence from the 2011 Eurobarometer survey,” European Demographic Research Papers 2012, No. 2. Vienna Institute of Demography, Austrian
Academy of Sciences. http://www.oeaw.ac.at/vid/download/edrp_2_2012.pdf.
Testa, Maria Rita and Stuart Basten. 2013. “Fertility intentions and the ‘Great Recession’:
Evidence from the Eurobarometer surveys,” paper presented at the Annual Meeting of
Population Association of America, New Orleans, 11–13 April. ­Accessed in September
2013 at http://paa2013.princeton.edu/abstracts/132162.
Testa, Maria Rita, Laura Cavalli, and Alessandro Rosina. 2011. “Couples’ childbearing behaviour in Italy: Which of the partners is leading it?” Vienna Yearbook of Population Research
2011 (volume 9): 157–178.
Testa, Maria Rita and Leonardo Grilli. 2006. “The influence of childbearing regional contexts
on ideal family size in Europe,” Population (English edition) 61(1–2): 99–127.
Trent, Roger B. 1980. “Evidence bearing on the construct validity of ‘ideal family size,’” Population and Environment 3(3-4): 309–327.
VID. 2014. “European Demographic Data Sheet,” Vienna Institute of Demography and IIASA
/ Wittgenstein Centre for Demography and Global Human Capital. http://www.oeaw.
ac.at/vid/datasheet/index.html.
Waite, Linda J., Frances K. Goldscheider, and Christina Witsberger. 1986. “Nonfamily living and
the erosion of traditional family orientations among young adults,” American Sociological
Review 51(4): 541–554.
Ware, Helen. 1974. “Ideal family size,” World Fertility Survey, Occasional Papers No. 13 (October).
Wilson, Chris. 2013. “Thinking about post-transitional demographic regimes: A reflection,”
Demographic Research 28(46): 1373–1388.