Working PaperPDF Available

How real are reproductive goals? Uncertainty and the construction of fertility preferences

Authors:
ESRC Centre for Population Change • Working Paper 73 • December 2015
Máire Ní Bhrolcháin
Éva Beaujouan
ISSN 2042-4116
CPC
centre for population chang
e
Improving our understanding of the key drivers and implications of population change
How real are reproductive
goals? Uncertainty and
the construction of fertility
preferences
i
ABSTRACT
The underlying reality of fertility intentions, expectations and preferences tends to be
taken for granted. This is a natural consequence of the assumption, implicit or explicit in
most demographic research, that fertility behaviour is governed by rational choice. We
question the reality of fertility intentions, expectations, and preferences, and propose
instead that they are by and large constructed. Drawing on behavioural economics,
psychology, and political science, we develop an outline theory of fertility intentions and
preferences that contrasts with the classical rational choice model assumed in most work
in this area.
We show that there is a relatively high frequency of uncertain responses to questions on
fertility intentions and expectations and argue that the uncertainty expressed is genuine. It
was largely in response to this finding that our theoretical approach was developed. The
presence of uncertain answers to preference and intentions questions is acknowledged in
most demographic surveys but their frequency and theoretical and empirical significance
has been largely neglected.
Preferred family size may, we suggest, be a discovery rather than a goal. Demographic
thinking about fertility decisions could be enriched by adopting the idea of constructed
preferences from behavioural economics and psychology, together with ideas and debates
in political science regarding survey response. The construction of fertility preferences
and intentions can account for some hitherto unexplained anomalies in survey findings on
fertility intentions and expectations. Preference construction theory provides a novel
perspective on fertility intentions and preferences and on family building behaviour, and
merits serious empirical investigation in this context.
KEYWORDS
Fertility intentions; fertility preferences; desired family size; ideal family size;
constructed preferences; uncertainty; ambiguity; ambivalence; preference construction;
rational choice.
ii
EDITORIAL NOTE
Máire Ní Bhrolcháin is a Visiting Professor in the Faculty of Social, Human and
Mathematical Sciences at the University of Southampton.
Éva Beaujouan is a Research Scientist at the Wittgenstein Centre for Demography and
Global Human Capital (IIASA, VID/ÖAW, WU), Vienna Institute of
Demography/Austrian Academy of Sciences.
Corresponding author: Máire Ní Bhrolcháin, mnb2@soton.ac.uk
ACKNOWLEDGEMENTS
Earlier versions of the paper were presented at several seminars and conferences,
including the conference “From Intentions to Behaviour: Reproductive Decision-Making
in a Macro-Micro Perspective”, Austrian Academy of Sciences, Vienna, December 2010,
the Population Association of America Annual Meeting, San Francisco, April 2012, and
the European Population Conference, Stockholm, June 2012. We thank participants at
these and other meetings for their thoughtful reflections and comments. The Centre for
Population Change GHS time series datafile on which the paper is partly based was
created in collaboration with Ann Berrington and with the assistance of Mark Lyons-
Amos. The research was funded by ESRC Grant number RES-625-28-0001.
iii
ESRC Centre for Population Change
The ESRC Centre for Population Change (CPC) is a joint initiative between the
Universities of Southampton, St Andrews, Edinburgh, Stirling, Strathclyde, in
partnership with the Office for National Statistics (ONS) and the National Records
of Scotland (NRS). The Centre is funded by the Economic and Social Research
Council (ESRC) grant numbers RES-625-28-0001 and ES/K007394/1.
This working paper series publishes independent research, not always funded
through the Centre. The views and opinions expressed by authors do not
necessarily reflect those of the CPC, ESRC, ONS or NRS.
The Working Paper Series is edited by Teresa McGowan.
Website | Email | Twitter | Facebook | Mendeley
Máire Ní Bhrolcháin and Éva Beaujouan all rights reserved. Short sections of text,
not to exceed two paragraphs, may be quoted without explicit permission
provided that full credit, including notice, is given to the source.
iv
HOW REAL ARE REPRODUCTIVE GOALS? UNCERTAINTY AND
THE CONSTRUCTION OF FERTILITY PREFERENCES
TABLE OF CONTENTS
1.INTRODUCTION .................................................................................... 1
1.1.HISTORICAL BACKGROUND ...................................................................... 2
2.THE PREVALENCE OF UNCERTAINTY .......................................... 5
3.ARE PEOPLE REALLY UNCERTAIN ABOUT THEIR
FUTURE FERTILITY? .......................................................................... 9
3.1.UNCERTAINTY IN THE CONTEXT OF PREFERENCES ...................... 10
3.1.1.INDIFFERENCE, WEAK OR UNCLEAR PREFERENCES ..................... 10
3.1.2.CLEAR POSITIVE PREFERENCES .......................................................... 11
3.1.3.CLEAR NEGATIVE PREFERENCES ........................................................ 14
3.2.FURTHER EVIDENCE THAT UNCERTAINTY IS GENUINE ............... 15
3.3.AMBIGUITY IN PREGNANCY INTENTIONS .......................................... 16
4.A NEW THEORY OF FERTILITY PREFERENCES AND
INTENTIONS ........................................................................................ 17
4.1.AN OUTLINE OF PREFERENCE CONSTRUCTION THEORY ............ 18
4.2.FERTILITY INTENTIONS AND PREFERENCES AS
CONSTRUCTED .............................................................................................. 19
4.3.RESPONSES TO SURVEY QUESTIONS .................................................... 23
5.DISCUSSION ......................................................................................... 25
5.1.TESTING THE HYPOTHESIS ...................................................................... 26
5.2.CONCLUDING COMMENTS ........................................................................ 28
REFERENCES ........................................................................................... 29
1
“Itisperhapsatestimonytothecoercivenessofinterviewsituationshowrarely
participantssaydon'tknow,muchlesstrytobolt…”(Fischhoff1991)
1. INTRODUCTION
Data on fertility preferences and intentions have been collected in demographic surveys for
many decades in a wide range of contexts. They are analysed for a variety of purposes, most
notably to interpret past trends, to gauge future prospects, and to develop an understanding of
reproductive decision making. However, it has long been known that intentions, expectations
and preferences can be quite inconsistent with eventual fertility outcomes, at both individual
and aggregate levels (Morgan 2001). This is at odds with the classical rational choice model
implicit in most of the literature on family formation. The rational choice model, with its
assumption that people have clear preferences and goals, has increasingly been challenged in
recent years not only in psychology and sociology but also in economics (Kahneman 1996,
Smelser 1998, Bruni and Sugden 2007).
The aim of the present paper is to develop an outline theory of fertility intentions and
preferences as an alternative to the classical rational choice model implicit in much of the
demographic literature on family formation. We adapt ideas from behavioural economics and
political science to argue that fertility intentions and preferences might usefully be thought of
as constructed.
Our approach is primarily motivated by the finding that the frequency of uncertain
answers to survey questions on intentions and expectations is relatively high, higher than is
generally appreciated. In Britain in recent years around two fifths of women aged under 35
were unsure whether they would have a (further) birth. This figure changed little between
1991 and 2007 and so is not an isolated estimate (Ní Bhrolcháin, Beaujouan and Berrington
2010; Ní Bhrolcháin and Beaujouan 2011). That so many should be unsure about their
prospective fertility appears at first glance surprising, since uncertainty is largely absent from
both theoretical and empirical accounts of reproductive decisions: the relatively high
prevalence of uncertainty in fertility intentions data has been largely overlooked in the recent
literature.
Although it was recognised at least as long ago as the 1955 Growth of American
Families study that women and couples may be uncertain in their fertility intentions,
2
uncertainty was not reported explicitly in the early American fertility surveys. It was present
only implicitly in tabulations of maximum, minimum and most likely expected births,
derived from answers expressed in terms of ranges. It was not until Morgan’s (1981, 1982)
pioneering work that the issue received serious demographic attention in its own right.
Morgan established that uncertain fertility intentions were not simply a form of nonresponse
but were meaningful in themselves. While many demographic surveys have since recognised
the need to record respondents’ level of certainty about their fertility expectations, Morgan’s
broader themes have been addressed by only a few demographic authors (Schaeffer and
Thomson 1992, Johnson-Hanks 2005).
Our paper builds on Morgan’s classic insights. We show that a relatively high
prevalence of uncertainty is a robust finding, and suggest that uncertainty may be even more
common than is indicated by standard questions. We then argue that uncertainty is a rational
response to the developing life course, and provide evidence in support of this view. Finally,
we propose a new theoretical approach to reproductive intentions and preferences. Our
approach can explain the prevalence of uncertainty, the instability of measured preferences
and intentions, and their inconsistency with outcomes. We discuss also more general
implications for ideas about reproductive decision-making. For economy, we focus
exclusively on developed country data, but our themes are relevant in a less developed
country context also (Agadjanian 2005; Johnson-Hanks 2005; Withers, Tayrow and Adinata
2011).
The legacy of several decades of analysis and debate has left its mark on current ideas
about reproductive intentions, and so we start with a brief historical background. Throughout,
we use the terms ‘fertility intentions’ and ‘fertility expectations’ interchangeably: while the
concepts differ in principle, individual survey responses to these questions are close to
identical (Ryder and Westoff 1971; Morgan 2001).
1.1. HISTORICAL BACKGROUND
Fertility expectations data have been collected since the early post-war surveys of fertility in
the US and in Britain. Interest in such data in the early American fertility surveys was largely
practical, driven by their potential for improving population forecasts. The need arose from
the then novel cohort component method of projection. Recognising that nothing was known
3
of the future fertility of the youngest cohorts, Whelpton proposed asking younger women
themselves for the information. Hence, the Growth of American Families surveys in 1955
and 1960 were undertaken primarily to assess the reliability of fertility expectations and their
utility for population projection (Freedman, Whelpton and Campbell 1959; Whelpton,
Campbell and Patterson 1966; Kiser 1967). The subsequent National Fertility Surveys of
1965 and 1970 were much less focussed on projection issues, though influential in
identifying the limitations of birth expectations data for the purpose (Ryder and Westoff
1971, Westoff and Ryder 1977). Even so, the potential utility of birth expectations for
forecasting motivated comparable surveys elsewhere into the 1970s and is suggested to have
been the basis for European support for the World Fertility Survey series (Woolf 1971;
Woolf and Pegden 1976; Ryder 1986).
By the 1980s, evidence from several decades of research was consistent on several
points, as follows. The agreement between fertility intentions and outcome was much better
in the aggregate than for individuals. Nevertheless, when aggregated, fertility intentions did
not perform well enough for use in forecasting—they appeared to reflect current fertility
conditions or those of the recent past rather than future prospects (Westoff and Ryder 1977;
Lee 1980,1981). Compared with other individual characteristics, intentions were strong
predictors of fertility outcomes at the individual level, but their predictive power was
nevertheless modest. Fertility intentions were not fixed but varied through the life course
(Westoff, Mishler and Kelly 1957; Bumpass and Westoff 1970; Freedman, Freedman and
Thornton 1980). Prospective and retrospective reporting of pregnancy intentions often
disagreed, and these could be so inconsistent with contraceptive practice that leading scholars
began to regard such data with considerable scepticism (see e.g. Ryder 1973, 1979).
This broad picture remains valid to the present1 and it is therefore unsurprising that a
study conducted in the late 1990s found that few national statistical agencies used fertility
intentions in formulating assumptions for population projection (Van Hoorn and Keilman
1997). Nevertheless, intentions data continue to attract demographic attention as predictors of
1 Thomson and Brandreth (1995), Van Hoorn and Keilman (1997), Trussell, Vaughan and Stanford
(1999), Morgan (2001), Santelli, Rochat, Hatfield-Timajchy et al. (2003)
4
individual-level fertility outcomes.2 Beyond their utility in this respect, data on reproductive
orientation have several further established roles for research and policy purposes: in
assessing how far social differentials in family size are due to varying family size desires, in
reflecting couple dynamics, in characterising the reproductive life cycle, in evaluating the
causes of aggregate change, in shaping and monitoring policy in relation to unintended
pregnancy, particularly in the United States, in measuring “unmet need” for family planning
in less developed countries, and in assessing the prospects for aggregate fertility.3
In sum, questions on fertility expectations originated as a practical statistical device—
one solution to the problem of forecasting fertility, especially that of younger cohorts. The
limited success of intentions data in predicting fertility outcomes, at both individual and
aggregate levels, has given rise to scepticism. Reproductive intentions and attitudes have
been subject to an array of criticisms: that they are meaningless, superficial, guarded
responses, biased by social desirability effects, that they represent “irresponsible” attitudes
or, are subject to measurement error, random answers, and even possibly mendacious.4
Nevertheless, sceptics have tended to assume that true reproductive intentions exist and that
the problem is that they are mis-measured or misreported in demographic surveys.
Conceptually, they lack a theoretical pedigree, being rooted in no formal psychological,
sociological, economic, or demographic representation of reproductive behaviour (Coombs
1974, Thomson and Brandreth 1995), though they currently feature in several social-
psychological theoretical frameworks (Ajzen 1991; Miller and Pasta 1995; Heckhausen,
Wrosch and Fleeson 2001). Their experimental origins, accumulated empirical performance,
and persisting contested status among scholars suggest a strong case for reconsidering the
underlying reality of fertility intentions.
2Rindfuss, Morgan and Swicegood (1988), Monnier (1989), Schoen, Astone, Kim et al. (1999),
Quesnel-Vallee and Morgan (2003), Toulemon and Testa (2005), Testa and Toulemon (2006),
Liefbroer (2009), Philipov (2009), Speder and Kapitany (2009), Morgan and Rackin (2010).
3 Ryder and Westoff (1971), Westoff and Ryder (1977), Morgan (1985), Williams and Thomson
(1985), Westoff (1988), Thomson, Mcdonald and Bumpass (1990), Dixon-Mueller and Germain
(1992), Brown and Eisenberg (1995), Thomson (1997), Thomson and Hoem (1998), Casterline and
Sinding (2000), Bongaarts (2002), Berrington (2004), Finer and Henshaw (2006), Hayford and
Morgan (2008), Hayford (2009), Liefbroer (2009), Musick, England, Edgington et al. (2009), Rosina
and Testa (2009), Iacovou and Tavares (2011).
4 Hauser (1967), Bumpass and Westoff (1970): 19, Ryder and Westoff (1971), Cartwright (1976),
Westoff and Ryder (1977), Ryder (1979): 118, Ryder (1985), Demeny (1988), Thomson and
Brandreth (1995), Bankole and Westoff (1998), Bachrach and Newcomer (1999), Hayford (2009),
Kodzi, Casterline and Aglobitse (2010), Demographic and Health Surveys (2011): 84
2. T
H
We lo
o
develo
p
Figure
2005/
Note:
“proba
childre
Source
respon
s
whethe
r
House
h
H
E PREV
A
o
k first at
p
ed countri
e
1:Proportion
7
.
Womenare
blynot”,or
n.Seenote1
2
:CentreforP
Figure 1 s
h
s
e (“proba
b
r
they thin
k
h
old Surve
y
A
LENC
E
the freque
n
e
s, and con
s
withuncerta
classifiedas
“don’tknow
2
fordetails.
opulationCh
a
h
ows the pr
o
b
ly yes”, “
p
k
they will
h
y
, 1979-20
0
E
OF UN
C
n
cy of un
c
s
ider briefl
y
infertilityint
e
havingunc
e
toaquest
a
ngeGHStim
e
o
portion of
p
robably n
o
h
ave any (
m
0
7; a chan
g
C
ERTAI
N
c
ertain fert
i
y
also some
e
ntionsbyag
e
rtainfertilit
y
iononwhet
eseriesdata
f
women, b
y
o
t”, “don’t
m
ore) child
r
g
e in answ
e
N
TY
i
lity intent
i
issues of d
e
eandperiod.
y
intentions
hertheytho
f
ile
y
age and ti
m
know” or
r
en, asked
a
e
r options
o
i
ons in Br
i
e
finition an
d
Womenage
d
iftheyans
w
ughttheyw
o
m
e period,
g
no answer
)
a
nnually in
o
ccurred in
i
tain, and
i
d measure
m
d
1844.GB,
G
w
ered“prob
a
ouldhavea
n
giving an
u
r
) to a que
s
the British
1991, and
5
i
n other
m
ent.
G
HS1991
a
blyyes”,
n
y(more)
u
ncertain
s
tion on
General
so data
6
from 1991 onwards only are shown in Figure 1.5,6 Two features are noteworthy in this graph.
First, the overall level of uncertainty is fairly substantial. Just over 30 per cent of all women
aged 18-44 are unsure whether they will have (more) children, this proportion being the
highest, close to 40 per cent, for women in each age group under 35. Second, we see also
from Figure 1 that there was little change between 1991 and 2005/7 in the level of
uncertainty, though a slight upward trend among women 35+. The high frequency of
uncertainty is, thus, consistent across 17 years of repeated surveys and is not just an isolated
observation.
The prevalence of uncertain fertility intentions in Britain is not unusual. Comparable
levels are found in a range of other developed societies. We present elsewhere a cross-
national compilation of published estimates of the frequency of uncertain responses to
fertility intentions and expectations questions (Ní Bhrolcháin and Beaujouan forthcoming).
Among the 33 studies surveyed, few give a frequency of uncertainty below 10 per cent, 23
record a frequency of 20 per cent or more and 18 a level of 30 per cent or more. The figure of
40 per cent found at ages up to the mid-30s in the British GHS is matched or exceeded by the
overall figures of half a dozen sources, covering the US and a range of European countries.
Thus, the relatively high prevalence of uncertainty given by the GHS is not exceptional or
implausible in the context of comparable studies, whether past or recent.
There are, furthermore, several reasons for thinking that the frequency of uncertain
fertility intentions and preferences has often been underestimated. First, in many earlier
surveys, particularly those predating the recognition of the significance of uncertainty
(Morgan 1981, 1982), a tentative or “don’t know” response to a question on intended or
expected family size was regarded as non-response (Werner 1986; Riley, Hermalin and
Rosero-Bixby 1993; Van Hoorn and Keilman 1997). People were assumed to have clear
5 The General Household Survey data series used in this paper are weighted throughout by a set of weights
constructed on a consistent basis for annual GHS rounds from 1979 to 2007, for use in analysis of individuals
responding to the Family Information section of the questionnaire with valid revised fertility histories. Further
details of the revisions to the fertility histories are given in (Ní Bhrolcháin, Beaujouan and Murphy (2011)) and
of the weights in (Beaujouan, Brown and Ní Bhrolcháin (2011).
6 The birth expectations question is: “Do you think that you will have any (more) children (at all) (after the one
you are expecting)?” The wording remained almost the same from 1979-2007 (with a minor change in 1995 and
1996; see Smallwood and Jefferies 2003); the words “at all” were omitted from 1998 on. From 1979-1990
precoded answer categories were “yes”, “no” and “don’t know”. From 1991 onwards, a showcard was used,
with answer options “yes”, “probably yes”, “probably not”, and “no”. Those initially answering “don’t know”
are probed further and recoded “probably yes” or “probably not” where possible. “Don’t know” and no answer
are a small group, just 1%-2% overall, and 2%-8% of those classified here as uncertain.
7
fertility intentions, and interviewers were instructed to elicit an unambiguous declaration of
these. The result is likely to have been an upward bias in the recorded level of certainty in
fertility expectations.
In addition, being asked about their fertility intentions may itself convey to
respondents that they ought to have clear-cut intentions (Cartwright 1976, Chapter 3,
Cartwright and Wilkins 1976: 7-8, Simons 1978). A social desirability effect may prompt
people who are uncertain to give a definite response (Westoff and Ryder 1977: 346).
Respondents in fertility surveys may sometimes overstate how sure they are (Wikman 2006),
or report what might be described as “nonintentions”, by analogy with nonattitudes in
political science and basic values in psychology (Converse 1964, Fischhoff 1991).
The reported level of certainty may also be overstated as a result of a restricted set of
precoded answers. Where “uncertain” or “don’t know” or equivalent pre-codes are not
explicitly offered as options in survey questions, “don’t know” answers are less frequent
(Converse 1974; Kaufmann, Morris and Spitz 1997; Krosnick 1999; Schaeffer and Presser
2003). We found evidence that fertility expectations suffer from this bias also in the annual
GHS: in the years following the introduction of explicit uncertain precodes into the question
on fertility expectations, the frequency of uncertain responses rose. Between 1990 and 1991,
when “possibly” answers became available, uncertain answers increased from 9% to 29% (Ní
Bhrolcháin and Beaujouan 2011: Figure 2).
Finally, the frequency of uncertain intentions may be underestimated for definitional
reasons. The intentions question used here occurs in a wide range of demographic surveys, in
this or closely similar form. It is nevertheless a rough and ready measure, chosen ad hoc, and
with little or no validation. It has not been designed to measure uncertainty per se, and just as
intentions are not dichotomous (Morgan 1981) so too certainty is unlikely to be a binary
state. We therefore explore two further definitions by adding to the uncertain group defined
above those answering “yes” to the intentions question who expect a birth in either (a) 5+
years’ time or (b) 3+ years’ time (fuller details are given in Ní Bhrolcháin and Beaujouan
2011).
Figure
age.W
Source
younge
r
double
s
include
years.
A
25-29,
3
three i
n
their f
u
7 To cou
n
justificat
i
time hor
i
and Gen
e
uncertai
n
time dis
p
feature i
s
1991-20
0
next birt
h
heaping
should b
uncertai
n
2:Proportion
omenaged1
8
:CentreforP
On these
r
women r
i
s
from an
o
d, and rise
s
A
t ages 20-
2
3
9 per cent
,
n
five of w
o
u
ture child
b
n
t as uncertai
n
i
on. Medium
i
zon for expe
c
e
rations Prog
r
n
ty. The distr
i
p
lays substan
t
s
less pronou
n
0
5/7, Whippl
e
h
is 1 to 2
y
observed is
p
e coded 22,
n
ty on the res
p
uncertain,a
c
8
44.GB,GH
S
opulationCh
a
expanded
i
ses drama
t
o
riginal 41
s
further to
2
4, compar
,
47 per ce
n
o
men under
b
earing.
7
I
n
those expec
t
to long time
c
tations in a
n
r
amme. Addi
t
i
bution of th
e
t
ial heaping,
n
ced, though
e
indices for t
h
y
ears, and for
p
artly due to
in the mid-2
0
p
ondent’s par
t
c
cordingtoth
S
19912005/
7
a
ngeGHStim
e
definitions
,
t
ically (Fig
u
per cent to
90 per ce
n
a
ble figure
s
n
t and 61 pe
r
35, and ab
o
I
f uncertai
n
t
ing a wait of
horizons mu
s
n
umber of sur
v
t
ional evidenc
expected ag
e
and the long
still present,
h
e distributio
n
3+ years be
t
i
nterviewer i
n
0
s 25, and s
o
t
as much as
w
reedefinitio
n
7
.
eseriesdata
f
,
the esti
m
u
re 2). At
a
o
84 per ce
n
n
t with the
i
s
are 39 pe
r
r
cent, resp
e
o
ut one in
f
n
ty in inte
n
3+ years to t
h
s
t involve so
m
v
eys with a f
o
c
e from the G
H
e
at next birt
h
er the expec
t
among those
n
of age at ne
x
t
ween 157 a
n
n
structions w
h
o
on. Howev
e
w
oul
d
direct r
e
n
s,astowhet
f
ile.
m
ated prev
a
a
ges 18-19
,
n
t when ex
p
i
nclusion o
f
r
cent, 67 p
e
e
ctively. O
n
f
ive of thos
e
n
tions is a
s
h
eir first/next
b
m
e uncertaint
y
o
llow-up com
p
H
S supports t
h
h
given by w
o
t
ed delay, th
e
expecting a
b
x
t birth are b
e
n
d 231, depe
n
h
ich specify
t
e
r, such ans
w
e
ports of 22,
2
h
ertheywill
h
a
lence of
u
,
the frequ
e
p
ected del
a
f
those exp
e
e
r cent and
n
the most i
e
aged 35-4
s
common
b
irth is arbitr
a
y
, and three
y
p
onent, such
a
h
e of 3+ year
o
men expecti
n
e
more heape
d
b
irth in 1-2 y
e
e
tween 115 an
n
ding on the
a
t
hat e.g. an a
n
w
ers are appr
o
2
5, 28, 30 an
d
h
aveafurthe
u
ncertainty
e
ncy of un
c
a
ys of 5+
y
e
cting a w
a
81 per ce
n
i
nclusive d
e
4
4, are uns
u
as this, si
g
a
ry and may
n
y
ears is the li
m
a
s the UN E
C
r
s criterion as
n
g a birth in
d
the distrib
u
e
ars’ time. I
n
n
d 140 when t
h
a
ge range ch
o
a
nswer in the
o
ximate, and
d
so on.
8
r
birth,by
among
c
ertainty
y
ears are
a
it of 3+
t, and at
e
finition,
re about
g
nificant
n
eed some
m
it of the
C
E Gender
reflecting
3+ years’
u
tion. The
n
the GHS
h
e time to
o
sen. The
early 20s
so reflect
9
questions arise about the nature and interpretation of fertility intentions, about the levels of
uncertainty measured in a wide range of fertility surveys, and about fertility decisions per se.8
In sum, the prevalence of uncertainty is fairly high in developed societies, and there
are indications that its frequency may be underestimated in existing sources. Being a
somewhat neglected issue, the relatively high prevalence of uncertainty suggests that it may
offer a route to a better understanding of the reproductive life course, if the phenomenon can
be shown to reflect something real. In the next section, we present arguments and evidence
that support the reality of uncertain fertility intentions. In a later section, we go on to outline a
theoretical approach to fertility preferences that can explain the level of uncertainty, the
instability of intentions, and their inconsistency with fertility outcomes.
3. ARE PEOPLE REALLY UNCERTAIN ABOUT THEIR FUTURE
FERTILITY?
How should we interpret the high prevalence of uncertainty documented in the previous
section? In much of the previous literature the high frequency of “don’t know” or ambiguous
answers is downplayed in a number of ways—either don’t know answers are overlooked and
omitted from analysis entirely or uncertainty is attributed to factors such as poor
measurement and lack of respondent motivation or knowledge. By contrast, we believe that
unsure answers should be taken seriously--that the uncertainty expressed about future fertility
is both genuine and well founded.
The GHS question elicits expectations regarding a future/the next birth. This style of
question is regarded as more reliable than questions on the number of intended or expected
births (Casterline and El-Zeini 2007). The question is simple and realistic. Women are asked
only whether they expect to have a (further) birth ever, rather than to express an imaginary
ideal, as in some demographic surveys. An uncertain response to this type of question
appears unlikely to be due to lack of knowledge or understanding. We propose that the
uncertainty expressed in answers to questions on intentions regarding a future/the next birth,
and reflected also in questions on expected age, is both real and reasonable. We suggest that
it is not primarily the result of mis-measurement, though some role for the latter cannot be
8 Substantial uncertainty about the timing of first birth is reported by Rindfuss et al (1988: 195-6), with 29% of
childless women and 43% of childless men in their early to mid-twenties answering “don’t know” to a question
on when they expected to have their first child.
10
ruled out. In the present section we draw on demographic and other evidence to argue this
case.
3.1. UNCERTAINTY IN THE CONTEXT OF PREFERENCES
Consider the conditions under which people could be reasonably sure about their future
fertility. Imagine a society in which some form of sexual union is universal, that sexual
unions begin at puberty and continue uninterrupted to menopause, that everybody has the
same level of fecundity, that no steps are taken to control fertility and that childbearing is not
in competition with other activities. In such a society the vast majority of people could be
fairly sure of their reproductive futures. Individuals in real-life developed societies
experience conditions far from this hypothetical scenario, and have multiple grounds for
being uncertain about their future fertility. Except for post-menopausal women and people
who know themselves to be sterile, most people in modern developed societies cannot be sure
of how their future reproductive lives will play out.
We elaborate below on the principal reasons why a person may (and should) be
unsure of their reproductive expectations in a modern developed society. These are
considered under three scenarios: where a person’s reproductive preferences are unclear,
where they are clear and positive, or clear and negative. We discuss fertility preferences and
intentions in general, and do not distinguish at this stage between preferences/intentions for
having a family per se, or for specific family sizes, or specific timing, since little is known
about the origin and inter-relation of these separate aspects.
3.1.1. INDIFFERENCE, WEAK OR UNCLEAR PREFERENCES
People may be unsure of what they want by way of family size, or timing. They may never
have thought about the issue and have no preference at all. They may alternatively have
thought about it and have weak preferences, or are ambivalent, or just do not know what they
want (Schaeffer and Thomson 1992). This does not mean that they do not care at all either
about having a family or about family size. Rather, the prospect may have little salience. This
could occur if they are at too early a life stage to have formed specific views and preferences
about childbearing, or too occupied with other activities to have considered the matter in
detail. They may be without a partner and the prospect of childbearing may therefore be an
11
abstract one. For all these reasons, fertility preferences may be ill-defined and this seems
particularly likely at younger ages.
Where people are either ambivalent about fertility, or unclear in what they want, or
have weak preferences, or no preferences at all, it would be unsurprising if they have no clear
intentions or expectations in relation to fertility. A lack of clear preferences may thus explain
what we saw earlier in Figure 2 at younger ages: that on our extended definition, the vast
majority of young women are uncertain about their fertility intentions. We saw that it is not
until the early 30s that a majority respond with certainty to the question on birth expectations.
3.1.2. CLEAR POSITIVE PREFERENCES
In contrast to the previous scenario, let us consider people who have clear, unambiguous
preferences for (further) childbearing. Does this imply that their fertility intentions will
necessarily be clear, unambiguous, unconditional? We believe not. Even where desires and
goals are clear, limitations of fecundity, of control, and of knowledge preclude complete
certainty about realizing these, and so limit intentions and expectations.
Those with a clear desire to have (more) children cannot be certain about their future
fecundity, particularly if they have never been exposed to the risk of pregnancy. We would
therefore expect uncertainty to be age related and this is indeed the case. Among women aged
under 30 the percent sterile and unable to have a live birth is low (below10 percent), based on
estimates from a number of sources. However, at age 30 the estimates range from 7 per cent
to 12 per cent, and at 35 from 13 per cent to 22 per cent (Leridon 2008: Table 3). Beyond
fecundity is the issue of finding a partner with whom to have children. People who already
have a partner cannot be sure that the union will remain intact. People who are not in a union
cannot be sure of finding a partner, or the right partner, with whom to have a child or children
(Zabin, Huggins, Emerson et al. 2000, Testa 2007). Thus, even where people have a clear
preference for childbearing, they cannot be sure that they will achieve this in the future, and
so cannot reasonably intend or expect to do so.
Data from the GHS of 2000-2007 give statistical substance to these points. Figure 3
shows the proportion of women having a birth within 10 years of initial observation, by
initial age and parity, among respondents to the General Household Surveys of 2000-2005/7.
Amon
g
greater
further
those
w
While
m
10-yea
r
there a
p
Figure3:
2005/7.
Note:
W
survey.
T
ageand
s
Source:
C
g
women ch
than abou
t
birth withi
n
w
ith 2+ chil
d
m
any more
r
figure rep
r
p
pears to b
e
Proportionh
a
W
omeninterv
i
T
hediagrams
s
tate.
C
entreforPo
p
ildless at a
n
t
three in f
i
n
10 years
i
d
ren, the p
r
younger
w
r
esents the
e
no sure th
i
a
vingabirth
w
i
ewedat20
0
howsthepro
p
ulationCha
n
n
y age, the
p
i
ve. Amon
g
i
s below 0.
8
r
obabilities
w
omen hav
e
immediate
i
ng, no rati
o
w
ithin10yea
r
0
02005/7su
r
portionofth
o
n
geGHStime
s
p
robability
g
those wit
h
8
at all age
s
are substa
n
e
a (further
)
and mediu
m
o
nal basis f
o
r
s,byinitiala
g
r
veysarecla
s
o
sewhohad
a
seriesdatafil
of having
a
h
one chil
d
s
, and is ju
s
n
tially lowe
r
)
birth by t
h
m
term fut
u
o
r a definit
e
g
eandparity.
s
sifiedbyth
e
a
tleastone(
f
e.
a
birth with
i
d
already, t
h
s
t 0.5 in th
e
r
, at all but
h
e time the
y
u
re, and ov
e
e
intention
o
Womenage
d
e
irageand
p
f
urther)birth
i
n 10 years
he probabi
l
e
early 30s.
t
the young
e
y
reach ag
e
e
r that time
o
r expectati
o
d
1844.GB,
G
p
arity10ye
a
beforesurve
y
12
is never
l
ity of a
Among
e
st ages.
e
45, the
horizon
o
n.
G
HS2000
rsbefore
y
byinitial
Figure4:
initialag
e
Note:W
o
befores
u
befores
u
Source:
C
women
Of 18-
2
the ne
x
alone a
t
at 34-3
7
those
w
years i
s
current
29, and
circum
s
be “rig
h
Pro
p
e
andstate.
W
omenintervi
e
u
rvey.Thed
i
u
rveybyiniti
a
C
entreforPo
p
Similar sta
t
experienci
n
2
1 year old
s
x
t 10 years
t
older age
s
7
formed a
u
w
ho are alre
s
not negli
g
at ages 18-
one in six
o
Beyond t
h
s
tances wil
l
h
t” (Testa
2
p
ortionofwo
W
omenaged
1
e
wedat200
0
i
agramshow
s
a
lageandsta
t
p
ulationCha
n
t
istical unc
e
n
g a chang
s
who are l
i
again, no
s
are well
b
u
nion with
i
ady in a u
n
g
ible and is
21 will bre
a
o
f those in
a
h
ese docu
m
l
be suit
a
b
l
e
2
007). Peo
p
menexperie
n
1
844.GB,G
H
0
2005/7surv
e
s
thepropor
t
t
e.
n
geGHStime
s
e
rtainty sur
r
e
of partne
r
i
ving alone
sure thing
elow this.
F
i
n 10 years.
n
ion. The c
h
particularl
y
a
k up withi
n
a
union in t
h
m
entable
r
e
for childb
p
le cannot
b
n
cingachan
g
H
S20002005
/
eysareclassi
f
t
ionofthose
seriesdatafil
r
ounds part
n
r
ship status
79 per cen
t
. The part
n
F
or exampl
Partnershi
p
h
ance that
a
y
high at t
h
n
10 years,
h
eir mi
d
-3
0
r
isks, peo
p
earing—th
e
b
e sure ab
o
g
eintheirpa
/
7.
f
iedbytheir
a
whoexperie
e.
n
ership. Fi
g
within 10
t
will form
n
ership pro
s
e
, just 40 p
p
is also so
m
a
n existing
u
h
e younges
t
over a qua
r
0
s, again fr
o
p
le cannot
e
re is no g
u
o
ut their o
w
r
tnershipsta
t
a
geandpart
n
nceachang
e
g
ure 4 sho
w
years, by i
n
at least on
e
s
pects of t
h
er cent of
w
m
ething of
a
u
nion will
t
t
ages. Ove
r
r
ter of thos
e
o
m the GH
S
be sure
u
arantee tha
t
w
n future p
r
t
uswithin10
n
ershipstatu
s
e
inpartners
h
w
s the prop
o
n
itial age a
n
e
partnershi
p
h
ose who a
r
w
omen livi
n
a
n unknow
n
t
erminate
w
e
r half of al
e
current at
a
S
survey hi
s
that thei
r
a
t the time
w
r
eferences:
13
years,by
s
10years
h
ipstatus
o
rtion of
n
d state.
p
within
r
e living
n
g alone
n
among
w
ithin 10
l unions
a
ges 26-
tories.
r
future
w
ill ever
sizeable
14
instability of preferences and intentions across the life course has been documented by
several decades of research. In addition, the future preferences of any current or future
partner are also unknown. Evidence that this matters is of two kinds. First, the preferences of
the partners in a union can differ appreciably and are independently associated with fertility
outcomes.9 Second, fertility expectations predict subsequent outcomes less well among
women who were not in a union at initial contact.10 In addition, women without a partner
more often express uncertain fertility intentions (O’Connell and Rogers 1983, Ní Bhrolcháin
and Beaujouan 2011). The very early fertility surveys asked fertility intentions questions only
of married women, but in recent decades these questions are put to all women, whether in a
union or not. Ryder (1984) was of the view that the fertility intentions of unmarried women
(and we can probably extend the point to the unpartnered) were particularly unlikely to be
trustworthy, precisely because fertility prospects are, in the main, subject to additional
uncertainty for women not in a union.
3.1.3. CLEAR NEGATIVE PREFERENCES
Finally, if we consider people who have clear and unambiguous preferences not to have
(further) children, comparable difficulties of knowledge and control arise. Those who want
no (more) children cannot, without abstinence, sterility, or sterilization, be sure that they will
avoid future pregnancy. A recent estimate puts the proportion of all pregnancies in developed
countries that are unintended at 47 per cent (Singh, Sedgh and Hussain 2010). Nor can people
know for certain how either they or a current or future partner would react in the event of
pregnancy. From longitudinal studies, the proportion of women initially stating an intention
not to have a birth who have had a birth on follow up ranges between 7 per cent and 25 per
cent, in studies with varying lengths of follow up.
To sum up, in all scenarios from weak or absent fertility preferences to strong and
unambiguous childbearing preferences, there are sound reasons for being uncertain in
intentions and expectations. Uncertainty in relation to fertility intentions is a perfectly
reasonable, indeed rational, response to the imperfect information people have about their
reproductive prospects.
9 Thomson, Mcdonald and Bumpass (1990), Thomson (1997), Berrington (2004), Jansen and Liefbroer (2006),
Liefbroer (2009), Rosina and Testa (2009)
10 Schoen, Astone, Kim et al. (1999), Quesnel-Vallee and Morgan (2003), Berrington (2004), Toulemon and
Testa (2005), Testa (2007), Morgan and Rackin (2010)
15
3.2. FURTHER EVIDENCE THAT UNCERTAINTY IS GENUINE
Two further types of evidence suggest that it is reasonable on statistical grounds to be
uncertain about future childbearing intentions: inconsistency between stated intentions and
completed fertility, and variance in the age at first birth.
Evidence that fertility expectations lack predictive validity at both individual and
aggregate levels has accumulated over several decades. Several longitudinal studies since the
late 1960s have examined the consistency of initial intentions/expectations with subsequent
outcomes. In most of these, between 40 per cent and 70 per cent of those stating an intention
to have a (further) birth at initial interview had not carried out their intention by the time of
follow up (ranging from 2 to 18 years later).11 These studies also show much greater
consistency between intention and outcome among those not intending a (further) birth.
There is thus a substantial mis-match between initial positive intentions and both shorter and
longer term outcomes, and lesser but still non-negligible levels of inconsistency among those
with negative intentions. These statistics in themselves underline it is reasonable to be
uncertain in relation to fertility expectations. Adding weight to this point, the uncertain
expectations of those initially declaring themselves unsure are borne out in subsequent years:
they are less likely to carry out the intention, whether positive or negative, than are those who
initially declared themselves certain.
The second statistical basis for uncertainty among childless women in relation to
childbearing is the wide dispersion in the age at first birth. In Britain in 2009, the 25th
percentile of the age specific first birth schedule was 23.7 and the 75th, 35.1, giving an
interquartile range of 11.4 years. While individual women cannot be aware of these
demographic quantities, it seems likely that the substantial variability in the age at first birth
is a known feature of women’s lives.
11 Sources on the subject include the following: (Bumpass and Westoff (1969), Cartwright and Wilkins (1976),
Westoff and Ryder (1977), Freedman, Freedman and Thornton (1980), Noack and Ostby (1985), Rindfuss,
Morgan and Swicegood (1988), Monnier (1989), Schoen, Astone, Kim et al. (1999), Noack and Ostby (2002),
Quesnel-Vallee and Morgan (2003), Berrington (2004), Toulemon and Testa (2005), Testa and Toulemon
(2006), Liefbroer (2009), Philipov (2009), Speder and Kapitany (2009), Morgan and Rackin (2010)
16
In all, there are strong empirical grounds for the proposition that uncertainty about
future fertility is a rational position for many people, much if not most of the time. That the
frequency of reported uncertainty is high when uncertain answer options are available
suggests that survey respondents are either explicitly or implicitly aware of this. Finally, the
advent of near-perfect contraception has not altered the predictive power of fertility
intentions and expectations data, indicating that there is a great deal more involved here than
simple failure to control conception.
3.3. AMBIGUITY IN PREGNANCY INTENTIONS
A final body of evidence of pervasive ambiguity and lack of certainty underlying
reproductive decisions is found in the field of family planning. In that literature, conventional
concepts and measures of intended, wanted, and planned pregnancies have been challenged
and debated in the last decade or so.12 Concern has arisen in particular from extensive
evidence of inconsistency between reports of intention, desires, contraceptive use, and
affective reaction to pregnancy. One pivotal study based on the 1995 National Survey of
Family Growth found that 31 per cent of women reporting a contraceptive failure said the
pregnancy was intended; of those with an unintended pregnancy, 25 per cent declared
themselves happy or very happy to be pregnant (Trussell, Vaughan and Stanford 1999).
Comparable discrepancies between stated intention, contraceptive use, and reaction to
pregnancy in studies over a period spanning several decades13 has led investigators to seek
more complex, multidimensional indicators of pregnancy intentions (Santelli, Rochat,
Hatfield-Timajchy et al. 2003, Santelli, Lindberg, Orr et al. 2009).
Qualitative studies addressing these issues suggest that women do not always think of
themselves, and cannot always be readily classified, as trying/not trying, or intending/not
intending, to become pregnant, and that the idea of planning a pregnancy either is not
understood or has negative connotations for some women (Barrett and Wellings 2002;
Gerber, Pennylegion and Spice 2002; Kendall, Afable-Munsuz, Speizer et al. 2005). An
unplanned pregnancy may furthermore be seen as advantageous, in removing the difficulty of
deciding whether and when to have a child (Luker 1999, Lifflander, Gaydos and Rowland
12 Petersen and Moos (1997), Bachrach and Newcomer (1999), Luker (1999), Zabin (1999), Barrett and
Wellings (2000, 2002), Santelli, Rochat, Hatfield-Timajchy et al. (2003), Esacove (2008)
13 Cartwright (1970, 1976), Ryder (1976, 1979), Cartwright (1988), Moos, Petersen, Meadows et al. (1997),
Sable and Libbus (2000), Petersen, Gazmararian, Clark et al. (2001), Higgins, Hirsch and Trussell (2008)
17
Hogue 2007). Ambiguity in pregnancy intentions has also been reported in larger scale
studies. In McQuillan et al’s (2010) telephone survey, 23 per cent of women classified
themselves as neither trying to become pregnant nor trying to avoid it, but were “okay either
way.” Prospective studies reveal, in addition, that reports of the intendedness of a birth can
change during the course of a pregnancy (Joyce, Kaestner and Korenman 2000, Poole,
Flowers, Goldenberg et al. 2000). The picture emerging from this highly applied and focused
body of work is that intentionality is more complex and fluid, and ambivalence more
prevalent, than has been allowed for, conceptually and methodologically, in mainstream
fertility surveys.14
To sum up, using a range of empirical evidence, we have argued in this section that
the uncertainty in fertility intentions expressed in fertility surveys is genuine, rather than that
uncertain responses reflect measurement error. We have also shown that uncertainty is a
rational response to individuals’ fertility prospects, in a statistical sense. Finally, strong
corroboration of these ideas is found in several decades of studies in the family planning
literature that demonstrate much ambivalence and ambiguity in attitudes to pregnancy.
4. A NEW THEORY OF FERTILITY PREFERENCES AND
INTENTIONS
The empirical findings presented here cannot be accommodated within existing theoretical
frameworks. Our findings are inconsistent with the rational choice account that is implicit or
explicit in theories of fertility decision making. In a means-end framework, preferences and
intentions are clear-cut, and there is little or no role for uncertainty and ambiguity of desire
and intent.15 We propose therefore a new theoretical approach to reproductive intentions and
preferences, adapting insights from recent research in psychology and economics. Our aim is
to present a realistic account of reproductive preference and choice that can accommodate the
high prevalence of uncertainty to which we have drawn attention. The ideas share
perspectives in common with developing ideas in the field, particularly those of Johnson-
Hanks (2005), Johnson-Hanks, Bachrach, Morgan et al. (2011), Morgan and Bachrach (2011)
and Bachrach and Morgan (2013).
14 Kaufmann, Morris and Spitz (1997), Klerman (2000), Stanford, Hobbs, Jameson et al. (2000), Santelli,
Rochat, Hatfield-Timajchy et al. (2003), Barrett, Smith and Wellings (2004), Finer and Henshaw (2006),
Santelli, Lindberg, Orr et al. (2009)
15 The pervasive survey practice of treating “don’t know” answers to intentions and preferences questions as
non-response, and attempting to recode them, is a concrete expression of the assumption that intentions and
preferences are clearly held.
18
4.1. AN OUTLINE OF PREFERENCE CONSTRUCTION THEORY
We suggest that fertility preferences and intentions are generated by a constructive process.16
Preference construction as a concept has developed at the intersection of psychology,
economics, and behavioural decision theory. It is a response to recurring evidence that in
many situations people’s preferences are changeable, context-dependent, and subject to
framing effects (for an overview, see Lichtenstein and Slovic 2006). In particular, extensive
experimental evidence of the reversal of preferences has challenged basic assumptions about
preferences in the classical rational choice model (Tversky and Thaler 1990; Kahneman
1994; Lichtenstein and Slovic 2006). Though a dynamic and evolving area of research in
economics, psychology, and decision theory, the ideas appear as yet to have made little
headway in sociology and demography.
If we view preferences as constructed, the high prevalence of uncertainty is no longer
anomalous, but inherent to fertility preference and choice. The basic premise of the approach
is that people do not always have clear preferences. Desires may be well-defined when it
comes to simple choices that are frequently encountered, in familiar contexts, or in relation to
somewhat more complex choices where social and cultural conditions have long been stable.
Many choices are not of this type. They may have been encountered either rarely or never
before, and may be one-off decisions. In these circumstances, people will often lack a clear
preference—they may not know what to want or how to choose. When called on either to
state a preference, or to act on one, they look for clues and make inferences as to what they
would like, and thus how to act, or what preference to declare. In other words, rather than
reading off their preference from a stored memory, they construct a preference from available
information (Lichtenstein and Slovic 2006). In traditional thinking the reporting of a
preference is like a form of archaeology but in the constructive account it is more akin to
architecture (Payne, Bettman and Schkade 1999). The constructive view holds that a person
identifying their own preference, or reporting a preference in a survey, is assembling a
preference when requested, rather than retrieving and declaring a pre-existing one. An often
16 Preferences and intentions are considered jointly here. While conceptually distinct, the two are closely linked
in practice, as desired family size seems to carry little information beyond that present in intended family size.
Using the National Fertility Survey follow-up data, Ryder (1981) shows that change in intended parity is a
significant predictor of change in desired parity 1970-1975, but that the reverse is not the case. In that sense he
considers intended family size “causally prior” to desired family size.
19
quoted analogy illustrates the distinction between old and new perspectives on preferences
and values via three baseball umpires:
"I call them as I see them," said the first. "I call them as they are," claimed the second. The
third disagreed, "They ain't nothing till I call them." Analogously, we can describe three
different views regarding the nature of values. First, values exist—like body temperature—
and people perceive and report them as best they can, possibly with bias (I call them as I see
them). Second, people know their values and preferences directly—as they know the
multiplication table (I call them as they are). Third, values or preferences are commonly
constructed in the process of elicitation (they ain't nothing till I call them).
Tversky and Thaler (1990: 210)
4.2. FERTILITY INTENTIONS AND PREFERENCES AS CONSTRUCTED
In the constructive interpretation, preferences are assembled in different ways for different
types of choice (Bettman, Luce and Payne 1998). In family formation there are, we suggest,
two distinct expressions of preferences—those operating in family building itself, and those
reflected in responses to survey questions. We term these effective preferences and stated
preferences, respectively17 and suggest that they result from different constructive processes.
Our account of effective preferences is rooted in the psychology and economics literature.
We look to political science for insights on stated preferences.
Proponents of preference construction do not see all preferences as constructed. In
simple, repeated and familiar choices, such as supporting one football team rather than
another, preferences are stored in memory. By contrast, when situations are unfamiliar, or
when trade-offs are necessary between elements of a choice situation, or when a person is
asked to express a preference numerically, a constructive process is hypothesised as operating
(Fischhoff 2006, Lichtenstein and Slovic 2006). Childbearing in developed societies fits these
conditions well. Building a family of their own is a novel experience for most people, and is
usually encountered only once. Trade-offs are certainly present, there being multiple
competing claims on a woman’s or couple’s time and resources besides bearing and rearing
children. Finally, preferences are elicited in numerical terms in survey questions on the
number of children desired/intended, or the expected timing of a birth.
17 The term “stated preferences” is also used in applied economics (see e.g. Carlsson 2010).
20
Reproductive orientation has several other features characteristic of constructive
preferences. We saw earlier that many women’s intentions are unclear, and so we have
explicit evidence that preferences are not always well-articulated. Key evidence for the
constructive nature of preferences is that preferences are both labile and differ systematically
depending on the method by which they are elicited. Both of these features are characteristic
of fertility preferences and intentions. Several decades of research has shown that they are
unstable,18 and can be quite inconsistent with fertility outcomes (note 19 above), and both
features continue to be found in recent sources.
The explanations offered in the demographic literature for labile preferences and
disparities between birth expectations and ultimate fertility include unrealistic preferences,
constraint, change in circumstances, change of mind, thoughtless responses, measurement
error, and the like. The constructive interpretation provides an alternative explanation of both
instability and inconsistency. Fertility preferences are unstable, on the constructive
hypothesis, because they are ill-defined, especially at young ages, and also because they are
context-dependent. They are inconsistent with outcomes both for this reason and because
actual childbearing and survey reports of preferences/intentions are quite different modes of
expressing preferences.
If childbearing preferences are entirely constructed, it follows that there is no “true”
underlying preference in relation to family size—at any rate before preferences converge on a
stable state. In the most extreme form of this idea, survey questions on intentions/preferences
are attempting to measure something that either does not exist or has only a shadowy
existence: people making choices about childbearing, or declaring a preference in this
respect, have to work out what they want, rather than simply consulting some mental master-
list of known desires (see also Greil and Mcquillan 2010: 140). Less extreme versions of the
idea can, of course, also be considered: for example, preferences may be inherent to some
extent, and may vary in degree of construction both between individuals and across the life
course (Simonson 2008).
18 Freedman, Coombs and Bumpass (1965), Bumpass and Westoff (1970), Cartwright and Wilkins (1976),
Freedman, Freedman and Thornton (1980), Berrington (2004), Heiland, Prskawetz and Sanderson (2008),
Liefbroer (2009), Iacovou and Tavares (2011)
21
If, as we suggest, fertility preferences are constructed, how does this occur, and what
influences the process? In the sections that follow, we sketch some outline ideas, based on
our findings to date. We conclude with some thoughts on how the subject can be pursued
empirically.
We saw above that almost all of the youngest women are uncertain about prospective
childbearing but that older women are surer of their expectations (Figure 2). We show
elsewhere that uncertainty varies largely with age and demographic life stage—partnership
status, parity, time since previous birth—and relatively little with socio-economic factors or
calendar time; net of other factors, women not in a union are more uncertain than others, as
are those of lower parity, and those who have had a birth more recently (Ní Bhrolcháin and
Beaujouan 2011). Similar associations are reported in other studies. O’Connell and Rogers
(1983) found that single women were more uncertain about birth expectations than were
married women. Declines in uncertainty with age or parity, or with both, are evident in the
studies of Morgan (1982), Monnier (1989), Wu and Wang (1998), Berrington (2004) and
Sobotka (2009).
These results are what we would expect from a constructive process in which young
people do not have well-defined childbearing preferences, but that preferences develop over
time. Young adults’ effective preferences—those influencing actual childbearing—are, we
hypothesise, generated by encounters with choices and contingencies, and by learning from
experience and observation, under the real conditions in which their lives play out. From
repeated exposure to reproductive choices, preferences and intentions ultimately stabilize.
Empirical evidence shows that fertility intentions are less volatile at older ages, as would be
expected on the constructive account.19
Our hypothesis differs from the proposition that people start off in young adulthood
with well-articulated preferences and simply change their minds over time (Iacovou and
Tavares 2011). Our view differs also from the idea that underlying preferences and intentions
exist, are badly measured by conventional questions but better identified by more complex
measuring instruments such as the Coombs scale (Coombs 1978, Coombs and Freedman
1979). We suggest instead that at young ages preferences are vague and unclear, but become
19 Berrington (2004), Heiland, Prskawetz and Sanderson (2008), Iacovou and Tavares (2011); see also
Kahneman (1996) on how we would expect an agent whose preferences are constructed to act.
22
more concrete with age, being generated through life-course experience. The changeability of
the preferences and intentions of the young is well documented as is their frequent
inconsistency with subsequent behaviour. This is often interpreted as reflecting a lack of
realism in the fertility expectations of young people. In our view, the lack of realism pertains
to the survey practice of asking questions on fertility preferences and expectations of people
who are at an age when their desires and expectations are vague or unformed. Preferences
and intentions are unstable, we suggest, because they are constructed, and this also explains
their often weak correspondence with subsequent outcomes.
Our hypothesis is similar to the shaping hypothesis of Loomes et al (2003) in relation
to economic choices. In Loomes et al’s hypothesis, the primary influences are market prices,
but in the fertility case the influences would, of course, be different. The shaping hypothesis
is distinct from Plott’s (1996) discovered preference hypothesis, in which people learn by
trial and error what their true preferences are. In Plott’s account, the preferences ‘discovered’
are pre-existing and inherent, rather than constructed and influenced by experience. Under
the shaping hypothesis and in a radical version of our proposal, there are no true, underlying
fertility preferences at the earliest ages. Preferences are, rather, literally generated by
experience as time goes by. At later ages, therefore, people’s preferences in relation to
fertility depend importantly on their experience before the point of choice or survey
observation.
Nevertheless, the term “discovery” expresses our meaning well. People arrive,
through a sequence of stages, at a family size that they “discover” to be right for them, their
preferences having been shaped and developed by experience. Thus, fertility preferences
could be said to be discovered but in a sense different from Plott’s. In essence, through the
life course people happen on a family size that is satisfactory and that becomes their
preference, rather than that they uncover a latent preference that has always been there. These
ideas are similar to those of Simons (1974) who suggested that the findings on intentions and
fertility outcomes could be interpreted as indicating that “individuals discover, by reaching it,
the family size at which they wish to cease childbearing.” Our approach has much in
common also with Morgan’s schematic model (Morgan 1982: 332), with Johnson-Hanks
(2005) and with the theory of conjunctural action (Johnson-Hanks, Bachrach, Morgan et al.
2011, Morgan and Bachrach 2011) as well as with the cognitive-social model of Bachrach
and Morgan (2013).
23
Our proposal can be encapsulated by saying that preferred family size is a discovery
rather than a goal. But that does not imply that family formation always results in a perfect
match between preferences and outcome. People may discover, too late, that they would have
preferred either fewer or more children than they eventually have. Such disparities can arise
because of lack of information. A particularly interesting type of information deficit is that
people may not know, or may be mistaken about, what will make them happy in the future—
variously referred to in the psychological literature as miswanting or failure of affective
forecasting or of hedonic prediction (Kahneman 1994; Gilbert and Wilson 2000; Wilson and
Gilbert 2005). In the case of childbearing, a person may find that a disparity between
preference and actuality cannot be corrected, either because there is no culturally acceptable
solution to the problem of having had too many children, or that age, fecundity or other
constraints limit the achievement of a preferred fertility outcome. Dissatisfaction of this type
need not result from failure to reach a clearly held goal, but from the realisation that the
outcome of a series of decisions, taken one at a time, is, in the event, less preferred than a
hypothetical alternative.20
4.3. RESPONSES TO SURVEY QUESTIONS
One criterion by which preferences are judged to be constructed is that the responses of the
same individual will differ systematically according to how they are elicited or expressed
(Slovic 1995). Thus, how people report preferences in response to survey questions—stated
preferences—may differ from how they identify their preferences when acting to have or not
have a (further) child—effective preferences—because the constructive process differs in
each case. Effective preferences will no doubt influence stated preferences, but the latter may
reflect other factors also. It has long been known in the demographic literature that the
desired family size reported by individuals can shift, either up or down, in response to the
occurrence or not of a birth. Preferences to some extent follow fertility performance, as
distinct from determining it. Furthermore, people appear to reinterpret their past preferences
to fit their accumulated behaviour. In a sample of engaged couples first contacted in the early
1930s and followed up in 1953-54, Westoff et al (1957) found that actual family size in 1953-
54 was more strongly correlated with their recall in 1953-54 of what they wanted 20 years
earlier than with the preferences they had actually declared two decades before. That is, they
20 On post-decision surprise, see (Goitein,1984), Harrison and March (1984)
24
misremembered their previous preferences as more consistent with their current
circumstances than was actually the case: they were rewriting the history of their preferences
to correspond with their achieved family size.
For insight into how stated preferences may be constructed, we look to the political
science literature. Evidence that individuals’ political attitudes are unstable over time has
long been debated in political science. That literature has much to offer in re-interpreting the
extensive evidence of changeability of reproductive attitudes and expectations. Contrasting
positions in the debate on political attitudes are those of Converse (1964) and Achen (1975).
Converse’s view is that while some respondents have and report genuine opinions, large
numbers do not have distinctive views on many subjects, but give polite more or less random
answers to survey interview questions. Achen, on the other hand, contends that short-term
opinion change is only apparent and is largely due to measurement error. Between these
extremes is the position of Zaller and Feldman (1992) who argue that people do not have
clear-cut views about every topic but are ambivalent about a range of political issues. In
answering survey questions, respondents are, in this perspective, expressing neither
meaningless “nonattitudes” nor unambiguous true attitudes. Rather, people will often have
multiple considerations in mind on any specific topic, not all consistent with each other.
Survey responses will be influenced by multiple factors, including the ideas and
considerations uppermost in a respondent’s mind when answering a question (Zaller and
Feldman 1992: 580). In this scheme, there need not and may not be an underlying “true”
attitude, unlike the Converse and Achen views. Responses inconsistent at two time-points
could reflect exactly the same set of underlying considerations, but that some are more salient
on the first occasion than on the second.
Zaller and Feldman’s approach seems a potentially useful account of how fertility
intentions are constructed in an interview situation. Both the high prevalence of uncertainty
in fertility intentions, and its sensitivity to answer options available, fit well within this
perspective. Stated preferences and intentions may be influenced by a variety of factors, both
distant and recent in time. For example, young people with little or no experience might
adopt a kind of Bayesian perspective (Elgamal and Grether 1995), stating as a preference
what is essentially a prediction of their likely behaviour, based on the best evidence available
to them at the time. Two candidate sources of information, that might function as anchors
(Payne, Bettman and Johnson 1992, Wilson, Lindsey and Schooler 2000), are the size of their
25
own family of origin and the family sizes of which they are most aware, e.g. the distribution
of contemporary family sizes or the modal family size in their social group or in society at
large. Régnier-Loilier (2006) found that the sibship size of their family of origin was
positively associated with desired family size among childless people aged under 30, but was
not so associated among those who were already parents. Heiland et al (2008) also report a
stronger association between family background and fertility preferences at younger than at
older ages. There is evidence also that intended family size is less dispersed at younger than
at older ages (Liefbroer 2009, Ní Bhrolcháin and Beaujouan 2011). Answers to fertility
intentions or preference questions are sometimes thought of as partly normative. If this is so,
the explanation could be that respondents look to the modal experience of others as the most
reasonable prediction of their own future behaviour, rather than that they aspire to do as
others do (Hayford 2009). A further possibility is that fertility intentions, both effective and
stated, may be influenced by the effect known in psychology as “mere exposure” (Zajonc
2001): repeated exposure to a particular stimulus, such as the size of the family of origin or of
families in a person’s social milieu, could induce a preference. Stated intentions and
preferences may also be influenced by current or recent considerations that are much more
transient.
5. DISCUSSION
We have been prompted to raise questions about the underlying reality of reproductive goals
by the high prevalence of uncertain fertility intentions both in the UK and in other developed
societies, in recent decades. Broadly defined, uncertainty is near universal at the youngest
ages, in our GHS sample, but less common in older age groups. The level of uncertainty
appears to have been underestimated by survey questions hitherto. Uncertainty is not, we
propose, mainly due to faulty measurement or poorly motivated respondents. Rather, a lack
of certainty is an inherent feature of fertility preferences and intentions. It is, furthermore,
rational to be unsure about prospective childbearing. Support for these ideas comes from a
range of family planning studies showing considerable ambiguity surrounding pregnancy
intentions. In all, uncertainty in reproductive orientation appears to be both real and
widespread in developed societies.
This array of findings, combined with long standing evidence on the instability and
limited predictive validity of reproductive intentions and preferences, has led us to a new
26
hypothesis. Drawing on psychology and economics, we propose that fertility preferences can
be seen as constructed—that is, generated at the point at which they are elicited or expressed,
rather than residing fully-formed in memory. The hypothesis suggests that insofar as people
have a family size goal, it is recognised when reached, rather than a pre-existing target.
Women and couples are, in other words, primarily goal-achieving rather than goal-directed
(Mcfarland 1989). An extreme version of the idea says that ultimate reproductive goals either
do not exist or have only a weak reality, at any rate in very early adulthood, though
preferences along the way may be sharper and clearer.
Our constructive approach to fertility intentions and preferences is currently a
hypothesis. Modifications might well be in order. For example, people may differ in the
certainty and clarity of fertility preferences. Some young adults may be unclear about
whether they want to have children, while others may have a well-articulated preference
either for some children sometime, or never to have children, ever. Also, insofar as fertility
preferences are constructed, they may not be wholly so—some preferences may be inherent
and stable from an early age. Debate currently surrounds the issue in the literature on
preference construction (see Simonson 2008 and associated commentary). Finally, fertility
preferences encompass at least three distinct questions: desire for a pregnancy/child now,
desire for a (further) child ever, and desired number of children. A person could, for example,
definitely wish not to have a child in the very short term, be fairly certain that they would like
a child sometime, and be quite unsure about their preferred ultimate family size. Certainty
and the degree of construction may well vary between these distinct preferences.
5.1. TESTING THE HYPOTHESIS
The construction hypothesis is an alternative to the account which says that survey
respondents have true underlying preferences but these vary over time and differ from
ultimate outcomes due to e.g. measurement error, constraint, and changes of mind. To specify
fully how to test this hypothesis is well beyond the scope of the present paper. Our aim is
rather to present the ideas as an alternative perspective that can help to make sense of a
disparate set of often puzzling and unsatisfactory findings regarding fertility preferences and
intentions. Investigating the ideas empirically is likely to be challenging. While ample
evidence exists of genuine uncertainty in childbearing preferences and intentions,
demonstrating empirically that these are constructed is a demanding task. In the psychology
27
and economics literature, evidence of preference construction has, thus far, been based
largely on experiment. Non-experimental methods for identifying preference construction in
natural settings have not been described hitherto, nor has survey data been deployed for the
purpose.
Of the two types of preference distinguished earlier—those reported in surveys (stated
preferences) and those that govern family formation itself (effective preferences)—stated
preferences can be the subject of experimental investigation via e.g. studies of priming,
variation in question style and order, and contextual effects.21 Effective preferences, however,
are not amenable to experiment. As we have conceptualised them thus far, they are
unobserved, and possibly unobservable. They could be thought of as a latent variable, whose
existence and properties are probably best evaluated by developing and testing formal
models. Several research areas in other disciplines could be drawn on for leads in this respect.
Methods used in political science to evaluate the Converse thesis could be adapted to develop
and interpret models of stability/instability in fertility preferences (see e.g. Brody 1986, Hill
and Kriesi 2001). Economic models of the subjective probabilities of significant personal
events such as survival are a further potential resource (Manski 2004, Hudomiet and Willis
2012). Finally, less formal empirical approaches could serve to document the essential
hypothesis, particularly the process by which preferences are constructed, which has received
less attention than has evidence of construction itself (Simonson 2008). Small-scale
qualitative investigation (Schaeffer and Thomson 1992) and ethnographic work (Johnson-
Hanks 2005) can be highly informative as to process. Johnson-Hanks suggests that the
reproductive attitudes of educated young women in Cameroon are characterised by what she
terms “judicious opportunism”, young women having a strong sense of contingency in their
future life course. Two lines of inquiry are suggested by these findings. First, follow up
studies could investigate how such scepticism is transformed as the life course progresses.
Second, the ethnographic approach might be usefully extended to developed country
contexts, to reach for meaning and mechanism behind the standard responses to large scale
survey questions.
21 See e.g. Feldman and Lynch (1988), Zaller and Feldman (1992), Wikman (2006, 2007), and Sturgis and
Smith (2010).
28
5.2. CONCLUDING COMMENTS
If reproductive preferences are constructed to any significant extent, several unexplained
aspects of fertility expectations become more explicable: why they are so unstable, why
sizeable discrepancies are found between intention and outcome, and why stated fertility
preferences and intentions, in the aggregate, reflect current rather than prospective fertility
conditions (Westoff and Ryder 1977, Lee 1980). The relevance of the ideas extends,
however, beyond these points. The hypothesis potentially opens a route to uncovering
hitherto poorly understood aspects of the process of family formation itself. The constructive
approach fits well with the reality of family building as a sequence of decisions (Namboodiri
1972, 1983, Udry 1983) and chimes well also with suggestions in the earlier literature that
couples have no need of a long-run view: all they need is to decide whether they want to have
or avoid a pregnancy in the very short term (Ryder 1973: 502, 1976: 289). Nevertheless, each
of these authors held that preferences are revised over time, rather than, as here, generated
over time, in the light of experience. Pursuit of the constructive preferences approach could
ultimately answer the question as to what kind of target people and couples have in relation
to childbearing: a fixed target, a moving one, or, perhaps, no target at all.
While we have drawn mainly on ideas from psychology, economics and political
science, our approach also reflects recent themes in sociology. In the theoretical arena,
Smelser’s (1998) focus on ambivalence and the difficulties it poses for rational choice models
and Sewell’s (1992) theory of structure, positing the interdependence of structure and human
agency, reinforce our focus on the constructive nature of preferences. Empirical sociological
work on foresight and planning in early adulthood gives a broader life-course context for the
ideas (Anderson, Bechhofer, Jamieson et al. 2002, Anderson, Bechhofer, Mccrone et al.
2005). The constructive approach allows of determinants of individual preferences at
multiple levels (Smith 1989, Hechter and Kanazawa 1997; Testa and Grilli 2006). The
constructive view can accommodate a range of influences on both stated and effective
preferences and intentions--from institutional, structural, historical, cultural and economic
forces, and cohort and period effects, through to social network and idiosyncratic personal
factors. Thus, as well as accounting for the many empirical anomalies surrounding fertility
preferences and intentions, the constructive process provides a mechanism linking the
aggregate with the individual level. The constructive approach to fertility intentions and
preferences has much potential explanatory power and merits serious empirical investigation.
29
REFERENCES
Achen,C.(1975).Masspoliticalattitudesandthesurveyresponse.AmericanPoliticalScience
Review,69,12181231.
Agadjanian,V.(2005).Fraughtwithambivalence:Reproductiveintentionsandcontraceptivechoices
inasubSaharanfertilitytransition.PopulationResearchandPolicyReview,24(6),617645.
Ajzen,I.(1991).Thetheoryofplannedbehavior.OrganizationalBehaviorandHumanDecision
Processes,50(2),179211.
Anderson,M.,Bechhofer,F.,Jamieson,L.,Mccrone,D.,Li,Y.andStewart,R.(2002).Confidence
amiduncertainty:Ambitionsandplansinasampleofyoungadults.SociologicalResearch
Online,6(4).
Anderson,M.,Bechhofer,F.,Mccrone,D.,Jamieson,L.,Li,Y.andStewart,R.(2005).Timespans
andplansamongyoungadults.Sociology,39(1),139155.
Bachrach,C.A.andMorgan,S.P.(2013).Acognitive–socialmodeloffertilityintentions.Population
andDevelopmentReview,39(3),459485.
Bachrach,C.A.andNewcomer,S.(1999).Intendedpregnanciesandunintendedpregnancies:
Distinctcategoriesoroppositeendsofacontinuum?FamilyPlanningPerspectives,31(5),
251252.
Bankole,A.andWestoff,C.F.(1998).Theconsistencyandvalidityofreproductiveattitudes:
EvidencefromMorocco.JournalofBiosocialScience,30(4),439455.
Barrett,G.,Smith,S.C.andWellings,K.(2004).Conceptualisation,development,andevaluationofa
measureofunplannedpregnancy.JournalofEpidemiologyandCommunityHealth,58(5),
426433.
Barrett,G.andWellings,K.(2000).Understandingpregnancyintentions:Aprobleminevidence
everywhere.FamilyPlanningPerspectives,32(4),194.
Barrett,G.andWellings,K.(2002).Whatisa'planned'pregnancy?EmpiricaldatafromaBritish
study.SocialScienceandMedicine,55(4),545557.
Beaujouan,Éva,Brown,J.J.andBhrolcháin,M.(2011)ReweightingtheGeneralHousehold
Survey,19792007.PopulationTrends,145:11945.
Berrington,A.(2004).Perpetualpostponers?Women’s,men’sandcouple’sfertilityintentionsand
subsequentfertilitybehaviour.PopulationTrends,117,919.
Bettman,J.R.,Luce,M.F.andPayne,J.W.(1998).Constructiveconsumerchoiceprocesses.Journal
ofConsumerResearch,25(3),187217.
Bongaarts,J.(2002).Theendofthefertilitytransitioninthedevelopedworld.Populationand
DevelopmentReview28(3),419443.
Brody,C.J.(1986).Thingsarerarelyblackandwhite:admittinggrayintotheconversemodelof
attitudestability.AmericanJournalofSociology,92(3),657677.
Brown,S.S.andEisenberg,L.,Eds.(1995).“TheBestIntentions:Unintendedpregnancyandthe
wellbeingofchildrenandfamilies.”Washington,D.C.,NationalAcademiesPress.
Bruni,L.andSugden,R.(2007).Theroadnottaken:Howpsychologywasremovedfromeconomics,
andhowitmightbebroughtback.EconomicJournal,117(516),146173.
Bumpass,L.andWestoff,C.F.(1969).Thepredictionofcompletedfertility.Demography,6(4),445
454.
Bumpass,L.andWestoff,C.F.(1970).“TheLaterYearsofChildbearing.”Princeton:Princeton
UniversityPress.
Carlsson,F.(2010).Designofstatedpreferencesurveys:Istheremoretolearnfrombehavioral
economics?Environmental&ResourceEconomics,46(2),167177.
Cartwright,A.(1970).“ParentsandFamilyPlanningServices.”London:RoutledgeKeganPaul.
Cartwright,A.(1976).“HowManyChildren?”London:Routledge&KeganPaul.
Cartwright,A.(1988).Unintendedpregnanciesthatleadtobabies.SocialScienceandMedicine,
27(3),249254.
30
Cartwright,A.andWilkins,W.(1976).Changesinfamilybuildingplans:Afollowupstudyto'How
manychildren?'StudiesonMedicalandPopulationSubjects,OPCSNo.33.
Casterline,J.B.andElZeini,L.O.(2007).Theestimationofunwantedfertility.Demography,44(4),
729745.
Casterline,J.B.andSinding,S.W.(2000).Unmetneedforfamilyplanningindevelopingcountries
andimplicationsforpopulationpolicy.PopulationandDevelopmentReview,26(4),691723.
Converse,P.E.(1964).“Thenatureofbeliefsystemsinmasspublics.”In:Apter,D.(ed.)Ideologyand
Discontent.NewYork:FreePress,206261.
Converse,P.E.(1974).NonattitudesandAmericanpublicopinion:Comment:Thestatusof
nonattitudes.AmericanPoliticalScienceReview,68(2),650660.
Coombs,L.C.(1974).Measurementoffamilysizepreferencesandsubsequentfertility.Demography,
11(4),587611.
Coombs,L.C.(1978).Howmanychildrendocouplesreallywant.FamilyPlanningPerspectives,10
(5),303308.
Coombs,L.C.andFreedman,R.(1979).Somerootsofpreference‐roles,activitiesandfamilial
values.Demography,16(3),359376.
Demeny,P.(1988).Socialscienceandpopulationpolicy.PopulationandDevelopmentReview,14
(3),451479.
DemographicandHealthSurveys(2011).“GuidelinesfortheMEASUREDHSPhaseIIIMainSurvey
Report.”Calverton,Maryland,USA:ICFInternationalInc.
DixonMueller,R.andGermain,A.(1992).Stalkingtheelusive'unmetneed'forfamilyplanning.
StudiesinFamilyPlanning,23(5),330335.
Elgamal,M.A.andGrether,D.M.(1995).ArepeopleBayesian?Uncoveringbehavioralstrategies.
JournaloftheAmericanStatisticalAssociation,90(432),11371145.
Esacove,A.(2008).Makingsenseofsex:Rethinkingintentionality.CultureHealth&Sexuality,10(4),
377390.
Feldman,J.M.andLynch,J.G.(1988).Selfgeneratedvalidityandothereffectsofmeasurementon
belief,attitude,intention,andbehavior.JournalofAppliedPsychology,73(3),421435.
Finer,L.B.andHenshaw,S.K.(2006).DisparitiesinratesofunintendedpregnancyintheUnited
States,1994and2001.PerspectivesonSexualandReproductiveHealth,38(2),9096.
Fischhoff,B.(1991).Valueelicitation‐isthereanythinginthere?AmericanPsychologist,46(8),835
847.
Fischhoff,B.(2006).“Constructingpreferencesfromlabilevalues.”In:Lichtenstein,S.andSlovic,P.
(eds.)TheConstructionofPreference.Cambridge:CambridgeUniversityPress,653667.
Freedman,R.,Coombs,L.C.andBumpass,L.(1965).Stabilityandchangeinexpectationsabout
familysize:Alongitudinalstudy.Demography,2,250275.
Freedman,R.,Freedman,D.S.andThornton,A.D.(1980).Changesinfertilityexpectationsand
preferencesbetween1962and1977‐theirrelationtofinalparity.Demography,17(4),365
378.
Freedman,R.,Whelpton,P.K.andCampbell,A.(1959).“FamilyPlanning,SterilityandPopulation
Growth.”NewYork:McGrawHill.
Gerber,A.,Pennylegion,M.andSpice,C.(2002).“'Ifithappens,ithappens'.Aqualitative
assessmentofunintendedpregnancyinSouthKingCounty.”Seattle:PublicHealthSeattle&
KingCounty.
Gilbert,D.T.andWilson,T.D.(2000).“Miswanting:Someproblemsintheforecastingoffuture
affectivestates.”In:Forgas,J.P.(ed.)FeelingandThinking.Theroleofaffectinsocial
cognition.Cambridge:CambridgeUniversityPress,178197.
Goitein,B.(1984).Thedangerofdisappearingpostdecisionsurprise:CommentonHarrisonand
March,"decisionmakingandpostdecisionsurprises".AdministrativeScienceQuarterly,29
(3),410413.
31
Greil,A.L.andMcquillan,J.(2010)."Trying"times:Medicalization,intent,andambiguityinthe
definitionofinfertility.MedicalAnthropologyQuarterly,24(2),137156.
Harrison,J.R.andMarch,J.G.(1984).Decisionmakingandpostdecisionsurprises.Administrative
ScienceQuarterly,29,2642.
Hauser,P.M.(1967).Review:'FamilyPlanningandPopulationPrograms':Abookreviewarticle.
Demography,4(1),397414.
Hayford,S.R.(2009).Theevolutionoffertilityexpectationsoverthelifecourse.Demography,46(4),
765783.
Hayford,S.R.andMorgan,S.P.(2008).ReligiosityandfertilityintheUnitedStates:Theroleof
fertilityintentions.SocialForces,86(3),11631188.
Hechter,M.andKanazawa,S.(1997).Sociologicalrationalchoicetheory.AnnualReviewof
Sociology,23,191214.
Heckhausen,J.,Wrosch,C.andFleeson,W.(2001).Developmentalregulationbeforeandaftera
developmentaldeadline:Thesamplecaseof"biologicalclock"forchildbearing.Psychology
andAging,16(3),400413.
Heiland,F.,Prskawetz,A.andSanderson,W.C.(2008).Areindividuals'desiredfamilysizesstable?
EvidencefromwestGermanpaneldata.EuropeanJournalofPopulationRevueEuropeenne
deDemographie,24(2),129156.
Higgins,J.A.,Hirsch,J.S.andTrussell,J.(2008).Pleasure,prophylaxisandprocreation:Aqualitative
analysisofintermittentcontraceptiveuseandunintendedpregnancy.PerspectivesonSexual
andReproductiveHealth,40(3),130137.
Hill,J.L.andKriesi,H.(2001).AnextensionandtestofConverse's"blackandwhite"modelof
responsestability.AmericanPoliticalScienceReview,95(2),397413.
Hudomiet,P.andWillis,R.J.(2012).“Estimatingsecondorderprobabilitybeliefsfromsubjective
survivaldata.”Cambridge,Mass.:NationalBureauofEconomicResearch.
Iacovou,M.andTavares,L.(2011).Yearning,learningandconceding:Reasonsmenandwomen
changetheirchildbearingintentions.PopulationandDevelopmentReview,37(1),87123.
Jansen,M.andLiefbroer,A.C.(2006).Couples'attitudes,childbirth,andthedivisionoflabor.
JournalofFamilyIssues,27(11),14871511.
JohnsonHanks,J.(2005).Whenthefuturedecides‐uncertaintyandintentionalactionin
contemporaryCameroon.CurrentAnthropology,46(3),363385.
JohnsonHanks,J.,Bachrach,C.,Morgan,S.P.andKohler,H.P.(2011).“Understandingfamily
changeandvariation:Towardatheoryofconjuncturalaction.”NewYork:Springer.
Joyce,T.,Kaestner,R.andKorenman,S.(2000).Thestabilityofpregnancyintentionsand
pregnancyrelatedmaternalbehaviors.MaternalandChildHealthJournal,4(3),171178.
Kahneman,D.(1994).Newchallengestotherationalityassumption.JournalofInstitutionaland
TheoreticalEconomics,150,1836.
Kahneman,D.(1996).“Comment.”In:Arrow,K.,Colombatto,E.,Perlaman,M.andSchmidt,C.
(eds.)TheRationalFoundationsofEconomicBehavior.LondonandNewYork:Macmillan/St
Martin'sPress,225250.
Kahneman,D.(1996).“Newchallengestotherationalityassumption.”In:Arrow,K.,Colombano,E.,
Perlman,M.andSchmidt,C.(ed.).TheRationalFoundationsofEconomicBehaviour,London:
Macmillan,203224.
Kaufmann,R.,Morris,L.andSpitz,A.M.(1997).Comparisonoftwoquestionsequencesfor
assessingpregnancyintentions.AmericanJournalofEpidemiology,145(9),810816.
Kendall,C.,AfableMunsuz,A.,Speizer,I.,Averya,A.,Schmidt,N.andSantelli,J.(2005).
UnderstandingpregnancyinapopulationofinnercitywomeninNewOrleans—resultsof
qualitativeresearch.SocialScienceandMedicine,60,297311.
Kiser,C.V.(1967).Review:TheGrowthofAmericanFamiliesstudies:Anassessmentofsignificance.
Demography,4(1),388396.
32
Klerman,L.V.(2000).Theintendednessofpregnancy:Aconceptintransition.MaternalandChild
HealthJournal4(3):155162.
Kodzi,I.A.,Casterline,J.B.andAglobitse,P.(2010).Thetimedynamicsofindividualfertility
preferencesamongruralGhanaianwomen.StudiesinFamilyPlanning,41(1),4554.
Krosnick,J.A.(1999).Surveyresearch.AnnualReviewofPsychology,50,537567.
Lee,R.D.(1980).Aimingatamovingtarget:Periodfertilityandchangingreproductivegoals.
PopulationStudies,30(2),205226.
Lee,R.D.(1981).“Amodelforforecastingfertilityfrombirthexpectationsdata.”In:Hendershot,
G.E.andPlacek,P.J.(eds.)PredictingFertility.Demographicstudiesofbirthexpectations.
Lexington,Mass.:LexingtonBooks,7599.
Leridon,H.(2008).Anewestimateofpermanentsterilitybyage:Sterilitydefinedastheinabilityto
conceive.PopulationStudies,62(1),1524.
Lichtenstein,S.andSlovic,P.(2006).“Theconstructionofpreference:Anoverview.”In:
Lichtenstein,S.andSlovic,P.(eds.)TheConstructionofPreference,Cambridge:Cambridge
UniversityPress,140.
Liefbroer,A.C.(2009).Changesinfamilysizeintentionsacrossyoungadulthood:Alifecourse
perspective.EuropeanJournalofPopulationRevueEuropeennedeDemographie,25(4),
363386.
Lifflander,A.,Gaydos,L.M.D.andRowlandHogue,C.J.(2007).Circumstancesofpregnancy:Low
incomewomeninGeorgiadescribethedifferencebetweenplannedandunplanned
pregnancies.MaternalandChildHealthJournal,11,8189.
Loomes,G.,Starmer,C.andSugden,R.(2003).Doanomaliesdisappearinrepeatedmarkets?
EconomicJournal,113(486),C153C166.
Luker,K.C.(1999).Areminderthathumanbehaviorfrequentlyrefusestoconformtomodels
createdbyresearchers.FamilyPlanningPerspectives,31(5),248249.
Manski,C.F.(2004).Measuringexpectations.Econometrica,72(5),13291376.
Mcfarland,D.(1989).“Goals,nogoalsandowngoals.”In:Montefiore,A.andNoble,D.(eds.)Goals,
NoGoalsandOwnGoals.London:UnwinHyman:3957.
Mcquillan,J.,Greil,A.L.andShreffler,K.M.(2010).Pregnancyintentionsamongwomenwhodonot
try:Focusingonwomenwhoareokayeitherway.MaternalandChildHealthJournal.15(2),
178187.
Miller,W.B.andPasta,D.J.(1995).Behavioralintentions‐whichonespredictfertilitybehaviorin
marriedcouples.JournalofAppliedSocialPsychology,25(6),530555.
Monnier,A.(1989).Fertilityintentionsandactualbehaviour.Alongitudinalstudy:1974,1979.
Population44(1):237259.
Moos,M.K.,Petersen,R.,Meadows,K.,Melvin,C.L.andSpitz,A.M.(1997).Pregnantwomen's
perspectivesonintendednessofpregnancy.Women'sHealthIssues7(6):385392.
Morgan,S.P.(1981).Intentionanduncertaintyatlaterstagesofchildbearing:TheUnitedStates
1965and1970.Demography,18(3),267285.
Morgan,S.P.(1982).Parityspecificfertilityintentionsanduncertainty‐theUnitedStates,1970to
1976.Demography,19(3),315334.
Morgan,S.P.(1985).Individualandcoupleintentionsformorechildren‐aresearchnote.
Demography,22(1),125132.
Morgan,S.P.(2001).“Shouldfertilityintentionsinformfertilityforecasts?TheDirectionofFertilityin
theUnitedStates.”Alexandria,Virginia,USCensusBureau.
Morgan,S.P.andBachrach,C.A.(2011).Isthetheoryofplannedbehavioranappropriatemodelfor
humanfertility?ViennaYearbookofPopulationResearch,9,1118.
Morgan,S.P.andRackin,H.(2010).Thecorrespondencebetweenfertilityintentionsandbehaviorin
theUnitedStates.PopulationandDevelopmentReview,36(1),91118.
Musick,K.,England,P.,Edgington,S.andKangas,N.(2009).Educationdifferencesinintendedand
unintendedfertility.SocialForces,88(2),543572.
33
Namboodiri,N.K.(1972).Someobservationsoneconomicframeworkforfertilityanalysis.
PopulationStudiesAJournalofDemography,26(2),185206.
Namboodiri,N.K.(1983).“Sequentialdecisionmakingandthelifecourse.”In:Bulatao,R.andLee,
R.D.(eds.)DeterminantsofFertilityinDevelopingCountries.London:AcademicPress,444
472.
Bhrolcháin,M.,andBeaujouan,É.(2011)."UncertaintyinfertilityIntentionsinBritain,1979
2007."ViennaYearbookofPopulationResearch,9,10134.
Bhrolcháin,Máire,andBeaujouan,É.forthcoming."Theprevalenceofuncertainfertility
intentionsindevelopedsocieties."
Bhrolcháin,M.,Beaujouan,É.andBerrington,A.(2010)."Stabilityandchangeinfertility
intentionsinBritain,19912007."PopulationTrends,141,123.
Bhrolcháin,M.,Beaujouan,É.andMurphy,M.(2011)."Sourcesoferrorinreportedchildlessness
inacontinuousBritishhouseholdsurvey."PopulationStudies,65,30518.
Noack,T.andOstby,L.(1985).Fertilityexpectations:Ashortcutordeadendinpredictingfertility?
ScandinavianPopulationStudies,7,4859.
Noack,T.andOstby,L.(2002).Freetochoose‐butunabletosticktoit?Norwegianfertility
expectationsandsubsequentbehaviourinthefollowing20years.DynamicsofFertilityand
PartnershipinEurope:InsightsandLessonsfromComparativeResearch,2,103116.
O'Connell,M.andRogers,C.C.(1983).AssessingcohortbirthexpectationsdatafromtheCurrent
PopulationSurvey,19711981.Demography,20(3),369384.
Payne,J.W.,Bettman,J.R.andJohnson,E.J.(1992).Behavioraldecisionresearch‐aconstructive
processingperspective.AnnualReviewofPsychology,43,87131.
Payne,J.W.,Bettman,J.R.andSchkade,D.A.(1999).Measuringconstructedpreferences:Towards
abuildingcode.JournalofRiskandUncertainty,19(13),243270.
Petersen,R.,Gazmararian,J.A.,Clark,K.A.andGreen,D.C.(2001).Howcontraceptiveusepatterns
differbypregnancyintention:Implicationsforcounseling.Women'sHealthIssues,11(5),
427‐435.
Petersen,R.andMoos,M.K.(1997).Definingandmeasuringunintendedpregnancy:Issuesand
concerns.Women'sHealthIssues,7(4),234‐240.
Philipov,D.(2009).Theeffectofcompetingintentionsandbehaviouronshorttermchildbearing
intentionsandsubsequentchildbearing.EuropeanJournalofPopulationRevueEuropeenne
deDemographie,25(4),525548.
Plott,C.R.(1996).“Rationalindividualbehaviourinmarketsandsocialchoiceprocesses:The
discoveredpreferencehypothesis.”In:Arrow,K.J.,Colombatto,E.,Perlman,M.andSchmidt,
C.(eds.)TheRationalFoundationsofEconomicBehaviour,London:Macmillan,225250.
Poole,V.L.,Flowers,J.S.,Goldenberg,R.L.,Cliver,S.P.andMcneal,S.(2000).Changesin
intendednessduringpregnancyinahighriskmultiparouspopulation.MaternalandChild
HealthJournal,4(3),179182.
QuesnelVallee,A.andMorgan,S.P.(2003).Missingthetarget?Correspondenceoffertility
intentionsandbehaviorintheUS.PopulationResearchandPolicyReview,22(56),497525.
RégnierLoilier,A.(2006).Influenceofownsibshipsizeonnumberofchildrendesiredatvarious
timesoflife.ThecaseofFrance.Population,61(3),193223.
Riley,A.P.,Hermalin,A.I.andRoseroBixby,L.(1993).Anewlookatthedeterminantsof
nonnumericresponsetodesiredfamilysize‐thecaseofCostaRica.Demography,30(2),
159174.
Rindfuss,R.R.,Morgan,S.P.andSwicegood,G.(1988).“FirstBirthsinAmerica.”London:University
ofCaliforniaPress.
Rosina,A.andTesta,M.R.(2009).Couples'firstchildintentionsanddisagreement:Ananalysisof
theItaliancase.EuropeanJournalofPopulationRevueEuropeennedeDemographie,25(4),
487502.
Ryder,N.B.(1973).AcritiqueoftheNationalFertilityStudy.Demography,10(4),495506.
34
Ryder,N.B.(1976).Thespecificationoffertilityplanningstatus.FamilyPlanningPerspectives,8(6),
283289.
Ryder,N.B.(1979).Consistencyofreportingfertilityplanningstatus.StudiesinFamilyPlanning,10
(4),115128.
Ryder,N.B.(1981).“Changesinparityorientationfrom1970to1975.”In:Hendershot,G.E.and
Placek,P.J.(eds.)PredictingFertility.Demographicstudiesofbirthexpectations.Lexington,
Mass.:LexingtonBooks,101127.
Ryder,N.B.(1984).“Expectationsandprogressiveanalysisinfertilityprediction.”In:UnitedNations
(eds.)Populationprojections:MethodologyoftheUnitedNations.NewYork:UnitedNations,
3338.
Ryder,N.B.(1985).Thestructureofpregnancyintervalsbyplanningstatus.PopulationStudies,39
(2),193211.
Ryder,N.B.(1986).ReviewofJohnClelandandJohnHobcraft(eds)'ReproductiveChangein
DevelopingCountries'.PopulationandDevelopmentReview,12(2),341359.
Ryder,N.B.andWestoff,C.F.(1971).“ReproductionintheUnitedStates1965.”Princeton:
PrincetonUniversityPress.
Sable,M.R.andLibbus,M.K.(2000).Pregnancyintentionandpregnancyhappiness:Arethey
different?MaternalandChildHealthJournal,191196.
Santelli,J.S.,Lindberg,L.D.,Orr,M.G.,Finer,L.B.andSpeizer,I.(2009).Towardamultidimensional
measureofpregnancyintentions:EvidencefromtheUnitedStates.StudiesinFamily
Planning,40(2),87100.
Santelli,J.S.,Rochat,R.,HatfieldTimajchy,K.,Gilbert,B.C.,Curtis,K.,Cabral,R.,Hirsch,J.S.,
Schieve,L.andUnintendedPregnancyWorkingGroup(2003).Themeasurementand
meaningofunintendedpregnancy.PerspectivesonSexualandReproductiveHealth,35(2),
94101.
Schaeffer,N.C.andPresser,S.(2003).Thescienceofaskingquestions.AnnualReviewofSociology,
29(1),6588.
Schaeffer,N.C.andThomson,E.(1992).Thediscoveryofgroundeduncertainty:Developing
standardizedquestionsaboutstrengthoffertilitymotivation.SociologicalMethodology,22,
37‐82.
Schoen,R.,Astone,N.M.,Kim,Y.J.,Nathanson,C.A.andFields,J.M.(1999).Dofertilityintentions
affectfertilitybehavior?JournalofMarriageandtheFamily,61(3),790799.
Sewell,W.H.(1992).Atheoryofstructure:Duality,agency,andtransformation.AmericanJournalof
Sociology,98,129.
Simons,J.(1974).ReviewofL.BumpassandC.Westoff"TheLaterYearsofChildbearing".Population
Studies,28,348350.
Simons,J.(1978).“Illusionsaboutattitudes.”In:CouncilofEurope(ed.).PopulationDeclinein
Europe.Implicationsofadecliningorstationarypopulation.London:EdwardArnold,197
214.
Simonson,I.(2008).WillIlikea"medium"pillow?Anotherlookatconstructedandinherent
preferences.JournalofConsumerPsychology,18(3),155169.
Singh,S.,Sedgh,G.andHussain,R.(2010).Unintendedpregnancy:Worldwidelevels,trends,and
outcomes.StudiesinFamilyPlanning,41(4),241250.
Slovic,P.(1995).Theconstructionofpreference.AmericanPsychologist,50(5),364371.
Smallwood,S.andJefferies,J.(2003).FamilybuildingintentionsinEnglandandWales:Trends,
outcomesandinterpretations.PopulationTrends,112,1528.
Smelser,N.J.(1998).Therationalandtheambivalentinthesocialsciences:1997presidential
address.AmericanSociologicalReview,63(1),116.
Smith,H.L.(1989).Integratingtheoryandresearchontheinstitutionaldeterminantsoffertility.
Demography,26,171184.
35
Sobotka,T.(2009).SubreplacementfertilityintentionsinAustria.EuropeanJournalofPopulation
RevueEuropeennedeDemographie,25(4),387412.
Speder,Z.andKapitany,B.(2009).Howaretimedependentchildbearingintentionsrealized?
Realization,postponement,abandonment,bringingforward.EuropeanJournalof
PopulationRevueEuropeennedeDemographie,25(4),503523.
Stanford,J.B.,Hobbs,R.,Jameson,P.,Dewitt,M.J.andFischer,R.C.(2000).Definingdimensionsof
pregnancyintendedness.MaternalandChildHealthJournal,4(3),183189.
Sturgis,P.andSmith,P.(2010).Fictitiousissuesrevisited:Politicalinterest,knowledgeandthe
generationofnonattitudes.PoliticalStudies,58(1),6684.
Testa,M.R.(2007).ChildbearingpreferencesandfamilyissuesinEurope:Evidencefromthe
Eurobarometer2006.ViennaYearbookofPopulationResearch,2007,357379.
Testa,M.R.andGrilli,L.(2006).Theinfluenceofchildbearingregionalcontextsonidealfamilysize
inEurope:Amultilevelanalysis.Population,61(12),107137.
Testa,M.R.andToulemon,L.(2006).FamilyformationinFrance:Individualpreferencesand
subsequentoutcomes.ViennaYearbookofPopulationResearch,2006,4175.
Thomson,E.(1997).Couplechildbearingdesires,intentions,andbirths.Demography,34(3),343
354.
Thomson,E.andBrandreth,Y.(1995).Measuringfertilitydemand.Demography,32(1),8196.
Thomson,E.andHoem,J.M.(1998).CouplechildbearingplansandbirthsinSweden.Demography,
35(3),315322.
Thomson,E.,Mcdonald,E.andBumpass,L.L.(1990).Fertilitydesiresandfertility‐hers,his,and
theirs.Demography,27(4),579588.
Toulemon,L.andTesta,M.R.(2005)Fertilityintentionsandactualfertility:Acomplexrelationship.
PopulationandSocieties,415.
Trussell,J.,Vaughan,B.andStanford,J.(1999).Areallcontraceptivefailuresunintended
pregnancies?Evidencefromthe1995NationalSurveyofFamilyGrowth.FamilyPlanning
Perspectives,31(5),246247,260.
Tversky,A.andThaler,R.H.(1990).Anomalies:Preferencereversals.JournalofEconomic
Perspectives,4(2),201211.
Udry,J.R.(1983).Docouplesmakefertilityplansonebirthatatime.Demography,20(2),117128.
VanHoorn,W.andKeilman,N.(1997).Birthexpectationsandtheiruseinfertilityforecasting.
EurostatWorkingPaperE4/19974,Eurostat.
Werner,B.(1986).Familybuildingintentionsofdifferentgenerationsofwomen:Resultsfromthe
GeneralHouseholdSurvey197983.PopulationTrends,44,1723.
Westoff,C.F.(1988).Thepotentialdemandforfamilyplanning:Anewmeasureofunmetneedand
estimatesforfiveLatinAmericancountries.InternationalFamilyPlanningPerspectives,14
(2),4553.
Westoff,C.F.,Mishler,E.G.andKelly,E.L.(1957).Preferencesinsizeoffamilyandeventualfertility
twentyyearsafter.TheAmericanJournalofSociology,62(5),491497.
Westoff,C.F.andRyder,N.B.(1977).“TheContraceptiveRevolution.”Princeton:Princetion
UniversityPress.
Westoff,C.F.andRyder,N.B.(1977).Thepredictivevalidityofreproductiveintentions.
Demography,14(4),431453.
Whelpton,P.K.,Campbell,A.andPatterson,J.(1966).“FertilityandFamilyPlanningintheUnited
States.”PrincetonUniversityPress.
Wikman,A.(2006).Reliability,validityandtruevaluesinsurveys.SocialIndicatorsResearch,78(1),
85110.
Wikman,A.(2007).Contexteffectsasanillustrationofresponseuncertainty‐acautionarytale.
SocialIndicatorsResearch,84(1),2738.
Williams,R.A.andThomson,E.(1985).Canspousesbetrusted‐alookathusbandwifeproxy
reports.Demography,22(1),115123.
36
Wilson,T.D.andGilbert,D.T.(2005).Affective