THE ROLE OF PARENTAL IDENTITY ON REACTION AND ADAPTATION TO THE
BIRTH OF A CHILD
By
Stevie C. Y. Yap

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Psychology – Doctor of Philosophy
2017

ABSTRACT
THE ROLE OF PARENTAL IDENTITY ON REACTION AND ADAPTATION TO THE
BIRTH OF A CHILD
By
Stevie C. Y. Yap
Parental identity is a conceptually important psychological construct, yet past work evaluating
this construct is scant and its causes, its correlates, and its trajectory over time are poorly
understood. Further, little is known about its consequences on well-being and the role parental
identity plays in how people respond to the birth of a child. The present research is the first to
provide a comprehensive look at the nature of parental identity, its correlates and its implications
on reactions to becoming a parent. This research is also the first to evaluate the trajectory of
parental identity over the life course and its pattern of change in response to the birth of a child.
Nationally representative panel studies from German, British, and Australian households were
used to examine these questions – utilizing data from over 55,000 respondents collected over 30
years. Results show that parental identity is associated with several important demographic
characteristics and outcomes, and that these associations are generally consistent across three
distinct national samples. Parental identity shows a consistent trajectory across the lifespan,
appearing to rise gradually from adolescence into early adulthood, then leveling off as
individuals move into their 30s and remaining relatively stable into older age. Childbirth is also
associated with large changes in parental identity, and these changes persist for many years
following birth. Finally, the results of this research indicates that parental identity is positively
associated with increased SWB in the years before birth, and is associated with an added, long
term boost in SWB following childbirth in Germany, Great Britain and Australia.

For my children.
May knowledge light your way.

iii

ACKNOWLEDGEMENTS

I would like to thank my committee members, Drs. Brent Donnellan, Emily Durbin, Isis Settles
and Richard Lucas for their mentorship and guidance during my time at Michigan State
University. I would also like to thank the Social Sciences and Humanities Research Council of
Canada for financially supporting this research. Finally, I would like to thank my spouse, Ivana
Anusic, and my children, Oliver, Charlie, and Avery for their never ending support and patience.
I owe much of this accomplishment to them.

iv

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................................ vii
LIST OF FIGURES ....................................................................................................................... ix
INTRODUCTION .......................................................................................................................... 1
Life Events and Subjective Well-Being ...................................................................................... 2
Parenthood and Subjective Well-Being ...................................................................................... 7
Variability in Reactions to Parenthood ....................................................................................... 9
Parental Identity and SWB........................................................................................................ 12
The Current Studies .................................................................................................................. 17
METHOD OVERVIEW ............................................................................................................... 21
Overview of Studies .................................................................................................................. 21
Overall Sample Characteristics ............................................................................................. 21
STUDY 1 ...................................................................................................................................... 23
Method ...................................................................................................................................... 23
Sample Selection ................................................................................................................... 23
Measures ............................................................................................................................... 23
Parental identity ................................................................................................................ 23
Demographic and other variables ..................................................................................... 25
Analytic Approach ................................................................................................................ 25
Modeling normative changes ............................................................................................ 31
Summary of Study 1 ............................................................................................................. 33
Results ....................................................................................................................................... 33
Sample Characteristics .......................................................................................................... 33
GSOEP .............................................................................................................................. 33
BHPS................................................................................................................................. 34
HILDA .............................................................................................................................. 34
Correlates of Parental Identity .............................................................................................. 34
Gender ............................................................................................................................... 34
Marital status ..................................................................................................................... 35
Education .......................................................................................................................... 40
Income............................................................................................................................... 44
Number of children ........................................................................................................... 45
Attitudes and values .......................................................................................................... 48
Personality......................................................................................................................... 51
Life satisfaction ................................................................................................................. 51
Parental Identity and its Pattern of Change across the Lifespan ........................................... 53
Parental Identity and Response to the Birth of a Child ......................................................... 58
Gender as a moderator ...................................................................................................... 62
Age as a moderator ........................................................................................................... 62

v

Marital status as a moderator ............................................................................................ 63
Income as a moderator ...................................................................................................... 63
Parental identity and normative change ............................................................................ 64
STUDY 2 ...................................................................................................................................... 67
Method ...................................................................................................................................... 67
Sample Selection ................................................................................................................... 67
Measures ............................................................................................................................... 67
Life satisfaction ................................................................................................................. 67
Parental identity ................................................................................................................ 68
Analytic Approach ................................................................................................................ 68
Results ....................................................................................................................................... 72
Sample Characteristics .......................................................................................................... 72
GSOEP .............................................................................................................................. 72
BHPS................................................................................................................................. 72
HILDA .............................................................................................................................. 73
Basic Models ......................................................................................................................... 73
Moderating Effect of Pre-Birth Parental Identity ................................................................. 78
Moderating Effect of Post-Birth Parental Identity ................................................................ 79
Does Parental Identity Matter more for Mothers or Fathers? ............................................... 80
GENERAL DISCUSSION ........................................................................................................... 83
Correlates of Parental Identity .................................................................................................. 84
Parental Identity Across the Lifespan ....................................................................................... 90
Parental Identity and Response to the Birth of a Child ............................................................. 91
The Role of Parental Identity in Reactions to the Birth of a Child ........................................... 94
Limitations and Future Directions ............................................................................................ 96
Conclusions ............................................................................................................................. 101
REFERENCES ........................................................................................................................... 102

vi

LIST OF TABLES

Table 1

Within- and between-person standard deviations ...................................................... 27

Table 2

Results of the model evaluating associations between marital status and
parental identity in the GSOEP .................................................................................. 38

Table 3

Results of the model evaluating associations between marital status and
parental identity in the BHPS..................................................................................... 39

Table 4

Results of the model evaluating associations between marital status and
parental identity in the HILDA .................................................................................. 39

Table 5

Results of the model evaluating associations between education level and
parental identity in the GSOEP .................................................................................. 42

Table 6

Results of the model evaluating associations between education level and
parental identity in the BHPS..................................................................................... 42

Table 7

Results of the model evaluating associations between education level and
parental identity in the HILDA .................................................................................. 43

Table 8

Model results predicting parental identity from income in the GSOEP and the
BHPS .......................................................................................................................... 45

Table 9

Number of people reporting each number of children in the BHPS and the
HILDA ....................................................................................................................... 46

Table 10

Model results evaluating association between number of kids and parental
identity in the BHPS .................................................................................................. 47

Table 11

Model results evaluating association between number of kids and parental
identity in the HILDA ................................................................................................ 47

Table 12

Correlations of parental identity and values and attitudes in the BHPS .................... 48

Table 13

Means of parental identity across responses to the question of who is
primarily responsible for care of the child(ren) ......................................................... 49

Table 14

Correlations of parental identity and values and attitudes in the HILDA .................. 50

Table 15

Correlations between parental identity (averaged across waves) and the
Big Five (averaged across waves when assessed in more than one wave) ................ 51

vii

Table 16

Model results predicting life satisfaction from parental identity in the
GSOEP and the BHPS ............................................................................................... 52

Table 17

Test-retest correlations of parental identity across waves in the GSOEP .................. 53

Table 18

Test-retest correlations of parental identity across waves in the BHPS .................... 53

Table 19

Model results for the association between age and parental identity in
the GSOEP ................................................................................................................. 56

Table 20

Model results for the association between age and parental identity in
the BHPS ................................................................................................................... 56

Table 21

Model results for the association between age and parental identity in
the HILDA ................................................................................................................. 56

Table 22

Results of reaction-adaptation models in the GSOEP and the BHPS ........................ 59

Table 23

Results of the reaction-adaptation model for parental identity including
control groups in the GSOEP and the BHPS ............................................................. 65

Table 24

Model results for basic life satisfaction reaction-adaptation model and
models with parental identity as moderators in the GSOEP ...................................... 75

Table 25

Model results for basic life satisfaction reaction-adaptation model and
models with parental identity as moderators in the BHPS......................................... 76

Table 26

Model results for basic life satisfaction reaction-adaptation model and
models with parental identity as moderators in the HILDA ...................................... 77

Table 27

Model results evaluating the role of gender and parental identity for SWB
responses to the birth of a child ................................................................................. 82

viii

LIST OF FIGURES

Figure 1

Overall means in parental identity across marital status categories in the
three samples .............................................................................................................. 37

Figure 2

Overall means in parental identity across educational achievement categories
in the three samples .................................................................................................... 41

Figure 3

Plot of parental identity vs. number of children in the BHPS and the HILDA ......... 46

Figure 4

Parental identity across the lifespan in the three datasets .......................................... 55

Figure 5

Model predicted trajectories of the basic reaction-adaptation model and the
models that included moderators (gender, age, marital status, and income) for
the GSOEP and the BHPS ......................................................................................... 61

Figure 6

Predicted parental identity trajectories from reaction-adaptation models
including non-parent matched controls ...................................................................... 66

Figure 7

Life satisfaction trajectories for basic model and models with parental identity
before and after birth of the first child as moderators in the three datasets ............... 74

ix

INTRODUCTION
There is little debate that reproduction is a fundamental human need (Kenrick,
Griskevicius, Neuberg, & Schaller, 2010). Indeed, many individuals view the birth and rearing of
children as important life experiences that require great effort and resources. There are many
reasons why individuals may choose to have children, and both scientific and intuitive theories
predict that satisfaction of this fundamental human need would ultimately lead to increases in
subjective well-being for many individuals (e.g., Lyubomirsky & Boehm, 2010). However,
recent research findings suggest that those who have children are not necessarily happier than
those that do not have children in the long term (e.g., Anusic, Yap & Lucas, 2014a; 2014b;
Dyrdal & Lucas, 2013; Yap, Anusic, & Lucas, 2012) and, in some instances, having children has
been linked to decreased well-being (e.g, Baumeister, 1991; Dyrdal & Lucas, 2013; Glenn &
Weaver, 1979).
Although past research may indicate that having children does not lead to long term
average increases in SWB, it is important to note that this research also indicates that substantial
variability exists between individuals in how they are affected by having children. For some
individuals the experience of childbirth is associated with permanent increases or decreases in
happiness, whereas others showed no long-term changes in happiness (e.g., Belsky, Spanier, &
Rovine, 1983; Galatzer-Levy, Mazursky, Mancini, & Bonanno, 2011). Relatively little empirical
work has been done to explain why individuals show such marked variation in their reactions to
major life events such as childbirth and identifying the factors that can account for this variation
is an important empirical concern. Although recent work has examined whether this variability in
reactions to childbirth may be accounted for by personality traits (Anusic et al.,2014a; Dyrdal &
Lucas, 2013; Yap et al., 2012), one construct that has not been examined as a potential moderator

1

is parental identity - the degree of importance one places in being a parent. Theoretically,
parental identity may be psychologically important, yet little is known about this construct and
its implications.
Thus, the goals of the current research project were two-fold. The first goal of this project
was to gain a better empirical understanding of parental identity, its stability and change over
time and the degree that life transitions (such as onset of parenthood) affect it. Second, this
project aimed to evaluate how parental identity affects the relationship between childbirth and
well-being, and whether this construct accounts for the variability in individuals’ response to the
birth of a child.
Life Events and Subjective Well-Being
Happiness, or subjective well-being (SWB) as it is known in the research literature, is
defined as the subjective evaluation of one’s own overall quality of life (Diener, 1984). SWB is
highly valued, and the possibility of obtaining high levels of well-being is a strong motivator of
behavior (Lyubomirsky, Sheldon, & Schkade, 2005) and understanding the factors that explain
SWB have been an important area of psychological study. However, this body of research has
generally shown that the relationships between life circumstances and SWB are relatively small
(e.g., physical health; Okun & George, 1984; see also Diener, Suh, Lucas, & Smith, 1999). As
such, some scholars have argued that the weak links between the objective circumstances of
one’s life and well-being suggest that individuals are able to adapt to most major positive and
negative life events over time.
Adaptation theories (e.g., Frederick & Loewenstein, 1999; see Diener, Lucas, & Scollon,
2006 for a review) posit that SWB is a homeostatic phenomenon that fluctuates around a stable,
genetically pre-determined set-point. These theories suggest that changes in an individual’s life

2

circumstances and the experience of major life events may be associated with temporary changes
in well-being levels, but individuals will inevitably adapt to these life events. In turn, adaptation
theories predict that SWB will quickly and inevitably return to genetically predetermined
baseline levels of SWB experienced prior to the experience of the life event (Lykken & Tellegen,
1996). One implication of these adaptation theories is that they suggest it is futile to attempt to
change one’s life circumstances in order to improve one’s quality of life as any gains in SWB
will inevitably disappear.
The proponents of these theories have suggested that the process of adaptation to life
events may serve several important functions. For example, adaptation to positive and negative
life events may help people cope with major life changes and buffer people from the potentially
negative emotional consequences of these life events. Adaptation to persistent aspects of the
environment may also be advantageous because these processes allow individuals to habituate to
changes in life circumstances and free one's cognitive resources to attend to novel environmental
stimuli (Frederick & Loewenstein, 1999). Indeed, the predictions of adaptation theories
contradict most intuitive theories about the long term effects that positive and negative major life
events (e.g., getting married; death of a family member) would have on individuals’ SWB.
Consequently, these theories raise the question of whether individuals adapt to major life events
(as adaptation theories predict) or are these events actually associated with lasting increases and
decreases in SWB.
Although some past research indicates that individuals do generally adapt to major life
events and that life circumstances are poor predictors of well-being, much of this research has
been restricted to relatively limited cross-sectional studies (e.g., Campbell, Converse, &
Rodgers, 1976; Okun & George, 1984; see Diener, 1984 for a review). Recent research,

3

however, has begun to examine questions of whether changes in life circumstances and major
life events are associated with changes in SWB using large, nationally representative panel
studies (for review, see Lucas, 2007a).
There are several advantages to using this type of design. First, these studies are
longitudinal studies that span many years and typically include relatively large samples of
participants. For example, the German Socio-Economic Panel (GSOEP), which has been used in
past studies examining reaction and adaptation to life events (e.g., Galatzer-Levy, Bonanno, &
Mancini, 2010; Lucas, 2007b; Pagán-Rodríguez, 2012; Stutzer & Frey, 2006), includes a
nationally representative sample of Germans numbering in the tens of thousands. These
participants have been assessed at yearly intervals for multiple decades. Using samples such as
this allows researchers to examine how life events (even relatively rare ones) affect SWB over a
long span of time in a large number of people.
A second advantage to using this type of data is that they are prospective. That is,
participants’ pre-event levels of life satisfaction are known and comparisons of pre-event and
post-event levels of life satisfaction are possible. The advantage of these prospective designs
over non-prospective alternative designs is that prospective designs eliminate the possibility that
any differences between those who have experienced the event and those who have not are due
to pre-existing differences (i.e., individuals who experience certain events might simply be
higher or lower in subjective well-being before the event occurred). Thus, prospective designs
can rule out at least one important confound of cross-sectional studies.
A third advantage of this type of design concerns the fact that participants are not selected
into the study precisely because they experienced the event. Because certain major life events
are very rare, researchers are often only be able to obtain large enough samples by explicitly

4

recruiting participants who have experienced (or are likely to experience) the event in question.
However, this recruiting strategy makes the focus of the study known, which can introduce
demand characteristics (e.g., Smith, Schwarz, Roberts, & Ubel, 2006). Specifically, when
participants know that they are being recruited on the basis of experiencing a particular life
event, the event may become particularly salient and the respondent may have ideas about how
their experience of the particular life event should have affected their subjective well-being. The
participant may also be motivated to appear better adjusted than they actually are. In turn, these
factors may cause individuals to over- or under-report their life satisfaction based on these
preconceived notions. In contrast, using large nationally representative samples minimizes the
possibility of demand characteristics, while still providing enough participants for meaningful
results.
Despite the strengths of using these national panel studies, use of such data is not without
its limitations. One of the most notable disadvantages of using these panel studies is that
individual researchers are limited to using measures and preexisting items that are already
included in the surveys. Generally speaking, researchers have no control of over the design of the
panel study used and the variables that are being assessed. Thus, the variables one evaluates,
one’s analytic approach, and design of an individual researcher’s study is limited to the existing
design and the items available at each wave of the panel being used. Another oft cited concern of
using data from national panel studies is that, because of the breadth of these studies, many
constructs are assessed with very brief measures – including many that are assessed with single
items. Use of brief measures in these national panel studies is usually done to minimize
participant burden given the wide breadth and length of these large-scale longitudinal studies.
The main criticism of using such measures is that they may not be as internally consistent as one

5

would typically expect in psychology, especially when compared to the internal consistency of
longer measures. Another concern about brief measures is that they may not have the breadth to
capture all the relevant or important aspects of many important constructs.
Although it is important to keep these issues in mind when interpreting results of such
studies, recent research suggests that these concerns may not be as problematic as expected. For
instance, Lucas and Donnellan (2011) used strategies for estimating reliability from longitudinal
data to show that the reliability of a single-item life satisfaction measure like the one that is
typically included in these panel studies is about .73, which is at a similar level to many
measures typically used in psychology.
Another criticism of using brief or single-item measures is that they potentially lack the
bandwidth necessary to adequately capture many of the psychological constructs researchers are
interested in studying. However, past research has shown that many constructs are captured
relatively well using a single-item and many commonly used multi-item measures used in
psychology (e.g., The Satisfaction with Life Scale; Diener et al., 1985) often contain items that
are simply rewordings of the same idea. Further, single-item measures of psychological
constructs often correlate strongly with multi-item measures of the same construct and in many
instances, single-item measures have been shown to be valid and used extensively in past
research (e.g., Cheung & Lucas, 2014; Kobau, Sniezek, Zack, Lucas, & Burns, 2010; Robins,
Hendin, & Trzesniewski, 2001).
Much of the research with panel studies is correlational and cannot definitively establish
causality. However, they often provide more descriptive information about the nature of the
associations between life circumstance and life event variables and SWB than can simpler crosssectional designs. Indeed, past research that has examined the effect of life events on SWB using

6

nationally representative panel data suggests that experiencing major positive and negative life
events may have substantial effects on an individual's life satisfaction (Anusic et al., 2014a;
2014b; Lucas, 2007a; Yap et al., 2012). However, the precise nature of these effects appears to
vary across different events. For instance, research using the GSOEP shows that individuals
typically do react to major life events (like marriage, divorce, disability, widowhood, and
unemployment), but the length of time that these reactions last varies across events (Lucas,
2007a).
Parenthood and Subjective Well-Being
One major life transition that has been the subject of study and debate in recent SWB
research is the experience of becoming a parent (e.g., Dyrdal & Lucas, 2013; Nelson, Kushlev, &
Lyubomirsky, 2014; Anusic et al., 2014a; 2014b; Yap et al., 2012). The notion that human
beings are intrinsically motivated to have children is relatively non-controversial (Kenrick et al.,
2010). However, the effect of satisfying this fundamental human need to reproduce on SWB and
psychological adjustment is more uncertain. Although past research has shown that satisfaction
of basic human needs such as hunger, self-protection and social affiliation are often associated
with increases in well-being and positive feelings (Macht & Dettmer, 2006; Smith, Kendrick,
Maben, & Salmon, 1994; Tay & Diener, 2011; Wong & Csikszentmihalyi, 1991), the extant
research literature suggests that the link between well-being and having children is more
ambiguous (e.g., Lyubomirsky & Boehm, 2010; Nelson et al., 2014). For instance, although
several studies indicate that parenthood is consistently associated with declines in marital
satisfaction, SWB, and higher levels of negative outcomes such as anxiety and depression (e.g.
Belsky et al., 1983; Belsky, Lang & Rovine, 1985; Glenn & McLanahan, 1982; Glenn &
Weaver, 1979; Lawrence, Nylen, & Cobb, 2007; McLanahan & Adams, 1987a; Twenge,

7

Campbell, & Foster, 2003; Umberson & Williams, 1999), the sizes of these effects are typically
small. There are also a number of studies that fail to find evidence that parenthood is associated
with substantial changes in global well-being (e.g., Rothrauff & Cooney, 2008), or report that
parenthood is associated with increased well-being (Aassve, Goisis, & Sironi, 2012; Nelson,
Kushlev, English, Dunn, & Lyubomirsky, 2013).
Much of the past literature evaluating the effects of parenthood on SWB has also relied
on cross-sectional designs, which are limited in several important ways (discussed above).
However, longitudinal studies on the impact of parenthood on well-being have also yielded
varied findings. For instance, a substantial body of longitudinal research indicates that
parenthood is consistently associated with long-term decreases in various well-being and
adjustment variables, mirroring much of the cross-sectional literature review above (e.g., Clark,
Diener, Georgellis, & Lucas, 2008; Doss, Rhoades, Stanley, & Markman, 2009; Dyrdal & Lucas,
2013; Dyrdal, Røysamb, Nes, & Vittersø, 2011; Grote & Clark, 2001; Rholes, Simpson,
Campbell, & Grich, 2001). However, other longitudinal research findings suggest that the
experience of becoming a parent itself is not linked to long term increases or decreases in SWB
(Anusic et al., 2014a; 2014b) and that overall decreases in SWB following childbirth may
actually reflect normative, age related changes in SWB that would have been observed even if
individuals remained childless (e.g., Yap et al., 2012).
Taken as a whole, much of the extant research literature demonstrates an ostensive
paradox. Parenthood may be associated with lower SWB despite the high value many individuals
place on achieving parenthood. Nevertheless, the research literature is mixed regarding the
impact of parenthood on SWB (Nelson et al., 2014) and further research evaluating the factors
that explain the discrepancies in this literature is needed.

8

Variability in Reactions to Parenthood
Research examining the impact of major life events on SWB typically reports the effects
of events on SWB as an average level effect across a given sample. However, past research also
indicates that there is a substantial amount of variability in how people respond to various life
events and these experiences are clearly not associated with uniform patterns of SWB change in
all people (e.g., Lucas, 2005, 2007b; Lucas, Clark, Georgellis & Diener, 2003; 2004). For
example, Lucas and colleagues (2003) reported that there was substantial variability in the
degree of adaptation following marriage, and observed that the trajectory of adaptation varied as
a function of one’s reaction to marriage. Some individuals remained happier than they were
before marriage, whereas others were less happy compared to their pre-marriage levels. Research
also indicates that there is substantial variability in individuals’ reactions of grief to traumatic
events and other major personal loss. These studies show that some individuals experience large
declines in well-being when faced with traumatic events, while others show little long term
change (e.g., Bonanno, 2004; Bonanno & Kaltman, 2001; Bonanno & Keltner, 1997).
In regards to childbirth, past research indicates that there is also a substantial amount of
variability in how people respond to the birth of a child (e.g., Belsky et al., 1983; Galatzer-Levy,
et al., 2011). For instance, although the results of a study by Belsky and colleagues (1983)
suggest that parenthood is associated with average declines in marital adjustment, these authors
also showed that there are individual differences in this negative effect. Similarly, Galatzer-Levy
et al. (2011) demonstrated that there were substantial individual differences in how people
reacted to the birth of a child. Using data from the German Socio-Economic Panel, these authors
found that although the onset of parenthood may be associated with average declines in SWB,
the vast majority of parents in their sample showed no long term changes in SWB in response to

9

childbirth. A relatively small portion of their sample showed evidence of a lasting decline
following childbirth, and some individuals showed evidence of long term gains in SWB
following the onset of parenthood.
To be sure, there is clear evidence that variability exists in how individuals react to
having children and whether these potential changes are long lasting. This evidence suggests that
reactions to parenthood are not homogeneous across all parents, and identifying the factors that
differentiate between individuals with different reactions to parenthood is an important concern.
However, past research that examines the characteristics and variables that may account for this
type of variability in reaction and adaptation to life events is relatively sparse.
Past work suggests that demographic characteristics play an important role and predict
parents’ reactions to having children. For instance, Galatzer-Levy et al. (2011) demonstrated that
demographic characteristics such as marriage, income, and educational attainment predicted the
response pattern of SWB exhibited by parents in their sample. In particular, parents who
displayed a more stable pattern of SWB in the years surrounding childbirth were more likely to
be married, have higher education, and have more income than parents who displayed other
response patterns. Similarly, Kohler and colleagues (2005) demonstrate that several important
demographic characteristics, including the age of parents, gender of parents, partnership status
and the number of children parents have moderate the pattern of SWB change one experiences in
response to having children.
Other research has demonstrated that variability in SWB reactions to parenthood are also
accounted for by stable individual differences among parents such as genetic factors (see Kohler
et al., 2005) and personality traits (Dyrdal & Lucas, 2013). For instance, a number of past studies
suggest that personality traits moderate individuals’ responses to various major life events (e.g.,

10

Anusic et al., 2014a; Boyce & Wood, 2011; Pai & Carr, 2010; Bonanno, 2004). In line with this
work, recent research by Dyrdal and Lucas (2013) and Yap et al. (2012) evaluated whether the
Big Five personality traits (Neuroticism, Extraversion, Openness to Experience, Agreeableness,
and Conscientiousness; McCrae & Costa, 1987; Goldberg, 1990) accounted for variability in
reaction and adaptation to the birth of a child in nationally representative samples of German and
British households, respectively. Taken together, these two studies indicate that Openness to
Experience appears to play a positive role in how having children impacts SWB. Indeed,
individuals higher in Openness to Experience appear to react more positively to the birth of a
child, but it is important to note that there were differences among these studies in the period of
time where the moderating effect of Openness to Experience emerged. Further, a third study of
Australian households failed to find evidence that Openness to Experience moderated reactions
to having children (Anusic et al., 2014a). The results of Dyrdal and Lucas (2013) and Yap et al.
(2012) also suggested that Neuroticism has a moderating effect on reactions to childbirth.
However, the nature of the moderating effect of Neuroticism on the association between
parenthood and SWB is less clear given that these two studies yielded inconsistent findings in
regards to the role of Neuroticism. Further, a recent study of Australian households (Anusic et
al., 2014a) failed to find a moderating effect of Neuroticism on parents’ responses to the birth of
their first child.
There may be other relevant individual differences that may account for the variability in
people’s responses to childbirth. For instance, differences in the characteristics of the experience
itself, such as differences in children’s temperament, difficulty and complications with the actual
birthing and variation in the availability of post-partum social support all could account for
variability in people’s responses to birth. Indeed, another possibility that has yet to be explored

11

in the research literature is that individuals’ responses to childbirth may be explained by
individual differences in the importance placed on being a parent - a notion referred to as
parental identity.
Parental Identity and SWB
Generally speaking, identification with a group or social role appears to serve important
social functions (Tajfel & Turner, 1986; Turner, Oakes, Haslam, & McGarty, 1994; Thoits,
1987). As such, major theoretical perspectives on identity and identity related processes (e.g.,
Tajfel & Turner, 1986) suggest that one of the main benefits of identity related processes is the
maintenance of one’s positive sense of self. According to this perspective, identification with
important groups and social roles is one of the central mechanisms an individual uses to bolster
self esteem and positive self image (Tajfel & Turner, 1986) and a substantial body of literature
has shown that various group identities are beneficial to psychological adjustment and subjective
well-being (e.g., Kiang, Yip, Gonzales-Backen, Witkow, & Fuligni, 2006; Phinney, Cantu, &
Kurtz, 1997; Settles, 2004; Yap, Settles, & Pratt-Hyatt, 2011). For instance, research evaluating
racial and ethnic identity indicates that racial and ethnic group membership is relatively salient
for many racial/ethnic minority group members and identification with one’s racial and ethnic
group is associated with outcomes such as life satisfaction (Yap et al., 2011), depression (Settles,
Navarrete, Pagano, Abdou, & Sidanius 2010) and self-esteem (Kiang et al., 2006; Phinney et al.,
1997).
People typically identify with several simultaneous social groups and roles, and research
suggests that parental roles are particular salient for many adults (Thoits, 1992). Identification
with these parental roles also appear to have important psychological implications. Parental
identity is defined as the degree to which an individual sees parenthood as important to the self.

12

In other words, it is the extent to which “being a parent” is central to an individual’s self image
(Maurer, Pleck, & Rane, 2001). Despite the theoretical and intuitive importance of this construct,
the research literature examining this construct is extremely small. Very little is known about the
construct itself, and no research has examined important questions such as how the importance
of being a parent develops and changes over the life span and whether it is affected by significant
life events such as the birth of a child. There are a handful of studies that show that parental
identity is associated with positive well-being outcomes. For example, a study by Martire and
colleagues (2000) reported that the self-rated importance of being a mother was associated with
life satisfaction in a sample of middle-aged women. Other research also suggests that parental
identity is linked to self-esteem (Reitzes & Mutran, 1994). However, other studies have also
failed to find evidence for such direct effects on individual well-being and adjustment (Noor,
2004; Thoits, 1992) and virtually nothing is known about the robustness and generalizability of
these existing effects given the limited research examining the consequences of parental identity
on well-being and adjustment.
The reasons for these discrepancies in the literature are unclear. One potential reason is
the way in which the notion of parental identity is conceptualized and measured across studies.
For instance, both studies that find evidence for associations between parental identity and wellbeing operationalize the construct of parental identity using continuous Likert scale measures
(i.e., Martire et al., 2000; Reitzes & Murtran, 1994). In contrast, the studies that failed to find
evidence of an association between parental identity and well-being operationalized the construct
in a different manner. Thoits (1992) measured parental identity by asking respondents to rank
order the importance of one’s parental role compared to two other roles (e.g., worker, spouse).
Noor’s (2004) scale items reflected broader notions of identification with and importance of

13

family and the questionnaire was in the context of work-family balance. Indeed, it is possible
that such differences in the operationalization of parental identity led to the observed differences
in results across studies. Overall, it is important to note that there is little consensus on the
optimal way to assess parental identity. There is also little consensus as to the core features of
this construct, and the specific feelings, attitudes and behaviors that would be considered part of
parental identity. Further, individual researchers often utilize measures of parental identity that
are unique to their own program of research and it appears that little effort has been made to
bring unique streams of research together to develop a unified model for thinking about and
assessing parental identity.
There could also be other reasons for these discrepancies. For instance, there could be
group differences in how parental identity and well-being are related among sampled
populations, but without a greater body of research in this area, it is difficult to predict the
robustness of these effects or the precise reasons for these discrepancies. Generally speaking,
much further research is needed before stronger conclusions can be made about whether there is
a link between parental identity and well-being.
The direct importance of parental identity on psychological adjustment is also
demonstrated by other studies linking parental identity to other important parenting behaviors,
such as caregiving behavior (Maurer & Pleck, 2006), and involvement with child rearing (Bruce
& Fox, 1999; Ihinger-Tallman, Pasley, & Buehler, 1993) which in turn may have positive wellbeing outcomes for both individual parents and their families and offsping. Indeed, the notion of
parental identity has been identified as one of the single most important theoretical antecedents
of parental involvement, particularly for fathers (e.g., Pleck, 1997). However, it is again worth
noting that multiple studies do not find evidence for associations between parental identity and

14

parental involvement (e.g., McBride & Rane, 1997; Minton & Pasley, 1996) and that this
literature is extremely small.
Other research indicates that parental identity may act primarily as a moderator of the
relationship between parental-role related stress and well-being outcomes. Under this
perspective, parental identity and the experience of parental-role related stressors are viewed to
have an interactive effect on well-being. Parental-role related stressors may have a particular
impact on individuals’ well-being if the domain of these experiences is especially important or
salient to the individual. Conversely, if being a parent is a domain that is relatively unimportant
to a person, his or her response and the relative permanence of this individual’s response to
stressors in this domain may be small.
Past evidence is mixed regarding the direction of the moderating effect of parental
identity, and whether it serves to exacerbate or protect against the negative effects of parental
stressors on various well-being outcomes. For example, a study by Simon (1992) suggested that
increased levels of parental identity were associated with increased vulnerability to parental role
strains and exacerbated the positive association of these strains on individuals’ depression and
anxiety. This finding is consistent with the notion that negative events and stressors that are
related to an important identity are going to be more detrimental to one’s overall well-being than
threats related to an unimportant identity (Thoits, 1992).
In contrast, other studies have found evidence showing parental identity to be a protective
factor against parental role related stressors. For example, Martire et al. (2000) found that
paternal identity buffered mothers from the negative effects of mother stress on depression. This
study showed that mother stress was related to depression for mothers low in parental identity,
but no relationship between mother stress and depression was observed for mothers high in

15

parental identity. Similarly, Luchetta (1995) also found evidence for a protective role of parental
identity against the negative effects of family related stressors on psychological distress. These
findings are consistent with past theory regarding the functions of identity (e.g., Tajfel & Turner,
1986). Under this perspective, individuals for whom being a parent is an important part of the
self will be especially motivated to feel good about their role as a parent, even in the face of
significant stressors associated with this role. These individuals may simply be more committed
to focus on positive attributes of being a parent, which may provide justification for the negative
aspects of being a parent.
Existing evidence regarding the role that parental identity plays in the association
between parental and family related stressors and well-being outcomes is mixed and does not
offer a consistent overall picture of the nature of this moderating effect. The reasons for these
discrepancies in this literature are unclear. One possible reason may be that there are differences
in the outcomes each of these studies are evaluating. Thus, it is possible that these discrepancies
may simply reflect real differences in how parental identity moderates the relation between
stressors and various outcomes. There are also differences in the specific predictor variables
being examined in each study. Again, although each of these predictors could be characterized as
stressors related to one’s role of being a parent, there may be subtle differences among these
constructs in how they interact with parental identity. Of course, there have been an extremely
small number of studies examining the moderating role of parental identity on the relation
between parental role related stressors and well-being, and it is clear that more work is needed
before stronger conclusions should be drawn and this area can move forward. Further, it is
possible that with more research evaluating this question, a more consistent picture of this effect
would emerge and the apparent discrepancies regarding the direction of this moderating effect

16

may fade away.
In summary, there are several unanswered questions regarding the development and
nature of parental identity, the role of parental identity on the effects of parental role related
stressors (such as childbirth) on well-being. Largely due the small number of existing studies, the
main issue in this area of research is that there is simply no consistent picture on the nature and
importance of parental identity, the role parental identity plays in individuals’ overall well-being
(for both parents and non-parents) and how individuals react to parenthood. The little research
that does exist suggests that there may be a direct positive relation between parental identity and
various well-being outcomes and that one’s level of parental identity may moderate how one
reacts to stressful events associated with one’s role of being a parent (such as childbirth). These
predictions would also be consistent with past theory and research evaluating other types of
identity, such as ethnic identity, and their implications. Clearly, the scientific understanding of
parental identity and its significance would benefit from more systematic research evaluating the
importance and overall implications of parental identity and whether parental identity affects
reaction and adaptation to new parenthood.
The Current Studies
Given these theoretical gaps, this project had two main goals. The first overall goal of
this project was to gain a better empirical understanding of parental identity. For instance, little is
known about the construct of parental identity itself, its stability and change over time, and its
correlates. Thus, this project will evaluate the extent to which parental identity changes over the
lifespan, and whether having a child leads to significant increases or decreases in self-rated
parental identity. Further, this project also evaluated the nature of parental identity and sought to
identify associations between parental identity and other important psychological constructs

17

(e.g., well-being, demographics).
Evaluating the degree to which parental identity changes over the course of one’s life is
an important developmental question as it speaks directly to the psychological process of
becoming a parent. For example, it is possible that individuals at different life stages may display
differences in the importance they place in being or becoming a parent, and understanding these
differences across the life span may lead to a better understanding of the development of
parenthood and its relevant psychological processes. Few unifying theories speak to the
psychological processes involved in one’s transition to parenthood and past scholars have
highlighted the need for a better scientific understanding of the psychological processes involved
in becoming a parent (Cabrera et al., 2000).
Further, it is also possible that the importance one places in having children or becoming
a parent is a relatively enduring individual difference across various stages in one’s life course.
As such, it is important to understand how individual differences in this construct predicts
various characteristics of one’s life, such as whether an individual has children, the number of
children one has and the age at which one became a parent. The second overall goal of this
project was to evaluate the degree to which individual differences in parental identity account for
variability in the association between having children and life satisfaction and variability in the
observed patterns of life satisfaction change following the onset of parenthood. In other words,
does parental identity moderate individuals’ short and long term responses to the birth of a child?
Although some past research has examined whether other individual differences (e.g.,
personality traits) moderate individuals’ responses to childbirth (e.g,, Anusic et al., 2014a;
Dyrdal & Lucas, 2013; Yap et al., 2012), the current paper extends this literature by examining a
construct that has yet to be explored as a moderator of individuals’ responses to the birth of a

18

child.
Past research suggests that parental identity may be important to several well-being and
parenting outcomes, and that it moderates the relationships between parental stress and wellbeing (e.g., Martire et al., 2000). Indeed, the literature described above supports two competing
predictions regarding the role of parental identity in individuals’ responses to childbirth. The first
prediction is that individuals high in parental identity may react more positively to the birth of a
child. This pattern of results could have multiple explanations. As discussed above, many
individuals are fundamentally motivated to have children (Kenrick et al., 2010). In addition to
these basic motivations, some individuals may also place high psychological importance on
being a parent. Fulfilling this important identity (i.e., actually having a child) may be particularly
beneficial to these individuals’ sense of self compared to other people who place lower
importance on being a parent. Becoming a parent is also considered a very stressful experience
for many individuals, and is rated among the most stressful life events one can experience
(Holmes & Rahe, 1967). Clearly, such levels of stress may have direct detrimental effects on an
individual’s well-being, but as some past research suggests, at least some of the potential
negative effects of the stress associated with parenthood may be mitigated by having a strong
parental identity (e.g., Luchetta, 1995; Martire et al., 2000).
Some past theories also suggest that individuals have a tendency to place particular
emphasis on the positive aspects of the identities one holds as a way to bolster one’s own self
image (e.g., Tajfel & Turner, 1986). Thus, individuals for whom being a parent is an important
part of the self may be especially motivated to feel good about their role as a parent, even in the
face of the high stress associated with events such as the birth of a child. In other words, these
individuals may simply be more committed to focus on positive attributes of their newfound role

19

as a parent and the psychological benefits that may come with this role.
A competing prediction regarding the role of parental identity on responses to the birth of
a child is that individuals high in parental identity may react more negatively to the birth of a
child. As some past research indicates (e.g., Simon, 1992), the stress of having a child could also
be particularly taxing for individuals who place high importance on being a parent, especially if
these individuals have difficulty coping with the disruptions associated with having children.
Despite the importance one places on being a parent, one may still have difficulty incorporating
their new parental responsibilities into their lives. As such, failure to adequately cope with the
stresses associated with this centrally important identity may be particularly threatening to these
individuals’ sense of self compared to threats and disruptions related to individuals’ unimportant
identities (Thoits, 1991).
In sum, I predicted that parental identity played a role in how individuals respond to the
experience of childbirth itself. However, given the mixed past evidence regarding the direction
of potential moderating role of parental identity in the association between childbirth and wellbeing, my predictions regarding the direction of this potential moderator of individuals’ reactions
and pattern of adaptation to childbirth were more tentative.

20

METHOD OVERVIEW
Overview of Studies
In order to address these questions, I conducted two interrelated studies. Study 1
evaluated the first goal of this project and sought to understand the nature of parental identity
itself, the degree of stability and change of parental identity over the life course, its relationship
to other constructs (e.g., income, marital age), and the extent to which mean levels of parental
identity were affected by a major life transition of becoming a parent. Study 2 evaluated the
second goal of this project, and explored the moderating role of parental identity on individuals’
responses to childbirth.
Each of these studies used data from large, nationally representative samples of German,
British and Australian households. The German Socio-Economic Panel Study (GSOEP), the
British Household Panel Survey (BHPS), and the Household, Income and Labour Dynamics in
Australia Survey (HILDA) are large studies of individuals from Germany, Great Britain and
Australia that span multiple decades and include large samples of people. One of the main
advantages to using these data was that I was be able to examine these questions using large
representative samples of people (over 50,000 in the GSOEP alone) from three distinct nations
over multiple decades.
Overall Sample Characteristics
The data used in these studies was drawn from waves 1 through 27 of the GSOEP, waves
1 through 18 of the BHPS, and waves 1 through 10 of the HILDA. The GSOEP is a nationally
representative study of German households that has been assessing households annually since
1984 (Haisken-DeNew & Frick, 2005). Households were selected using clustered multistage
random sampling. Members of each selected household, 16 years of age and older, were

21

surveyed using face-to-face interviews. General household-related information was collected
from the head of each household using a questionnaire. Initially, the GSOEP included
individuals residing in West Germany, but was later expanded to include households in East
Germany as well. Over the course of the study, there was some attrition and refreshment samples
of households were added. Thus, many individuals did not participate in every wave of the
GSOEP. Overall, the GSOEP includes data from over 50,000 individuals that participated in at
least one wave of data collection.
The BHPS is a nationally representative study of households in Great Britain that has
been assessing individuals annually since 1991 (Institute for Social and Economic Research,
University of Essex, 2010; Taylor, Brice, Buck, & Prentice-Lane, 2009). The BHPS sampled
households using a multistage clustered probability design and systematic sampling (Taylor et
al., 2009). Members of each sampled household age 16 or over were asked to participate. As
with the GSOEP, some attrition occurred in the BHPS sample and there were new subsamples of
participating households added. As such, many respondents in the BHPS did not participate in
every wave of data collection. Overall, the BHPS includes over 30,000 individuals that
participated in at least one wave.
The HILDA is a nationally representative study of households in Australia which began
in 2001(Summerfield et al., 2012). The HILDA sampled households using a probability design
(see Summerfield et al., 2012 for more details). Members of each sampled household age 16 or
over were invited to participate. Eligible members of selected households were asked to
complete a face-to-face interview and a paper questionnaire. Overall, over 19,000 individuals
participated in at least one wave of the HILDA.

22

STUDY 1
The goal of this first study was to evaluate the nature of parental identity and its
correlates, its stability and change over time, and how it changes in response to becoming a
parent.
Method
Sample Selection
To examine the characteristics of parental identity, I selected a sample of individuals
from each study that responded to the parental identity item in at least one wave. These samples
of people in the GSOEP, BHPS and HILDA served as the basis for all analyses in Study 1.
Measures
Parental identity. One of the biggest challenges to achieving the goals of this research
was the lack of clarity regarding the assessment of parental identity. As the literature on parental
identity is relatively small, there is not a clear set of well-validated, standard measures used to
assess this construct. Past research has assessed parental identity using various instruments,
ranging from multi-item scales assessing various theorized dimensions of parental identity (e.g.,
Maurer et al., 2001) to single-items assessing the importance of being a parent or having children
(e.g., Martire et al., 2000). In each of the datasets used in this project, parental identity was
assessed using a single-item, capturing this notion of importance.
In relation to other measures, these single item measures are narrow in scope. For
instance, Maurer et al.’s Caregiving and Breadwinning Identity and Reflected-Appraisals
Inventory is a 70-item measure that assesses importance of parenthood, but also captures
evaluations and self-rated importance of various caregiving and breadwinning behaviors of
respondents, their partners, and their beliefs about how partners see respondents in terms of their

23

caregiving and breadwinning roles. Maurer et al.’s measure offers researchers greater breadth in
the features they may wish to evaluate when considering the causes and consequences of parental
identity. Other measures like Thoits’ (1992) role sorting task assessed relative importance of
parental roles compared to other roles the respondent identified with and casts the importance of
this role in relativistic terms. Thoits’ measure captures and quantifies importance, which is
theoretically congruent with the measure used in the current study, but operationalizes the idea of
importance in a different way.
Although not much is known about the measurement of this construct, nor is there
consensus as to the ideal way to assess it, the way parental identity has been theoretically defined
in the literature is conceptually narrow – often simply characterized by the extent to which being
a parent or having children is important to an individual’s self image (e.g., Maurer et al., 2001).
Further, the notion of importance is also an area of convergence in most existing measures of
parental identity and is not only viewed as a core feature of each of these operationalizations of
parental identity but many major theoretical frameworks of identity as well (e.g., Sellers, Smith,
Shelton, Rowley, & Chavous, 1998). Thus, parental identity is likely to be captured reasonably
well when assessed as the importance of parenthood or having children in this manner.
In the GSOEP, parental identity was measured using a single item that asked individuals
to rate the importance of having children on a Likert scale ranging from 1 (very important) to 4
(unimportant), which I reverse coded so that higher numbers represent higher importance. This
item was assessed at 5 waves of data collection (1990, 1992, 1995, 2004, and 2008).
In the BHPS, parental identity was measured with a single question that asked
respondents to rate the importance of having children on a Likert scale ranging from 1 (not
important at all) to 10 (very important). This item was assessed at 3 waves of data collection

24

(waves 8, 13, 18).
In the HILDA, parental identity was measured with a single item that asked respondents
to rate the current importance of having children on a Likert scale ranging from 0 (not at all
important) to 10 (very important). These items were administered only at wave 4 of the survey
and administered only to individuals aged 15 to 29.
Demographic and other variables. The GSOEP, BHPS and HILDA each include
questions at each wave assessing gender, marital status, education, income. BHPS and HILDA
also include a question about number of children at each wave, whereas the GSOEP simply
keeps a biographical datafile with total number of children in the family, and the birth year of
youngest and oldest child. There are also variables in each study assessing various psychological
characteristics including measures of personality and life satisfaction. Attitudes and values
related to work and family are also assessed in certain waves of the BHPS and HILDA.
Generally speaking, questions about basic demographics (e.g., marital status) are asked to all
respondents and at every wave. Psychological characteristics and attitudes, however, are
measured in some waves and not others.
Analytic Approach
The first step in these analyses was a descriptive analysis describing the correlates of
parental identity to evaluate the association between parental identity and several important
demographic and psychological constructs. As past research evaluating this construct is scant,
little empirical work has explored the individual differences and characteristics of one’s life that
predict parental identity in great detail. Indeed, descriptive analyses identifying the important
constructs that are part (and not part) of parental identity’s nomological net is an important step
to understanding parental identity, its causes, and its consequences. Thus, I evaluated the

25

association between parental identity and demographic characteristics such as gender, education,
income, number of children, marital status, and age. I also evaluated the association between
parental identity and some psychological variables, including Big Five personality traits, overall
life satisfaction, and various measures of attitudes and values related to family, children,
marriage and gender roles.
Covariation between parental identity and other constructs can be conceptualized in
multiple ways. It can be conceptualized as a within-person effect – to what degree does a does a
variable in question change as an individual’s parental identity changes across the lifespan (e.g.,
does a person’s life satisfaction change as their own parental identity changes?). Covariation can
also be conceptualized as a between-person effect – what is the difference between people with
different levels of parental identity (e.g., is there a difference in life satisfaction between people
with different levels of parental identity?). This latter conceptualization is often operationalized
as a cross-sectional bivariate correlation.
In order to take advantage of the longitudinal nature of these datasets, whenever possible
I specified a series of regression models for each variable of interest that separated within-person
and between-person effects. Such analyses account for the longitudinal structure of the data and
allow one to estimate the unique contribution of within-person and between-person processes to
an association between the variable in question and parental identity. Following this logic, each
model specifies one variable as a function of within-person change in the other variable,
between-person change in the other variable, an intercept parameter and error.
The main disadvantage to this approach is that there’s no clear way to standardize
regression coefficients and compare the magnitude of within-persons and between-persons
effects. One way is to use within- and between-person standard deviations obtained from an

26

intercept-only model with random intercept effects for each variable being evaluated to estimate
standardized effect size of the association between the two variables. For example, one can use
the following formula to compute a standardized estimate of association similar to a correlation
from the regression coefficient and the estimated standard deviations of the two variables:
standardized effect = regression coefficient * (SDX / SDY)
where X and Y are the variables being evaluated and the SDs used are the within-person SDs or
between-person SDs (depending whether one is evaluating a within- or between-person effect). I
applied this approach to each analyses where comparisons of within-person and between-person
effects are made. Evaluation of variables where within-person effects are not meaningful (e.g.,
gender) or were not possible due to limited data availability (i.e., in HILDA where only one
wave of parental identity was available), associations with parental identity were assessed with
traditional correlational methods. Within- and between-person standard deviations estimated in
this way are presented in Table 1.
Table 1
Within- and between-person standard deviations.

Parental identity
Life satisfaction
Income/10,000

GSOEP
BetweenWithinperson SD person SD
0.666
0.596
1.201
1.353
5.186
2.601

BHPS
BetweenWithinperson SD person SD
2.293
1.709
0.884
0.900
1.611
1.945

HILDA
Betweenperson SD
3.702
1.481
3.900

The next step was to evaluate the pattern of change in parental identity over the life
course. To understand the association between age and parental identity, I first evaluated a
regression model separating within-person and between-person linear effects of age. To account
for the possibility that parental identity is characterized by a curvilinear trajectory over the life
span, I also estimated a quadratic growth model, where within-person and between-person
estimates of curvature in the trajectory of parental identity can be evaluated. Higher-order

27

models of change in parental identity (e.g., cubic models) were evaluated in an exploratory
fashion. However, the added complexities of these models placed more demands on the data
needed to estimate the models (Singer & Willet, 2003), and it was not possible to estimate these
more complex models with these data.
The next major step of this analytic plan was to evaluate the degree to which parental
identity changes in response to the birth of a child. The critical test of this question in these
analyses involved comparing one’s levels of parental identity after the birth of a child to parental
identity levels before the experience of birth (see Lucas, 2007a). Models that allow for this
within-person comparison can test whether a particular event or experience is associated with
change in parental identity. For instance, if one found evidence that parental identity levels prior
to the birth of a child were similar to parental identity levels in the years that followed the birth,
then one could conclude that birth did not elicit growth or decline in parental identity.
Alternatively, if one observed significant increases in parental identity in the years surrounding
birth compared to pre-birth levels, and these increases persisted in the years that followed birth,
then one could take this as evidence that the birth of a child was associated with lasting increases
in parental identity.
There are multiple ways one could model these changes in parental identity. In ideal
circumstances, one could try to capture changes in parental identity using non-linear models that
estimate a baseline level of parental identity before an event followed by gradual, curvilinear
change in parental identity, eventually leading to a new stable asymptote level following the
major life transition being evaluated (see Yap et al., 2012). However, these types of models
require large amounts of data to accurately estimate the two asymptotic parameters in this model
(i.e., baseline and long term adaptation). Given that I had limited numbers of waves where

28

parental identity was assessed (five at most in the GSOEP), it was necessary in these data to test
theoretically simpler models. Simpler models, like those used in past research evaluating how
subjective well-being is affected by major life events (see Lucas 2007a for review) may be less
precise in describing the trajectory of parental identity over time, but are easier to estimate while
still offering an adequate description of the data.
As such, following models used in past research (Lucas, 2007a), I simplified these
nonlinear models by dividing the time surrounding the experience of childbirth into three distinct
periods. The baseline period reflects the years before birth of the first child, the reaction period
reflects the years around the time of childbirth, and the adaptation period reflects the years
following childbirth. Such models are often referred to as reaction-adaptation models because
they simply capture average levels of parental identity before, during, and after the experience of
an event, they are advantageous due to their relative ease of interpretation and estimation. These
models are especially accurate if the change in parental identity following the event being
evaluated occurs quickly. However, if the change in parental identity happens more gradually in
the years before and after the event being evaluated, then the accuracy of these reactionadaptation models in describing patterns of change in parental identity is compromised.
In these reaction-adaptation models, parental identity was modeled as a function of an
overall intercept (which reflects the average baseline level of parental identity), a reaction
parameter (which reflects average change from baseline during the reaction period), and an
adaptation parameter (which reflects average change from baseline during the adaptation period).
Following past research (e.g., Lucas, 2007a; Lucas et al., 2003), I defined the reaction period as
the year before the event, the year of the event and the year after the event. The baseline period
was defined as the period before the reaction period and the adaptation period was defined as the

29

period after the reaction period in all models.
To estimate these models, dummy variables were created at the within-person level and
coded for each time period in this model. The reaction dummy variable was coded as 1 in the
year before, the year of, and the year after birth, and 0 for all other years. The adaptation dummy
variable was coded as 1 for all years that were at least 2 years following birth, and 0 in all other
years. I included estimates of random effects where possible to allow for between-person
variability in the baseline, reaction and adaptation parameters.
Once I estimated the overall pattern of change in parental identity in the years that
surround the birth of a child using the reaction-adaptation model, I proceeded to evaluate
whether gender, age, marital status and income can meaningfully explain variability in people’s
response in parental identity to the birth of a child. These questions are important because they
can help us understand and identify which characteristics predict individual differences in the
trajectory of parental identity, individual differences in how it changes in response to life events,
and the factors that underlie the development of this construct over time. For example, although
it is clearly not necessary for individuals to be married to have children, marital status may be
important to parental identity because establishing long-term romantic partnerships with others
often precedes having children (e.g., Bumpass & Lu, 2000) and the experience of marriage (at
least in Western society) is often associated with a transition to a new life stage where the
possibility of having children becomes more salient (e.g., Erikson, 1950). Further, one’s
decisions about parenting and child rearing are often made in conjunction with a partner – and
establishing a long-term partnership with another individual could predict differences in how
parental identity changes in response to becoming a parent as a result of this process.
To evaluate the moderating effect of these demographic variables, a new reaction-

30

adaptation model was specified in each dataset for each of the four moderators. Each new model
included parameters estimating moderating effects on each time period in the model (i.e.,
baseline, reaction, adaptation).
Modeling normative changes. One disadvantage of within-person designs such as the
reaction-adaptation model is that they confound changes that can be attributed to an event with
normative, age related changes that may have been observed even if the event had not occurred.
For instance, finding that parental identity scores increased following childbirth may indicate
that parental identity changes in response to the onset of parenthood, but it also could reflect a
normative, developmental increase as one ages, which would have been observed even if
individuals did not ultimately have children. Thus, although these within-person models can help
us understand trajectories of parental identity before and after major life transitions such as the
birth of a child, they confound change due to the experience of life events with change due to
other developmental processes. It is important to account for this analytically.
In order to evaluate whether respondents have higher or lower parental identity after the
experience of childbirth than they would have had if they did not experience the event, one needs
to separate changes in parental identity due to the experience of childbirth from normative
changes that would have happened in the absence of the birth. As such, my next set of analyses
included a matched control group of people who did not become parents during the study to
estimate this normative change (which should occur in both event and non-event groups). The
matched controls of non-parents were selected to be similar to parents in ways that are
potentially relevant to parenthood. Specifically, I selected people who reported having no
children throughout the duration of the study, and used propensity score matching (Gelman &
Hill, 2009) to match individual control participants to the target childbirth group on important

31

demographic characteristics at the start of the study. This procedure involves generating a
propensity score for each person in the sample using a logistic regression model. First, I
specified a model that predicts the occurrence of childbirth (using a variable coding the
childbirth group as 1 and the control group as 0) from their gender, age, age squared, marital
status, education, household income, and the initial parental identity level. This model results in
a propensity score for each individual that reflects the probability that a given individual (with a
particular set of demographic characteristics) will experience childbirth while in the study. The
matching function of the arm package (Gelman et al., 2011) in R software (R Development Core
Team, 2010) was used to match each individual in the childbirth group to a corresponding
individual in the control group with the closest propensity score. The goal of this procedure is to
create two groups that are, on average, similar on the important demographic characteristics used
in matching, and differ only in whether or not they experienced childbirth.
Following this, I estimated the reaction-adaptation models again, but this time in addition
to the people who experienced childbirth, these models also included matched comparison
groups in each dataset. These statistical models will include an additional parameter for both
event and control groups estimating a linear trajectory for parental identity to account for time
related changes that are not due to the experience of the event. This yearly change parameter will
be estimated using the number of years the participant has been in the study as the time variable.
To do this, I included a time variable that is set to zero in the first year in which the individual
was in the study and increases by one in each subsequent year. These models will also include a
group effect parameter which will reflect pre-existing differences between people in each event
group and those in each respective control group. This will be estimated with a dummy variable
coded as 0 for people in the event group (i.e., parents) and 1 for people in the control group (i.e.,

32

non-parents). As such, parental identity will be modeled as a function of the intercept (which
reflects the average parental identity of parents in the first year of participation in this study), the
group effect (initial difference in parental identity between parents and non-parents), the linear
trend in parental identity (i.e., normative changes in identity that exist in both groups), a reaction
parameter (change from the linear year trend in parental identity for the parent group in the years
around childbirth) and an adaptation parameter (change from the linear tread in the years that
follow in the parent group). Reaction and adaptation periods in these models will be defined the
same way as in the traditional reaction-adaptation models described earlier using dummy codes
where the reaction period is coded as 1 in the year before, year of and year after the event. The
adaptation period will be coded as 1 in all the years that follow the reaction period. For the nonparent control group, these will be set to 0 in all years.
Summary of Study 1
Overall, Study 1 makes several important contributions to the scientific understanding of
the psychological implications of parenthood. This study will allow researchers to gain a better
understanding the nature of parental identity, the degree of stability and change of parental
identity over the life course, the extent to which mean levels of parental identity are affected by
the major life transition of becoming a parent.
Results
Sample Characteristics
GSOEP. In total, N = 32,028 people completed at least one wave of parental identity. Of
these participants, 52% identified as female and 48% identified as male. N = 2,681 completed all
5 waves. The mean age at time of survey was 46.3 years (SD = 17.5 years; Range = 16-99 years).
On average, participants completed 2.2 waves of parental identity.

33

BHPS. In total, N = 20,457 people completed at least one wave of parental identity. Of
these participants, 54% identified as female and 46% identified as male. N = 5,296 completed all
three waves. The mean age at time of survey was 46.1 years (SD = 18.8 years; Range = 16-99
years). On average, participants completed 1.8 waves of parental identity.
HILDA. Parental identity was measured only in wave 4 of the HILDA, and was only
asked of participants 15-29 years of age. In total, N = 3,058 people completed parental identity in
wave 4. Of these participants, 51% identified as female and 49% identified as male. The mean
age at time of survey was 21.5 years (SD = 4.3 years; Range = 15-29 years). Thus, the GSOEP
and BHPS samples were very similar in basic demographic characteristics, whereas the HILDA
sample was substantially younger.
Correlates of Parental Identity
First, I evaluated the association between some basic demographic constructs that are
associated with parental identity. As discussed previously, very little research has evaluated this
construct and how it relates to basic demographic variables and understanding the characteristics
of one’s life that predict parental identity is important to understanding this construct and its
implications.
Gender. In order to evaluate gender differences in parental identity, I computed an
overall parental identity score for each person, which was an aggregate of all available waves of
parental identity information. The results were quite consistent across the three studies. In the
GSOEP, women reported higher parental identity than men (3.27 vs. 3.08; t (32,026) = 20.46, p
< .05 d = 0.23 (95% CI = 0.21, 0.25)). In the BHPS, women also reported higher parental
identity that men (7.95 vs. 7.24; t (20,454) = 18.76, p < .05 d = 0.26 (95% CI = 0.24, 0.29)).
Similar results were observed in the HILDA, with women rating parental identity higher than

34

men (4.21 vs. 3.46; t (3,056) = 5.66, p < .05 d = 0.20 (95% CI = 0.13, 0.28)). Taken as a whole,
there is evidence that in general, women experience higher parental identity than men. Data
suggest that this gender difference is relatively small, but consistent in both size and direction
across three distinct nationally representative samples, with Cohen ds ranging from .20 to .26.
Marital status. Figure 1 shows a simple plot of parental identity by marital status in the
GSOEP, BHPS, and HILDA. I separated within- and between-person effects of marital status in
the GSOEP and the BHPS because there were multiple waves of parental identity and people’s
marital status could change over these waves. Because the HILDA included only one wave of
parental identity, within-person effects could not be assessed in this dataset. To estimate
between-person effects in the GSOEP and BHPS, I used marital status in the first year in which
parental identity was reported. To estimate within-person effects I included marital status at each
wave for each person. I also allowed for random effects in the intercept. Dummy variables were
used to represent marital status categories, with the ‘Married’ category used as the reference in
all analyses. Results from the analysis of GSOEP data can be seen in Table 2. Married
individuals reported higher parental identity than all other groups when evaluating between- and
within-person differences. Table 2 also reports the model estimates in units of standard
deviations, for comparison purposes.
An interesting finding is that within-person changes associated with going from singlemarried are larger than corresponding between-person differences. One explanation for this may
be that overall age differences between the group of single and married people were larger than
within-person changes in age as people go from single to married. Indeed, the data showed that
married people were typically older than single people (mean age for single individuals = 26.1,
for married = 48.5). To test the possibility that this is the reason for differences in sizes of

35

within- and between-person effects I included age in the model, modeling both within- and
between- person age effects. For between-person age effects I used age at first wave in which
parental identity was reported, and which I centered around the grand mean. For within-person
age effects I used age reported at each wave and centered around age at the first wave for each
person. I also include quadratic age terms at both within-and between-person level. The
estimates of model that adjusted for age can be seen in the last column of Table 2. Although the
difference between within- and between-person effects of single vs. married status were reduced,
they still persisted suggesting that factors other than age are responsible.

36

Figure 1
Overall means in parental identity across marital status categories in the three samples.

Note: Error bars represent 95% bootstrapped confidence intervals around the mean.

37

Table 2
Results of the model evaluating associations between marital status and parental identity in the
GSOEP.
SD of PI
SD of
(adjusted
Estimate
SE
t
PI
for age)
Intercept
3.450*
0.005
643.6
Between-person
-.480*
Single
-0.132*
0.016
-8.2
-.198*
-.393*
Separated
-0.231*
0.037
-6.3
-.347*
-.282*
Divorced
-0.181*
0.025
-7.2
-.272*
-.177*
Widowed
-0.195*
0.029
-6.8
-.293*
Within-person
-1.086*
Single
-0.636*
0.015
-42.9 -1.067*
-.285*
Separated
-0.174*
0.025
-7.1
-.292*
-.394*
Divorced
-0.247*
0.020
-12.6
-.404*
-.107*
Widowed
-0.126*
0.023
-5.6
-.211*
Note: * p < .05.

Results from the analysis of BHPS data can be seen in Table 3. As in the GSOEP, the
largest differences were seen between single and married groups both at the within- and the
between-person level, with larger differences occurring at the within-person level. Adjusting for
age diminished difference between within- and between-person effects of comparison between
single and married individuals, but it did not eliminate it. Differences between other marital
status groups were typically smaller than those in the GSOEP and less consistent across withinand between-person level.
Results from the analysis of HILDA data can be seen in Table 4. As in the other two
datasets, married individuals report higher parental identity than individuals that were never
married. The married group also reported higher parental identity than those in common-law
partnerships, a difference that was seen at the within-person but not the between-person level of
the BHPS. As in the GSOEP and the BHPS, those that were divorced reported lower parental
identity than married individuals, however caution should be taken in weighing this finding too

38

heavily as only 7 people were divorced in this sample.
Table 3
Results of the model evaluating associations between marital status and parental identity in the
BHPS.
SD of PI
SD of
(adjusted
Estimate
SE
t
PI
for age)
Intercept
8.487*
0.025
342.2
Between-person
-0.633*
Never married
-0.753*
0.074
-10.2 -0.328*
0.030
Common law
0.149
0.623
0.2
0.065
-0.030
Separated
0.008
0.153
0.1
0.003
-0.135*
Divorced
-0.378*
0.097
-3.9 -0.165*
0.135*
Widowed
-0.008
0.121
-0.1
0.003
Within-person
-0.858*
Single
-1.595*
0.068
-23.5 -0.933*
-0.544*
Common law
-0.864*
0.261
-3.3 -0.506*
-0.115
Separated
-0.197
0.113
-1.8 -0.115
-0.180*
Divorced
-0.252*
0.080
-3.2 -0.147*
-0.072
Widowed
-0.069
0.102
-0.7 -0.040
Note: * p < .05.

Table 4
Results of the model evaluating associations between marital status and parental identity in the
HILDA.
SD of PI
SD of
(adjusted
Estimate
SE
t
PI
for age)
Intercept
8.010*
0.159
50.4
-1.197*
Never married
-5.462*
0.173
-31.6 -1.475*
-0.508*
Common law
-2.265*
0.207
-11.0 -0.612*
-0.829*
Separated
-2.972
0.632
-4.7 -0.803*
-1.092*
Divorced
-3.868
1.189
-3.3 -1.045*
Note: * p < .05.

When all three sets of results are taken as a whole, we see consistent evidence that
marriage is associated with higher parental identity than other marital status groups, particularly
in comparisons to singlehood and divorce. Moreover, these effects persist even when age is
included in each model to account for the possibility of these effects being the result of age

39

related changes, suggesting that there are factors other than age that are responsible for the
observed differences in parental identity across marital status categories.
Education. Figure 2 shows a simple plot of parental identity by education level in the
GSOEP, BHPS, and HILDA. Because the HILDA included only one wave of parental identity,
within-person effects of education could not be assessed in this dataset, but was assessed in the
GSOEP and BHPS. To evaluate between-person effects in the GSOEP and BHPS, I used
education level in the first year of participation. In the final model I controlled for age in the
same way as described in the marital status analyses section. Education was operationalized
using a different categorical variable (see Figure 2) in each dataset, however the order in which
they are presented can be interpreted to range from less education to more education. As a result,
direct comparisons of these results cannot be made across samples. In each dataset, the most
common educational level completed was used as a reference category. For the GSOEP, the
reference category was ‘Middle vocational’. In the BHPS ‘Low secondary-vocational’ was used.
In the HILDA, ‘Year 11 or below’ was used as the reference category. Results from the analysis
of GSOEP data are presented in Table 5, BHPS results are shown in Table 6, and HILDA results
are shown in Table 7.

40

Figure 2
Overall means in parental identity across educational achievement categories in the three
samples.

Note: Error bars represent 95% bootstrapped confidence intervals around the mean.

41

Table 5
Results of the model evaluating associations between education level and parental identity in the
GSOEP.
SD of PI
(adjusted
Estimate
SE
t
SD of PI for age)
Intercept
3.213*
0.006 496.9
Between-person
In school
-0.350*
0.032 -10.9
-0.526*
-0.146*
No education completed
-0.025
0.051
-0.5
-0.038
0.087
General elementary
-0.006
0.018
-0.4
-0.009
0.117*
Vocational plus abi
-0.056
0.033
-1.7
-0.084
-0.062
Higher vocational
0.013
0.028
0.5
0.020
-0.038
Higher education
-0.054*
0.026
-2.0
-0.081*
-0.131*
Within-person
In school
-0.189*
0.030
-6.2
-0.317*
-0.216*
No education completed
-0.174*
0.045
-3.9
-0.292*
-0.329*
General elementary
-0.118*
0.015
-7.8
-0.198*
-0.221*
Vocational plus abi
0.057*
0.027
2.1
0.096*
0.096*
Higher vocational
0.090*
0.022
4.1
0.151*
0.151*
Higher education
0.157*
0.024
6.6
0.263*
0.237*
Note: * p < .05

Table 6
Results of the model evaluating associations between education level and parental identity in the
BHPS.
SD of PI
SD of (adjusted
Estimate
SE
t
PI
for age)
Intercept
7.643*
0.034 224.74
Between-person
No education completed
-0.194
0.241
-0.80 -0.085
0.078
Primary
0.514*
0.193
2.67 0.224*
0.090
Low secondary
0.715
0.698
1.02 0.312
0.101
Hisec-mivoc
-0.428*
0.135
-3.18 -0.187*
-0.111
Higher voc
0.110
0.181
0.61 0.048
-0.088
First degree
-0.207
0.151
-1.37 -0.090
-0.143*
Higher degree
-0.548*
0.270
-2.03 -0.239*
-0.278*
Within-person
No education completed
-0.478*
0.220
-2.17 -0.280*
-0.225
Primary
-0.218
0.189
-1.15 -0.128
-0.112
Low secondary
-0.474
0.687
-0.69 -0.277
-0.081
Hisec-mivoc
-0.036
0.126
-0.29 -0.021
0.019

42

Table 6 (cont’d)
Higher voc
First degree
Higher degree
Note: * p < .05

-0.017
0.064
0.348

0.176
0.136
0.239

-0.10 -0.010
0.47 0.037
1.46 0.204

0.026
0.035
0.122

Table 7
Results of the model evaluating associations between education level and parental identity in the
HILDA.
SD of PI
SD of (adjusted
Estimate
SE
t
PI
for age)
Intercept
2.937*
0.107
27.42
Year 12
0.853*
0.163
5.22 0.230*
-0.085*
Cert not defined
4.813*
1.809
2.66 1.300*
1.044*
Cert I/II
1.686*
0.549
3.07 0.455*
0.139
Cert III/IV
2.151*
0.204
10.56 0.581*
0.026
Diploma
2.014*
0.321
6.27 0.544*
-0.176*
Bachelors/honours
1.597*
0.216
7.40 0.431*
-0.272*
Graduate
2.234*
0.538
4.16 0.603*
-0.263
Postgraduate
1.382
0.777
1.78 0.373
-0.527*
Note: * p < .05

Taken together, these analyses do not offer strong evidence that education level is
associated with parental identity. In the GSOEP, there appears to be a slight within-person trend
toward increased parental identity at as people move towards higher levels of education (in
comparison to middle vocational as a reference point). This trend persists even when the effects
of age are taken into account. However, this pattern was not observed in the BHPS – once age
was taken into account, none of the within-person effects reached significance. Results were also
inconsistent in the HILDA, and initial analysis suggests that people with more than 11 years of
education have higher parental identity than those with fewer years, but after controlling for age
the direction of these effects switches for those with higher education levels). Indeed, across
three datasets an inconsistent picture emerges regarding the association between education and

43

parental identity, and further work would be needed before conclusions could be made about an
association between these constructs.
Income. Because household income range was much larger than that of parental identity
I transformed the income variable by dividing it by 10,000 in each of the datasets. I then created
a model that evaluated between- and within-person effects of income on parental identity. To
model between-person effects I first took the mean of income across waves for each person and
centered these scores about the grand mean. For within-person effects I used reported income at
each wave centered about the individual’s mean taken across waves. Thus, the intercept in this
model can be interpreted as average parental identity for a person of average income at the time
that their income is at their average. Between-person effects can be interpreted as difference in
parental identity between individuals who differ in income by 10,000 monetary units (Euros in
the GSOEP, British pounds in the BHPS, Australian dollars in the HILDA). Within-person
effects can be interpreted as changes in parental identity for each 10,000 monetary units change
in income. Standardized estimates were obtained by using appropriate person-level standard
deviations (see Table 1) and are conceptually similar to correlations. Model results evaluating the
association between income and parental identity in the GSOEP and the BHPS are shown in
Table 8. Because the HILDA did not have a longitudinal component I used the Pearson
correlation between income and parental identity, r = -.063, p < .05. Because of the skewed
nature of income I evaluated both raw income scores and log transformed scores in these models.
Results showed minimal differences when raw scores or transformed scores were used, so
estimates using raw scores are reported for ease of interpretation. Overall, across all three
national samples, results indicated that association between income and parental identity are
extremely low (i.e., smaller than .1). Thus, there is no strong evidence here that income is

44

associated with parental identity.
Table 8
Model results predicting parental identity from income in the GSOEP and the BHPS.
GSOEP
BHPS
Std.
Std.
Estimate
SE
t
estimate
Estimate SE
t
estimate
Intercept
3.189* 0.004
712.9
7.643 * 0.019 407.9
Between-person
0.001
0.001
0.7
.008
0.022* 0.008
2.6 .015*
Within-person
0.002
0.001
1.3
.009
0.025* 0.007
3.6 .028*
Note: * p < .05.

Number of children. Table 9 displays number of participants reporting each number of
children at each wave where parental identity is assessed in the BHPS and HILDA.1 For BHPS I
used number of children in the first wave of participation to model between-person effects, and
the reported number of children at each wave for within-person effects. These variables were not
centered because the natural zero in this case lends to an interpretable intercept (i.e., parental
identity for people with no children). Model estimates are differences in parental identity that are
observed for each additional child. Results of the model evaluating the association between
number of children and parental identity are shown in Tables 10 and 11. Standardized estimates
were obtained by dividing the model estimate by the within- or between-person standard
deviation of parental identity and it represents the difference in parental identity for each
additional child in units of standard deviations. Plot of parental identity vs. number of children
also suggested a quadratic trend (see Figure 3), leading me to also evaluate a quadratic model.
Results in the BHPS revealed significant linear and quadratic effects at both the within-person
and between-person levels (where number of children at the first wave in the study was used as
the between-person variable) indicating that parental identity increases as individuals go from
childlessness to one child, and levels off once people have more than two children. Inspection of
1

Wave-to-wave information about number of children was not available in the GSOEP.

45

Figure 3 shows that the quadratic model fits the data well, particularly where more data are
available (i.e., number of children <6).
Table 9
Number of people reporting each number of children in the BHPS and the HILDA.
0
BHPS
Wave 8
Wave 13
Wave 18
HILDA
Wave 4

1

3

4

5

6

7

8

7,464 1,263 1,172
9,043 1,500 1,499
9,776 1,636 1,451

458
450
463

92
83
86

13
11
22

12
6
1

4
0
2

2
0
2

2,138

29

8

5

1

1

0

231

2

133

Figure 3
Plot of parental identity vs. number of children in the BHPS and the HILDA.

Notes: Points represent raw means obtained from the data. Error bars represent 95%
bootstrapped confidence intervals around the mean. Solid gray line shows model estimated
parental identity from the quadratic model.

46

Table 10
Model results evaluating association between number of kids and parental identity in the BHPS.
Linear
Quadratic
SD of
SD of
Estimate
SE
t
PI
Estimate
SE
t
PI
Intercept
7.240* 0.020 354.4
7.162* 0.021 345.0
Between-person
Number of kids
0.394* 0.026
14.9 0.172*
0.770* 0.058
13.3 0.336*
Number of
kids2
-0.131* 0.018
-7.2 -0.057*
Within-person
Number of kids
0.437* 0.021
20.7 0.256*
0.794* 0.045
17.6 0.465*
Number of
kids2
-0.126* 0.014
-8.9 -0.074*
Note: * p < .05.

Table 11
Model results evaluating association between number of kids and parental identity in the
HILDA.
Linear
Quadratic
SD of
SD of
Estimate
SE
t
PI
Estimate
SE
t
PI
Intercept
3.585* 0.075
47.9
3.446* 0.074
46.4
Number of kids
1.619* 0.102
15.8 0.437*
3.595* 0.206
17.4 0.971*
2
Number of kids
-0.689* 0.063 -11.0 -0.186*
Note: * p < .05.

Analyses of the HILDA data also revealed significant linear and quadratic effects at the
between-person level (estimates of within-person effects were not possible given the lack of
longitudinal parental identity data), indicating that parental identity was positively associated
with number of children, however the estimates of quadratic effects in the HILDA is less
meaningful in this analysis due to the fact that very few participants had more than two children
in this sample (see Table 8). Overall, we see clear, consistent evidence in both datasets that those
with children report higher parental identity than those without children, and there is some
evidence that parental identity levels off once individuals have multiple children.

47

Table 12
Correlations of parental identity and values and attitudes in the BHPS.

Pre-school child is likely to suffer if his or her mother works.
All in all, family life suffers when the woman has a full-time job.
A woman and her family would all be happier if she goes out to
work.
Both the husband and wife should contribute to the household
income.
Having a fulltime job is the best way for a woman to be an
independent person.
A husband’s job is to earn money; a wife’s job is to look after home
and family.
Children need a father to be as closely involved in their upbringing
as the mother.
Employers should make special arrangements to help mothers
combine jobs and childcare.
A single parent can bring up children as well as a couple.
Adult children have an obligation to look after their elderly parents.
It is better to divorce than to continue an unhappy marriage.
It is alright for people to live together even if they have no interest in
considering marriage.
When there are children in the family, parents should stay together
even if they don’t get along.
It makes no difference to children whether their parents are married
to each other or just living together.
Homosexual relationships are always wrong.
Note: * p < .05.

Wave
8

Wave
13

Wave
18

-

.16*
.07*

-

-

-.04*

-

-

-.03*

-

-

.01

-

-

.11*

-

-

.10*

-

.15*
-.06*

-.07*
-.21*
-

.14*
-.11*

.01

-

-.01

.18*

-

.23*

-.03*
.17*

-

-.05*
.21*

Attitudes and values. Measures of attitudes and values were available in the BHPS and
the HILDA. However, in the BHPS not all variables were assessed at each wave. To evaluate
relationships with parental identity I used measures that were available in the same waves at
which parental identity was assessed. Namely, some of the variables were assessed at waves 8
and 18, and some at wave 13 (see Table 12). As Table 12 shows, correlations tended to be close
to zero and were greater than .2 in magnitude in only 3 cases (single parents good, parents stay
together, gay wrong). However, there was a general pattern of parental identity being positively
associated with endorsement of some items that represent stereotypically ‘traditional’ values

48

(e.g., Pre-school child is likely to suffer if his or her mother works) and negatively associated
with less traditional values (e.g., A single parent can bring up children as well as a couple),
although these correlations were small in magnitude.
Another relevant item that was available during all three waves at which parental identity
was assessed in the BHPS asked parents “Who is mainly responsible for looking after the
child(ren)?”. Response options were “Mainly you”, “Mainly your husband/wife/partner”,
“Jointly with your husband/wife/partner”, or “Someone else”. To evaluate how parental identity
was related to provision of childcare I compared the means of parental identity across the
response categories. Table 13 shows these means and the overall ANOVA tests that showed that
means differed across the response categories. I followed these analyses with multiple t-tests
comparing each of the response categories with each other and applied the Bonferroni correction
to account for multiple comparisons. As the table shows, the main consistent finding was that
parents who are mainly responsible for childcare reported higher parental identity than parents
whose partners primarily care for the children.
Table 13
Means of parental identity across responses to the question of who is primarily responsible for
care of the child(ren).
Wave
Wave
Wave
8
13
18
Means
Mainly respondent
9.18 a
9.15 a
9.43 a
Mainly partner
8.77 b
8.84 b
9.02 b
b
a, b
Joint with partner
8.83
8.98
9.29 a
Someone else
8.24 b
8.28 b
8.66 b
Overall ANOVA test
F
10.04*
7.96*
15.12*
df
3, 2,250 3, 2,670 3, 2,753
Notes: * p < .05. Same superscripts indicate no significant difference between groups based on ttests and after applying Bonferroni correction for multiple comparisons.

In the HILDA, attitudes and values items were assessed at waves 5 and 8. I used the

49

average score taken across these two waves and correlated them with parental identity measured
at wave 4. These correlations are shown in Table 14, as are the test-retest correlations between
the attitudes and values items at waves 5 and 8. As in the BHPS, correlations tended to be close
to zero, with only two items exceeding .2 in magnitude ( “Whatever career a woman may have,
her most important role in life is still that of being a mother” and “Whatever career a man may
have, his most important role in life is still that of being a father.”). However, taken together,
analyses of two large datasets reveal that parental identity is not strongly related to measures of
attitudes and values that concern family issues.
Table 14
Correlations of parental identity and values and attitudes in the HILDA.
r

Whatever career a woman may have, her most important role in life is still
that of being a mother
Whatever career a man may have, his most important role in life is still that
of being a father
Mothers who don’t really need the money shouldn't work
As long as the care is good, it is fine for children under 3 years of age to be
placed in child care all day for 5 days a week
A working mother can establish just as good a relationship with her
children as a mother who does not work for pay
It is better for everyone involved if the man earns the money and the
woman takes care of the home and children
If parents divorce it is usually better for the child to stay with the mother
than with the father
A working father can establish just as good a relationship with his children
as a father who does not work for pay
Many working mothers seem to care more about being successful at work
than meeting the needs of their children
A pre-school child is likely to suffer if his/her mother works full-time
Many working fathers seem to care more about being successful at work
than meeting the needs of their children
If both partners in a couple work, they should share equally in the
housework and care of children
A father should be as heavily involved in the care of his children as the
mother
Children often suffer because their fathers concentrate too much on their
work

50

Testretest

.20*

.45*

.20*
.13*

.43*
.42*

-.09*

.42*

-.08*

.41*

.07*

.52*

.05*

.41*

-.04*

.34*

.04*
.03

.34*
.41*

-.03

.31*

.02

.39*

-.02

.36*

.02

.38*

Table 14 (cont’d)
On the whole, men make better political leaders than women do
-.01
.50*
It is not good for a relationship if the woman earns more than the man
-.01
.35*
Children do just as well if the mother earns the money and the father cares
for the home and the children
.00
.44*
Notes: * p < .05. Test-retest correlations are between attitudes and values items at waves 5 and 8.

Personality. Correlations between parental identity and measures of Big Five Personality
traits and parental identity in the GSOEP, BHPS and HILDA are shown in Table 15. In the
GSOEP, personality was measured at waves 22 & 26. In the BHPS, it was assessed at wave 15,
and in the HILDA at wave 5. To simplify presentation of results I averaged all available waves of
parental identity, averaged all available waves (if there were multiple waves available) of
personality, and correlated these two scores. The correlations are shown in Table 15.
Table 15
Correlations between parental identity (averaged across waves) and the Big Five (averaged
across waves when assessed in more than one wave).
GSOEP
BHPS
HILDA
Neuroticism
.050*
.015
-.012
Extraversion
.053*
.037*
-.013
Openness to experience
.017*
-.032*
-.093*
Agreeableness
.148*
.119*
.078*
Conscientiousness
.162*
.100*
.131*
Note: * p < .05.

The results were fairly consistent across datasets. Correlations were close to zero between
parental identity and Neuroticism, Extraversion, and Openness, and somewhat higher, but still
smaller than .2 for Agreeableness and Conscientiousness in all three samples. Overall, results do
not suggest robust evidence that parental identity is strongly linked to the Big Five traits.
Life satisfaction. Both life satisfaction and parental identity were assessed longitudinally
in the GSOEP and the BHPS, and I was able to use the model that separated within- and
between-person associations between the two variables in these two datasets. The models

51

predicted life satisfaction from parental identity. At the between-person level I first took the
mean of parental identity for each person across all available waves and centered this score about
the grand mean. At the within-person level I used parental identity scores at each wave and
centered them about each person’s mean parental identity score. I also allowed for random
variation in the intercept in both datasets. Table 16 shows the results of these models. As
specified, the intercept can be interpreted as life satisfaction level for people of average parental
identity level, and regression coefficients can be interpreted as change in life satisfaction for each
unit change in parental identity. I also provide standardized estimates that were obtained by using
appropriate standardized deviations of both variables, and can be interpreted as a correlation (i.e.,
change in life satisfaction in units of standard deviations for each standard deviation change in
parental identity). The results showed that both between-person and within-person associations
between parental identity and life satisfaction are significant, but relatively small in the GSOEP
and BHPS. Further, in both GSOEP and BHPS data, the association between parental identity
and life satisfaction is slightly higher at the between-person level than at the within-person level.
Table 16
Model results predicting life satisfaction from parental identity in the GSOEP and the BHPS.
GSOEP
BHPS
Std.
Std.
Estimate
SE
t
estimate
Estimate SE
t
estimate
Intercept
6.929* 0.009
797.5
5.262 * 0.008 650.3
Between-person
0.168* 0.011
15.5 0.093*
0.031* 0.003 10.2 0.080*
Within-person
0.108* 0.013
8.5 0.048*
0.016* 0.004
3.7 0.030*
Note: * p < .05.

In the HILDA, life satisfaction was also assessed in the wave in which parental identity
was assessed and so I used the Pearson correlation, which was r = .041 (p < .05). Overall, these
results indicate that there is a consistent, but weak positive association between parental identity
and life satisfaction. Life satisfaction tends to be higher in people with higher average levels of

52

parental identity and appears to increase as individuals’ experience parental identity increases.
Parental Identity and its Pattern of Change across the Lifespan
One of the central goals of this study was to gain a better understanding of the nature of
parental identity itself, the degree of stability and change of parental identity over the life course.
To gain a better understanding of this, I first evaluated the interwave correlations of parental
identity within the GSOEP and BHPS2. Correlations across time within each dataset are shown in
Tables 17 and 18. As these tables show, test-retest correlations across time are relatively strong
across datasets, and range between .41-.66. Notably, the retest correlations remain above .40
even over relatively long periods of time across each dataset (e.g., r = .61 across a 10 year retest
interval in the BHPS). These correlations suggest that there is stability across time in individuals’
reports of parental identity across time, and that stability of parental identity across time may be
comparable with established similar measures of other important psychological constructs.
Table 17
Test-retest correlations of parental identity across waves in the GSOEP.
Wave 7 Wave 9 Wave 12 Wave 21
Wave 9
.61*
Wave 12
.54*
.58*
Wave 21
.45*
.46*
.50*
Wave 25
.41*
.44*
.46*
.61*
Note: * p < .05.

Table 18
Test-retest correlations of parental identity across waves in the BHPS.
Wave 8 Wave 13 Wave 18
Wave 13
.64*
Wave 18
.61*
.66*
Note: * p < .05.
Next, I evaluated the trajectory of parental identity across the lifespan separately in the
GSOEP, BHPS and HILDA. The aim here was to gain a better understanding of how parental
2

HILDA was not included here because this dataset only included parental identity in a single wave.

53

identity changes across the lifespan. First, I plotted the average level of parental identity across
age in each sample (see Figure 4). Visual inspection of the plots suggests that parental identity
follows a curvilinear (quadratic) trend in the GSOEP and the BHPS. I proceeded with attempts to
fit a model separating the within- and between-effects for age (including quadratic age terms) in
the GSOEP and BHPS. In these datasets, for the between-person variables I used mean age for
each person that was also grand-mean centered. For the within-person variables I used age at
each wave centered around personal mean. I also divided age by 10 because regression
coefficients were otherwise too small, and so the unit of age in the regression was a decade.
Thus, the regression included a grand-mean centered personal mean (and its quadratic term), and
wave-specific age centered around personal mean (and its quadratic term) as well as the random
intercept to allow for any additional variation in parental identity across individuals. As such, the
intercept in these analyses can be interpreted as parental identity for a person of average age, the
between-person regression estimate is the difference in parental identity between people who are
10 years apart in age, and a within-person estimate is the change in parental identity for each
decade that a person gets older. The standardized estimates are differences in parental identity in
units of standard deviation for each 10-year change in age. These results can be seen in Table 19
(for GSOEP) and Table 20 (for BHPS). Table 21 shows the model results for HILDA, which
included only one wave of parental identity, and thus did not include within-person effects.
Rather, the regression included a centered age variable and its quadratic term.

54

Figure 4
Parental identity across the lifespan in the three datasets.

Notes: Points are raw means obtained from the data. Gray lines are model predictions of parental
identity from the quadratic models for GSOEP and BHPS and the linear model for HILDA.

55

Table 19
Model results for the association between age and parental identity in the GSOEP.
Linear
Quadratic
Estimate
SE
t
SD PI
Estimate
SE
t
Intercept
3.187* 0.004 716.3
3.335* 0.006 524.7
Between-person
Age
0.056* 0.003
22.1 0.084*
0.084* 0.003
32.3
2
Age
-0.046* 0.001 -34.1
Within-person
Age
0.020* 0.006
3.3 0.034*
0.025* 0.006
3.9
Age2
-0.032* 0.012
-2.7
Note: * p < .05.

Table 20
Model results for the association between age and parental identity in the BHPS.
Linear
Quadratic
SD of
Estimate
SE
t
PI
Estimate
SE
t
Intercept
7.635* 0.018 413.2
7.961* 0.027 291.3
Between-person
Age
0.239*
0.01
24.9 0.104*
0.309* 0.010
30.0
Age2
-0.091* 0.005 -18.4
Within-person
Age
0.433* 0.029
14.8 0.253*
0.433* 0.029
14.8
Age2
0.132
0.105
1.3
Note: * p < .05.

Table 21
Model results for the association between age and parental identity in the HILDA.
Linear
Quadratic
Estimate
SE
t
SD PI
Estimate
SE
t
Intercept
3.843* 0.061
63.5
3.943* 0.091
43.3
Age
3.669* 0.140
26.2 0.991*
3.702* 0.142
26.1
Age2
-0.535
0.366
-1.5
Note: * p < .05.

SD PI

0.126*
-0.069*
0.042*
-0.054*

SD of
PI

0.135*
-0.040*
0.253*
0.077

SD PI
1.000*
-0.145

Overall, the plots and the models from three distinct national samples show marked
consistencies in the shape of the trajectory and pattern of change that parental identity takes over
the life span. Generally, parental identity appears to, on average, rise steadily from adolescence
into early adulthood, leveling off when individuals move into their 30s and remains relatively

56

stable into older adulthood.3 Results of the quadratic model in the GSOEP suggest that every 10
years of change in age is associated with a between-person linear change of 0.126 SD in parental
identity. I observed a similar between-person linear effect in the BHPS, where 10 years of change
in age was associated with 0.135 SD change in parental identity. The between-person
standardized quadratic effects were -0.069 GSOEP and -0.040 in the BHPS, indicating a similar,
significant curvilinear effect in both national samples.
Model results also suggest that there are significant within-person linear effects in both
the GSOEP and BHPS, but there is divergence in the size of the effects. In the German sample, a
10 year increase in age is associated with 0.042 SD within-person change in parental identity. In
contrast, the BHPS linear within-person effect is higher, and suggests that a 10-year increase in
age is associated with 0.253 SD within-person change in parental identity. The within-person
quadratic effect in the GSOEP was significant (.042), but the quadratic term for within-person
change in the BHPS was similar in magnitude and direction to the within-person quadratic effect
from the GSOEP, but failed to reach significance. One possible explanation is that in the BHPS
there were less waves available than in the GSOEP (3 vs. 5) and thus the model may have had
more difficulty estimating more complex curvilinear trends from more sparse data that was
available in the BHPS.
Results from the HILDA show a strong linear effect, indicating that a 10 year increase in
age is associated with a whole standard deviation increase in parental identity. Results indicate
that there is no quadratic effect in the HILDA. A likely reason for this is that the HILDA
surveyed only people 15 to 29 years of age, during which time there is a substantial linear

3

Although there appear to be larger changes in the HILDA across age than in the the BHPS and GSOEP during the
comparable age groups in the latter datasets, the between-person SD in parental identity in the HILDA is larger than
in the GSOEP and BHPS. As such, age related changes during the 15-29 year age range in each sample are similar
when evaluated in units of SD during this period.

57

increase in parental identity that is evident in both the GSOEP and the BHPS. Thus, parental
identity change during this time period can be captured with a linear model.
In sum, parental identity appears to show substantial growth during early adulthood,
tapering off in the early 30s. This change was evident in all three datasets. After the age of 30,
parental identity appears to remain fairly stable on average across the lifespan. This is consistent
with identity theories (e.g., Erikson, 1968; Marcia, 1966; Phinney, 1990), which note that
adolescence and young adulthood are the periods during which most of identity development
occurs.
Parental Identity and Response to the Birth of a Child
A theoretically important life event for parental identity is birth of the first child. Here I
present the analyses of trajectories of parental identity during this time in two steps. First, I
discuss trends in parental identity change during the period before, during and after the
experience of birth of a child across the GSOEP and BHPS. These analyses will allow one to
evaluate the degree to which parental identity changes in response to the birth of a child and
evaluate the replicability of these effects across two national samples. Second, I present results
on analyses evaluating the potential moderating role of gender, age, marital status and income on
individuals’ response to childbirth.
In both the GSOEP and BHPS, I modeled response to childbirth using a reactionadaptation model. As described earlier, the basic model included an intercept (interpreted as level
of parental identity before childbirth), a reaction parameter (change in parental identity in the
years surrounding childbirth), and an adaptation parameter (change in parental identity in the
years that followed). I also allowed for random variation in all three parameters in the GSOEP. In
the BHPS, it was not possible to allow for random variation in all three parameters, likely

58

because the data in this dataset were sparse, so I allowed for random variation in baseline and
adaptation parameters.
Each model used data for individuals who had no children at the beginning of the study,
but had their first child at some point of the study. Further, I only included participants who had
at least one wave of parental identity data before the birth of their child, and one wave after. In
total, the final sample in the GSOEP was 1,611 people (54% women) and 895 people (57%
women) in the BHPS. Model results for GSOEP and BHPS are shown in Table 22. Figure 5
displays parental identity plots as a function of years from the birth of a child in the GSOEP and
BHPS for the overall sample and with gender, age, marital status and income as moderators. As
the table and the top panel of the figure show, parental identity in both datasets increased during
the reaction period by 0.58 points (0.97 within-person standard deviations) in the GSOEP, and by
1.88 points (1.10 within-person standard deviations) in the BHPS. Rather than returning to
baseline, parental identity remained higher than the pre-childbirth levels in the adaptation period,
by 0.82 points (1.38 standard deviations) in the GSOEP and 2.35 points (1.38 standard
deviations) in the BHPS. Thus, across both datasets I found a consistent increase in parental
identity that occurred around the time of birth of the first child, and that remained even several
years afterwards.
Table 22
Results of reaction-adaptation models in the GSOEP and the BHPS.
GSOEP
BHPS
Estimate SE
t
Estimate SE
t
Basic model
Baseline
2.89* 0.02 138.5
6.95* 0.07 93.1
Reaction
0.58* 0.03 21.9
1.88* 0.09 20.0
Adaptation
0.82* 0.02 36.1
2.35* 0.09 27.4
Gender as moderator
Baseline
Estimate
Gender

2.89* 0.02 139.1
-0.09* 0.02
-4.3

59

6.93* 0.07 92.5
-0.20* 0.07 -2.7

Table 22 (cont’d)
Reaction
Estimate
Gender
Adaptation
Estimate
Gender
Age as moderator
Baseline
Estimate
Age
Age2
Reaction
Estimate
Age
Age2
Adaptation
Estimate
Age
Age2

0.57* 0.03
0.00 0.03

21.6
-0.1

1.89* 0.09 20.0
0.11 0.09 1.1

0.82* 0.02
0.02 0.02

36.1
0.9

2.37* 0.09 27.4
0.15 0.09 1.7

2.89* 0.02 119.7
-0.13* 0.05
-2.9
0.02 0.05
0.3

7.08* 0.09 81.2
0.05 0.14 0.3
-0.31* 0.14 -2.2

0.62* 0.03
0.03 0.06
-0.18* 0.07

19.6
0.6
-2.5

1.93* 0.11 17.4
0.57* 0.18 3.3
-0.22 0.19 -1.1

0.83* 0.03
0.09 0.05
-0.05 0.06

31.4
1.7
-0.9

2.28* 0.10 22.6
0.05 0.16 0.4
0.18 0.16 1.1

0.03
0.04

80.6
5.5

6.43* 0.12 53.2
0.84* 0.15 5.5

0.04
0.05

12.5
0.9

1.71* 0.16 10.8
0.25 0.20 1.3

0.04
0.05

24.6
-3.7

2.71* 0.14 19.1
-0.57* 0.18 -3.2

2.89* 0.02 138.3
0.00 0.01
0.5

6.96* 0.08 92.7
0.03 0.03 0.8

Marital status as moderator
Baseline
Estimate
2.74*
Married
0.23*
Reaction
Estimate
0.55*
Married
0.05
Adaptation
Estimate
0.93*
Married
-0.18*
Income as moderator
Baseline
Estimate
Income
Reaction
Estimate
Income
Adaptation
Estimate
Income
Note: *p < .05.

0.58* 0.03
-0.01 0.01

21.8
-1.0

1.87* 0.09 19.8
0.08* 0.04 2.1

0.82* 0.02
0.01 0.01

36.1
1.3

2.36* 0.09 27.3
0.02 0.04 0.5

60

Figure 5
Model predicted trajectories of the basic reaction-adaptation model and the models that included
moderators (gender, age, marital status, and income) for the GSOEP and the BHPS.

Notes: For models moderated by age, average trajectories are estimates for those of average age,
and younger/older are estimates for people 1SD lower/higher in age than the average. For models
that include income as a moderator average trajectories are estimates for people with average
household income, and lower/higher trajectories are for people with household incomes that are
1SD lower/higher than the average.

61

Gender as a moderator. Gender was coded -1 for women and 1 for men. When gender
is included as a moderator in each model, results show that although there is a gender difference
in parental identity during the baseline period (where men report lower parental identity than
women) in the GSOEP and BHPS, the moderating effect of gender did not reach statistical
significance during the reaction and adaptation periods in either sample. Thus, gender differences
persist to some degree regardless of parental status.
Age as a moderator. Age at the year first child was born is used as the moderator. Age
was centered on the sample mean and then divided by 10 (to be consistent with the analyses
above). Results in the GSOEP show that there is a small linear effect of age at baseline,
indicating that older people having lower parental identity when they are childless than younger
childless people. In the BHPS, the corresponding linear effect at baseline was not significant.
There was, however, a small quadratic trend of age at baseline indicating that older people have
lower parental identity, and that these differences increase with age. During the reaction period
in the GSOEP, a significant quadratic effect of age emerged, suggesting that older people do not
increase as much in parental identity in the years surrounding childbirth, and that these
differences increase with age. This quadratic trend at reaction was not replicated in the BHPS,
where only the linear effect of age was significant, suggesting that older people increase more on
parental identity than younger people in the years surrounding birth. As for the adaptation period,
these analyses indicate that there are no moderating effects of age in the adaptation period for
either GSOEP or BHPS participants, so the age differences seen in the reaction period do not
persist long-term in the years that follow birth. Figure 5 visualizes these age effects in each
dataset with three lines, one for people of average age, one for people 1 SD older than the mean,
and one for people with 1 SD lower than the mean. As can be seen, any age differences during

62

childless years are quite small, and during parenthood become not practically significant.
Marital status as a moderator. Majority of people in these samples were either married
(GSOEP: 62%, BHPS: 63%) or never married (GSOEP and BHPS: 33%) in the year of
childbirth. Only 1% were separated in both datasets, 3% in GSOEP and 2% in BHPS were
divorced, and 1% of individuals in the BHPS were in a common-law partnership. Because the
initial analyses above indicated that the married group tended to report higher parental identity
than other groups, I combined never married, separated, divorced, and common-law partnership
groups and contrasted this “not married” group (coded as 0) with the group of married
individuals (coded as 1). Model results for both GSOEP and BHPS are shown in Table 23.
Results were remarkably consistent across both the GSOEP and BHPS, and indicated that people
who were married in the year in which they had a child reported higher parental identity at
baseline in both samples. There were no differences between married and not married groups in
changes in parental identity from baseline during the reaction period. However, in both the
GSOEP and BHPS, the differences between the two groups were largely reduced in the
adaptation period because people who were not married in the year of childbirth showed larger
increases in this period from their pre-childbirth baseline levels of parental identity.
Income as a moderator. Household income reported in the year of childbirth was used
to evaluate the moderating role of income in individuals’ responses to the birth of a child.
Income was centered on the sample mean and divided by 10,000. Model results can be viewed in
Table 20. Overall, results across the GSOEP and BHPS were fairly consistent and indicated that
income differences account for little variability in how people respond to the birth of a child.
There was no differences in baseline parental identity based on income in either the GSOEP and
BHPS, but people with higher income tended to report somewhat larger increases in parental

63

identity around the time their first child was born in the BHPS. However, this effect was not
observed in the GSOEP. In the years that followed childbirth these differences were reduced, as
evidenced by the lack of moderating effects of income on change in parental identity during the
adaptation period in the GSOEP and BHPS. Figure 5 shows a plot of these income effects in the
GSOEP and BHPS. Again, each plot contains three lines: one for people of average income, one
for people with household income 1 SD lower than the mean, and one for people with household
income 1 SD higher than the mean. The figure shows visually that any differences in reaction to
childbirth between people of differing income levels are small in the BHPS and virtually nonexistent in the years that follow in both datasets.
Parental identity and normative change. As described above, it is important to
consider and account for normative changes in constructs such as parental identity when
evaluating this construct’s pattern of change and its implications across the lifespan. For
instance, the results of these analyses in the GSOEP and BHPS suggest that there are substantial
increases in the years surrounding the birth of a first child, and that these increases persist in the
years that follow birth. This pattern of change could be attributable to the experience of
childbirth, with the pattern of change resulting from something unique about having children.
However, it is also possible that these findings are attributable to normative development, and
this pattern of change would have been observed regardless of whether one experienced
childbirth or not. To account for normative change, I identified group of people who are similar
in important ways the group who does have children, but do not have children during the course
of the study using propensity score matching. This control group was matched on gender, age
(and age2), marital status, educational attainment, income, and the initial parental identity level. I
included the initial parental identity level as one of the matching variables because it may be a

64

variable that plays an important role in whether people decide to have children. Thus, I wanted
the control group to be initially as close as possible to the group of parents in their feelings of
importance of having children. I used this group, who are on average similar to those who
become parents in order to get an estimate of what normative change for this group of parents
(i.e., if they did not have children) would have been.
I modified the reaction-adaptation models above to include information about this control
group. Namely, the new models include an intercept, which is now interpreted as the initial
parental identity, a non-parent control variable (coded as 0 for parents and 1 for non-parents) that
reflects the difference in the initial parental identity levels between parents and non-parents, a
linear change parameter that captures year-to-year change in parental identity in both groups, and
reaction and adaptation parameters, coded in the way described above for parents and set to zero
for non-parents. I also included all variables used for matching as covariates except the initial
parental identity level. I excluded this variable from the covariates because it was used as an
outcome variable and its inclusion would complicate the interpretation of the regression
coefficients.
Table 23
Results of the reaction-adaptation model for parental identity including control groups in the
GSOEP and the BHPS.
BHPS
GSOEP
Estimate SE
t
Estimate
SE
t
First year
3.07* 0.05 57.0
7.20* 0.39 18.7
Non-parent control
-0.03 0.03 -1.0
-0.65* 0.12 -5.6
Linear change
-0.03* 0.00 -6.8
-0.14* 0.05 -2.8
Reaction
0.59* 0.03 21.1
1.96* 0.11 17.1
Adaptation
0.96* 0.03 32.0
2.56* 0.12 21.0
Note: * p < .05.

Results for these reaction-adaptation models with control groups in the GSOEP and
BHPS are shown in Table 23. Plots of each of these models in the GSOEP and BHPS are shown

65

in Figure 6. Taken together, model results in the GSOEP and the BHPS are largely consistent.
Reaction and adaptation parameters were statistically significant and positive in both samples,
suggesting that those who become parents in the GSOEP and BHPS reported higher ratings of
parental identity than non-parents after an equivalent amount of time, even after accounting for
expected normative increases in parental identity. In other words, these results suggest that
parents report higher parental identity than they would have had if they had not had children In
contrast with the GSOEP, there was a significant difference between the two groups on the initial
levels of parental identity in the BHPS (as evidenced by a significant “non-parent control”
estimate). This is notable because it suggests that the matching procedure was not completely
successful in matching the two groups in the BHPS, although matching did substantially reduce
this initial difference.
Figure 6
Predicted parental identity trajectories from reaction-adaptation models including non-parent
matched controls.

Notes: Black solid lines represent predicted parental identity trajectory for parents in the study.
Dashed line shows predicted trajectory for non-parent controls. Gray line shows predicted
trajectory that the parent group would have if they did not have any children in the study.

66

STUDY 2
Method
This study evaluated the second main goal of this project, which was answering the
question of whether parental identity moderates people’s SWB responses to having children. As
with the first study, Study 2 used data from the GSOEP, BHPS and HILDA. This data allowed
for a prospective design evaluating whether parental identity moderates the pattern of pre- to
post-childbirth SWB change in new parents.
Sample Selection
To examine the effect of childbirth on well-being, I selected a sample of individuals from
each dataset that indicated that they had no children at the beginning of their participation in the
study and had a first child at some point during the study. In order to get better estimates of
within-person changes in life satisfaction in the years surrounding childbirth, I also limited each
sample to individuals who rated their life satisfaction at least one wave before and one wave after
childbirth (following the procedures of previous studies by Yap et al., 2012 and Anusic et al.,
2014a).
Measures
Life satisfaction. In the GSOEP, life satisfaction was measured with a single item, “how
satisfied are you at present with your life as a whole?” Participants responded to this question
with a Likert scale ranging from 0 (totally unsatisfied) to 10 (totally satisfied). This item was
assessed at every wave of the GSOEP. Past research has shown that life satisfaction scores in the
GSOEP are subject to panel conditioning effects (Baird, Lucas & Donnellan, 2010). These
effects are unique to the GSOEP data and indicate that scores decline by about .03 points every
year as a function of length of time in the study. Baird et al. revealed this effect to be

67

independent of age or calendar year, instead reflecting a unique testing effect observed in the
GSEOP data. To adjust for the general changes associated with panel conditioning effects, I
followed the method of Dyrdal and Lucas (2013) and added .03 for each wave that a participant
was in the study. Past research does not indicate that there are similar panel conditioning effects
in the BHPS and HILDA (e.g., Baird et al., 2010).
In the BHPS, life satisfaction was measured with a single item that asked participants to
rate how dissatisfied or satisfied they were with their life overall on a Likert scale from 1 (not
satisfied at all) to 7 (completely satisfied). This construct was assessed from 1996 onwards,
excluding 2001. As such, my analyses included data from 12 waves of data collection, taken over
13 years (1996-2008, omitting 2001).
In the HILDA, life satisfaction was measured with the following item, “all things
considered, how satisfied are you with your life?” Respondents rated how dissatisfied or
satisfied they were with their life overall on a Likert scale from 0 (totally dissatisfied) to 10
(totally satisfied).
Parental identity. Parental identity was assessed as described in Study 1 in all three
datasets.
Analytic Approach
The main goal of this analysis is to examine the extent to which people’s parental identity
explains the variability in patterns of SWB change that are associated with the experience of
childbirth.
For the sake of consistency across datasets and interpretability of results, I used the
reaction-adaptation model to estimate the pattern of change in life satisfaction in the years
surrounding childbirth in each dataset. Although recent past research has relied on an alternative,

68

nonlinear model (e.g., Yap et al., 2012) to examine reaction and adaptation to various life events
(including childbirth) with some success, these more complex models place larger demands on
the data than simpler models of change. A major factor that affected my ability to use this model
is sample size. Modeling such nonlinear changes in a multilevel framework requires a large
amount of data to get meaningful estimates. This was less of a challenge in the GSOEP and
BHPS, but the HILDA includes fewer waves of data (10 waves) and a substantially smaller
sample than the GSOEP and BHPS. Coupled with the fact that the parental identity item was
only administered to individuals aged 15 to 29 years of age during a single wave, the HILDA
sample was much smaller than in the other datasets and a simpler model was necessary.
Indeed, the pattern of SWB change in the years surrounding the birth of a child has
already been evaluated in great detail in previous research using these datasets (GSOEP, Dyrdal
& Lucas, 2013; BHPS, Yap et al., 2012; HILDA, Anusic et al., 2014a) and the main aim of the
present analyses were the evaluate whether parental identity moderates previously established
associations between childbirth and patterns of SWB change in the years surrounding the event.
As such, I applied the reaction-adaptation model to all three datasets to estimate changes in
SWB associated with childbirth to ease interpretability and the ability to contrast results across
samples in order to best evaluate the role of parental identity in responses to the birth of a first
child.
This statistical model allowed me to evaluate two questions regarding the role of parental
identity. First, I evaluated the moderating effect of parental identity on immediate reactions to
childbirth by looking at whether the extent to which people’s SWB changes from baseline differ
as a function of their parental identity levels. Second, this model allows evaluation of whether
parental identity helps (or hinders) adaptation to childbirth by comparing change in SWB in the

69

years that follow birth between individuals reporting different levels of parental identity.
One important issue to consider when evaluating the moderating effect of parental
identity in prospective longitudinal data such as these is the operationalization of the construct
and how it changes across different periods of one’s life. For instance, one approach could be to
use average parental identity score across all available waves of data as a moderator. However,
this approach has some important limitations. First, this approach does not account for the fact
that past research has shown that major life events such as childbirth are associated with
subsequent changes in personality traits and other individual differences (Löckenhoff,
Terracciano, Patriciu, Eaton, & Costa, 2009; Mroczek & Spiro, 2003; Roberts, Caspi, & Moffit,
2003; Roberts, Walton, Bogg, & Caspi, 2006; Specht, Egloff, & Schmukle, 2011; Vaidya, Gray,
Haig, & Watson, 2002), an effect that findings from Study 1 of this paper suggest may occur
with parental identity as well. By using average parental identity (which may also include postbirth ratings of parental identity), one cannot be certain whether associations between average
parental identity and post-birth changes in SWB reflect a protective/exacerbating effect of
parental identity or simply the tendency of parents who are most affected by childbirth (in terms
of their SWB) to experience the most subsequent change in parental identity.
Another approach to operationalizing parental identity would be to limit analyses to
include only prospective measures of parental identity. In other words, to focus only on the
moderating effects of parental identity ratings measured prior to childbirth on the reaction and
adaptation to the subsequent birth. Clearly, this approach is not without its limitations, the major
limitation being the relative infrequency to which the parental identity items were assessed in
each dataset. The major implication of this for our analyses is that the subsample of people who
completed measures of parental identity prior to experiencing childbirth are likely to be much

70

smaller than they would be otherwise. This approach also makes the explicit assumption that
parental identity is a meaningful construct to individuals that do not yet have children.
Given these issues, I used two separate operationalizations of parental identity at two
relevant periods (pre-childbirth and post-childbirth) to evaluate this construct’s role in responses
to birth. Specifically, this approach involved fitting four distinct models to each dataset to
examine the main questions of the study. All analyses were conducted using the R Statistical
Software (R Development Core Team, 2010). The first model was a basic reaction-adaptation
model to estimate the basic trajectory of life satisfaction before and after childbirth. This model
was examined simply to verify what was already known from previous research about childbirth
and SWB in these datasets. The second model evaluated the moderating effects of pre-childbirth
parental identity on SWB during the baseline, reaction and adaptation periods in the years
surrounding the birth of a first child. A third model evaluated the moderating effects of postchildbirth parental identity on SWB during the baseline, reaction and adaptation periods in the
years surrounding the birth of a first child. Finally, I specified a fourth model that included both
pre-childbirth and post-childbirth parental identity simultaneously in order to test the unique
effects of parental identity before and after childbirth. In essence, this model allows one to
account for the stable “trait” aspects of parental identity and consider the unique influence of
pre-childbirth and post-childbirth parental identity on the extent that people react and adapt to the
birth of a child. In the table reported below, I show estimates from all 4 models for each dataset
(with the exception of the HILDA), but focused my reporting of the results of the fourth model
as they were the most relevant to the target question of this study. With this model, I could
evaluate the moderating effect of two operationalizations of parental identity, from two
theoretically distinct periods play a role in SWB change following birth.

71

In the HILDA, it was not possible to include the model that evaluated both parental
identity before and after childbirth as moderators because parental identity was measured in only
one wave. Thus, I report and interpret the moderating effects of pre-childbirth and post-childbirth
parental identity from estimates of the second and third models where pre-childbirth and postchildbirth parental identity were specified singly.
Finally, to examine the question of whether there are differences in the moderating effect
of pre-childbirth and post-childbirth parental identity among mothers and fathers, I evaluated a
final reaction-adaptation model where I specified moderating effects of gender, pre-childbirth
parental identity, post-childbirth parental identity and the interaction between gender and each
parental identity effect in each dataset. The critical test here was whether any moderating effects
of parental identity across the baseline, reaction and adaptation periods differ among men and
women in a consistent way across the GSOEP, BHPS and HILDA. Note that in the GSOEP and
the BHPS the gender as a moderator was included in the full model (model 4) that included both
pre-childbirth and post-childbirth parental identity. In the HILDA, that model was not available
so gender was included in the two models that tested the moderating effects of pre-childbirth and
post-childbirth parental identity separately.
Results
Sample Characteristics
GSOEP. In total, N = 3,729 people reported having their first child at some time during
the study and completed at least one wave of life satisfaction before and one wave after
childbirth. Of these participants, 57% identified as female and 43% identified as male. The mean
age at time of birth of first child was M = 28.7 years (SD = 5.4 years; Range = 18-60 years).
BHPS. In total, N = 1,518 people reported having their first child at some time during the

72

study and completed at least one wave of life satisfaction before and one wave after childbirth.
Of these participants, 57% identified as female and 43% identified as male. The mean age at time
of birth of the first child in was M = 29.3 years (SD = 6.4 years; Range = 17-56 years).
HILDA. In total, N = 1,137 people reported having their first child at some time during
the study and completed at least one wave of life satisfaction before and one wave after
childbirth. Of these participants, 51% identified as female and 49% identified as male. The mean
age at the time of birth of the first child was M = 30.2 years (SD = 7.1 years; Range = 17-84
years).
Basic Models
All model results for the GSOEP can be seen in Table 24. Model results for the BHPS are
shown in Table 25. Results for the HILDA are displayed in Table 26. All model estimates are
visually presented in Figure 7. First, I specified a basic reaction-adaptation model to evaluate
participants’ overall response to the birth of a child in each dataset. As expected, results in the
GSOEP suggest that participants react positively to childbirth, but these changes are not longlasting and people eventually adapt to baseline (see Figure 7; see also Dyrdal & Lucas, 2013).
Results in the BHPS indicate that participants react positively to childbirth, but in contrast to the
results in the GSOEP, BHPS participants report experiencing a small, but lasting negative drop
in SWB in the years that follow childbirth. Results in the HILDA suggest that SWB remained
stable from baseline into reaction period, but in the long run, childbirth is associated with longterm decline in the years that follow.

73

Figure 7
Life satisfaction trajectories for basic model and models with parental identity before and after birth of the first child as moderators in
the three datasets.

Notes: Average trajectories are representative of people with average parental identity scores before/after childbirth. Low/high
trajectories are representative of people with parental identity scores 1SD lower/higher than average.

74

Table 24
Model results for basic life satisfaction reaction-adaptation model and models with parental identity as moderators in the GSOEP.
PI before and after
Basic model
PI before childbirth
PI after childbirth
childbirth
Estimate SE
t
Estimate SE
t
Estimate SE
t
Estimate SE
t
Baseline
Estimate
7.41* 0.02 344.0
7.37* 0.03 276.1
7.40* 0.02 316.8
7.38* 0.03 250.1
PI before childbirth
0.14* 0.04
3.9
0.12* 0.04
3.1
PI after childbirth
0.30* 0.05
5.9
0.13* 0.07
2.0
Reaction
Estimate
0.24* 0.02 11.1
0.30* 0.03 11.5
0.24* 0.02 10.3
0.30* 0.03 10.6
PI before childbirth
-0.03 0.03
-0.9
-0.06 0.04
-1.6
PI after childbirth
0.12* 0.05
2.3
0.20* 0.06
3.1
Adaptation
Estimate
0.01 0.02
0.6
0.11* 0.03
3.5
0.01 0.02
0.6
0.10* 0.03
3.1
PI before childbirth
0.01 0.04
0.4
-0.04 0.04
-0.9
PI after childbirth
0.22* 0.05
4.1
0.29* 0.07
4.0
Notes: * p < .05. PI = parental identity.

75

Table 25
Model results for basic life satisfaction reaction-adaptation model and models with parental identity as moderators in the BHPS.
PI before and after
Basic model
PI before childbirth
PI after childbirth
childbirth
Estimate SE
t
Estimate SE
t
Estimate SE
t
Estimate SE
t
Baseline
Estimate
5.36* 0.03 212.5
5.35* 0.03 187.9
5.37* 0.03 204.4
5.36* 0.03 181.3
PI before childbirth
0.04* 0.01
3.6
0.04* 0.01
3.1
PI after childbirth
0.10* 0.02
5.0
0.08* 0.02
3.6
Reaction
Estimate
0.12* 0.02
5.0
0.11* 0.03
4.4
0.11* 0.02
4.6
0.11* 0.03
4.0
PI before childbirth
0.01 0.01
0.9
0.01 0.01
0.8
PI after childbirth
0.04* 0.02
2.2
0.04 0.02
1.9
Adaptation
Estimate
-0.07* 0.03
-2.5
-0.06 0.03
-1.8
-0.07* 0.03
-2.5
-0.06* 0.03
-2.0
PI before childbirth
-0.02 0.01
-1.4
-0.03* 0.01
-2.0
PI after childbirth
0.03 0.02
1.4
0.08* 0.03
2.8
Notes: * p < .05. PI = parental identity.

76

Table 26
Model results for basic life satisfaction reaction-adaptation model and models with parental identity as moderators in the HILDA.
Basic model
Estimate SE
t
Baseline
Estimate
7.95* 0.03 244.7
PI before childbirth
PI after childbirth
Reaction
Estimate
0.05 0.03
1.8
PI before childbirth
PI after childbirth
Adaptation
Estimate
-0.13* 0.04
-3.5
PI before childbirth
PI after childbirth
Notes: * p < .05. PI = parental identity.

PI before childbirth
Estimate SE
t
7.97* 0.05 160.0
0.03* 0.01
2.1
-0.01
0.00
-

PI after childbirth
Estimate SE
t
7.90* 0.14
0.15* 0.06

55.9
2.5

0.04
0.01
-

-0.2
0.2
-

0.22
0.01

0.13
0.06

1.6
0.2

-0.17* 0.06
0.00 0.02
-

-2.7
-0.1
-

-0.13
0.00

0.14
0.06

-0.9
-0.1

77

This finding is consistent with similar analyses in past studies of HILDA, which suggest
that this pattern of change could be due to the fact that people in this age range tend to show
normative declines in life satisfaction (e.g., Anusic et al., 2014a). Nevertheless, the positioning
of these within-person estimates relative to baseline are not as crucial in the current study
because the main question of interest here is not the nature of change itself, but to see whether
parental identity moderates change over time.
Moderating Effect of Pre-Birth Parental Identity
Model results in the GSOEP indicate that there is a statistically significant (but small)
effect of parental identity on baseline that persists to some degree throughout the years. In
particular, those with higher parental identity before childbirth tend to have slightly higher life
satisfaction before they become parents and in the years that follow birth (see Figure 7).
In the BHPS, results differed from those garnered in the GSOEP in important ways.
Consistent with the GSOEP, BHPS results suggest that there is a difference before people
become parents (at baseline) – those with higher parental identity appear to report higher life
satisfaction. Results suggest that this difference persists (but is not further amplified) during the
reaction period (evidenced by the non-significant moderation by parental identity in this period),
but eventually fades in the years that follow birth (evidenced by negative moderating effect of
parental identity on SWB change during adaptation). Indeed, regardless of parental identity
before childbirth, results in the BHPS suggest that everyone ends up at the same levels of SWB
in the years that follow birth, and there is no evidence for lasting boosting effect of parental
identity on life satisfaction.
Results in the HILDA4 suggest that there is a small moderating effect of parental identity

4

As noted previously in the analytic approach, it was not possible to include the model that evaluated both parental
identity before and after childbirth as moderators because parental identity was measured in only one wave in this

78

before childbirth at baseline, and this small difference persists through time. Indeed, the effect of
parental identity did not change during reaction or adaptation period, which was evidenced by
non-significant moderating effects of pre-childbirth parental identity on reaction and adaptation
parameters. This means that people who start off higher on parental identity do not increase any
more or less on life satisfaction after having a child.
Overall, there is no strong evidence that pre-childbirth parental identity plays a
meaningful role in how individuals respond to birth of a child and whether the experience affects
an individual’s SWB in the years surrounding the event. Pre-childbirth parental identity does
appear to consistently positively predict SWB differences at baseline, but there is no evidence
that pre-childbirth parental identity is associated with any lasting meaningful added boosts or
declines in SWB in the years that surround birth and the years that follow.
Moderating Effect of Post-Birth Parental Identity
In the GSOEP, there is a small effect of post-birth parental identity at baseline, indicating
that those that report increased in parental identity after childbirth report being more satisfied
with their lives before childbirth than people with lower post-birth parental identity. Larger
positive effects are seen in reaction and adaptation periods, suggesting that these differences
between participants who report higher parental identity and those who are lower in parental
identity are amplified in the years surrounding birth and the years that follow. Indeed, those who
report higher parental identity levels after becoming parents appear to show bigger increases in
life satisfaction as a result of having a child and this added boost appears to persist following
birth.
Results in the BHPS are consistent with the GSOEP. The BHPS sample also shows a

dataset. As such, unlike the GSOEP and BHPS results, these results are obtained from estimates from separate
models for before/after parental identity and must be interpreted with this in mind.

79

difference at baseline, again suggesting that people who report increased in parental identity after
childbirth report being more satisfied with their lives before childbirth. These differences persist
in the reaction phase, and then continue to be amplified in the adaptation phase. Those who
report higher parental identity after becoming parents also tend to report added increases in life
satisfaction in the years that follow birth.
HILDA data showed there is an association between post-childbirth parental identity and
baseline levels of SWB, and this difference persists throughout the reaction and adaptation
phases (see Figure 7). However, there were no significant moderating effects of post-childbirth
parental identity on the extent to which parents reacted or adapted to the birth of their first child.
Thus, people who report higher parental identity after birth of a child are also those who start off
highest in SWB, and these individuals (who are highest in SWB to begin with) maintain this
difference throughout birth and the years that follow.
Overall, there is some evidence that post-childbirth parental identity is associated with an
added, long term boost in SWB following the birth of a first child. This suggests that people who
report higher parental identity not only start out with higher life satisfaction, but also are
especially likely to report higher life satisfaction in response to the experience of birth compared
to those who are lower in parental identity. The moderating effects of post-childbirth parental
identity also appear to be somewhat robust across samples, and I observed some consistency
across distinct national samples of parents.
Does Parental Identity Matter more for Mothers or Fathers?
Model results evaluating the interactive effects of gender and parental identity on SWB
and responses to parenthood in the GSOEP, BHPS and HILDA are presented in Table 27. When
taken as a whole, results indicate that there is little evidence that the role of parental identity in

80

how people respond to the birth of a child differs for mothers and fathers. There was no
consistency in the findings across samples. For instance, in the GSOEP, men reacted less
positively to childbirth than women as evidenced by the negative moderating effect of gender on
reaction to childbirth in this dataset, but there were no gender differences any datasets during
baseline or adaptation periods. There were no significant interactions between parental identity
prior to childbirth and gender during baseline, reaction, or adaptation periods in any of the
datasets. In the BHPS, there was some evidence that parental identity after childbirth mattered
less for men overall (i.e., significant moderating effect of parental identity after childbirth on
baseline), but this did not replicate in other datasets. There were no other significant interactions
between post-childbirth parental identity and gender. As such, there is little reason to suggest that
the role that parental identity plays in responses to childbirth differs meaningfully among
mothers and fathers.

81

Table 27. Model results evaluating the role of gender and parental identity for SWB responses to the birth of a child.
GSOEP
BHPS
HILDA
Estimate SE
t
Estimate SE
t
Estimate SE
t
Estimate SE
Baseline
Estimate
7.38* 0.03 245.2
5.36* 0.03 178.0
7.98* 0.05 159.6
7.87* 0.15
Gender
0.03
0.03
0.9
0.02
0.03
0.6
-0.05
0.05 -1.0
-0.06 0.15
PI Before
0.13* 0.04
3.2
0.04* 0.01
3.0
0.03
0.01
1.9
PI Before x
Gender
0.01
0.04
0.2
0.01
0.01
0.8
0.02
0.01
1.6
PI After
0.14* 0.07
2.1
0.09* 0.02
3.7
0.19* 0.07
PI After x Gender
-0.02
0.07 -0.3
-0.05* 0.02 -2.2
0.07 0.07
Reaction
Estimate
0.30* 0.03 10.4
0.10* 0.03
3.7
-0.02
0.04 -0.4
0.20 0.14
Gender
-0.06* 0.03 -2.1
-0.03
0.03 -1.1
-0.05
0.04 -1.2
-0.05 0.14
PI Before
-0.07
0.04 -1.8
0.01
0.01
0.6
0.00
0.01
0.1
PI Before x
Gender
0.02
0.04
0.4
-0.01
0.01 -1.0
-0.01
0.01 -1.1
PI After
0.18* 0.07
2.8
0.04
0.02
1.7
0.01 0.07
PI After x Gender
0.01
0.07
0.2
0.00
0.02 -0.2
-0.01 0.07
Adaptation
Estimate
0.10* 0.03
3.2
-0.06
0.03 -1.8
-0.18* 0.06 -2.7
-0.13 0.15
Gender
0.02
0.03
0.7
0.02
0.03
0.5
0.00
0.06 -0.1
0.03 0.15
PI Before
-0.04
0.04 -0.8
-0.03* 0.01 -2.2
0.00
0.02 -0.1
PI Before x
Gender
0.00
0.04
0.1
-0.01
0.01 -1.0
0.00
0.02 -0.3
PI After
0.29
0.07
3.9
0.07* 0.03
2.5
0.01 0.07
PI After x Gender
0.06
0.07
0.8
0.00
0.03 -0.1
0.02 0.07
Notes: * < .05. PI = parental identity.

82

t
53.2
-0.4
2.8
1.1
1.4
-0.4
0.2
-0.1
-0.9
0.2
0.1
0.3

GENERAL DISCUSSION
There is a notable lack of empirical research aimed at understanding the construct of
parental identity and its theoretical underpinnings, leading to a clear lack of understanding of this
conceptually important characteristic. Past research has discussed the importance of this
construct for parental well-being and sense of self at a conceptual level (e.g., Rane & McBride,
2000), but this body of work has presented limited evidence that parental identity is associated
with important psychological outcomes. Indeed, the present work is an important contribution to
this area of research, and offers the first systematic evaluation of the trajectory of parental
identity over the life course, its patterns of change, and how it relates to important demographic
and psychological characteristics.
This work is also the first to evaluate the role of parental identity in how individuals react
and subsequently adapt to a theoretically important life event, that is, the birth of a first child.
Indeed, there is a healthy body of recent literature exploring the link between parenthood and
well-being outcomes (e.g., Nelson et al., 2013) and the moderators of these processes (e.g.,
Anusic et al., 2014a; Yap et al., 2012) and the present research adds to this literature by
introducing and evaluating a conceptually important moderator of this important research
question. Overall, these studies are valuable because they give us a better understanding of
parental identity, its causes and consequences, and the role it plays in how people respond to the
onset of parenthood. All of which are important factors to consider in future research as this area
of study moves forward.
Beyond its implications for basic research in this field, this work also has clear applied
value, and has potential to inform public policy decisions regarding support for parents in the
years surrounding the birth of a child and the years that follow. As policy makers and

83

organizational leaders in the public and private sectors move towards a greater emphasis on datadriven decision making, studies like these are vital to inform decisions regarding the most
efficient ways to allocate resources and efforts in supporting new parents. Determining what
types of interventions and policies (and what specific constructs and processes they could affect)
that would lead to the greatest returns in individuals’ adaptation to parenthood needs to be a
central goal of this area of research. Moving forward, I hope research in this area continues with
these applied goals in mind.
Correlates of Parental Identity
The results of these studies clearly indicate that parental identity is associated with
several important demographic characteristics and outcomes. First, there is a clear gender
difference in parental identity where women, on average, report modestly higher parental identity
levels than men. Further, this effect shows remarkable consistency among all three national
samples. Given the replicability of these findings across three distinct national samples in the
current study, coupled with the consistency of the present findings with those in the previous
literature (e.g., Maurer et al., 2001; Reitzes & Murtran, 1994) one can have a fair degree of
confidence that this is a reliable effect.
It is hard to say at this point what processes underlie gender differences in parental
identity, and past theories suggest that this difference could be a result of socialization, a
biological difference, or some combination of the two. For example, societal expectations
continue to prescribe the primary responsibilities of childcare to women, who continue to be
more likely to be the primary caregiver of children in western households, regardless of
employment status (Yap & Anusic, 2016). Past studies also suggest that women report more
competence and confidence in their ability to care for children (Reitzes & Mutran, 1994), which

84

is may also lead to greater parental identity. Other research also suggests that there may have
been selective pressures for female humans (among many other mammalian organisms) to
display a greater degree of parental investment than males (e.g., Trivers, 1972). Indeed, through
the course of human evolution, one way that this could have psychologically manifested itself in
people is feelings of parental identity, and future research could benefit by evaluating the link
between parental investment and parental identity.
These studies also revealed some interesting findings regarding the association between
parental identity and marital status. Perhaps the most important finding here is that people who
are married, whether one is evaluating within-person or between-person comparisons, report the
highest levels of parental identity compared to all other marital status groups. This finding is
makes theoretical sense. Although one does not necessarily need to be legally married to have
children, marriage is often considered a rite of passage in many cultures where individuals
commit to long term romantic partnership, a characteristic that often precedes having children.
Further, this major life transition often is seen as the point where you enter the life stage where it
is normative to have children, and for some people in our society, a necessary condition before it
is morally appropriate to bear children (Thornton & Young-DeMarco, 2001). For these reasons,
it makes sense that married individuals place greater importance than other groups on having
children. The causal direction, of course, is unclear in these findings, and it is possible that
marriage predicates increases in parental identity, or that individuals higher in parental identity
are more likely to seek out marriage.
Another interesting finding is that the within-person effects of going from single to
married are larger than the between-person effect across individuals who start off married vs.
start off single in both the GSOEP and BHPS. Similarly, these analyses also indicate that the

85

within-person effects of going from married to divorced are larger than the corresponding
between-person effects. One possible explanation for this finding is that there is something
unique about the transition or process involved with changing marital status (i.e., the actual
process of establishing a long term romantic partner or dissolving a marriage) that affects
parental identity and gives individuals an added boost or reduction in parental identity over and
above the change in status itself. Analyses also reveal that these marriage effects are mitigated,
but not eliminated, when age is taken into account, suggesting that this effect is not wholly
attributable to just normative effects of age in parental identity. Again, it is noteworthy to
mention that several of these effects, although small, were consistent in magnitude and direction
across datasets, suggesting a robust relationship between marital status and parental identity.
It is important to understand the interplay between marital status, parental identity, and
the underlying mechanisms through which these constructs are related. From an applied
perspective, an understanding of these mechanisms could shed some light on when and how
people ultimately decide to have children and the factors that predicate this important decision.
In this case, identifying the unique features of long-term relationship formation that could predict
parental identity, for example, may help demographers understand and develop strategies to
curtail declining birth rates in many industrialized western countries (Bongaarts, 1999). Overall,
there are many contexts where an understanding of the demographic factors that predict parental
identity, and the processes through which these constructs affect one another could have
tremendous applied value.
Another robust correlate of parental identity appears to be number of children. The most
interesting findings here are the within-person effects of number of children on parental identity
in the BHPS. These results suggest that having more children is associated with a rise in parental

86

identity, leveling off after the second child (refer to Figure 3). This suggests that not only does
individuals’ self-rated parental identity increase when individuals initially become parents, but
that there is an added boost when individuals become parents of multiple children. But once
people have multiple children, data suggest that having additional children is not associated with
additional boosts in parental identity.
These findings have clear implications on the understanding of transition to parenthood
and the psychological implications of having additional children. For instance, this is the first
study to evaluate the question of how people’s parental identity and the self-rated importance of
having children changes as a function of the number of children born to the individual. Indeed, a
better understanding of the processes through which individuals transition to parenthood, and
how these transitional processes differ depending on the number of children an individual
already has could be of great value to policy makers interested in supporting parents and families
through these major life transitions. To give a concrete example, there have been large scale
campaigns in the United States targeting fathers to promote greater involvement in their
children’s lives (Levere, 2010). These ad campaigns were, in essence, publically sponsored
efforts (by the United States Department of Health and Human Services) to actively boost
fathers’ parental identity and encourage men to increase the value and importance they place in
their role as a parent as a means to improve the lives of families across the nation. Knowledge of
the dynamics of parental identity change and how number of children is related to parental
identity change could have informed the design of this campaign and what audience segments
were specifically targeted by this campaign to get the best public return on investment.
It is important to note that there were relatively few people with more than three children
in these data, and although there is a clear, consistent trend that going from no children to having

87

a child is associated with substantial increases in parental identity, conclusions about the pattern
of change in parental identity as individuals have more than three kids should be interpreted with
caution. Data from the HILDA was also more limited than that of the BHPS because parental
identity was limited to only one wave of data and the measure was only administered to
participants between 15-29 years of age. As a result, there are very few individuals in this dataset
that have two or more children and only between-persons comparisons are possible.
Overall, these data do show clear differences in parental identity between childless and
those with a single child – but it is unclear in the HILDA what pattern takes shape beyond having
a single child. Further, just as it is plausible that parental identity predicts whether or not an
individual chooses to have children or remains childless, it is also plausible that the experience of
becoming a parent is associated with subsequent changes in parental identity. Thus, strong
conclusions cannot be made about the causal direction of this association.
These results also shed some light on how it relates to psychological characteristics and
outcomes as well. For instance, I was also able to evaluate the association between parental
identity and several measures of attitudes and values related to parenthood, childrearing and
attitudes towards work-life balance. When these analyses are taken as a whole, the main finding
that is consistent across waves is that people who are mainly responsible for childcare report
higher parental identity than people whose partner is primarily responsible for childcare. Of
course, the causal direction here is unknown, and it is unclear whether individuals who are
primarily responsible for their children do so because they are higher in parental identity, or they
are higher in parental identity because they have assumed the role of a primary caregiver.
Regardless of the causal direction, this finding is consistent with past literature linking parental
identity to caregiving behavior (Maurer & Pleck, 2006), and involvement with child rearing

88

(Bruce & Fox, 1999; Ihinger-Tallman et al., 1993). This association could also partially account
for the gender differences in parental identity. Indeed, if women assume the role of primary
caregiver more often than men (which data suggests is still the case in modern western society;
Yap & Anusic, 2016), then regardless of the causal direction of this effect women would report
higher parental identity as a result.
This study also suggests that parental identity is related to other important outcomes, and
shows a small but consistent positive association with life satisfaction in each national sample.
Overall, these findings suggest that placing importance in parenthood is associated with
increased life satisfaction both when you evaluate the association between people with different
overall levels of parental identity and when you evaluate how within-person variability in
parental identity relates to changes in one’s own life satisfaction across time. Although these
effects are very small (r < .1), it is also noteworthy how consistent they are across samples.
Referring back to Table 16, one can see that the between-person and within-person correlations
between life satisfaction and parental identity in the GSOEP are each within .02 of the
corresponding estimates in the BHPS. The HILDA did not include a longitudinal component for
parental identity since there was only one wave of parental identity data. Thus, I was not able to
separate between-person and within-person effects in this sample. However, zero-order
correlations between parental identity and life satisfaction in HILDA were consistent in
magnitude to the results shown in the GSOEP and BHPS.
These findings are consistent with findings in the past literature showing that parental
identity is associated with positive well-being outcomes (e.g., Martire et al., 2000) and offer
some of the first evidence of the robustness and generalizability of these effects. As discussed
earlier, the past literature evaluating the link between parental identity and well-being is

89

extremely limited and relied solely on small, unrepresentative samples (e.g., middle-aged
mothers). In terms of resolving the discrepancies in the past literature, the size and
representativeness of these national samples gives us evidence that the link between parental
identity and well-being that is not just limited to small subpopulations of society, and is observed
across the populations of three distinct nations. However, the magnitude of the associations
between life satisfaction and parental identity observed in the present study are very small
compared to the effect sizes observed in past research (e.g., Martire et al., 2000) and further
studies would be needed before I would draw strong conclusions about the nature of the
discrepancies in the literature and the robustness of these effects.
Parental Identity Across the Lifespan
Results regarding the trajectory of parental identity across the lifespan were remarkably
consistent across the datasets. Indeed, parental identity appears to rise gradually from the period
of adolescence into early adulthood, leveling off when individuals move into their 30s and
remaining relatively stable into older adulthood. In the HILDA, results at first glance appear to
differ from findings in the GSOEP and BHPS. However, assessments of parental identity were
limited only to participants aged 15-29 in the HILDA. Thus, the parental identity plot in the
HILDA only captures the period of adolescence and early adulthood, which one can see is
consistent with the period of relatively rapid positive change observed in the GSOEP and BHPS.
As this is the first longitudinal study of parental identity in the literature and the first to
evaluate the trajectories of parental identity over time, these results are noteworthy because they
provide first insight into the stability and pattern of change of this construct. Indeed, having
information about the pattern of change in parental identity is central to the understanding of
parental identity, its implications, and the understanding of the development of parenthood and

90

its relevant psychological processes. Not surprisingly, individuals at different life stages do differ
in the importance they place in having children – and it appears that adolescence and early
adulthood are the periods of greatest change, consistent with theories of identity formation and
development across the lifespan (e.g., Erikson, 1950) However, once individuals reach 30, it
appears that parental identity becomes a relatively enduring individual difference, a pattern of
change that is consistent with past research evaluating the stability of other many other important
personality traits and individual differences (e.g., Terracciano, Costa & McCrae, 2009). From an
applied perspective, this type of information may give practitioners insight into the normative
pattern of change for parental identity across the lifespan. By helping them understand how the
importance of parenthood typically changes over the course of one’s life, this work could inform
counseling and therapy strategies surrounding issues related to parenthood and parental mental
health.
Parental Identity and Response to the Birth of a Child
Another novel aspect of the current study was that it is also the first study to evaluate
whether the experience of having children is associated with changes in parental identity in the
years that surround birth and the years that follow – questions which have not been addressed in
the past research literature. Overall, the results of this study clearly show that childbirth is
associated with large changes in parental identity, and these changes persist for many years
following birth. These findings are also very consistent across datasets, which suggest this
pattern of change is replicable and a robust description of how parental identity is affected by
childbirth.
I also evaluated several potential demographic moderators of how parental identity
changes in response to the birth of a child, but results revealed little consistent evidence that

91

people vary in how parental identity changes in response to birth along the lines of gender, age,
and income. Lack of a moderating effect of gender is particularly surprising as the importance of
gender for parental identity is clear from past research suggesting that parental identity has
different implications across men and women (e.g., Martire et al., 2000; Maurer et al., 2001).
This would lead one to predict that the experience of this construct (and its pattern of change
following major life events) would differ meaningfully across these groups.
The only consistent moderator of parental identity change in response to birth appears to
be marital status. In both the GSOEP and BHPS, it appears that people that are married when
they have children report higher initial levels of parental identity before the birth of their first
child than unmarried respondents. However, in the years that follow birth, unmarried people
increase in parental identity to a greater extent than married respondents, and these initial
differences between married and unmarried people are largely reduced. Like our findings that
suggest the process of getting married is associated with increases in parental identity, this
pattern of results also suggest that something about marriage, whether it is just existence of a
long-term partner or social expectations regarding what being married means, leads to increased
in parental identity in the years before an individual even becomes a parent. These findings are
also consistent with past research suggesting that marriage is associated with changes in attitudes
towards having children (e.g., Cox, Owen, Lewis, & Henderson, 1989). It is clear however, that
regardless of marital status, having children leads to clear increases in parental identity for all
groups, and any differences due to “preparatory” parental identity experienced by those who are
married before having children is largely dissolved.
Evaluating this question with this type of within-person model is advantageous for
several reasons. For instance, if the critical comparison was a between-person comparison, this

92

study would not be able to rule out the possibility of selection effects. That is, finding differences
in parental identity (and its trajectory) between those who have children and those who do not
does not rule out the possibility that these differences in parental identity trajectory do not result
from the experience of childbirth itself and instead are due to a preexisting, dispositional
difference. In contrast, this within-person approach is able to rule out the possibility of selection
effects because the critical comparison in this design is the contrast between one’s baseline levels
of parental identity in the years before birth of a first child and one’s levels of parental identity in
the years that follow birth.
Using a propensity score matched control group, I was also able to rule out normative,
age related change as an explanation of this pattern of parental identity change in the years that
follow birth. Across both the GSOEP and BHPS, childbirth was still associated with increases in
parental identity even after accounting for the estimated change that would have occurred even if
childbirth had not happened. Indeed, this can be interpreted as evidence that this pattern of
change in parental identity is not due to normative change and instead provides first evidence
that the experience of childbirth itself is associated with substantial, long lasting increases in
parental identity. However, when using such techniques, it is important to consider the
assumptions are being made about the appropriateness of this control group and the conclusions
that can be drawn about normative change using controls groups matched in this way. For
instance, a control group selected in this manner will inevitably be diverse in the reasons they
have not had children. Some individuals in this group are people who have chosen not to have
children at any time during their lives, whereas others may plan to have children, but just not had
children yet. Still others may be strongly motivated to have children, but for some reason are not
able to have children (e.g., fertility, finances, lack of partner). By using this approach to

93

identifying a matched control group, each of these heterogeneous subsamples would be included
in this control group that is used to assess normative change. This issue clearly raises the
question of whether estimates derived from this group are really “normative” and whether it
would serve as an effective comparison group (i.e., are members of this group really alike in
every way other than the fact that they do not have a child?). For instance, the person who was
highly motivated to have children throughout their lives, but ultimately did not due to the lack of
an adequate partner is likely to have other characteristics that make them unique. Even in the
results reported here, there was a significant difference between the event and matched control
groups on the initial levels of parental identity in the BHPS. This suggests that the propensity
score approach to matching groups still resulted in important differences between the two groups
on a key variable of interest. As a result, care should be taken when interpreting these results and
drawing conclusions about how these findings could be used in applied settings.
The Role of Parental Identity in Reactions to the Birth of a Child
The final goal of this study was to evaluate whether parental identity plays a meaningful
role in how people respond to the birth of a child and predicts variability in patterns of SWB
change surrounding birth. Generally speaking, the findings of these studies suggest that parental
identity does in fact play an important role in SWB responses to the birth of a child, but that it
also depends on how one operationalizes this construct. In particular, both pre and postchildbirth parental identity are positively associated with increased SWB in the years before
birth, but only post-childbirth parental identity is associated with an added, long term boost in
SWB following childbirth in three national samples.
These findings are most consistent with the theoretical perspective suggesting that
constructs like parental identity serve a positive function, and may be a protective factor against

94

parental role related stressors (e.g., Luchetta, 1995; Martire et al., 2000). Under this perspective,
individuals high in parental identity will be especially motivated to feel good about their role as a
parent, which in turn buffers individuals from the stressors associated with this role and other
negative aspects of being a parent. Drawing from past theory regarding the functions of identity
constructs (e.g., Erikson, 1950; Tajfel & Turner, 1986), these individuals may simply be more
committed to focus on positive attributes of being a parent, which may provide added
justification for assuming this role when faced with the negative aspects of parenthood in the
years that follow birth. As a result, people’s SWB and their judgements of their quality of life
may be more protected from declines.
Although more research would be needed before one could say definitively that the
apparent discrepancies regarding the direction of this moderating effect are resolved. The
consistency of these findings across three distinct national samples, as well as the size and
representativeness of these samples lends weight to these results as strong evidence for a
buffering model of parental identity on response to childbirth. Of course, one still cannot make
definitive statements about causality from this design and it remains possible that people who
experience parenthood more positively report higher parental identity as a result, or that being
highly identified as a parent leads one to experience parenting more positively.
Another issue to consider when evaluating the strength of conclusions that can be drawn
from these findings is the meaningfulness of measuring parental identity with individuals who do
not have children. Critics of this approach might argue that when one assesses the importance of
having children with a non-parent you are actually assessing a distinct construct that is not an
evaluation of about their identity as a parent (since they are not a parent) but a more general
value or belief about the role of children in the respondent’s life. On the other hand, a case could

95

be made that even for most non-parents, parenthood is a possible role that most people
eventually choose to fulfill. Indeed, the psychological process through which one decides to
become a parent likely involves a meaningful increase in the importance of that role (or at least
their precocieved notions of that role) and what that role means to them. Further, this role is
unique that there is an expectation that most people will eventually occupy this role, thus notions
about the role and what it will mean for an individual’s sense of self likely begin to form long
before one actually becomes a parent.
Limitations and Future Directions
An important advantage of this study was that it was the first study to evaluate change in
parental identity longitudinally in large, nationally representative samples. The data used in this
study allowed me to examine parental identity in samples of people that were approximately
representative of the general populations of Germany, Great Britain and younger populations of
Australia. Previous studies of parental identity have often focused on small samples of middleaged parents, and often only include mothers. As the results of the current study indicate, the
implications of parental identity are not limited to mothers, and this construct has important
implications for people of many groups across the lifespan. Indeed, these panel studies provided
a unique opportunity to explore stability and change in parental identity, its correlates over time
and its long term implications on responses to birth for the first time in the literature.
Despite the strengths of using these large scale, nationally representative datasets to
evaluate the nature of parental identity, the use of such datasets are not without their limitations.
Use of preexisting data limits the control I had over the data that was available at each wave, the
items used to assess key variables and how important constructs were operationalized. For
instance, although parental identity is typically conceptualized as a narrow construct, little work

96

has gone into understanding the assessment of this construct, its dimensions, and the
psychometric properties of existing assessment tools. Importance of a role is often seen as a part
of identity and considered a good indicator of one’s strength of identification with a role or social
group, which is a core aspect of most big theories of identity. (e.g., Sellers et al., 1998; Tajfel &
Turner, 1986). However, it does remain possible that simply asking about the importance of
having children may not adequately capture the construct of parental identity, or simply does not
capture the full breadth and complexity of this construct, which may include dimensions such as
commitment to parenthood or general evaluations and attitudes towards the role of being a
parent.
The implications of this narrow focus for the interpretation of this work are important to
consider because it speaks to the conceptualization of the construct itself and the conclusions we
can draw from this research. For instance, it is plausible that some of the null findings emerging
out of the present analyses were not due to the absence of an effect (e.g., absence of a robust
association between income and parental identity) but instead could be due to the fact that these
measures of parental identity are simply too narrow to capture what would be robust effects had
other aspects or features of parental identity been considered (e.g., income may be associated
with other aspects of parental identity that are not captured by just the importance of the role).
Alternatively, some of the effects and associations that did emerge from this work could be
unique to the notion of importance, and might not have been observed in other aspects of
parental identity. Granted, the notion of importance may be considered a core aspect of parental
identity, but it is important to consider that it may not be the only aspect of parental identity of
consequence. Further, the item we used to measure parental identity was also not included at
each wave of data collection, and at times was not asked of everyone in the sample, which

97

limited the analyses I was able to perform to examine this study’s main research questions.
Another potential limitation is of this study involves the use of single item measures of
parental identity. Typically, single item measures are criticized because they are not as internally
consistent as expected in traditional assessment tools, especially when compared to other
measures of greater length. Use of brief measures in large scale panel studies is often necessary
to minimize participant burden given the usual length of these types of studies. These issues are
important to keep in mind when interpreting the results as there has been no research examining
the reliability of this single item measure of parental identity, nor has there been research
examining the validity of such a measure of parental identity. As such, this literature would
benefit tremendously from more research being devoted to understanding how to assess parental
identity. Other longer measures of parental identity (e.g., Maurer et al. 2001) do capture more
breadth in the features of the parental identity that may be of psychological consequence, but
such measures were not available in these panel datasets. Even if such measures were included in
future primary research evaluating these questions, the added length of these measures would
place substantially more burden on participants and researchers would have to trade off on other
potential strengths in their research designs.
Finally, it is possible that the results of this study are biased by selective attrition. In other
words, there could have been important differences between individuals that remained in the
study versus those who dropped out, and it is possible that these selective differences may
influence the conclusions one could draw from these findings. Nonetheless, this issue and others
discussed above are typically seen as trade-offs inherent in any study that involves secondary
data analysis where one is trading off lack of control over the design of the study for the benefits
associated with use of large nationally representative samples without the burden of collecting

98

such samples.
The results of these studies help to increase the scientific community’s understanding of
parental identity and its role in the lives of parents, and this work also raises other important
questions about the causes and consequences of healthy parenthood. For instance, these panel
studies are rich sources of data, and future research could continue to take advantage of these
datasets to evaluate other potential factors that moderate individuals’ response to the birth of a
child. These panel datasets include measures of many characteristics that are potentially relevant
to how people react and adapt to childbirth, such as post-partum social support, perceived
physical health, and relationship satisfaction. Further research using these data could also make
use of some of the variables already evaluated in the present studies as potential moderators of
individuals’ responses to childbirth. For example, the variables related to division of labor and
childcare may play an important role in how individuals react (and adapt) to having a child.
Indeed, it is possible that the burden of childcare, or the balance of childcare responsibilities
between partners may ultimately affect how the birth of a child impacts parent’s long term well
being.
One of the key challenges for future research is refining the assessment of parental
identity, developing more robust measures of this construct, and more depth in the work to
validate existing measures of the construct. Assessment of parental identity was a key limitation
of this study, and will continue to be a challenge in this area of research until further progress is
made in the measurement of this construct. Although the indicators of parental identity used in
the current studies do assess the importance of having children, these items could have multiple
interpretations to respondents. For example, rather than assessing the importance of having
children to individuals’ own identity or sense of self, respondents may interpret these items as

99

asking about the larger societal importance of having children. Thus, individuals may feel that
having children is important in a general, societal sense, but do not necessarily see having
children or being a parent to be a central part of who they are. This is important to consider when
interpreting these results, as each respective interpretation of these items lend themselves to
vastly different implications one would draw from the findings of these studies. One
interpretation frames what is measured with these items as an identity construct, where the other
interpretation frames what is measured as a more general belief about the importance of children
to societal progression. Moreover, these items assess the importance of “having children” rather
than the importance of “being a parent” which are related, but could be argued to be theoretically
distinct constructs. Secondary data analysis limits the control a researcher has on the variables
that are included in the study and the way in which various constructs are measured.
Efforts to develop a valid, psychometrically sound measure of parental identity are
necessary to allow for greater precision in the measurement of this construct. These measures
could take multiple forms. For instance, they could be designed to capture more breadth and
assess other dimensions of parental identity (e.g., extent of caregiving behaviors; issues related to
work-life balance) and used as multi-dimensional scales. However, they could also continue to
focus on the notion of importance or some other narrowly defined operationalization of parental
identity. The path this work takes largely depends on how researchers in this field continue to
conceptualize parental identity and define the nature of the construct. This is vital to the future of
this area of research, and these efforts would need to move beyond the limitations of panel data.
Additional studies could also evaluate other potential moderators of people’s reaction and
subsequent adaptation to the birth of a child. For instance, physical complications due to
birthing, or complications with the health of one’s newborn baby may also play an important role

100

how people’s well-being changes in response to birth. Sleep deprivation is also known to be an
important factor in post partum well-being in parents (Dennis & Ross, 2005). However, little is
known about whether the negative effects of sleep deprivation on parental well-being persist in
the years beyond infancy and affect long-term adaptation to the birth of a child. Other, more
policy driven factors should also be evaluated as potential moderators of reaction and adaptation
to childbirth in future research. For instance, parental leave is a multifaceted, often contentious
area of public policy. As past research suggests that the availability of parental leave is
associated with an array of parental and child outcomes (Galtry & Callister, 2005),
understanding the long term well-being implications of parental leave on parental well-being
could have huge implications on policy in this area.
Conclusions
This study speaks to the scientific understanding of parental identity, but it also clearly
speaks to the larger issue of parental well-being and this area of research can have far reaching
implications for public policy. For policy makers tasked with improving quality of life for new
mothers, new fathers and growing families, having a clear understanding of the factors that may
contribute to long term well-being and adjustment to new parenthood is critical. Creating policies
that serve to improve parental adjustment, and ultimately contribute to the development of
healthy children and healthy societies need to rely on data describing the factors that relate to
well-being and improve outcomes for parents and children. Overall, this work identifies parental
identity as one of these vital factors that may contribute to long-term adjustment to parenthood,
and considering ways to foster parental identity and create feelings of importance for this role in
expectant parents, particularly those that are at risk, may be a fruitful applied avenue for this area
of research to pursue.

101

REFERENCES

102

REFERENCES

Anusic, I., Yap, S. C. Y., & Lucas, R. (2014a) Does personality moderate reaction and
adaptation to major life events? Analysis of life satisfaction and affect in an Australian
national sample. Journal of Research in Personality, 51, 69-77.
Anusic, I., Yap, S. C. Y., & Lucas, R. E. (2014b). Testing set-point theory is a Swiss national
sample: Reaction and adaptation to major life events. Social Indicators Research, 119,
1265-1288.
Aassve, A., Goisis, A., & Sironi, M. (2012). Happiness and childbearing across Europe. Social
Indicators Research, 108, 65–86.
Baird, B. M., Lucas, R. E., & Donnellan, M. B. (2010). Life satisfaction across the life span:
Findings from two nationally representative panel studies. Social Indicators Research,
99, 183-203.
Baumeister, R.F. (1991). Meanings of life. New York: Guilford Press.
Belsky, J., Lang, M. E., & Rovine, M. (1985). Stability and change in marriage across the
transition to parenthood: A second study. Journal of Marriage and the Family, 47, 855865.
Belsky, J., Spanier, G. B., & Rovine, M. (1983). Stability and change in marriage across the
transition to parenthood. Journal of Marriage and the Family, 45(3), 567-577.
Blanchflower, D. G., & Oswald, A. J. (2008). Is well-being U-shaped over the life cycle? Social
Science & Medicine, 66, 1733-1749.
Bonanno, G. A. (2004). Loss, trauma, and human resilience: Have we underestimated the human
capacity to thrive after extremely aversive events? American Psychologist, 59(1), 20-28.
Bonanno, G. A., & Kaltman, S. (2001). The varieties of grief experience. Clinical Psychology
Review, 21, 705–734.
Bonanno, G. A., & Keltner, D. (1997). Facial expressions of emotion and the course of conjugal
bereavement. Journal of Abnormal Psychology, 106, 126–137.
Bongaarts, J. (1999). Fertility decline in the developed world: Where will it end? The American
Economic Review, 89, 256-260.
Boyce, C. J., Wood, A. M., & Brown, G. D. A. (2010) The dark side of conscientiousness:
Conscientious people experience greater drops in life satisfaction following
unemployment. Journal of Research in Personality, 44, 535-539.

103

Boyce, C. J. & Wood, A. M. (2011). Personality prior to disability determines adaptation:
Agreeable individuals recover lost life satisfaction faster and more completely.
Psychological Science, 22, 1397-1402.
Bruce, C., & Fox, G. L. (1999). Accounting for patterns of father involvement: Age of child,
father-child coresidence, and father role salience. Sociological Inquiry, 69(3), 458-476.
Bumpass, L., & Lu, H. H. (2000). Trends in cohabitation and implications for children’s family
contexts in the United States. Population Studies, 54, 29-41.
Cabrera, N. J., Tamis-LeMonda, C. S., Bradley, R. H., Hofferth, S., & Lamb, M. E. (2000).
Fatherhood in the twenty-first century. Child Development, 71(1), 127-136.
Cambell, A. Converse, P. E., & Rodgers, W. L. (1976). The quality of American life:
Perceptions, evaluations and satisfactions. New York: Russell Sage Foundation.
Cheung, F. & Lucas, R. E. (2014). Assessing the validity of single-item life satisfaction
measures: Results from three large samples. Quality of Life Research, 23, 2809-2818.
Clark, A. E., Diener, E., Georgellis, Y., & Lucas, R. E. (2008). Lags And Leads in Life
Satisfaction: A test of the baseline hypothesis. The Economic Journal, 118(529), F222F243.
Cox, M. J., Owen, M. T., Lewis, J. M., & Henderson, V. K. (1989). Marriage, adult adjustment,
and early parenting. Child Development, 1015-1024.
Deaton, A. (2008). Income, health and well-being around the world: Evidence from the Gallup
World Poll. Journal of Economic Perspectives, 22, 53–72.
Dennis, C. L., & Ross, L. (2005). Relationships among infant sleep patterns, maternal fatigue,
and the development of depressive symptomatology. Birth, 32, 187-193.
Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95, 542-575.
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The satisfaction with life scale.
Journal of Personality Assessment, 49, 71–75.
Diener, E., Lucas, R. E., & Scollon, C. N. (2006). Beyond the hedonic treadmill: Revising the
adaptation theory of well-being. American Psychologist, 61(4), 305-314.
Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective well-being: Three
decades of progress. Psychological Bulletin, 125(2), 276-302.
Doss, B. D., Rhoades, G. K., Stanley, S. M., & Markman, H. J. (2009). The effect of the
transition to parenthood on relationship quality: An 8-year prospective study. Journal of
Personality and Social Psychology, 96(3), 601-619.

104

Dyrdal, G. M., & Lucas, R. E. (2013). Reaction and adaptation to the birth of a child: A couplelevel analysis. Developmental Psychology, 49(4), 749-761.
Dyrdal, G. M., Røysamb, E., Nes, R. B., & Vittersø, J. (2011). Can a happy relationship predict a
happy life? A population-based study of maternal well-being during the life transition of
pregnancy, infancy, and toddlerhood. Journal of Happiness Studies, 12(6), 947-962.
Erikson, E. H. (1950). Growth and crises of the “healthy personality”. In M. J. E., Senn (Ed.),
Symposium on the healthy personality (pp. 91-146). Oxford, England: Josiah Macy, Jr.
Foundation.
Erikson, E. H. (1968). Identity, youth, and crisis. New York: Norton.
Frederick, S., & Loewenstein, G. (1999). Hedonic adaptation. In D. Kahneman, E. Diener, & N.
Schwarz (Eds.), Well-being: The foundations of hedonic psychology (pp. 302–329). New
York: Russell Sage Foundation.
Galatzer-Levy, I. R., Bonanno, G. A., & Mancini, A. D. (2010). From marianthal to latent
growth mixture modeling: A return to the exploration of individual differences in
response to unemployment. Journal of Neuroscience, Psychology, and Economics, 3(2),
116-125.
Galatzer-Levy, I. R., Mazursky, H., Mancini, A. D., & Bonanno, G. A. (2011). What we don't
expect when expecting: Evidence for heterogeneity in subjective well-being in response
to parenthood. Journal of Family Psychology, 25(3), 384-392.
Galtry, J., & Callister, P. (2005). Assessing the optimal length of parental leave for child and
parental well-being: How can research inform policy? Journal of Family Issues, 26, 219246.
Gelman, A., & Hill, J. (2009). Data analysis using regression and multilevel/hierarchical
models. New York: Cambridge University Press.
Gelman, A., Su, Y., Yajima, J., Hill, J., Pittau, M. G., Kerman, J., & Zheng, T. (2011). arm:
Data analysis using regression and multilevel hierarchical models. R package version
1.4-14. http://cran.r-project.org/web/packages/arm/arm.pdf.
Glenn, N. D., & McLanahan, S. (1982). Children and marital happiness: A further specification
of the relationship. Journal of Marriage and the Family, 44, 63-72.
Glenn, N. D., & Weaver, C. N. (1979). A note on family situation and global happiness. Social
Forces, 57, 960–967.
Grote, N. K., & Clark, M. S. (2001). Perceiving unfairness in the family: Cause or consequence
of marital distress? Journal of Personality and Social Psychology, 80(2), 281-293.

105

Haisken-DeNew, J. P., & Frick, R. (2005). Desktop companion to the German Socio-Economic
Panel (SOEP). Berlin: German Institute for Economic Research.
Holmes, T. H., & Rahe, R. H. (1967). The social readjustment rating scale. Journal of
Psychosomatic Research, 11, 213-218.
Ihinger-Tallman, M., Pasley, K., & Buehler, C. (1993). Developing a middle-range theory of
father involvement postdivorce. Journal of Family Issues, 14(4), 550-571.
Institute for Social and Economic Research, University of Essex (2010). British Household
Panel Survey: Waves 1–19, 1991–2009 (computer file). Colchester, England: UK Data
Archive (distributor).
Kenrick, D. T., Griskevicius, V., Neuberg, S. L., & Schaller, M. (2010). Renovating the pyramid
of needs: Contemporary extensions built upon ancient foundations. Perspectives on
Psychological Science, 5(3), 292-314.
Kiang, L., Yip, T., Gonzales-Backen, M., Witkow, M., & Fuligni, A. J. (2006). Ethnic identity
and the daily psychological well-being of adolescents from Mexican and Chinese
backgrounds. Child Development, 77, 1338–1350.
Kobau, R., Sniezek, J., Zack, M. M., Lucas, R. E., & Burns, A. (2010). Well-being assessment:
An evaluation of well-being scales for public health and population estimates of wellbeing among US adults. Applied Psychology: Health and Well-Being, 2, 272–297.
Kohler, H. P., Behrman, J. R., & Skytthe, A. (2005). Partner+ Children= Happiness? The effects
of partnerships and fertility on well-being. Population and Development Review, 31(3),
407-445.
Lawrence, E., Nylen, K., & Cobb, R. J. (2007). Prenatal expectations and marital satisfaction
over the transition to parenthood. Journal of Family Psychology, 21(2), 155-164.
Levere, J. L. (2010). Ads Urge Fathers to ‘Take Time’ to Be a Dad. Retrieved from
http://www.nytimes.com/2010/10/19/business/media/19adnewsletter1.html.
Löckenhoff, C. E., Terracciano, A., Patriciu, N. S., Eaton, W. W., & Costa, P. T. (2009). Selfreported extremely adverse life events and longitudinal changes in five-factor model
personality traits in an urban sample. Journal of Traumatic Stress, 22, 53-59.
Lucas, R. E. (2005). Time does not heal all wounds: A longitudinal study of reaction and
adaptation to divorce. Psychological science, 16(12), 945-950.
Lucas, R. E. (2007a). Adaptation and the set-point model of subjective well-being: Does
happiness change after major life events? Current Directions in Psychological Science,
16(2), 75-79.

106

Lucas, R. E. (2007b). Long-term disability is associated with lasting changes in subjective wellbeing: Evidence from two nationally representative longitudinal studies. Journal of
Personality and Social Psychology, 92(4), 717-730.
Lucas, R. E., Clark, A. E., Georgellis, Y., & Diener, E. (2003). Reexamining adaptation and the
set point model of happiness: Reactions to changes in marital status. Journal of
Personality and Social Psychology, 84(3), 527-539.
Lucas, R. E., Clark, A. E., Georgellis, Y., & Diener, E. (2004). Unemployment alters the set
point for life satisfaction. Psychological Science, 15(1), 8-13.
Lucas, R. E., & Donnellan, M. B. (2011). Estimating the reliabiloty of single-item life
satisfaction measures: Results from four national panel studies. Social Indicators
Research, 105, 323-331.
Luchetta, T. (1995). Parental and work role salience, everyday problems, and distress: A
prospective analysis of specific vulnerability among multiple-role women. Women &
Health, 22(4), 21-50.
Lykken, D., & Tellegen, A. (1996). Happiness is a stochastic phenomenon. Psychological
Science, 7(3), 186-189.
Lyubomirsky, S., & Boehm, J. K. (2010). Human Motives, Happiness, and the Puzzle of
Parenthood Commentary on Kenrick et al.(2010). Perspectives on Psychological Science,
5(3), 327-334.
Lyubomirsky, S., Sheldon, K. M., & Schkade, D. (2005). Pursuing happiness: The architecture
of sustainable change. Review of General Psychology, 9(2), 111–131.
Macht, M., & Dettmer, D. (2006). Everyday mood and emotions after eating a chocolate bar or
an apple. Appetite, 46(3), 332-336.
Marcia, J. E. (1966). Development and validation of ego identity status. Journal or Personality
and Social Psychology, 3, 511-558.
Martire, L. M., Stephens, M. A. P., & Townsend, A. L. (2000). Centrality of women's multiple
roles: Beneficial and detrimental consequences for psychological well-being. Psychology
and Aging, 15(1), 148-156.
Maurer, T. W., & Pleck, J. H. (2006). Fathers' caregiving and breadwinning: A gender
congruence analysis. Psychology of Men & Masculinity, 7(2), 101-112.
Maurer, T. W., Pleck, J. H., & Rane, T. R. (2001). Parental identity and reflected-apprasials:
Measurement and gender dynamics. Journal of Marriage and Family, 63(2), 309-321.

107

McBride, B. A., & Rane, T. R. (1997). Role identity, role investments, and paternal involvement:
Implications for parenting programs for men. Early Childhood Research Quarterly,
12(2), 173-197.
McLanahan, S., & Adams, J. (1987). Parenthood and psychological well-being. Annual Review
of Sociology, 13, 237-257.
Minton, C., & Pasley, K. (1996). Fathers' parenting role identity and father involvement A
Comparison of nondivorced and divorced, nonresident fathers. Journal of Family Issues,
17(1), 26-45.
Mroczek, D. K., & Spiro, A., III (2003). Modeling intraindividual change in personality traits:
Findings from the normative aging study. Journal of Gerontology: Psychological
Sciences, 58, 153-165.
Nelson, S. K., Kushlev, K., English, T., Dunn, E. W., & Lyubomirsky, S. (2013). In defense of
parenthood: Children are associated with more joy than misery. Psychological Science,
24, 3–10.
Nelson, S. K., Kushlev, K., & Lyubomirsky, S. (2014). The pains and pleasures of parenting:
When, why, and how is parenthood associated with more or less well-being?
Psychological Bulletin, 140, 846-895.
Noor, N. M. (2004). Work-family conflict, work-and family-role salience, and women's wellbeing. The Journal of Social Psychology, 144(4), 389-405.
Okun, M. A., & George, L. K. (1984). Physician-and self-ratings of health, neuroticism and
subjective well-being among men and women. Personality and Individual Differences,
5(5), 533-539.
Pagán-Rodríguez, R. (2012). Longitudinal analysis of the domains of satisfaction before and
after disability: Evidence from the German Socio-Economic Panel. Social Indicators
Research, 108, 365-385.
Pai, M., & Carr, D. (2010). Do personality traits moderate the effect of late-life spousal loss on
psychological distress? Journal of Health and Social Behavior, 51, 183-199.
Phinney, J. S. (1990). Ethnic identity in adolescents and adults: A review of research.
Psychological Bulletin, 108, 499-514.
Phinney, J. S., Cantu, C. L., & Kurtz, D. A. (1997). Ethnic and American identity as predictors
of self-esteem among African American, Latino, and White adolescents. Journal of Youth
and Adolescence, 26(2), 165-185.
Pleck, J. H. (1997). Paternal involvement: Levels, sources, and consequences. In M. E. Lamb
(Ed.). The Role of the Father in Child Development (3rd ed., pp.66-103). Hoboken, NJ:

108

John Wiley & Sons Inc.
R Development Core Team (2010). R: A language and environment for statistical computing.
Vienna, Austria: R Foundation for Statistical Computing.
Reitzes, D. C., & Mutran, E. J. (1994). Multiple roles and identities: Factors influencing selfesteem among middle-aged working men and women. Social Psychology Quarterly, 57
(4), 313-325.
Rholes, W. S., Simpson, J. A., Campbell, L., & Grich, J. (2001). Adult attachment and the
transition to parenthood. Journal of Personality and Social Psychology, 81(3), 421-435.
Roberts, B. W., Caspi, A., & Moffit, T. (2003). Work experiences and personality development
in young adulthood. Journal of Personality and Social Psychology, 84, 582-593.
Roberts, B. W., Walton, K., Bogg, T., & Caspi, A. (2006). De-investment in work and nonnormative personality trait change in young adulthood. European Journal of Personality,
20, 461-474.
Robins, R. W., Hendin, H. M., & Trzesniewski, K. H. (2001). Measuring global self-esteem:
Construct validation of a single-item measure and the Rosenberg Self-Esteem Scale.
Personality and Social Psychology Bulletin, 27, 151-161.
Rothrauff, T., & Cooney, T. M. (2008). The role of generativity in psychological well-being:
Does it differ for childless adults and parents? Journal of Adult Development, 15, 148159.
Sellers, R. M., Smith, M. A., Shelton, J. N., Rowley, S. A. J., & Chavous, T. M. (1998).
Multidimensional Model of Racial Identity: A reconceptualization of African American
racial identity. Personality and Social Psychology Review, 2, 18–39.
Settles, I. H. (2004). When multiple identities interfere: The role of identity centrality.
Personality and Social Psychology Bulletin, 30(4), 487-500.
Settles, I. H., Navarrete, C. D., Pagano, S. J., Abdou, C. M., & Sidanius, J. (2010). Racial
identity and depression among African American women. Cultural Diversity and Ethnic
Minority Psychology, 16(2), 248-255.
Singer, J. D., & Willett, J. B. (2003). Applied Longitudinal Data Analysis: Modeling Change and
Event Occurrence. New York: Oxford University Press.
Simon, R. W. (1992). Parental role strains, salience of parental identity and gender differences in
psychological distress. Journal of Health and Social Behavior, 33, 25-35.
Smith, A., Kendrick, A., Maben, A., & Salmon, J. (1994). Effects of breakfast and caffeine on
cognitive performance, mood and cardiovascular functioning. Appetite, 22(1), 39-55.

109

Smith, D.M., Schwarz, N., Roberts, T. R., & Ubel, P. A. (2006). Why are you calling me? How
survey introductions change response patterns. Quality of Life Research, 15, 621-630.
Specht, J., Egloff, B., & Schmukle, S. C. (2011). Stability and change of personality across the
life course: The impact of age and major life events on mean-level and rank-order
stability of the Big Five. Journal of Personality and Social Psychology, 101, 862-882.
Stutzer, A., & Frey, B. S. (2006). Does marriage make people happy, or do happy people get
married? The Journal of Socio-Economics, 35(2), 326-347.
Summerfield, M., Freidin, S., Hahn, M., Ittak, P., Li, N., Macalalad, N, & Wooden, M. (2012).
HILDA User Manual – Release 11. Melbourne Institute of Applied Economic and Social
Research, University of Melbourne.
Tajfel, H. & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S.
Worchel & W. G. Austin (Eds.), Psychology of Intergroup Relations (2nd ed., pp.7-24).
Chicago: Nelson-Hall.
Tay, L., & Diener, E. (2011). Needs and subjective well-being around the world. Journal of
Personality and Social Psychology, 101(2), 354-365.
Taylor, M. F., Brice, J., Buck, N., & Prentice-Lane, E. (2010). British Household Panel Survey
user manual. Introduction, technical report, and appendices (Vol. A). Colchester,
England: University of Essex.
Terracciano, A., Costa Jr, P. T., & McCrae, R. R. (2006). Personality plasticity after age
30. Personality and Social Psychology Bulletin, 32(8), 999-1009.
Thoits, P. A. (1987). Negotiating Roles. In F. J. Crosby (Ed.), Spouse, Parent, Worker: Gender
and Multiple Roles (pp. 11-22). New Haven: Yale University Press.
Thoits, P. A. (1992). Identity structures and psychological well-being: Gender and marital status
comparisons. Social Psychology Quarterly, 55, 236-256.
Thornton, A., & Young-DeMarco L. (2001) Four decades of trends in attitudes toward family
issues in the United States: The 1960s through the 1990s. Journal of Marriage and
Family, 63, 1009-1037.
Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, C. (1994). Self and collective: Cognition
and social context. Personality and Social Psychology Bulletin, 20, 454-463.
Trivers, R. (1972). Parental investment and sexual selection. In Sexual selection and the decent
of man: 1871-1971. Chicago: Aldine de Gruyter.
Twenge, J.M., Campbell, W.K. & Foster, C.A. (2003). Parenthood and marital satisfaction.
Journal of Marriage and Family, 65(3), 574–583.

110

Umberson, D., & Williams, K. (1999). Family status and mental health. In Handbook of the
Sociology of Mental Health (pp. 225-253). Springer US.
Vaidya, J. G., Gray, E. K., Haig, J., & Watson, D. (2002). On the temporal stability of
personality: Evidence for differential stability and the role of life experiences. Journal of
Personality and Social Psychology, 83, 1469-1484.
Wong, M. M., & Csikszentmihalyi, M. (1991). Motivation and academic achievement: The
effects of personality traits and the duality of experience. Journal of Personality, 59(3),
539-574.
Yap, S. C. Y., Anusic, I., & Lucas, R. E. (2012). Does personality moderate reaction and
adaptation to major life events? Evidence from the British Household Panel Survey.
Journal of Research in Personality, 46, 477-488.
Yap, S. C. Y., & Anusic, I. (2016). Women still take up the bulk of household labour and
childcare. Retrieved from http://insightaresearch.com/2016/05/17/time-use-by-gender/.
Yap, S. C. Y., Settles, I. H., & Pratt-Hyatt, J. S. (2011). Mediators of the relationship between
racial identity and life satisfaction among a community sample of African American
women and men. Cultural Diversity and Ethnic Minority Psychology, 17, 89-97.

111