MATERNAL SEPARATION ANXIETY: A LONGITUDINAL ANALYSIS WITH
WELL-EDUCATED, WORKING MOTHERS
Amy Delaplane Gee
BA, Duquesne University, 1988
MS, University of Pittsburgh, 2004
Submitted to the Graduate Faculty of
The School of Education in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
University of Pittsburgh
UNIVERSITY OF PITTSBURGH
SCHOOL OF EDUCATION
This dissertation was presented
Amy Delaplane Gee
It was defended on
April 7, 2008
and approved by
Christina J. Groark, PhD, School of Education
Lindsay Clare Matsumura, PhD, School of Education
Catherine G. Greeno, PhD, School of Social Work
Dissertation Advisor: Joan I. Vondra, PhD, School of Education
MATERNAL SEPARATION ANXIETY: A LONGITUDINAL ANALYSIS WITH
WELL-EDUCATED, WORKING MOTHERS
Amy Delaplane Gee, M.S.
University of Pittsburgh, 2008
This associational study examines Maternal Separation Anxiety in a sample of at least college
educated, employed mothers with young children in non-maternal care selected from Phase I
of the National Institutes of Child Health and Development study of early child care. The
study provides descriptive data on maternal separation anxiety from 6 months to 24 months
after birth of a child, its association with indices of mother-child relations, and whether these
associations are merely a manifestation of general neuroticism/anxiety or depressive
symptoms, or if there is something unique about maternal separation anxiety of relevance to
mother-child relations. Self-report and observational data on 147 mothers indicated that
maternal separation anxiety showed high individual stability, but on average decreased over
time. Maternal personality characteristics and maternal education level showed modest
associations with MSA. However, MSA was associated with neither observed maternal
behavior at day care drop-offs and pick-ups, nor infant-mother attachment in the expected
patterns. Results suggest that MSA may not be a fruitful approach to thinking about parenting
in a White, well-educated sample of women.
TABLE OF CONTENTS
1.0 INTRODUCTION........................................................................................................ 1
2.0 LITERATURE REVIEW............................................................................................ 5
2.1 ETIOLOGY OF MATERNAL SEPARATION ANXIETY............................ 6
2.2 WHAT DOES MATERNAL SEPARATION ANXIETY LOOK LIKE?.... 11
2.2.1 Maternal Separation Anxiety over time ................................................... 11
2.3 LINKS TO PERSONALITY AND DEPRESSIVE SYMPTOMS ................ 13
2.3.1 Maternal personality dimensions.............................................................. 13
2.3.2 Depressive symptoms.................................................................................. 16
2.4 MATERNAL SEPARATION ANXIETY AMONG WELL-EDUCATED
2.5 LINKS BETWEEN MATERNAL SEPARATION ANXIETY AND
MOTHER-CHILD RELATIONS..................................................................................... 22
2.5.1 Maternal separation and reunion behavior.............................................. 22
2.5.2 Mother-infant attachment security........................................................... 26
2.6 OTHER CONSTRUCTS POSSIBLY RELATED TO MATERNAL
SEPARATION ANXIETY ................................................................................................ 31
2.6.1 Maternal temperament............................................................................... 31
2.6.2 Child temperament..................................................................................... 31
2.6.3 Quality of non-maternal care..................................................................... 32
2.7 STUDY HYPOTHESES AND ANALYTIC PLAN........................................ 33
3.0 METHODS................................................................................................................. 39
3.1 SAMPLE............................................................................................................. 39
3.2 MEASURES....................................................................................................... 44
3.2.1 Maternal Separation Anxiety..................................................................... 44
3.2.2 General anxiety........................................................................................... 45
3.2.3 Depressive symptoms.................................................................................. 46
3.2.4 Maternal education..................................................................................... 46
3.2.5 Maternal behavior ...................................................................................... 47
3.2.6 Attachment .................................................................................................. 48
4.0 RESULTS ................................................................................................................... 54
4.1 DESCRIPTIVE STATISTICS ......................................................................... 57
4.2 MATERNAL SEPARATION ANXIETY OVER TIME............................... 60
4.2.1 Mean level of maternal separation anxiety............................................... 60
4.2.2 Individual differences in maternal separation anxiety over time........... 61
4.2.3 Maternal personality dimensions.............................................................. 61
4.2.4 Maternal depressive symptoms ................................................................. 62
4.3 MATERNAL SEPARATION ANXIETY AND MATERNAL EDUCATION
4.4 LINKS BETWEEN MATERNAL SEPARATION ANXIETY AND
MOTHER-CHILD RELATIONS..................................................................................... 64
4.4.1 Maternal separation and reunion behavior.............................................. 64
4.4.2 Mother-infant attachment security........................................................... 64
4.4.3 Adding controls for maternal personality................................................ 65
5.0 DISCUSSION............................................................................................................. 67
6.0 CONCLUSION........................................................................................................... 77
6.1 LIMITATIONS.................................................................................................. 77
6.2 FUTURE RESEARCH...................................................................................... 79
LIST OF TABLES
Table 1. Levels of Maternal Separation Anxiety by Employment Preference and Status............ 10
Table 2. Sample descriptive statistics........................................................................................... 42
Table 3. Construct measurement time points, locations, and administrators................................ 44
Table 4. Frequencies of forced attachment classification at 15 months (Strange Situation)........ 50
Table 5. Descriptive statistics for the independent and dependent variables ...............................51
Table 6. Internal validity (Cronbach’s alpha) for the study variables .......................................... 53
Table 7. Intercorrelations among study variables......................................................................... 55
Table 8. Comparison of MSA means in current study with means of previous research (working
mothers only)................................................................................................................................ 59
Table 9. Descriptive statistics for maternal separation anxiety at 6, 15, & 24 months................. 60
Nearly 73% of infants and toddlers of working mothers in the United States are in some form of
non-parental care, and approximately 39% are in full-time (35 hours or more per week) non-
parental care. The average amount of childcare for all children is currently 25 hours per week
(Macomber, Adams, & Tout, 2001). This constitutes almost 5 million children experiencing non-
parental care on a weekly basis, which in turn means that for millions of families, balancing
employment and childcare is a salient issue (Macomber, Adams & Tout, 2001). Mothers are
traditionally responsible for the majority of childcare responsibilities. Therefore, for mothers
who work, one consequence of this balancing act is that millions of mothers may be experiencing
some degree of anxiety related to the process of trying to maintain equilibrium between home
and work demands.
In fact, there are more American mothers in the paid workforce today than in the past two
to three decades (U.S. Census Bureau, 2005). More mothers in the workforce necessitates the use
of more non-maternal care, and there are currently more children in non-maternal care than in
past decades (U.S. Department of Education, 2005). This increase in non-maternal care is, in
part, responsible for a large body of literature related to the effects of non-maternal care on child
outcomes. However, there has been very little research conducted on the emotional aspects for
mothers of using non-maternal care (Hock & Lutz, 1998). This lack of attention is problematic
because maternal psychological well-being is also linked to child outcomes (see Belsky, 1984).
Likewise, although there is a relatively large body of literature on parenting, little
of it concerns parenting beliefs in general or beliefs about childcare in particular (Mistry,
Chaudhuri & Diez, 2005). Instead, most of these studies investigate parent beliefs about their
child’s development. Garcia-Coll and colleagues, (1998) noted this trend:
Clinical and developmental psychology have historically focused on child
development, with little attention paid to the lifelong relational development
of mothers and children; even less attention is paid to the psychological
development of mothers. The absence of the mother’s perspective and
experience in clinical formulations, developmental theory, and psychological
research is striking (p. 2).
In particular, there is very little research on mothers’ views about separating from their
infant on a regular basis (McBride, 1990).
Hock, McBride and Gnezda (1989) operationalized the construct of ‘maternal separation
anxiety’ (MSA) as negative emotions related to separating from one’s child, including feelings of
worry, sadness, and guilt. Maternal separation anxiety is conceived as being associated
specifically with the experience of separation from one’s child and as being temporary distress
rather than a constant emotional state (Hock, DeMeis, & McBride, 1987). In a sample of 400
middle-class, first-time mothers, Hock, McBride and Gnezda (1989) found that MSA can be
measured reliably through maternal self-report, and using a sample of 87 White, middle-class
mothers, Hock and Schirtzinger (1992) established that maternal reported separation anxiety is
positively related to maternal depressive symptoms and a mother’s reported reluctance to allow
her child to individuate from her. Little empirical research beyond this is available on the
The few childcare studies that include a mother’s perspective, or her views related to
childcare, deal primarily with good communication between non-maternal caregivers and parents
(Owen, Ware & Barfoot, 2000) and associations between childcare and mother-child interaction
quality (NICHD, ECCRN, 1999). Therefore, a mother’s perspective is relatively neglected in
work on the effects of childcare on the family. How do various maternal and family
characteristics and child-related circumstances relate to maternal belief systems and concerns
about non-maternal care? How is maternal separation anxiety related to other aspects of a
mother’s psychological functioning?
Hock, McBride, and Gnezda (1989) argued that the guilt a mother feels about separating
from her child might result in her reporting a greater number of depressive symptoms. In fact,
Hock and Schirtzinger (1992) found a strong, positive association between MSA and reported
depressive symptoms. This suggests that the degree of maternal separation anxiety a mother
experiences reflects and/or has implications for her mental health. A mother’s psychological
well-being has implications for maternal caregiving and, in turn, child outcomes (Belsky, 1984).
For instance, a mother who experiences a high level of MSA may have difficulty allowing her
child the independence, both physically and emotionally, he/she needs to achieve age-
appropriate developmental milestones, such as autonomy and exploration. Maternal separation
anxiety, in particular, may influence the interplay between maternal employment and child
outcomes (Hock, DeMeis, & McBride, 1987). Maternal separation anxiety, for instance, has been
associated with an infant’s coping ability and adaptability at center-based care (Hock, 1984).
However, the construct of MSA has been largely ignored by researchers (Hock & Lutz, 1998).
The work of Hock and her colleagues (1984, 1986, 1987, 1989, 1992, 1995, & 1998)
serves as the foundation for this study. The proposed study is designed as a correlational,
exploratory study to examine MSA in a sample of 147 well-educated (a college degree or
higher), employed mothers with young children in non-maternal care. All participants were
involved in Phase One of the National Institutes of Child Health and Development study of early
child care. The following questions serve as the basis for this study: (1) What does maternal
separation anxiety look like in a sample of well-educated mothers? (2) How does maternal
separation anxiety change over time? (3) Is MSA merely a manifestation of mothers’ general
anxiety and/or tendency toward depressive symptoms? (4) Is MSA more typical of mothers
without an advanced degree? Finally, (5) Is MSA related to mother-child relations? The answers
to these questions are relevant to both maternal functioning and child development because a
mother’s anxiety about separating from her child can likely be transmitted to the child, resulting
in less optimal child outcomes (Hock, 1984).
These questions will be examined by generating descriptive statistics of MSA and
calculating correlations between MSA and various maternal characteristics (education,
separation and reunion behavior, and personality). A repeated measure ANOVA will be
computed in order to examine MSA over time. A Fisher z-test will be performed comparing
correlations of mothers of firstborns and mothers of laterborns in order to determine if
experience as a parent has any relevance to the MSA a well-educated mother experiences. T-tests
will be performed in order to examine whether MSA has any bearing on mother-infant
attachment security. Finally, regression analyses will be run controlling for maternal personality
and depressive symptoms in order to determine if MSA has meaning for maternal behavior and
mother-child relations beyond general anxiety and depressive symptoms.
2.0 LITERATURE REVIEW
Separating from one’s child, whether at center-based care, in neighbor care, or with a nanny,
may cause some mothers to experience worry, sadness, and guilt related to the separation (Hock
et al., 1989). Hock and her colleagues conceptualized MSA (in the 1980’s) during a period of
societal transition related to families and employment when a great number of young mothers
were entering the workforce. In this review of the literature, studies bearing on the incidence and
evolution of MSA are first considered. Attention is then given to the question of whether MSA is
merely a manifestation of general maternal anxiety or tendency to experience depressive
symptoms. Next, the focus is on whether maternal separation anxiety is related to level of
maternal education. Following this, the association between MSA and mother-child interaction
and attachment are explored. Finally, the potential role of variables not included in the present
study receives attention. There are, as yet, no clear answers to these questions, and a greater
understanding of the concept of maternal separation anxiety would expand the existing literature
on maternal attitudes and beliefs about her use of childcare. It might also improve understanding
of how employed mothers struggle to balance work and home lives, and how this balance relates
to child adaptability to and adjustment in non-maternal care.
2.1 ETIOLOGY OF MATERNAL SEPARATION ANXIETY
Working mothers balance the roles of mother and employee in many different configurations,
depending on characteristics of their job, their family make-up, financial need, and number of
children (Hock, DeMeis, & McBride, 1987). The fact that there are no set strategies or formulas
for accomplishing this task makes it more difficult (Hock, DeMeis, & McBride, 1987). The
myriad differences in women themselves and how each one navigates the balance between home
and employment makes it difficult to identify the root of maternal separation anxiety. Indeed,
according to Hock, DeMeis, and McBride (1987), there is no single cause of maternal separation
One source of maternal separation anxiety is theorized to be maternal personality (Hock,
DeMeis, & McBride, 1987). A key component of a mother’s personality (as it relates to maternal
separation anxiety) is her level of general anxiety and depressive tendencies, or neuroticism. A
mother’s personality has implications for her behavior, which can have implications for her
child’s behavior (Hock, 1984). A mother who is anxious about separating from her child may
behave in a manner that signals to her child that there is something to fear about the separation
(Hock, 1984). If initial separations go badly, this may lead to increased anxiety on the part of
both the mother and child during subsequent separations (Hock, 1984).
Hock, McBride, and Gnezda (1989) found that their Maternal Separation Anxiety Scale
as a whole was only modestly associated with general anxiety (as assessed by the Taylor
Manifest Anxiety Scale, r = .34). This constitutes only 12% shared variance between these two
constructs. These researchers interpreted this result to indicate that maternal separation anxiety is
something different from general anxiety.
Other variables that may contribute to MSA arise from the society and culture in which a
mother functions. In other words, the socio-cultural context in which a mother lives can alter the
maternal separation anxiety she experiences (Hock, 1984; Hock, DeMeis, & McBride, 1987).
One’s culture defines the various roles of its members. As it relates to motherhood, culture
creates expectations of the maternal role (Hock, DeMeis, & McBride, 1987). In particular,
depending on the culture, there may be strong views related to how much time a mother spends
away from her child, or whether she should separate from her infant at all. In some cultures (e.g.,
Japanese) these cultural values include having an infant sleep in the parents’ bed, in some cases
for years, and using very little non-maternal care. In most cultural groups, there is a definite
belief that mothers are the best caregivers for their infants. Mothers who have grown up with
such cultural values, but who choose to use non-maternal care (e.g., in order to pursue a career)
may experience maternal separation anxiety to a higher degree, due to internalized cultural
A third proposed source of maternal separation anxiety is the combination of maternal
preference for employment and the degree of importance a mother gives to the maternal role
(Hock, DeMeis, & McBride, 1987). DeMeis, Hock, and McBride (1986) found that mothers who
wanted to work had lower MSA scores than mothers who preferred to stay at home when their
child was 8 months and again at 13½ months. Hock and DeMeis (1990) found that mothers who
preferred to be at home reported significantly more MSA than did mothers who preferred to
work, and that the greatest differences existed between mothers who wanted to stay at home and
did (highest on MSA) versus mothers who wanted to work and did (lowest on MSA).
Mothers who prefer to work may view motherhood as just one in a number of women’s
roles that provide fulfillment and may have crafted a view of motherhood that is not mutually
exclusive with employment, consequently experiencing MSA to a lesser degree (Hock, DeMeis,
& McBride, 1987). Wille (1998) reported no connection between a mother’s employment
preference and MSA assessed when infants were 6 and 18 months of age in a sample of 66
White, lower to middle class, employed mothers. However, this investigator found significant
results with a closely related construct – “Employment-Related Separation Concerns.” The
“Maternal Separation Anxiety” subscale of the Maternal Separation Anxiety Scale measures
maternal feelings about separating from her child in general, whereas the “Employment-Related
Separation Concerns” subscale measures maternal feelings about separating from her child from
a specific point of view: for reasons of employment. The number of hours a mother worked per
week (at child age 6 and 18 months), the amount of child-care a father performed (at child age 6
months), and the perception a mother had of the maternal role (at child age 6 and 18 months)
predicted a mother’s Employment-Related Separation Concerns. Mothers reported lower
employment-related separation concerns when they worked more hours per week, fathers helped
more with childcare, and mothers preferred to be employed.
At the other end of the spectrum of home-to-work preference are mothers who prefer to
stay home with their children and do stay home. In several studies, mothers committed to an
exclusive parenting role–who wanted to stay at home with their children and did so–reported
higher levels of MSA than both conflicted mothers (those with a mismatch between employment
preference and status) and employed mothers who wanted to work (DeMeis, Hock, & McBride,
1986; Hock & DeMeis, 1990; Booth, Clarke-Stewart, Vandell, McCartney and Owen, 2002).
Women who view the role of mother as paramount in their lives and as their sole source of role
fulfillment may hold more traditional beliefs about maternal roles and choose not to separate
from their children often. When they do separate, it may trigger anxiety for these mothers. In
fact, mothers who have a match between wanting to stay at home with their children and do stay
at home report higher MSA than mothers who are in conflict about working versus staying at
home with their children.
In turn, mothers whose employment preference conflicts with their employment status
tend to report higher levels of MSA than mothers who want to work and are working (Hock &
Schirtzinger, 1989; Hock & DeMeis, 1990). In one investigation, mothers who preferred to work
but stayed at home reported higher MSA than mothers who preferred to be employed and were
employed (Hock & Schirtzinger, 1989). The former may feel that they should stay home with
their child in order to promote the best outcomes for their child, ignoring or setting aside their
own desire to be employed and find fulfillment in another role. The emphasis on child needs
over the mother’s may manifest itself as higher reported MSA for this group of mothers.
Likewise, mothers who preferred to stay at home but were employed (e.g., for financial
reasons) also reported higher MSA than mothers who preferred to be employed and were
employed (DeMeis, Hock & McBride, 1986). Mothers who prefer to stay at home tend to view
the maternal role as paramount in their lives (DeMeis, Hock & McBride, 1986). When these
women move into the workforce they may feel that their employment will result in poorer
outcomes for their child. Consequently, the separation process may be a source of anxiety for
them. The levels of MSA by employment preference and status are listed in Table 1.
Table 1. Levels of Maternal Separation Anxiety by Employment Preference and Status
Situation Prefer to stay at home Prefer to be employed
Stay at home Highest Medium
Employed Medium Lowest
While the match between employment preference and status does explain some
variability in maternal separation anxiety, it does not explain all or even most of the variation
between mothers in MSA. Otherwise, mothers who want to stay at home and do (indicating a
match between employment preference and status) would report the same low levels of maternal
separation anxiety that mothers who want to work and do report, but they do not. Stay-at-home
mothers who want to be at home report the highest MSA.
In sum, mothers who presumably struggle psychologically to balance their roles at home
and at work (i.e., their preference and circumstances do not match) tend to experience higher
MSA than mothers who prefer employment and are employed, but less than mothers who prefer
to stay at home with their children and do. These examples illustrate the importance of the
mother’s employment preference, in addition to her status, for predicting MSA. Indeed, the
mother’s employment preference has been hypothesized to influence the association between a
mother’s work status and outcomes for her child (Hock, DeMeis, & McBride, 1987). Maternal
satisfaction with and positive approach to motherhood presumably result in better parenting
which, in turn, results in better child outcomes, regardless of whether a mother is employed or
stays at home. Therefore, maternal employment preference alone is a necessary, but not
sufficient, component for explaining maternal separation anxiety.
Put succinctly, maternal separation anxiety is a multifaceted construct affected by a
variety of factors (Hock, DeMeis & McBride, 1987). Women experience MSA differently
depending on their background and experience. Therefore, a first step in understanding MSA is
to describe how it manifests itself in different maternal populations.
2.2 WHAT DOES MATERNAL SEPARATION ANXIETY LOOK LIKE?
Using a sample of women from a wide range of income levels (M = $32,366, range = $13,500-
$103,000), ages (M = 26, range = 19-39), and education (M = 14.38 years, range = 11 – 19
years), Hock, McBride, and Gnezda (1989) found that maternal separation anxiety is a
significant issue for many mothers. To what extent that is true along the entire continuum of
income and education, however, has yet to be determined. The present study in part addresses
2.2.1 Maternal Separation Anxiety ove
Over time, as children get older and become more self-sufficient, they spend increasing amounts
of time away from their mothers (e.g., non-maternal care, pre-school or an early education
program), providing their mothers with more experience separating from them. The increased
maturity of the child in combination with the greater experience separating from one’s child may
decrease a mother’s feelings of MSA over time.
McBride and Belsky (1988) discerned a significant decrease in MSA between infant ages
3 and 9 months for employed mothers only (M = 21.1 versus 19.6, n = 33). Hock and Lutz
(1998) found that from first pregnancy to child age 18 months, there was a significant main
effect for time (decline in MSA) for first-time mothers (over 80% were working). Hsu (2003)
also reported that MSA decreased significantly from infant age 1 month (M = 22) to 6 months (M
= 21) for first-time mothers (over 50% were working). Using a repeated measures ANOVA,
Hock, DeMeis and McBride (1987) found that over four measurement points (child ages birth, 7
weeks, 8 months, and 14 months), self-reported MSA decreased significantly in a sample of first
time, working (56%) mothers. Means at each measurement point were 21.5 (birth), 20.5 (7
weeks), 19.9 (8 months), and 19.5 (14 months). Similarly, Hock and Schirtzinger (1992) found
that over three measurement points (child ages 8 months, 3 years, and 6 years), self-reported
MSA decreased significantly from 8 months (M = 19.4) to 3 years (M = 18.8), to 6 years (M =
17.9). Therefore, over time, there is a statistically significant decrease in MSA, although in
practical terms, the actual decrease is very small (possible scale range is 7 to 35). There is
relative consistency in the means each researcher reports: from birth to infant age 1 month (M =
22); from 2 to 6 months (M = 21); from 8 to 14 months (M = 20); at 3 years (M = 19); and at 6
years, (M = 18). The present study focuses on children from birth until age 2 years, therefore it is
expected that mean MSA scores will range from about 21 (at 6 months) to 19 (at 2 years).
Hock and Schirtzinger (1992) found that individual rank order in maternal separation
anxiety scores were highly stable across measurement points (at child age 8 months to 3 years, r
= .72; 3 years to 6 years, r = .74) in a sample of 87 White mothers with a relatively wide range of
income (M = $54,251, range = $16,000 - $153,000), age (M = 33, range 26 – 40), and an average
education of 15.3 years. Hock, DeMeis and McBride (1987), likewise, found a great deal of
individual stability in reported maternal separation anxiety over time (average r = .67) in a
sample of 130 first time, White mothers, with an average age of 26.4 years (SD = 3.4), and
education of 15 years (SD = 2.0). These studies established maternal separation anxiety as a
relatively common phenomenon among mothers that shows rank order stability over time,
meaning that mothers who score higher than average on MSA at one particular time will tend to
score higher on MSA at another. Other studies show that, on average, levels of maternal
separation anxiety decrease over time.
The stability of individual differences in maternal separation anxiety may indicate that
these differences are related to stable personality traits. A woman who is more anxious by nature
may tend to express more MSA at any one point in time. The same may be true of women with a
psychologically based tendency to experience symptoms of depression. Of particular interest to
this study is whether maternal separation anxiety is merely a proxy for mothers’ general anxiety
or neuroticism and/or depressive symptoms, or if there is something unique about MSA for
maternal behavior and mother-child relations.
2.3 LINKS TO PERSONALITY AND DEPRESSIVE SYMPTOMS
sions 2.3.1 Maternal personality dimen
In their conceptualization of maternal separation anxiety, Hock and Schirtzinger (1989)
envisioned characteristics of mothers’ personality having some bearing on their ability to
navigate the process of maintaining a balance between separation from and proximity to their
child, as he/she grows older. Women who score higher on measures of neuroticism (worrying,
nervousness, emotionality, and insecurity) are expected to score higher on measures of maternal
separation anxiety because they tend to experience more psychological distress, may have more
unrealistic ideas about parenting, and/or may possess poorer coping skills. In contrast, women
who are more extraverted (more sociable, optimistic, and affectionate) and/or agreeable (trusting,
modest, compliant, and straightforward) may have better interpersonal skills and experience
more positive social interactions. Greater optimism and interpersonal trust from positive social
interactions may, in turn, result in lower reported MSA when leaving a child in care. In other
words, mothers who have a more positive outlook in general and who have faith that social
interaction will be a positive experience may feel less guilt and anxiety about leaving their child
in non-maternal care, and may even perceive how their child could benefit from the experience.
Further, these characteristics of personality are independent from each other (Caspi, 1998);
therefore, each variable should account for unique variance in an analysis of maternal
It is important to point out the caveat that a more inhibited and/or negatively reactive
infant or child is likely to elicit more maternal separation anxiety. But maternal perceptions of
and reactions to child inhibition/negative reactivity are, themselves, a likely product of maternal
personality (Sameroff, Seifer & Elias, 1982). So there is an inherent confound between MSA and
maternal personality when considering how child temperament/behavior might affect a mother’s
level of MSA.
One study that supports the link between maternal personality and MSA is McBride and
Belsky’s (1988) examination of a sub-sample of the Pennsylvania Infant and Family
Development Project. Using regression analysis, these researchers found that scores from two
personality measures (Personality Factor Questionnaire and the Interpersonal Affect and Self-
Esteem subscales of the Jackson Personality Inventory) accounted for a significant portion
(19%) of the variance in self-reported MSA. The Personality Factor Questionnaire evaluates the
quality and magnitude of empathy a person has, with higher scores suggesting lower self-esteem
and affect. Mothers reporting higher levels of MSA also described themselves as lower in self-
esteem and having less empathy toward others. The association between low self-esteem and
MSA indicates that mothers who doubt and criticize themselves report experiencing greater
distress about leaving their child in the care of someone else.
Hock and Lutz (1998) found a significant, positive correlation between self-criticism and
MSA (r = .25, p<.01) when children were 24 months of age. Using hierarchical regression
analysis, Hock and Lutz (1998) found that maternal scores on the personality traits of self-
criticism and dependency components of the Depressive Experiences Questionnaire accounted
for a significant portion (15%) of the variance in MSA. A mother highly dependent on her infant
or child for her sense of well-being may experience anxiety related to separations, and leaving
her child in non-maternal care could provoke feelings of guilt, expressed as MSA. As an
alternative, the authors suggested that self-critical (neurotic) mothers may feel unworthy of love,
and the act of separating from their child may heighten these fears, manifested in higher MSA
scores (Hock & Lutz, 1998). The results of this research establish a link between the personality
characteristics of self-criticism and dependency and MSA. Other researchers have found
associations between MSA and other personality characteristics, such as self-esteem.
In Hock and Schirtzinger’s (1992) longitudinal study of mothers of six-year-olds, women
reporting MSA in the highest 20% of the sample (61.8 or higher out of a possible 105, n = 17)
described themselves as “less effective mothers” and “poorer wives” and as having self-esteem
lower than mothers reporting MSA in the lowest 20% of the sample (43.8 or less, n = 17). These
findings recall those of McBride and Belsky (1988), who found an inverse association between
MSA and self-esteem.
As noted earlier, Lutz and Hock (1995) found that mothers who remembered their own
mothers as “rejecting” and as not encouraging independence reported higher MSA when their
infants were 2 months of age. These mothers were also more often categorized as having an
insecure working model of attachment (as assessed by the Adult Attachment Interview). In
general, then, women who are more negative about themselves (lower self-esteem), more
dependent, more negative about their own mothers, and less trusting about close relationships in
general (adult attachment) seem to score higher on MSA. This certainly suggests the possibility
that MSA may simply be a manifestation of an overall tendency to experience negative affect
(neuroticism) or, more specifically, to experience symptoms of depression. The latter possibility
is discussed in the next section.
2.3.2 Depressive symptoms
There is not much literature on the association between maternal separation anxiety and
depressive symptoms, and the findings of two studies that directly address this association
contradict each other. Using a sample of 87 White, mothers with an average income of $54,251
(range $16,000 - $153,000), average age 33 years (range 26 – 40), and average education of 15.3
years, Hock and Schirtzinger (1992) found that depressive symptoms (measured using the Center
for Epidemiological Studies Depression Scale) were correlated with MSA at child age 6 years (r
= .36, p=.001), but not at 8 months (r = .01), or 3 years (r = .17). The greater a mother’s self-
reported MSA, when her child was 6, the more depressive symptoms she reported. An analysis
of the differences between the coefficients at the three measurement points showed a significant
difference between the first and third (z = 2.46, p=.02). Therefore, no significant association
between higher MSA and depressive symptoms existed until the children in this sample were
older. Higher than usual MSA scores when a child has entered elementary school may signal
adjustment issues on the part of the mother.
Using a sample of 53 White, “middle-class” mothers with an average age of 27 years (SD
= 5.51), and average education of 15 years (SD = 2.24), Hsu (2003) found that maternal
depressive symptoms (measured with the Beck Depression Inventory) were not significantly
associated with MSA at 1 month (r = .16) or at 6 months (r = .29), although the 6-month
association could be significant in a larger sample. This study supports Hock and Schirtzinger’s
(1992) finding of no significant association between MSA and maternal depressive symptoms
during infancy. Therefore, it appears that initially, MSA and depressive symptoms are not
significantly associated. But when mothers report more MSA than their peers by the time their
child has started school, they are more likely to report depressive symptoms.
Based on the two available studies (Hock & Schirtzinger, 1992; Hsu, 2003) it appears
that there may not be a significant association between MSA and depressive symptoms for the
current sample of mothers of children ages 6 to 24 months. However, the previous lack of
significant findings may be related to their relatively small samples (N = 87 and N = 53,
respectively). It may be the case that these samples were not large enough to detect significant,
moderate correlations, whereas the proposed study has a sufficient sample (n = 147). Therefore,
based on the conflicting research results, it remains an empirical question how (or even whether)
there will be an association between depressive symptoms and MSA.
2.4 MATERNAL SEPARATION ANXIETY AMONG WELL-EDUCATED
nd adjusting their hours of employment to
n. Therefore, the idea of “well-educated” currently
means more education than it did previously.
Maternal education is a broad indicator of many kinds of differences among mothers. On the one
hand, it is an indicator of maternal social class and the myriad of differences associated with it.
Whether the construct of interest is maternal intelligence and verbal fluency, encouragement of
child cognitive growth and competence, knowledge of early developmental milestones, attitudes
about child rearing, use of disciplinary strategies, or emphasis on encouraging versus punishing
child behavior, maternal education and social class translate into predictable patterns of
differences. On the other hand, the best-educated women, those with degrees beyond college, are
more likely to embark on self-fulfilling careers that are not particularly tolerant of family needs.
College-educated women who do not go on to advanced degrees, in contrast, may feel the most
flexibility about moving in and out of employment a
perceived needs of their family and children.
In their study of maternal separation anxiety among well-educated mothers, DeMeis,
Hock and Gnezda (1986) defined “well-educated” mothers as those with 14 or more years of
education. Fourteen was chosen because at the time many “traditional female professions” (e.g.,
nursing) required only two years of post high school education. With the societal changes that
have occurred over the past 20 years, education requirements for all employees, not just women,
have increased. Today, many workers are required to have at least a four-year college degree for
an entry-level job. In addition, with the increased entry of women into previously male-
dominated fields (e.g., engineering, medicine), more women are achieving higher levels of
education than previous generations of wome
There are education and income differences in choices mothers make regarding childcare
arrangements. According to the 2002 National Survey of America’s Families (NSAF), children
with higher income parents are more likely to be in (generally costly) center-based care, whereas
children with lower income parents (less than 200 percent of the poverty level) are more likely to
be cared for by (generally inexpensive) non-parental family members (Capizzano & Adams,
2003). Furthermore, children from higher income homes tend to spend more time in non-
maternal care than their lower income counterparts (Capizzano & Main, 2005). Higher income
mothers are, on average, more highly educated mothers (Macomber, Adams & Tout, 2001).
Therefore, more highly educated mothers are more likely to utilize non-family care and require
this care for more hours per week. If there is a match between a highly educated, employed
mother’s employment preference and status, it can be hypothesized that she may experience low
levels of MSA; conversely, a mismatch is expected to produce higher levels of self-reported
However, existing literature is contradictory in its findings related to MSA and maternal
education. Some studies have not found a reliable link between maternal education and MSA
(Gee & Vondra, under review; Hock & Schirtzinger, 1992), but others have established an
association (Hsu, 2003; McBride & Belsky, 1988; NICHD ECCRN, 1999). These conflicting
results could be a product of sample differences. Conceptually, there are at least two rationales
for expecting links between maternal education and MSA.
One rationale concerns the environments in which well-educated, working mothers
function. The more educated a mother, the more likely that she was socialized to value
education, career goals, and/or personal and economic independence. Secondly, highly educated
mothers who work may also be socialized by their co-workers or others in their social networks
to focus on positive aspects of center-based care and social benefits children receive from the
experience. This socialization should reduce a working mother’s feelings of guilt and distress
related to putting her child in center-based care, thereby reducing her separation anxiety.
Despite these rationales, both Gee and Vondra (under review), and Hock and Schirtzinger
(1992) found no significant association between MSA and maternal education. The sample in the
Gee and Vondra (under review) study, however, was limited in size (N = 46), socio-economic
diversity (combined family income ranged from $50,000 to over $150,000, with a mean of
$100,000), maternal age (M = 33.51, SD = 3.64), and ethnicity (White only). The average age of
the children in this investigation was 2½ years at the time MSA was measured. The Hock and
Schirtzinger (1992) sample was slightly larger (N = 87), consisting of working- and middle-class
mothers (mean combined family income = $54,251), though, again relatively older (M = 33 years
of age), White women with an average of 15 years of education. The average age of the children
in their investigation was 6 years when MSA was measured.
Using a younger sample of 53 White, middle-class mothers (M age = 27 and M education
= 15 years), however, Hsu (2003) found that mothers with more education reported lower levels
of maternal separation anxiety when their infants were only 1 month (r = -0.28, p<.05) and 6
months old (r = -0.30, p<.10). McBride and Belsky (1988) also assessed MSA in a somewhat
younger sample of 63 predominantly White, working- to middle-class women (M age = 27, and
average education = 16 years) when infants were only 3 and 9 months of age. Using a regression
model, the investigators found that including education (combined with maternal age and family
income) in a block of variables explained a significant amount (17%) of concurrent variance in
maternal self-reported separation anxiety (F(3,62) = 4.36, p<.01). Further, the only variable that
made a significant, unique contribution to the prediction of MSA was maternal education (β = -
1.34, p<.01). The more years of education a mother reported, the lower her self-reported MSA.
A theme that emerges from these studies is that child age may be one critical variable
related to maternal separation anxiety and its connection to maternal education. The studies that
did not find an association between MSA and maternal education had mean child ages of 2½
years (Gee & Vondra, under review) and 6 years (Hock & Schirtzinger, 1992). Conversely, the
studies that did find an association between MSA and maternal education had mean child ages of
3.5 months (Hsu, 2003), 3 and 9 months (McBride & Belsky, 1988), and 6, 15, 24 and 36 months
(NICHD, ECCRN, 1997). Therefore, it is more likely that an association with maternal education
will be found if the sample of mothers have children who are infants, as opposed to preschool
and school-aged children. The sample in the present investigation is a sub-sample of Wave 1 of
the NICHD Study of Early Child Care, which spans target child ages from birth to 36 months.
So, it remains an empirical question how (or even whether) maternal education will relate to
maternal separation anxiety.
Maternal separation anxiety thus appears to be a more salient issue to mothers who prefer
to be at home with their child, have a first born in infancy, and have no more than a college
education. But is it an issue with significant consequences for the mother-child relationship?
Unless the construct has bearing on mother-child interaction and/or mother-child relationship
quality, does it hold much meaning in the field of developmental psychology? In the next
section, the association between MSA and a potentially key component of mother-child
interaction, maternal behavior during actual separations and reunions at the child care site, and
between MSA and patterns of infant-mother attachment, will be discussed.
2.5 LINKS BETWEEN MATERNAL SEPARATION ANXIETY AND MOTHER-
When a mother leaves her child in non-maternal care, the mother-child dyad separates. However,
much of the literature on separation anxiety has focused exclusively on the child, ignoring the
mother’s perspective on the separation and her contribution to the process (Hock, 1984). Despite
the importance of the mother’s contribution to the separation, there is very little literature about
the feelings a mother has at separation, and how these feelings may affect her behavior. Yet
maternal separation and reunion behavior may ultimately shape the quality of attachment her
infant makes with her. In this study, mother-child relations are operationalized as maternal
separation and reunion behavior and attachment security.
2.5.1 Maternal separation and reunion beh
Despite the wealth of parenting studies, little relates to maternal separation anxiety. It is unclear
whether and how MSA is associated with various parenting behaviors. MSA may manifest itself
at the childcare setting as lingering for an extended period of time at drop-off, discussing and
questioning the caregivers, controlling interactions with the child, and/or not focusing on what
the child has to say about his/her experiences during the day. In other words, drop-off at
childcare (when the mother and child separate) and pick-up from childcare (when the mother and
child reunite after an extended separation) are times when MSA may be particularly salient.
Documenting that a mother’s self-reported MSA is associated with her behavior during childcare
separations and reunions assists in establishing the validity of the MSA measure.
If a mother’s MSA interferes with her ability to navigate distance and closeness with her
child, her behavior may prevent the child from reaching developmentally appropriate
independence and exploration. For example, mothers who experience a great deal of separation
anxiety may maintain a degree of physical and/or psychological closeness, protection, and/or
control that interferes with their child’s attempts at independence. Conversely, mothers who feel
very little MSA may not be receptive to actual child neediness and dependency. Therefore, self-
reported MSA at either end of the range in a given population may reflect a mother-child
relationship that is out of psychological and/or developmental balance.
According to Hsu (2003), a mother experiencing higher levels of separation anxiety
(negative affect) may generally be more aware of and responsive to her infant’s negative versus
positive affect. This researcher found maternal reported MSA and response to negative infant
affect correlated at 1 month (r = .34, p<.05), and at 6 months (r = .40, p<.05) after birth when
measured in the home during free play. Hsu (2003) defined maternal responsiveness as a mother
modifying her behavior to attend to her infant’s signal (verbal and/or physical) within three
seconds of the child’s signal during free play in the home. A mother’s own negative affect and
increased feelings of “guilt, sadness and/or worry” related to separation from her infant may lead
mothers high in MSA to be more responsive to their infant’s negative affect (fussiness, crying) in
general. Alternatively, a mother higher in generalized anxiety or depressive symptoms may both
endorse more questionnaire items indicative of MSA and respond differentially to infant negative
Hsu (2003) also noted a non-significant trend for higher levels of self-reported MSA to
be associated with lower response rates to infant positive social signals, again at 1 month (r = -
.25, p<.10) and at 6 months (r = -.31, p<.10). The difference in associations of MSA with
maternal responsiveness to negative versus positive behaviors was significant at both 1 month
and at 6 months.
In contrast to the selective responding to infant affect documented by Hsu (2003), Hock,
McBride and Gnezda (1989) found differences in maternal responsivity to be contextually based
in a sample of 53 White, middle-class, mothers. Hock and her colleagues created a lab-based
observational rating measure of MSA concerning maternal behavior at separation and reunion.
This measure consists of a series of Likert-type ratings, where 1 indicates lower levels of
observed MSA and 5 indicates higher levels of observed MSA. Examples of behavior that
indicate higher observed MSA on this scale include: at separation, a mother lingering long or
asking questions related to the caregiver’s ability to care for her child; at reunion, a mother
verbally expressing anxiety, or comforting a child who is not in distress. The average correlation
between self-reported MSA and observed departure behavior (in the laboratory) was not,
unfortunately, significant (r = .12). The average correlation between self-reported MSA and
reunion behavior, in contrast, was modest to moderate in size (r = .40, p<.05). Mothers anxious
about separation may resist the urge to respond to and/or hover over their infants during
departure at center-based care, in order to present a calm front and/or reassure their child during
the separation process. At reunion, mothers high in self-reported MSA may, however, attempt to
“compensate” for the separation by demonstrating especially solicitous behavior toward their
Hock and her colleagues reasoned that a mother’s behavior during reunion is a better
indicator of her true anxiety related to the separation than her (perhaps more self-conscious)
behavior at departure. When a mother picks up her child from day care, according to their logic,
she may display more of her feelings of guilt and distress related to the earlier separation from
her infant because she is no longer trying to prepare her infant for a separation. At reunion,
highly anxious mothers may display more anxious, hovering, and/or overly solicitous behavior
because they no longer feel the need to present a calm facade. Alternatively, the anxiety and/or
guilt related to leaving their child may intensify as work is set aside and reunion nears. At
reunions, mothers may display more anxious behavior (trying to sooth a child who is not in
distress, questioning the child about whether the child missed her while she was gone) with their
child because they are not in the physical and psychological process of separating from them and
no longer feel a need to monitor their own behavior as carefully. If this is the case, the
association between self-reported maternal separation anxiety and reunion behavior will be
stronger than the association between MSA and departure behavior.
These two studies (Hsu, 2003; Hock et al., 1989) established a link between MSA and
maternal behavior. Importantly, Hock (1984) theoretically extended this link from MSA to
maternal behavior, to child adjustment and adaptability in non-maternal care. She hypothesized
that the way early separations are handled by the dyad influences how subsequent separations are
experienced. If early separations go smoothly, and the infant associates separations with neutral
or even positive emotions, subsequent separations are also expected to be relatively easy. If,
however, early separations do not go well, and fear and/or apprehension are connected with the
process of separating, over time the heightened anxiety of the infant will influence maternal
behavior, and the mother’s behavior will, in turn, increase the anxiety both mother and child
experience at separations. Thus, according to Hock (1984), MSA has implications not only for a
mother’s behavior in separation/reunion contexts, but also for her child’s adjustment in the
relevant care context. This suggests that the relationship between infant and mother may be
involved as well. In the next section this possibility is explored further.
2.5.2 Mother-infant attachment security
Attachment security refers to the emotional bond an infant and his/her primary caregiver, usually
the mother, have (Bowlby, 1969). The quality of the attachment an infant develops with his/her
primary caregiver is believed to influence the degree of independence an infant demonstrates,
and how he/she deals with new situations and people. This first bond is also believed to serve as
a blueprint for future relationships, and has theoretical implications for child (and adult)
development. For instance, children with a secure attachment are more likely to be rated as
optimistic and empathic towards others, showing greater competence with peers, and better at
developing friendships (Sroufe, 1988).
Attachment theory includes a place for a moderate amount of maternal separation
anxiety. Benedek (1970) considered moderate amounts of maternal separation anxiety as a
“normal” aspect of motherhood. Early on, this involves mothers being in close proximity to their
infant and continuously monitoring their infant’s behavior. As the infant grows older, a mother’s
role involves recognizing the child’s developmentally appropriate need for independence.
McBride (1990) stated that mothers of infants tend to view them as emotionally vulnerable, but
as the infant gets older mothers come to realize that their infant can handle such experiences as
regular separations. Therefore, a mother’s MSA should decline over time. Hock and Schirtzinger
(1992) found that mean self-reported MSA scores did diminish between birth and the child age
of 6 years, but that individual differences remained relatively stable over repeated measures.
High self-reported level of MSA by the time one’s child reaches the age of 6 years may reflect
some non-adaptive aspect of the mother’s personality.
In point of fact, higher than average maternal-reported MSA has been linked with a
mother’s insecure adult working model of attachment. Lutz and Hock (1995) reported that those
mothers higher in MSA who tended to remember their own mothers as “rejecting” and not
encouraging independence were also more often categorized as having an insecure working
model of adult attachment (as assessed by the Adult Attachment Interview) at infant age two
months. Lutz and Hock (1995) demonstrated that higher MSA scores are directly associated with
insecure adult attachment models, but did not clarify which (or both) sub-classifications of
insecure attachment, because the researchers dichotomized the attachment variable into secure
and insecure classifications (overall, 8% dismissing, 13% preoccupied) in order to increase the
size of the groups used in their multivariate analysis (Lutz & Hock, 1995).
Findings related to particular categories of insecure attachment are provided by McBride
and Belsky (1988). These researchers found that the mean level of MSA reported by mothers of
securely attached infants was in the mid-range for their sample, but that higher reported MSA
and lower reported MSA were linked to different forms of insecure attachment, avoidant and
resistant, respectively. However, McBride and Belsky (1988) did not clarify what constitutes
“higher” in their middle- and working-class, White sample. MSA scores ranged from 21 to 105,
a broader range than the scores in Hock and Schirtzinger’s (1992) research with a sample of
exclusively working-class, White mothers (28 to 84, with scaling adjusted for comparison
purposes). According to the results of their research, “mothers of secure infants expressed a
moderate degree of anxiety – less than mothers of avoidant (Type “A” attachment) infants but
more than mothers of resistant (Type “C” attachment) ones” (McBride & Belsky, 1988, p. 409).
This result was contrary to their research hypothesis; mothers of insecure-avoidant infants are
usually seen as not admitting to negative feelings (George & Solomon, 1999) and often, in
middle-class samples, projecting a competent demeanor (Cassidy & Kobak, 1988). There are,
however, studies that link maternal separation anxiety with intrusive parenting (Berger & Aber,
1986; McBride, 1983) and, in separate studies, link this type of parenting with insecure-avoidant
attachment (Belsky, Rovine & Taylor, 1984; Isabella, Belsky & von Eye, 1987; Lewis & Feiring,
McBride and Belsky (1988) hypothesized that the MSA measure may avoid the defense
mechanisms of insecure-avoidant mothers to extract some of their true feelings about separating
from their child. The wording of the questions and the reverse coding of some items (e.g., item #
5: “If a child is independent and outgoing, he/she will make friends easily without his/her
mother’s help,” Hock, McBride & Gnezda, 1989) may have provided the means to bypass the
defensive reactions of some mothers. This focus on the child could glean more accurate
responses from mothers who tend to be defensive about their own competence. Therefore, either
the MSA measure is designed in such a way as to avoid triggering the defense mechanisms of
mothers whose infant has an insecure-avoidant attachment with her, or McBride and Belsky’s
(1988) findings are specific to their sample of 63 White, middle- and low-income, mothers living
in rural Pennsylvania, who had, on average, almost a college education (M = 15.6 years, SD =
Stifter, Coulehan, and Fish (1993) found mixed results regarding the association between
MSA and the quality of infant attachment, some of which support McBride and Belsky’s (1988)
findings, and others which conflict. In a sample of 73 employed and non-employed, middle-
class, mostly White mothers (mean age 29 years, average education 15 years), they assessed
MSA shortly before infants reached 5 months of age, and attachment when infants reached 18
months of age (using the Strange Situation assessment). Stifter, Coulehan and Fish (1993) found
no association between MSA at 5 months and infant attachment at 18 months when analyzed
using the A (avoidant), B (secure), and C (resistant) classification system or by dichotomizing
attachment into secure and insecure categories. However, the investigators did find that (only)
employed mothers who self-reported high MSA had a disproportionate number of infants with
insecure-avoidant attachments. According to Stifter, Coulehan, and Fish (1983), this result could
indicate that employed mothers who report high levels of maternal separation anxiety may be
more likely to be “out of synch” with their infants and these mothers also may be very
controlling when they spend time with their infants.
Using multiple regression analysis in a sample of 66 White, lower to middle class,
employed mothers and fathers (average age of 29, mean education of 14 years), however, Wille
(1998) found neither linear nor curvilinear relations between MSA and infant attachment
(assessed using the Attachment Behavior Q-set) when infants were 6 and 18 months of age.
It is important to note that the findings for infant-mother attachment, at best, may relate
only to types of attachment insecurity, and not the overall frequency of security or insecurity.
This is different from findings for mothers’ own working model of attachment, using the Adult
Attachment Interview, wherein maternal insecurity (of any form) covaries with higher MSA. So,
it remains an empirical question how (or even whether) infant-mother attachment will relate to
maternal separation anxiety.
The inconsistencies apparent in the results of these four studies have a variety of
explanations. Sampling bias could be one of them. The samples in all of these studies are
relatively small, varying from 49 to 73 cases. In most cases, these samples were then divided
according to attachment classification for comparison purposes, resulting in very small sub-
samples. Small samples size is associated with restricted power in statistical analysis, which can
lead to a Type II error, or failing to reject a false null hypothesis (Pallant, 2006). This would lead
one to believe no difference exists, when, in fact, there is one (though typically modest in size).
Despite the inconsistencies in results, higher maternal self-reported MSA has been linked in two
of four studies with an insecure infant-mother relationship.
In sum, the research relating maternal behavior or infant-mother attachment insecurity
with self-reported maternal separation anxiety is scanty, inconsistent, and based on generally
small sample sizes. The samples themselves are often unrepresentative, even of White, middle-
class, employed mothers. It seems likely that such broad variables as maternal employment,
social class, and race might all have some cultural influence on a mother’s tendency to
experience (and report) separation anxiety with her infant. In the present investigation, maternal
employment and race, and to some extent education, are held constant through selective
sampling from a large, national sample of mothers. Both convergent and discriminant validity
need to be examined for the MSA construct. Ultimately, the value of understanding MSA is in its
importance for mother-child interactions and relationships (convergent validity). But if there is
an association between MSA and either neuroticism or depressive symptoms, one must then ask
whether MSA can explain variance in maternal behavior over and above these broader indices of
a mother’s (characterological) tendency to experience negative affect (i.e., discriminant validity).
The discussion thus far has focused on constructs associated with MSA and investigated
in the current research. However, there are others that will not be examined empirically. While
the following constructs are hypothetically important to the idea of maternal separation anxiety,
their omission from this analysis was undertaken specifically to maintain the focus of this
research on the maternal perspective, and to minimize, to the extent possible, measurement
2.6 OTHER CONSTRUCTS POSSIBLY RELATED TO MATERNAL SEPARATIO
2.6.1 Maternal tempera
Maternal temperament (inhibition, negative reactivity, and sociability) may be related to
maternal separation anxiety. Maternal perceptions of and reactions to child inhibition and/or
negative reactivity are, themselves, a likely product of maternal personality (Sameroff, Seifer &
Elias, 1982). A mother who is highly reactive may respond more strongly to an infant’s cry at
departure, exacerbating her own and her infant’s negative affect. A mother who is inhibited may
be overly solicitous with her infant or toddler at separations, possibly increasing her child’s
distress (and her separation anxiety), or may simply interpret normative infant wariness as more
extreme inhibition or negative reactivity. So there may be an inherent confound between MSA
and maternal temperament when considering correlates of MSA.
2.6.2 Child tempera
Another likely contributor to maternal separation anxiety is the child’s temperament. Having a
truly inhibited and/or negatively reactive child may contribute to a mother’s MSA. The mother
of an inhibited child may worry that her child is not adapting to non-maternal care once she
leaves. The mother of a negatively reactive child, who displays high levels of distress at
separation, may become increasingly anxious as she witnesses the child’s negative emotional
displays. Thus, there is theoretical support for an association between child temperament and
MSA (McBride, 1990).
There is, however, a measurement issue related to assessing infant temperament that must
qualify conclusions drawn from the results of these two studies. If mothers complete the child
temperament assessments, there is a data source confound with self-reported MSA. Results may
be biased because of the subjective nature of both mother’s report of her own separation anxiety
and her perception of the child. Furthermore, after 6 months or so, a child’s observed
“temperament” reflects both genotype and rearing environment, including quality of attachment
to mother. So, although the construct of child temperament is theoretically important to consider,
methodologically it is problematic. Partly for this reason, the focus of this investigation will not
be broadened beyond maternal personality, behavior and mother-child attachment security.
2.6.3 Quality of non-maternal care
Another issue of likely importance to maternal separation anxiety is the quality of care available.
The type of care a mother chooses is associated with maternal education (NICHD ECCRD,
1997). More highly educated, higher income mothers are more likely to choose center-based care
(Capizzano & Adams, 2003), and their economic situation enables these mothers to afford better
care than lower income, less educated mothers (NICHD ECCRN, 1997). McBride (1990) found
that mothers who indicated that they had trouble finding quality childcare reported higher levels
of MSA. The ability to secure quality care for one’s child will tend to make mothers feel better
about their decision to put their child into non-maternal care, presumably resulting in lower
MSA. Not only can better-educated mothers afford higher quality care, but they have more
choices in the type of care they choose. Some employed mothers want center-based care so that
their child–especially firstborns–can benefit from the socialization available at these centers.
Others choose in-home care so that their child has the advantage of the familiarity of his or her
own home. Still others desire neighbor care, with their child integrated into the family life of the
caregiver. The issue of perceived quality of care and its relation to MSA is potentially a key one,
but is not the focus of the proposed research, and is not included in the analysis. This limitation
will be addressed more fully in the discussion section.
2.7 STUDY HYPOTHESES AND ANALYTIC PLAN
This cross-sectional and longitudinal research is a descriptive study of maternal separation
anxiety in a sample of well-educated, married White mothers. First, a descriptive analysis of
maternal separation anxiety will be conducted. The mean and range of reported maternal
separation anxiety scores and the distribution of these scores will be analyzed and reported.
Then, in an attempt to replicate previous research, an examination of maternal separation anxiety
over time (from infant age 6 to 24 months) will be performed by repeated measure ANOVA.
Based on previous research (Hock & Lutz, 1998; Hock & Schirtzinger, 1992; Hsu, 2003;
McBride & Belsky, 1988), it is expected that, over the course of the three measurement
points, there will be a significant decrease in MSA. The present study will also attempt to
replicate stability coefficients by correlational scores across study time points. Based on Hock
and Schirtzinger (1992), it is expected that stability coefficients will be on the order of .70 to
Following this, MSA will be correlated with both overall maternal neuroticism as well as
depressive symptoms. Based on existing literature, it is predicted that the association between
MSA and neuroticism at 6 months (and between MSA and extraversion and agreeableness
at 6 months; as an exploratory analysis) will be on the order of .3 to .4 based on associations
reported by Hock and Lutz (1992), Hock and Schirtzinger (1992), and McBride and Belsky
(1988). As far as maternal depressive symptomatology, there are only two studies that examined
its association with MSA. The results of these two investigations contradict each other, and their
samples are dissimilar. Hsu (2003) using a sample of mothers of infants (ages 1 and 6 months of
age) found no significant association between MSA and maternal depressive symptoms. Hock
and Schirtzinger (1992) found a significant association between MSA and maternal depressive
symptoms at child age 6 years, but not earlier. Therefore, it remains an empirical question
how (or even whether) maternal depressive symptoms will relate to maternal separation
Next, the association between a mother’s level of education and her self-reported
separation anxiety will be examined. Previous literature indicates that education has a modest to
moderate, inverse association with self-reported MSA, depending on child and/or mother’s age
(Hsu, 2003; NICHD ECCRN, 1999; McBride & Belsky, 1988). Based on these results, it is an
empirical question how (or even whether) maternal education will relate to maternal
separation anxiety. Whether maternal education is concurrently associated with self-reported
MSA at 6, 15 and 24 months will be tested by correlation. Additionally, in order to examine
whether experience as a parent has any relevance to the MSA a college-educated mother
experiences, an examination of the associations between MSA and maternal education will be
performed in two sub-groups: mothers of firstborns and mothers of laterborns. Mothers with
more than one child, and therefore more experience as a parent and potentially with non-
maternal care, are expected to report lower MSA, regardless of education level in this well-
educated sample. However, again, it is an empirical question how (or even if) the
associations between MSA and education in these two groups of mothers will differ.
Correlations will be compared by Fisher z-test.
In addition to documenting its prevalence and longitudinal course among well-educated
women, this investigation tests whether maternal separation anxiety is related to mother-infant
relations (maternal separation and reunion behavior and infant-mother attachment security), and
whether those relations are merely a manifestation of general neuroticism and/or depressive
symptomatology on the part of mothers. Previous research indicates that maternal reunion
behavior in a childcare setting (e.g., whether or not mother seems rushed at pick-up, whether or
not mother listens when child speaks at pick-up) will moderately correlate with self-reported
separation anxiety. Maternal separation behavior in the same setting (e.g., how long mother stays
at center before leaving, whether or not she tries to engage child in an activity before she leaves)
will be correlated only modestly with self-reported maternal separation anxiety. Based on Hock,
McBride and Gnezda’s (1989) single study on the topic, it is expected that concurrent
associations between MSA and reunion behavior will be in the range of .3 to .5, and that
associations between MSA and separation behavior will be in the range of .1 to .2 at both
15- and 24-month assessments.
In order to detect a small effect (.1 to .2) in a correlational analysis at p<.05, a sample of
783 participants is required (Cohen, 1992). Unfortunately, the full NICHD dataset only includes
observations of maternal behavior for 521 mother-child dyads (NICHD, Child Care Data Report,
1994). However, for a medium effect size (.3 to .5), a sample of only 85 participants is required
(Cohen, 1992). The sub-sample selected for this analysis consists of a total of 147 participants,
with a range in number from 97 to 147 for different analysis (depending on the measure), and an
average sample size of 120. Therefore, it is unlikely that a small effect size (i.e., the expected
correlation with maternal separation behavior) will be detected. However, there should be
sufficient power to detect a moderate effect size (i.e., the expected correlation with maternal
The next analyses will test whether a mother’s self-reported separation anxiety at 6
months (and then again at 15 months) has any bearing on the subsequent insecurity of attachment
her baby demonstrates in relation to her at 15 months. Based on the conflicting results of
previous literature, it remains an empirical question how (or even whether) maternal
separation anxiety has any connection to insecure infant-mother attachment classification
(i.e., avoidant and resistant).
Finally, analyses of mother-child relations will be re-run controlling for general maternal
neuroticism and tendency to experience symptoms of depression. If MSA has meaning for
maternal behavior and mother-child attachment beyond more general anxiety and depressive
symptoms, MSA should continue to covary with maternal behavior even after maternal
personality traits are statistically controlled. Linear regression will be used to test this hypothesis
for maternal behavior. Logistic regression will be used for attachment classifications.
This investigation is unique in that the women represent a narrow, but often over-looked
segment of the American population that uses non-maternal childcare: affluent, well-educated
professional women. Much developmental research relies on samples of middle-class children,
with little attention paid specifically to high-income families, parents in particular. Luthar (2003)
argued that this may be due to the misconception that middle-class families are very similar to
affluent families, and/or that affluent families do not have problems. However, the limited
research that does exist suggests that the kinds of pressure affluent women experience, combined
with the importance of privacy and maintaining a ‘gracious’ facade, result in increased rates of
self-medication with alcohol and/or prescription drugs, as well as eating disorders such as
anorexia, bulimia, and/or extreme dieting (Luthar, 2003). In addition, there are statistics showing
high rates of depressive symptoms among affluent, professional women (Luthar, 2005). These
problems have important implications for mother-child relationship quality as well as other
aspects of family and child well-being.
Well-educated, high earning mothers may feel pressured to work long hours for a variety
of reasons, such as the desire to achieve personal or professional success or to maximize
resources for the next generation (Luthar, 2003). However, work time can detract from family
life. Working longer hours can lead to less time for family, and some erosion in family
relationships. The latter may increase feelings of unhappiness and self-doubt even if, for some
mothers, longer work hours represent an escape from problems at home and/or an effort to gain
self-esteem. A sub-set of mothers maintain the time they spend with their children by working
part-time (Bianchi, Robinson, & Milkie, 2006). But many mothers do not feel they have the
option to work part-time, whether for economic, personal, or professional reasons.
There is empirical justification, then, for a descriptive, partly exploratory examination of
maternal separation anxiety in a sample of affluent, well-educated, professional women. First,
there has been relatively little research into Hock’s construct of maternal separation anxiety
(MSA). A review of the literature produced less than 50 peer-reviewed journal articles that
utilized her self-report measure since it was published in the mid-1980’s, and few of those
specifically focused on a maternal perspective. Even fewer have focused on any particular sub-
group of mothers, instead combining women across sometimes widely disparate social classes,
job conditions, and family circumstances. Therefore, the proposed study is an attempt to redress
that imbalance by investigating maternal separation anxiety in a sample of affluent, well-
educated, employed mothers who utilize non-maternal care for their children.
E 3.1 SAMPL
The participants for this investigation are taken from the NICHD Study of Early Childcare,
Phase I (NICHD ECCRN). The initial NICHD sample was recruited beginning in 1991 in 10
cities across the United States (Little Rock, AR; Irvine, CA; Lawrence and Topeka, KS; Boston,
MA; Philadelphia, PA; Pittsburgh, PA; Charlottesville, VA; Morgantown and Hickory, NC;
Seattle, WA; and Madison, WI). Recruitment was not intended to be representative of the United
States general population; however, subsequent comparisons of this sample with the
demographic characteristics of the general population indicate that this sample is similar to the
general public on a number of key demographics, for example average family income and ethnic
group representation (NICHD ECCRN, 1999). Initially, 8,986 families were contacted related to
participating in the study; however, 134 (2%) refused the initial interview, and 310 (4%) refused
to be interviewed during a follow-up telephone call at 2 weeks postpartum, therefore
approximately 6% or the original pool of participants refused to participate. Further, 3,721 (41%)
were ineligible because they did not meet study criteria (mother over 18, native English speakers,
home visit at 1 month post-partum). The resultant pool of eligible participants was 5,265 (59%).
Then, 1,364 eligible families were randomly selected for participation in the study, resulting in a
sample of 26% of eligible participants.
Mothers were initially contacted regarding participation while still in the hospital shortly
after they had given birth. The first data collection occurred when infants were one month old.
Data across the duration of the study were collected in the home, over the telephone, in a lab and
at center-based care. The various methods of data collection include maternal self-report,
interview, and observation (by both research staff and childcare workers). Time points for data
collection were 1, 6, 15, 24, and 36 months of infant age.
The population of interest for the present investigation is well-educated, married mothers
who worked at least 20 hours per week consistently throughout Phase I of the NICHD study
(birth to child age 36 months). Further, because of the theoretical influence of culture on MSA
(Hock, DeMeis & McBride, 1987), only Caucasian mothers are included. These steps were taken
in order to produce the most functionally homogeneous sample of mothers to study (Richters,
1997). Functional homogeneity is important because there are then fewer confounds in
associational analyses and because it allows greater insight about the behavior of a specific group
of people. For example, including single mothers in this analysis (a functionally different group
of mothers from married mothers) could result in a great deal of variability in reported MSA, but
it could be the stress of being a single mother, as opposed to her experience of leaving her child
in non-maternal care, that accounts for the variance in MSA.
The various filtering activities resulted in a sub-sample of 147 mothers, or 11% of the
NICHD sample. Among these women, the range of education (highest degree completed)
extended from at least a bachelor’s degree (16 years of education, n = 111), to a Master’s degree
(18 years of education, n = 69), law degree (19 years of education, n = 7), and doctoral degree
(MD, PhD, EDD, n = 11). The range of hours worked per week was 20 to 70 (M = 36 hours per
week, SD = 10). Total family income for this sample ranged from $20,000 to $315,000 (M =
$80,662, SD = $50,508) at 6 months and from $22,500 to $252,500 (M = $77,445, SD =
$38,340) at 15 months. Although these ranges contain some relatively low incomes, at each
measurement point the percentage of families making above $50,000 (4 times the poverty rate at
the time of measurement) ranged from 78% (at 6 months) to 83% (at 24 months), making this a
predominantly affluent sample. All mothers were stably married over the course of Phase I of the
NICHD data collection (i.e., married at 1, 6, 15, 24 and 36 months). Mothers’ ages at the 1
month assessment ranged from 24 to 46 years (M = 32, SD = 4, Mode = 30). Sample descriptive
statistics are provided in Table 2.
Table 2. Sample descriptive statistics
Variable N Minimum Maximum Mean SD
Maternal Age 147 24 43 32.30 4.0 years
Maternal Education 147 16 21 17.18 1.53 years
Income-Total Family at 6 mo. 143 $20,000 $315,000 $80,662 $50,508
Worked/Week 6 mo. 147 20 70 37.64 9.69
Worked/Week 15 mo. 147 20 56 38.22 8.25
Worked/Week 24 mo. 147 20 64 38.00 8.88
Primary – 5 months 138 6 55 36.01 11.84
Secondary – 5 months 26 4 38 15.00 9.65
(Table 2 continued)
Tertiary – 5 months 6 4 10 7.17 2.32
Primary – 14 months 146 4 59 36.25 12.31
Secondary – 14 months 40 4 45 12.73 10.31
Tertiary – 14 months 5 2 8 6.20 2.50
Primary – 23 months 146 2 55 37.48 12.54
Secondary – 23 months 50 1 60 11.02 11.63
Tertiary – 23 months 5 4 16 8.60 4.45
Data were collected using self-reports, observations, and interviews. A complete listing of
construct measurement time points, locations where assessments took place, and who completed
the instruments can be found in Table 3.
Table 3. Construct measurement time points, locations, and administrators
Construct Administrator 1 Month 6 Months 15 Months 24 Months 36 Months
Maternal Separation Anxiety Scale Self-report Home Home Home Lab
NEO Personality Inventory:
Neuroticism, Extraversion, and
Symptomatology (CES-D) Self-report Home Home Home Home Home
Maternal Education Interviewer Home
Maternal Behavior: Child Care
Separation & Reunion Scale (Ratings
of Drop-off and Pick-up Behaviors)
Caregivers Childcare Childcare Childcare
Mother-infant Attachment (Strange
Situation) Research Team Lab Lab
Timepoints generating data for the
3.2.1 Maternal Separation Anxiety
The Maternal Separation Anxiety Scale (Hock, McBride & Gnezda, 1989) consists of 35 items
separated into three subscales, each with a possible range of 7 to 35. The Maternal Separation
Anxiety subscale is a 21-item scale and the only subscale of the Maternal Separation Anxiety
Scale utilized by the NICHD ECCRN study. The Maternal Separation Anxiety subscale and was
administered at 1, 6 and 15 months in the home and at 24 months in the lab. It measures the
degree to which a mother experiences anxiety during separations with their child, with a high
score indicating that a mother experiences concern about the separation event itself. Responses
are rated on a Likert scale with 1 = Strongly Disagree and 5 = Strongly Agree. The scale covers
issues of importance to parents, and includes statements such as, “I miss holding or cuddling my
child when I am away from him/her” and “My child is happier with me than with babysitters or
teachers.” Sub-scale totals are calculated by reverse coding particular items and summing
3.2.2 General anxiety
The NEO Personality Inventory (NEO PI) and NEO five-Factor Inventory (NEO-FFI) (Costa &
McCrae, 1989) were completed by mothers in the family home when children were 6 months
old. The complete measures were not administered; only certain aspects of personality were
assessed. From the NEO PI, neuroticism and extraversion were measured; from the NEO FFI,
agreeableness was measured. The neuroticism scale measures a person’s emotional stability and
adjustment. High scores on this measure indicate nervousness, insecurity, and feelings of
inadequacy; low scores indicate calmness, feelings of security, and low emotional reactions. The
extraversion scale measures the amount and degree of a person’s interaction with others, level of
activity, and tolerance of stimulation. High scores on this measure indicate outgoing, positive,
and energetic individuals; low scores indicate unfriendly, calm, and “standoffish” individuals.
The agreeableness scale measures various aspects of an individual’s interpersonal style such as
expectations of others, conformity, and self-sacrifice. High scores indicate empathy and
collaboration, whereas low scores indicate competition with others and stubbornness.
3.2.3 Depressive symptoms
Mothers’ reported depressive symptoms over the previous week were measured using the Center
for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977). This 20-item scale has a
possible range of 0 to 60, with higher scores indicative of more symptomatology. Scores falling
between 0 – 16 indicate no depressive symptoms, those between 16 – 30 indicate mild to
moderate depressive symptoms, and any score over 30 indicates severe depressive symptoms
(Radloff, 1977). Item examples include, “Your sleep was restless,” “You felt lonely,” and “You
did not feel like eating; your appetite was poor.” Internal consistency (Cronbach’s alpha) for this
sub-sample is .88 (at child age 6 months), .86 (at child age 15 months), and .89 (at child age 24
months). Scores for this measure ranged from 0 to 30, M = 7, SD = 6.74 (at 6 months) to 0 to 41,
M = 7, SD = 6.68 (at 24 months).
3.2.4 Maternal edu
Maternal education was measured by asking mothers how many years of education they have
completed. Sixteen years of education was coded as a bachelor’s degree, 18 years as a master’s
degree, 19 years as a law degree, and 21 years as a doctoral degree (PhD, MD, EDD). The range
of responses for this sub sample is 16 to 21 years. Maternal education was assessed at the time of
the baby’s birth.
3.2.5 Maternal behavior
The Child Care Separation/Reunion Scale (McCartney & Beauregard, unpublished) was
completed by caregivers at the childcare site when the child was 15, 24 and 36 months of age.
This measure consists of 17 items–10 for departures, 7 for reunions–scored from 1 (Never) to 4
(Always) with higher scores indicative of more involved, responsive parent behavior (Bub &
McCartney, 2004). Examples of items at departure include, “Before leaving, parent tries to
interest child in an activity,” and “Parent stays too long before leaving.” Examples of items
related to parent behavior at reunion include, “Parent listens attentively when child speaks (or
makes sounds),” and “In general, parent seems rushed.” Scores are calculated by summing
For the original NICHD dataset, 692 caregivers assessed departure and reunion behavior
at center-based care. Instructions provided to caregivers was that they, “Please read each item
below and rate how often the behavior occurs on a scale from 1 (never) to 4 (always).” In other
words, the measures may not have been completed during actual departures and reunions, but are
likely composites of caregiver recollections of parental departure and reunion behaviors. No
reliability information is available.
Hock’s (1989) maternal behavior scale assessed the amount of positive behavior a mother
demonstrated during departures and reunions (helping her infant/child adapt to the environment
prior to departure, solicitous behavior demonstrated with the child, and/or exchanges with
caregiver). For the purpose of replicating Hock et al.’s (1989) findings, two composite scores
were computed. One composite used only the positive behavior items from the Child Care
Separation/Reunion Scale that are similar to Hock and colleagues’ ratings, (#14, “Before
leaving, parent tries to interest child in an activity.”). The other composite that was created
included relevant items that address “negative” behavior, but were significantly correlated
(average r = .43) with the positive items (#13, “Parent seems uncomfortable when child gives
attention to a caregiver.”). The Hock departure composite used items 14, 15, 16, and 17; the
Hock departure and negative items composite added items 13, 18, 20, and 22. The Hock reunion
composite used items 14 and 15; the Hock reunion and negative items composite added items 13,
The attachment classification assigned to infant-mother relationships was based on their behavior
in the Strange Situation (Ainsworth, Blehar, Water, & Wall, 1978) at 15 and again at 36 months.
The Strange Situation is an observational measure used to assess security of attachment an infant
has established with the primary caregiver. The procedure takes place in an unfamiliar
environment and consists of one introductory play episode and two separation and reunion
episodes, each lasting approximately three minutes. During the process, a stranger enters/exits
the room and interacts with the child at various points throughout the assessment. The entire
procedure is videotaped. Trained coders view the tapes and rate the infant’s behavior in terms of
seeking proximity with mother, maintaining contact with mother, so-called resistant behavior
(e.g., pulling back from mother’s embrace, pushing away toys mother offers), and avoidant
behavior (e.g., greeting mother, then averting gaze, veering off during an approach to mother).
Securely attached infants explore the room, periodically checking back with mother
visually. They may be distressed at separation but, if so, seek out and are comforted by mother.
Infants with an avoidant attachment reference mother less during exploration, often are not
distressed at separation, and show subtle or overt avoidance of mother at reunion. Infants with a
resistant attachment tend to hover around mother during exploration, are upset at separation, and
at reunion demonstrate both contact seeking and resistant behavior. Attachment using the
Strange Situation was assessed at both 15 and 36 months, but only the 15-month data are used
here because validity is more established for infant attachment. In addition, data from the 15-
month assessment are more consistent with previous research on MSA and infant-mother
There are no published data about the fourth, “disorganized,” classification in relation to
maternal separation anxiety, and a fourth category further limits attachment cell sizes for
analysis. For these reasons, only the secure, avoidant and resistant “forced” classifications are
used in analyses. The frequency and percentage of the attachment (forced) classifications are
listed in Table 4. In this subsample of well-educated, working mothers attachment classifications
approximate the frequencies found in typical middle-class, U.S. samples, (65% secure, and 15%
resistant), except for the low percentage of avoidant attachments (9% here versus 20% typically).
Relatively little avoidance was observed in this sample of well-educated, working mothers and
Table 4. Frequencies of forced attachment classification at 15 months (Strange Situation)
Classification Frequency Percent
Secure 101 69%
Insecure-Avoidant 13 9%
Insecure-Resistant 27 18%
Unclassified 6 4%
Total: 147 100%
Descriptive statistics were computed for the independent and dependent variables.
A list of descriptive statistics is provided in Table 5.
Table 5. Descriptive statistics for the independent and dependent variables
Variable N Min. Max. Mean SD
Maternal Separation Anxiety
6 mo. 147 25.00 96.00 58.70 11.85
15 mo. 147 29.00 91.00 57.82 11.47
24 mo. 141 31.00 97.00 56.40 11.10
Neuroticism 6 mo. 147 15.00 52.00 28.17 6.70
Extraversion 6 mo. 147 27.00 56.00 44.07 5.62
Agreeableness mo. 147 39.00 58.00 47.00 4.30
Maternal Depressive Symptoms
6 months 147 0 30 7 6.74
15 months 147 0 28 6.5 6.12
24 months 141 0 41 9 6.68
Maternal Education 147 16 21 17.17 1.53
Departure 15 mo. 88 2.29 4.0 3.52 .32
Reunion 15 mo. 76 2.57 3.43 3.12 .24
Departure 24 mo. 100 2.57 4.0 3.51 .33
Reunion 24 mo. 83 2.64 4.0 3.58 .32
(Table 5 continued)
Departure 36 mo. 100 1.57 4.0 3.50 .38
Reunion 36 mo. 84 2.43 4.0 3.54 .32
Further, internal validity for the measures used in this study was, on average, very high.
Cronbach’s alpha for each measure was at least .7. Internal validity values are listed in Table 6.
Table 6. Internal validity (Cronbach’s alpha) for the study variables
Variable N Cronbach’s alpha
Maternal Separation Anxiety
6 mo. 148 .91
15 mo. 147 .91
24 mo. 141 .91
Maternal Behavior at 15 Months
Hock Composite 100 .71
Composite & Negative Items 99 .85
Hock Composite 100 .89
Composite & Negative Items 100 .83
Neuroticism 6 mo. 147 .82
Extraversion 6 mo. 147 .76
Agreeableness mo. 147 .70
Maternal Depressive Symptoms
6 Months 146 .88
15 Months 147 .86
24 Months 136 .89
This study examined maternal separation anxiety in a sample of well-educated, married, White
mothers, its interface with maternal personality factors, its bearing on mother-child relations, and
whether it continues to covary with mother-child relations once maternal personality/depressive
symptoms are statistically controlled. The intercorrelations among the study variables are
reported in Table 7.
Table 7. Intercorrelations among study variables
MSA Mat. Maternal Maternal Depressive
6M 15M 24M Ed. Behavior Personality Symptoms
[ 1 2 3] 4 [5 6 7 8 9 10 11 12] [13 14 15] [16 17 18]
Maternal Separation Anxiety (self-report)
1) 6 mo.
2) 15 mo.
3) 24 mo.
.72*** .75*** 1
n 141 141
4) Maternal Education
-.06 -.16* -.08 1
n 147 147 141
Departure at 15 Months:
5) Hock Composite
.11 .05 .08 -.14 1
n 100 100 97 100
6) Composite & Negative Items
-.01 .02 -.01 -.17† .86***1
n 99 99 97 99 99
Reunion at 15 Months:
7) Hock Composite
-.12 -.11 -.06 .06 .09 .07 1
n 100 100 97 100 99 98
8) Composite & Negative Items
-.17† -.08 -.08 -.01 .05 .17 .88*** 1
n 100 100 97 100 99 98 100
(Table 7 continued)
Departure 24 Months:
9) Hock Composite:
-.01 -.43 .01 -.15 .18 .06 .07 -.13 1
n 52 52 51 52 35 35 35 35
10) Composite & Negative Items:
-.006 -.03 -.04 -.12 .08 .001 .05 -.14 .82*** 1
n 51 51 50 51 34 34 34 34 51
Reunion at 24 Months:
11) Hock Composite:
.02 .14 .04 -.14 -.08 -.07 .02 -.03 .54*** .69*** 1
n 54 54 53 54 37 37 37 37 52 51
12) Composite & Negative Items:
.18 .11 .07 .17 -.01 -.05 .07 -.07 .45** .63*** .78*** 1
n 54 54 53 54 37 37 37 37 52 51 54
.20* .12 .18* .14† .10 .02 -.13 -.18† -.16 -.06 -.11 .02 1
n 146 146 140 146 99 98 99 99 52 51 54 54
-.01 .00 -.03 -.02 -.02 -.02 -.34** -.26** .33* .30* .14 .13 -.31***1
n 146 146 140 146 99 98 99 99 52 51 54 54 146
-.02 -.01 -.03 -.15† .16 .12 .04 .05 .23 .005 -.01 .02 -.33*** 11 1
n 146 146 140 146 99 98 99 99 52 51 54 54 146 146
Maternal Depressive Symptoms
16) 6 months
.11 .03 .07 -.02 -.03 -.05 -.13 -.15 -.07 -.04 -.01 -.09 .67*** -.30*** .23** 1
n 147 147 141 147 100 99 100 100 52 51 54 54 146 146 146
(Table 7 continued)
17) 15 months
.05 .12 .19* -.05 -.06 -.12 -.08 -.15 -.09 -.11 -.09 -.22 .45*** -.17* -.29*** .60*** 1
n 147 147 141 147 100 99 100 100 52 51 54 54 146 146 146 147
18) 24 months
.16† .13 .26** -.03 .08 .00 -.09 -.13 -.10 -.10 -.09 -.13 .48*** -.11 -.09 .55*** .56*** 1
n 141 141 141 141 97 97 97 97 51 50 53 53 140 140 140 141 141
*p<.05, **p<.01, ***p<.001, †=approaching significance
4.1 DESCRIPTIVE STATISTICS
In order to examine maternal separation anxiety, descriptive statistics were computed, graphs
generated, and correlations run among reports of maternal separation anxiety at 3 measurement
points (6, 15 and 24 months after birth of child). The sample size at 6 and 15 months is 147
mothers, and at 24 months is 141. At all ages the range in MSA scores is relatively high (e.g., at
6 months, 25-96 out of a possible range of 21 to 105). The distribution of scores at each age is
normal. An example of the distribution of scores is provided in Figure 1, which illustrates
maternal separation anxiety at 6 months.
Figure 1. Maternal separation anxiety at 6 months
Comparing maternal separation anxiety means in the present study (highlighted) with the
MSA means of published studies reveals that the MSA means in the present study are lower than
those of other studies. Means appear in Table 8.
Table 8. Comparison of MSA means in current study with means of previous research
(working mothers only)
Child Age at Administration of MSA to Mothers
Study N Pre
Current 147 20 19 18.
Vondra 47 19
Lutz 133 22 22 21 20 20
McBride 49 19
McBride 26 21 20 18 18
r 87 19 19 18
Hock 49 25
McBride 130 22 21 20 19
Hsu 53 22 21
Belsky 33 21 20
Wille 70 23 22
& Fish 73 20
ECCRN* 609 21 21 20 20
study): 22 22 21 22 24 21 20 22 19 21 19 21 21 20 19 20 18
*Only mothers using center-based care
4.2 MATERNAL SEPARATION ANXIETY OVER TIME
nxiety 4.2.1 Mean level of maternal separation a
Next followed an examination of mean levels of maternal separation anxiety over time. Based on
previous research (Hock & Lutz, 1998; Hock & Schirtzinger, 1992; Hsu, 2003; McBride &
Belsky, 1988), it was expected that, over the course of the three measurement points, there would
be a significant decrease in MSA. To test this prediction, a repeated measures ANOVA followed
by a priori comparison of the 6-, 15-, and 24-month means was computed. A significant effect
for time was detected, Wilks’ Lambda = .92, F(2,139) = 5.72, p<.01, multivariate partial eta
squared = .076. Over time, there is a statistically significant decrease in MSA. The means and
standard deviations are reported in Table 9.
Table 9. Descriptive statistics for maternal separation anxiety at 6, 15, & 24 months
Time N Mean SD
Maternal Separation Anxiety
6 mo. 147 59a 12
15 mo. 147 58b 11
24 mo. 141 56a, b 11
ab Means with a different superscript are significantly different at p<.05.
The post hoc analysis revealed that there was a significant decrease in MSA between 6
and 24 months, t = 2.41, p<.05, and between 15 and 24 months, t = 3.35, p<.01. The decrease in
MSA between 6 and 15 months was not significant (t = 1.34, p=.18). Significant change appears
to occur sometime after 15 months. In this sample, mothers of 6- and 15-month olds were similar
in their reported MSA. The average MSA score before 24 months is 22; the average after 24
months is 19.
4.2.2 Individual differences in maternal separation anxiety over time
Correlational analyses indicated relatively strong stability of maternal separation anxiety over
time. MSA at each measurement point was highly correlated r = .78, p<.01 (6 & 15 months), r =
.71, p<.01 (6 & 24 months), and r = .75, p<.01 (15 & 24 months). The high correlation of the
MSA scores over the 3 measurement points indicates that there is considerable individual
stability in MSA over time in this sample. The fact that mothers retain their higher or lower
anxiety level over time suggests that some sort of personality feature in mothers may be involved
in the MSA they report.
4.2.3 Maternal personality dimen
The association between both individual and multiple facets of a mother’s personality and her
self-reported separation anxiety were examined. Based on existing literature, it was expected that
there would be a moderate, positive association (in the range of .3 to .4) between the personality
characteristic of neuroticism and both concurrent (6 months) and subsequent (15 and 24 months)
MSA, and moderate, inverse associations (in the range of -.3 to -.4) between the personality
characteristics of extraversion and agreeableness and both concurrent and subsequent MSA. A
significant association emerged between concurrent neuroticism and MSA at 6 months (r = .20,
p<.05). The higher a mother’s level of general anxiety at 6 months, the higher her self-reported 6
month MSA. Additionally, a significant association emerged between 6 month neuroticism and
subsequent MSA at 24 months (r = .18, p < .05). The higher a mother’s general anxiety at 6
months, the higher her self-reported MSA at 24 months. The overall sizes of the correlations,
however, were lower than expected. There were no significant associations between MSA and
either extraversion or agreeableness. In summary, maternal personality factors are generally not
associated with MSA, with the exception of very modest, direct associations for neuroticism.
4.2.4 Maternal depressive symptoms
Based on the conflicting results of previous research, it was an empirical question how (or even
whether) depressive symptoms would relate to maternal separation anxiety. A significant
association emerged for 24-month MSA and 15-month depressive symptoms (r = .19, p<.05 and
concurrent maternal separation anxiety and maternal depressive symptoms at 24 months (r = .26,
p < .01). The higher a mother’s self-reported separation anxiety at 24 months, the higher her self-
reported depressive symptoms at both 15- and 24-months. The magnitude of the correlations
were lower than expected, however.
In sum, there was only limited evidence that maternal separation anxiety covaries with
either general neuroticism or symptoms of depression. At 24 months only, both showed modest
associations with MSA. Prior to that age, MSA correlated modestly with concurrent neuroticism
only at 6 months.
4.3 MATERNAL SEPARATION ANXIETY AND MATERNAL EDUCATION
Next, the association between a mother’s level of education and her self-reported separation
anxiety was examined. Based on the conflicting results of previous research, it was an empirical
question how (or even whether) maternal education would relate to maternal separation anxiety.
The only significant association between MSA and maternal education occurred at the 15-month
measurement point, but was extremely small, r = -.16, p < .05. At 15 months, the higher a
mother’s education level, the lower her separation anxiety. Therefore, this question was not
given much support in this sample of college–and graduate–educated mothers.
In order to examine whether the association between MSA and maternal education is
stronger for first-time mothers, two sub-groups were identified: mothers of firstborn children and
mothers of later born children. It was an empirical question how (or even if) the average MSA of
the two groups would differ. A Fisher z-test was performed on the correlations between MSA
and maternal education for the two groups (firstborn mothers’ M = 57, laterborn mothers’ M =
58). Results indicate that the correlations for the 2 groups do not significantly differ. Therefore,
in this sample of well-educated women, prior experience with a firstborn infant has no effect on
the association between current MSA and maternal education.
4.4 LINKS BETWEEN MATERNAL SEPARATION ANXIETY AND MOTHER-
4.4.1 Maternal separation and reunion beh
To test whether a mother’s separation and reunion behaviors in the actual childcare setting are
related to her self-reported separation anxiety, correlations were computed between maternal
behavior composites at 15 and 24 months (2 at departure and 2 at reunion) and Maternal
Separation Anxiety at 15 months and again at 24 months. It was expected that concurrent
associations between MSA and reunion behavior would be in the range of .3 to .5, and that
associations between MSA and separation behavior would be in the range of .1 to .2. No
significant results were found in this analysis at either 15 or 24 months. In this sample of well-
educated, married, White, working mothers, there does not appear to be any association between
self-reported MSA and either departure or reunion behavior in the child care setting.
4.4.2 Mother-infant attachment se
The next test concerned whether a mother’s 6- and 15-month self-reported separation anxiety has
any bearing on the concurrent and/or subsequent security of attachment her baby demonstrates
with her. It was an empirical question how (or even whether) average MSA scores of mothers of
infants with an Avoidant attachment classification would be higher than the average MSA scores
of mothers of infants with a Resistant attachment classification.
Independent samples t-tests were run on MSA at 6 months (n = 141) and again at 15
months (n = 141) based on insecure attachment classification (Avoidant and Resistant ) at 15
months. Neither test detected a significant difference between group means. Therefore, there was
not much support for this question.
As a follow-up to this finding of no differences in attachment insecurity related to MSA,
all attachment classifications were compared on MSA. It is possible, as McBride and Belsky
(1988) found, that MSA differences lie across the secure/insecure dimension, rather than in the
type of insecurity. ANOVAs were computed on Maternal Separation Anxiety (MSA) at 6 months
and again at 15 months by Attachment Classification at 15 months (Strange Situation). There
was a statistically significant difference in MSA at 15 months (but not at 6 months) across the
three attachment classifications (Insecure-Avoidant, Secure, Insecure-Resistant), F(2,138) =
3.11, p<.05. Post hoc comparisons revealed a significant difference between Secure (n = 101, M
= 59) and Insecure-Resistant classifications (n = 27, M = 54). The Secure group of mothers
scored higher than the Insecure-Resistant group. These results are opposite to the results of Lutz
and Hock (1995), who found that mothers with an insecure adult working model of attachment
had the highest MSA. MSA at 6 months was not significant, F(2,138) = 2.43, p=.092, although
means were in the same direction.
4.4.3 Adding controls for maternal personality
Finally, tests were run to determine whether maternal separation anxiety is simply a proxy for
general anxiety or depressive symptoms in its association with mother-child relations. There was
only one instance when MSA had any correlation (in this case, a non-significant trend) with
maternal behavior (6 month MSA and 15 month negative composite of reunion behavior). A
hierarchical regression controlling for 6-month neuroticism and average depressive symptoms,
using 6-month MSA to predict maternal reunion behavior at 15 months was not significant. None
of the 3 predictors (neuroticism, depressive tendency, MSA) significantly contributed to the
prediction of maternal reunion behavior. Therefore, controlling for general anxiety and
depressive tendency does not alter the poor predictive ability of MSA for maternal behavior.
This descriptive study offers insight into the prevalence and trajectory of maternal separation
anxiety as well as its association with maternal personality, education, and mother-child
relations. The data put the issue of maternal separation anxiety into the ecological context of
well-educated, married, White, working mothers. The findings include descriptive statistics
related to what MSA looks like in a sample of well-educated mothers, how it manifests itself
over time, the extent to which it correlates with maternal personality characteristics, and whether
MSA is uniquely related to mother-child relations.
Results of this investigation indicate that there is substantial variability in the maternal
separation anxiety reported by well-educated, high-earning mothers. Despite the perception that
this group of mothers have the resources available to them that would theoretically limit their
maternal separation anxiety (higher education, greater access to quality care, and for some,
higher income), there exists a full range of reported maternal separation anxiety. Therefore, even
among college and graduate educated women, there is real concern expressed by some mothers
about separating from their infant/young child.
In this affluent, well-educated sample, the average means, however, are lower than those
found in studies with more demographically heterogeneous samples. Well-educated mothers may
be more likely to doubt media reports about the negative aspects of non-maternal childcare,
thereby reducing their MSA. Alternatively, well-educated mothers may rationalize about child
separations while completing the MSA measure and under-report their feelings about separating.
Questions from the MSA measure such as, “My child will benefit from group experiences (i.e.,
nursery school, day-care, kindergarten) since they will provide him/her social experiences that
he/she could not get at home,” and “It is good for my child to spend time away from me so that
he/she can learn to deal independently with unfamiliar people and new situations,” may be
endorsed at higher rates by mothers who have chosen to pursue higher education and greater
economic independence themselves.
In this sample, significant decline in MSA after birth did not occur until sometime after
15 months. Mothers of 6- and 15-month-olds were similar in their reported MSA. The fact that a
significant decline in MSA was detected between 15 and 24 months could be, in part, due to
reduced infant separation anxiety, which lessens after 15 months (Kagan, 1974). If childcare
drop-offs go more smoothly due to the child’s reduced separation anxiety and accompanying
reduction in anxious behavior (crying, clinging, general upset), as well as the dyad’s increased
experience with separating, mothers may feel less anxious about leaving their child in non-
maternal care, and therefore report lower MSA.
The high correlation of MSA scores over the 3 measurement points (6, 15, and 24
months) indicates that there is great individual stability in MSA over time in this sample, which
supports Hock and Schirtzinger’s (1992) findings (8 months, 3 and 6 years), with the size of the
correlations in the present study (average r = .75) being comparable to Hock and Schirtzinger’s
(average r = .73). Mothers hold onto their higher—or lower—rank order of anxiety, which could
imply that some kind of personality feature in mothers maintains the stability.
In this sample of high-earning, well-educated mothers, maternal neuroticism and
depressive symptoms has a small, but consistent, association with MSA, whereas other facets of
maternal personality did not. The overall magnitude of the associations for personality and
depressive symptoms, however, was lower than expected (average r = .21). Generally, it appears
that MSA is not simply an indicator of either maternal neuroticism or depressive tendencies. If
this is the case, does MSA have something to do with traditional, stereotypical beliefs that are
less evident in better-educated mothers? Even in this sample of well-educated mothers, the lower
a mother’s MSA at 15 months, the more likely she was to have a graduate degree. It may be that
women who seek an advanced degree have been socialized to believe that they should pursue a
career and not merely a job and/or that independence (personal and financial) is crucial to their
The association between MSA and independence socialization has empirical support.
Hock and Lutz (1998) found that maternal self-reported dependency accounts for a significant
portion of the variance in MSA, and Lutz and Hock’s (1995) research determined that women
who remembered their own mothers as not encouraging independence reported higher MSA.
This socialization process towards independence could weaken a mother’s stereotypical ways of
thinking about infant needs, and the role of a mother.
Or it could be that colleagues, partners, and other associates, both at the place of
employment and within the greater social networks associated with the profession (professional
organizations, networking opportunities) socialize or reinforce well-educated women to pay
attention to the constructive aspects of non-maternal care and the positive socialization provided
to children from non-maternal care. These foci challenge traditional beliefs about infant needs,
potentially resulting in a decrease in the MSA well-educated women report. A final possibility is
that for the mothers in this sample there is a good match between employment preference and
employment status that has resulted in the lower than average MSA scores reported by this
sample. Overall, the well-educated mothers in this sample reported, on average, lower MSA than
those in published studies with more heterogeneous samples, with the lowest scores more typical
of the women with advanced degrees.
In the current study, there was no significant difference between mothers of firstborns
and laterborns in the association between MSA and maternal education. Experience as a parent
does not appear to be a factor for college-educated women in the association between maternal
education and MSA. Socialization by parents and peers, knowledge and skepticism of childcare
research, income differences, and accessibility to quality center-based care all may be more
important than experience as a parent in regard to the association between maternal education
and MSA. In other words, you are who you are regardless of experience as a parent: when a new
baby is born, the mechanism(s) that produced MSA in response to the first child trigger those
same feelings with subsequent births. Mothers who seek advanced degrees either start out as—or
subsequently become less anxious about—separating from their infants and toddlers, regardless
of previous experience with older children. Education does seem to be one key element or
marker in the etiology of maternal separation anxiety.
From the perspective of construct validity, this study had little support to offer. There was
no association between self-reported MSA and either maternal departure or reunion behavior in
the childcare setting. A number of possible explanations can explain this outcome. The first is
that the MSAS measure has problems capturing a mother’s actual feelings about separation from
her infant/young child. For example, some mothers may report that they feel one way, but
display behaviors indicating that they feel differently when actually separating from their child.
Some mothers may intellectualize their feelings about leaving their infant/child when responding
to a questionnaire in a calm environment/context, but when in the child-care setting, during an
actual separation, they may display behaviors suggesting anxiety or guilt. Questions from the
MSA measure such as, “My child will benefit from group experiences (i.e., nursery school, day-
care, kindergarten) since they will provide him/her social experiences that he/she could not get at
home,” and “It is good for my child to spend time away from me so that he/she can learn to deal
independently with unfamiliar people and new situations,” may especially appeal to mothers
whose baby has problems with separations (producing lower MSA scores). However, when
mothers are in the actual process of separating from their infant/child at center-based care (as
opposed to leaving the child with a nanny at home, or with a neighbor), the emotions produced in
the mother (perhaps triggered or heightened by the child’s reaction to the separation) may result
in behaviors that indicate a higher level of MSA. This would explain the lack of association
between the self-reported MSA scores and the observation data.
A second possibility is that the observation measures have problems capturing behaviors
a mother shows who scores high on the MSA questionnaire. Observational items include,
“Before leaving, parent tries to interest child in an activity,” and “Parent seems rushed,” which
may not be appropriate or accurate in trying to capture maternal separation anxiety. These
behaviors may relate more to time constraints than to issues related to maternal separation
Additionally, the behavior data in the larger study were collected by 692 different care
providers on 594 infants and toddlers (NICHD CCDR-61). Especially without training, multiple
caregivers no doubt see events differently and may provide disparate ratings of behavior. No
information is provided about the amount of training, if any, these care providers received on the
measure, only nominal instructions were provided to the caregivers, and there are no reliability
data. The great number of caregivers involved in collecting data, along with the lack of
information related to their training and reliability, call into question the accuracy of the maternal
drop-off and reunion data collected from caregivers at these center-based facilities.
Objective, but psychometrically strong, ratings of maternal behavior are required to
validate the MSA measure. However, obtaining these ratings may be difficult. In center-based
care, teachers have pre-existing relationships with the children in their care and the parents of
these children. These relationships, with both the children and the parents, may create
preconceived notions about the mothers’ behavior, or bias the teachers in their opinions of the
mothers’ behavior in some other way. Previous maternal behavior, positive or negative, may
influence ratings of the mothers’ behavior during separations and reunions. Additionally, the
main priority of teachers is caring for the children in their care. If other children have difficulty
at departure, ratings of the observed child may be tainted, or the observation may be given short
shrift by observers whose main goal is to keep a classroom running smoothly.
It could even be that separation and reunion behavior is not the best indicator of parenting
that is affected by maternal separation anxiety. There are a number of reasons that a mother’s
behavior, particularly public behavior, may not reflect her self-reported MSA. First, an inhibited
mother may not demonstrate a high level of MSA because her emotional and behavioral response
style manifests itself in a reserved and composed nature in public, despite how she feels inside.
Second, a mother of an infant/young child with a difficult temperament (fussy, does not handle
separations easily), may actively downplay her emotions in order to present a calm demeanor at
separations in an attempt to prevent upsetting the child further, which could also mask her true
feelings about separating.
Finally, the lack of a relation between the self-report and observational measures could
indicate that the anxiety a mother reports feeling when she separates from her child at center-
based care is simply not an influential and/or useful concept for parenting. Separation concerns
or guilt may simply not have much bearing on the day-to-day style of care and interaction a
mother provides her infant or young child. This lack of an association may be due to the
transitory nature of separations at center-based care in relation to the greater amount of time
mothers spend in the parenting role. Separations and reunions are merely brief periods in the
course of a day with dozens of other mother-child bouts of interaction that have a greater bearing
on the mother-child relationship.
In terms of infant-mother attachment classification, the pattern found in McBride and
Belsky’s (1988) research (mothers in avoidant relationships reporting the highest MSA; mothers
in resistant relationships reporting the lowest) did not clearly emerge here. Neither did the
finding in Stifter, Coulehan, and Fish’s (1993) research that employed mothers who reported
high MSA had a disproportionate number of infants with an avoidant attachment. In the present
study MSA did not differ for mothers of toddlers with secure or avoidant attachments, although
mother of toddlers with resistant attachments did score lowest.
A review of the multiple discrepant results highlights the fact that there is little consensus
in the studies that have looked at maternal separation anxiety and mother-infant attachment
classification. One explanation may be that over the span of the past two decades, maternal
attitudes have changed related to the use of non-maternal care. Another explanation may be
sampling. The McBride and Belsky (1988) sample (avoidant>secure>resistant), for example, was
obtained in a rural area where there may be greater endorsement of traditional maternal roles and
fewer professional women. The timing of MSA data collection could also be an issue. McBride
and Belsky (1988) collected their MSA data at child ages 3 and 9 months, Stifter, Coulehan, and
Fish (1993) at child ages 5, 10 and 18 months, and Wille (1998)—who found no relation
between MSA and infant-mother attachment—at child ages 6 and 18 months. Perhaps the
differences in developmental milestones achieved by the infants and toddlers in these various
studies could have influenced MSA scores, leading to disparate results. The measure of mother-
infant attachment differed as well. McBride and Belsky (1988) and Stifter, Coulehan, and Fish
(1993) both used the Strange Situation, whereas Wille (1998) used the Attachment Q-sort. These
contradictory results seem to suggest that there is either error variance in the attachment/MSA
data or sampling fluctuation that capitalizes on random differences in how attachment patterns
relate to self-reported MSA.
A theoretical alternative is that there is a moderating variable at work that none of these
studies has investigated. One moderating variable could be the degree of investment professional
women have in the role of parent. Perhaps it is a positive sign for mother-infant attachment when
professional women report disagreement (i.e., higher “anxiety”) with MSA questions such as, “It
is good for my child to spend time away from me so that he/she can learn to deal independently
with unfamiliar people and new situations.” This suggests a valuing of attachment among
mothers who actually do have their child away from them for many hours each week. The
reverse may be true of mothers who are busy with careers but express little anxiety about the
impact of their separations on their child. Among women who work primarily out of economic
necessity, as well as for women who choose to raise their children themselves, higher expressed
separation anxiety may signal the enmeshment issues raised in the literature.
Another moderating variable could be child temperament. Mothers who report higher
levels of MSA and have an infant/young child who is inhibited (has problems separating from
his/her mother) may be more sensitive and responsive to their child’s needs. Therefore, higher
levels of self-reported MSA may be more adaptive to the attachment process when they reflect
actual child needs, rather than strictly maternal beliefs.
In conclusion, the inconsistency of links between MSA and infant-mother attachment
security argues that there may not be a direct association between MSA and mother-infant
attachment, and that one or more moderating variables (in conjunction with sampling
differences) account for the discrepant results.
Controlling for general anxiety and depressive tendency did not alter the (in this case
poor) predictive ability of MSA for mother-child relations, either in terms of maternal separation
and reunion behavior or infant-mother attachment. Thus, statistically removing more stable
aspects of a mother’s tendency to experience anxiety and distress did not yield a “purer”
indicator of maternal ambivalence about her child’s independence that had bearing on either her
own behavior or the security of their relationship. There simply is not much connection between
the feelings educated, White, working mothers report on the MSA subscale and their behavior
around daily separations, and only a limited, perhaps even random, connection to attachment
Despite the lack of significant findings in the present study, the results do tell us
something about maternal separation anxiety among well-educated, working mothers--
specifically what variables it is not associated with--and provide some clues as to what other
constructs MSA may be associated with, which does advance the knowledge base in the field of
parenting research. In developmental psychology, very little research addresses parenting beliefs
in general, or maternal beliefs in particular. Specifically, there is little research related to how a
mother feels about separating from her child. In the present study, maternal separation anxiety
does not seem to be capturing something about mother-child relations, but it may tell us
something about how well-educated mothers achieve in their careers. Well-educated mothers
who report higher levels of MSA may choose to work fewer hours, arrange to work flex time, or
choose slower career tracks in order to spend more time with their children (i.e., taking longer to
make partner at a law firm, or achieve tenure at an academic institution). MSA may not be
evident in a well-educated mother’s parenting behavior, but may manifest itself in the career
choices these mothers make.
In this sample of well-educated mothers, the construct of maternal separation anxiety appears
somewhat problematic, with no strong or consistent links to any of the theoretically relevant
variables examined in this study, other than, ironically enough, level of maternal education. The
lack of consensus in the results of studies examining MSA implies that there may either be
moderating variables at work in the association between MSA and maternal functioning, that the
MSAS measure may not be a valid measurement tool of MSA, and/or that the construct of
maternal separation anxiety may not be highly relevant to parenting within a well-educated and
more or less well-off sample of mothers. One must consider the possibility that MSA is merely
an indication that a mother wants to care for her own child at home, even if she also wants (or
has) a career, or simply works in order to contribute to the household income.
The data from this investigation provide only a set of snapshots of how a modest-sized sample of
well-educated, married, White, working mothers report feelings of anxiety related to separating
from their infant/young child at non-maternal care. The generalizability of these results is
obviously limited to this population of well-educated, married, White, working mothers. The size
of the sample and its homogeneity likely limited its power to detect associations that characterize
a broader segment of the population of working mothers.
The process by which the maternal behavior data were collected is also a limitation of
this study. The large number of observers, question of whether training was provided to
observers, caregiver pre-existing relationships with children and their parents and/or
distractibility of caregivers, and the lack of inter-rater reliability statistics could have resulted in
unreliable and/or inaccurate data being collected. Therefore, the data collected may not have
been the accurate, objective data required to validate the self-report MSA measure.
Another limitation is that no child characteristics or circumstances were taken into
consideration in this analysis. Child temperament may have an influence on the MSA a mother
feels. Mothers of inhibited or difficult temperament infants/young children may experience more
MSA than mothers of easy temperament infants/young children. Additionally, quality of center-
based care may have an influence on the MSA a mother feels. Mothers leaving their child in a
high quality center (e.g., centers having a positive reputation in the community, a long waiting
list, accreditation by the National Association for the Education of Young Children (NAEYC),
and/or the recommendation of a director from another center) may report lower MSA than those
mothers leaving their children in a center that is of questionable quality. These center- and child-
related variables may explain some of the MSA variance in this study. Therefore, the results of
this study need to be interpreted with care.
6.2 FUTURE RESEARCH
If the results of this study are valid, maternal separation anxiety does not seem to merit more
empirical attention, at least among well-educated mothers, and in relation to parenting. Maternal
separation anxiety appears to be more relevant to less educated mothers who either stay home or
work at lower paying jobs. However, there are two caveats to this position. First, if additional
research is undertaken, researchers may need to look at day-to-day parenting in the home,
especially independence socialization. MSA may be inversely related to independence-
promoting behavior in the home. Second, although maternal separation anxiety may not be an
especially useful concept for parenting among well-educated, working women, MSA may be a
useful construct for gaining insight into maternal career development and/or employment
behavior. If future research is conducted, it may be helpful to look at maternal employment
decisions and career tracks.
Any future research should also be designed to examine aspects of both the mother’s and
the infant/child’s temperament in order to detect any moderating influences of either variable on
the association between MSA and mother-child interaction. Including these variables into
analyses of MSA and mother-child relations may reveal moderating effects.
In may be possible to obtain a better understanding of maternal separation anxiety if
future research were designed as a mixed-methods study of employed mothers. The MSA
subscale could still be used, but could be administered as part of a more in-depth interview, so
that researchers could provide item clarification when necessary and mothers could explain and
elaborate on their responses. This approach allows the subjective meaning of the MSA questions
to be explored rather than inferred. Finally, in order to conduct an accurate analysis of the
validity of the MSA measure, trained observers could collect observational data of mothers and
children at home. Data from the home may explain differences in associations across samples
between MSA and maternal characteristics by providing a basis for comparison to maternal
disposition demonstrated at separation and reunion from one’s child at non-maternal care.
Designing future studies this way may provide information that explains the inconsistent findings
in previous research on MSA and provide clues to the etiology and possible moderators of
maternal separation anxiety that further understanding of this construct.
Belsky, J., Taylor, D. G., & Rovine, M. (1984). The Pennsylvania infant and family development
project, II: The development of reciprocal interaction in the mother-infant dyad. Child
Development, 55, 706-717.
Bianchi, S. M., Robinson, J. P., & Milkie, M. A. (2006). Changing rhythms of American life.
New York: Russell Sage Foundation.
Booth, C. L., Clarke-Stewart, K. A., Vandell, D. L., McCartney, K. & Owen, M. T. (2002).
Child-care usages and mother-infant “quality time”. Journal of Marriage and Family, 64,
Bowlby, J. (1969). Attachment and loss (Vol. 1). London: Hogarth Press.
Capizzano, J. & Main, R. (2005). Many young children spend long hours in child care.
Snapshots of America’s Families. Urban Institute: Washington, DC. Retrieved November
28, 2007 from: http://www.urban.org
Capizzano, J. & Adams, G. (2003). Children in Low-income families are less likely to be in
center-based care. Snapshots of America’s Families. Urban Institute: Washington, DC.
Retrieved November 28, 2007 from: http://www.urban.org
Caspi, A. (1998). Personality development across the life course. In W. Damon (Series Ed.) & R.
B. Cattell (Volume Ed.), The description of personality: Basic traits resolved into
clusters. Journal of Abnormal and Social Psychology, 38, 476-506.
Cassidy, J. & Berlin, L. J. (1994). The insecure/ambivalent pattern of attachment theory and
research. Child Development, 65, 971-991.
Cassidy, J. & Kobak, R. R. (1988) Avoidance and its relation to other defensive processes. In J.
Belsky & T. Nezworski (Eds.), Clinical Implications of Attachment (pp. 300-323).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112 (1), 155-159.
DeMeis, D. K., Hock, E. & McBride, S. L. (1986). The balance of employment and motherhood:
Longitudinal study of mothers’ feelings about separation from their first-born infants.
Developmental Psychology, 22(5), 627-632.
Galinsky, E. (1981). Between generations: The six stages of parenthood. New York: Times
Garcia-Coll, C., Surrey, J. L. & Weingarten, K. (1998). Mothering against the odds. New York:
The Guilford Press.
George, C. & Solomon, J. (1999) Attachment and caregiving: The caregiving behavioral system.
In J. Cassidy & P. R. Shaver (Eds.), Handbook of Attachment, (pp. 649-670). New York:
The Guilford Press.
Hock, E. (1980). Working and nonworking mothers and their infants: A comparative study of
maternal caregiving characteristics and infant social behavior. Merrill Palmer Quarterly,
Hock, E. (1984). The transition to day care: Effects of maternal separation anxiety on infant
adjustment. In R. C. Ainslie (Ed.), The child and the day care setting (pp. 183-203) New
Hock, E. & DeMeis, D. K. (1990). Depression in mothers of infants: The role of maternal
employment. Developmental Psychology, 26(2), 285-291.
Hock, E., DeMeis, D., & McBride, S. (1987). Maternal separation anxiety: Its role in the balance
of employment and motherhood in mothers of infants. In A. Gottfried & A. Gottfried
(Eds.), Maternal employment and children's development: Longitudinal research (pp
191-229). New York: Plenum.
Hock, E., Gnezda, M. T., & McBride (1984). Mothers of infants: Attitudes toward employment
and mother hood following birth of first child. Journal of Marriage and the Family, 46,
Hock, E. & Lutz, W. (1998). Psychological meaning of separation anxiety in mothers and
fathers. Journal of Family Psychology, 12(1), 41-55.
Hock, E., McBride, S. L., & Gnezda, M. T. (1989). Maternal separation anxiety: Mother-infant
separation from the maternal perspective. Child Development, 60(4), 793-802.
Hock, E. & Schirtzinger, M. B. (1992). Maternal separation anxiety: Its developmental course
and relation to maternal mental health. Child Development, 63(1), 93-102.
Hsu, H. C. (2003). Antecedents and consequences of separation anxiety in first-time mother:
infant, mother, and social-contextual characteristics. Infant Behavior & Development, 27,
Johnston, D. (2003). Invisible mothers: a content analysis of motherhood ideologies and myths in
magazines. Retrieved January 9, 2006, from:
Kagan, J. (1974). Discrepancy, temperament, and infant distress. In M. Lewis & L. Rosenblum
(Eds.), The origins of fear. New York: Wiley.
Lamb, M. E. (1999). Parental behavior, family processes, and child development in
nontraditional and traditionally understudied families. In M. E. Lamb (Ed.), Parenting
and child development in non-traditional families (pp. 1-14). Mahwah, NJ: Lawrence
Lewis, M., Feiring, C. & Rosenthal, S. (2000). Attachment over time. Child Development, 71(3),
Lutz, W. J. & Hock, E. (1995). Maternal separation anxiety: Relations to adults’ attachment
representations in mothers of infants. Journal of Genetic Psychology, 156(1), 57-73.
Macomber, J. E., Adams, G. & Tout, K. (2001). Who’s caring for our youngest children? Urban
Institute: Washington, DC. Retrieved November 28, 2007 from:
McBride, S. (1990). Maternal moderators of child care: The role of maternal separation anxiety.
New Directions for Child Development, 49, 53-70.
McBride, S. & Belsky, J. (1988). Characteristics, determinants, and consequences of maternal
separation anxiety. Developmental Psychology, 24(3), 407-414.
Mistry, J., Chaudhuri, J. H., & Diez, V. (2005). Ethnotheories of parenting: At the interface
between culture and child development. In R. M. Lerner, F. Jacobs, & D. Wertlieb (Eds.),
Applied Developmental Science (pp. 201-206). Thousand Oaks, CA: Sage Publications.
NICHD Early Child Care Research Network (1999). Childcare and the mother-child interaction
in the first 3 years of life. Developmental Psychology, 35 (6), 1399-1413.
NICHD Study of Early Child Care. (1997). Familial factors associated with the characteristics on
nonmaternal care for infants. Journal of Marriage and the Family, 59, 389-408.
Owen, M. T., Ware, A. M. & Barfoot, B. (2000). Caregiver-mother partnership behavior and the
quality of caregiver-child and mother-child interactions. Early Childhood Research
Quarterly, 15(3), 413-428.
Pallant, J. (2006) SPSS survival manual. New York: McGraw-Hill.
Richters, J. E. (1997). The Hubble hypothesis and the developmentalist's dilemma. Development
and Psychopathology, 9(2), 193-229.
Rogoff, B. (2003) The cultural nature of human development. New York: Oxford University
Sameroff, A. J., Seifer, R. & Elias, P. K. (1982). Socio-cultural variability in infant temperament
ratings. Child Development, 53, 164-173.
Sroufe, L. A. (1988). The role of infant-caregiver attachment in development. In J. Belsky & T.
Nezworski (Eds.), Clinical Implications of Attachment (pp. 18-38). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Stifter, C. A., Coulehan, C. M. & Fish, M. (1993). Linking employment to attachment: The
mediating effects of maternal separation anxiety and interactive behavior. Child
Development, 64, 1451-1460.
Storm, H. A. & Ridley-Johnson, R. (1995, August). Maternal separation anxiety and child care:
Effects on maternal behavior. Paper presented at the 103rd Annual Convention of the
American Psychological Association, New York, NY.
U. S. Census Bureau (2005). We the people: Women and men in the United States. U. S. Census
Bureau: Washington, D.C. Retrieved October 16, 2007 from:
U. S. Department of Education (2005). Early Childhood Program Participation Survey of the
National Household Education Surveys Program (NHES). National Center for Education
Statistics: Previously unpublished tabulation. Retrieved October 16, 2007 from:
Waters, E., Merrick, S., Treboux, D. Crowell, J. & Albersheim, L. (2000). Attachment security
in infancy and early adulthood: A twenty-year longitudinal study. Child Development,
Wille, D. E. (1998). Longitudinal analysis of mothers’ and fathers’ responses on the maternal
separation anxiety scale. Merrill-Palmer Quarterly, 44(2), 216-233.