Change and Stability in Attachment Security from Age 1 to Age 2: The Effects of Maternal
Mental Health, Stressful Life Events, and Economic Hardship
Katherine L. Harris, B.A., Alissa C. Huth-Bocks, Ph.D., Sarah Ahlfs-Dunn, M.S., Maegan Calvert, M.S., Christine Wall, B.A., & Kate McLeod
Eastern Michigan University
INTRODUCTION
METHOD continued
RESULTS
The importance of secure attachment on later
child development has been well documented.
While attachment is considered to be relatively
stable over time, significant environmental
experiences may influence change in the quality
of attachment (Belsky et al., 2002). Early
researchers have demonstrated stability of
attachment across time in middle-class samples
where the environment is assumed to remain
stable (Waters, 1978), and instability in high-risk
samples where the environment may be unstable
or stressful (Vaughn et al., 1979). Such
environmental experiences include those more
distal to the child, such as his or her neighborhood
or family income, or more proximal factors such
as parental psychopathology, the effects of which
are transmitted through the intimate bonds
children share with their caregivers. Disruptive
life events, domestic violence, and income have
been associated with stability and change in
attachment insecurity in high-risk samples in
more recent work (Vondra et al., 1999;
Levendosky et al., 2011). However, few studies
have examined the trajectories of both proximal
and more distal aspects of the caregiving
environment in relation to the stability/instability
of attachment over time.
Procedures
120 pregnant women were recruited through the
posting of flyers in pregnancy agencies and area
community organizations. The first interview took
place during the third trimester of pregnancy (T1),
with subsequent interviews at 3-months postpartum (T2), 1-year postpartum (T3), and 2-years
postpartum (T4). Retention (n): T2 (119), T3 (115),
T4 (99).
Repeated measures ANOVAs were used to examine within-subject and between-groups main effects, as well as Group x Time
interactions in order to explore how changes in the caregiving environment were related to attachment patterns over time. A
significant Time x Group interaction for total PTSD symptoms, F(6, 86)=2.23, p < .05, η2=.07, with a significant quadratic
effect, F(3, 86)=5.78, p < .01, η2=.17 was found (see Figure 1). While maternal PTSD symptoms steadily decreased overtime
for the S-S, S-I, and I-I groups, for the I-S group, symptoms started out rather high in pregnancy, dropped considerably at age
1, and increased again at age 2. There was a main effect for time for negative life events, F(2, 86)=4.08, p < .05, η2=.05, with
a significant quadratic effect, F(1, 86)=6.07, p < .05, η2=.07; there was an increase from pregnancy to age 1 and a decrease
from age 1 to age 2 for all groups. Additionally, there was a trend-level Group x Time interaction for negative life events, F
(6, 86)=1.83, p = .09, η2=.06, with a trend-level linear effect, F(3, 86)=2.70, p = .05, η2=.09 (see Figure 2). The were no
significant results regarding the association between monthly income and attachment stability/instability (see Figure 3). Post
hoc analyses revealed significant change across time in regards to PTSD symptoms in the I-S group, F(2, 11)=3.90, p < .05,
η2=.25, with a significant quadratic effect F(1, 12)=10.47, p < .01, η2=.47. PTSD symptoms in this group started high, then
dropped, then increased again.
Research Aim:
 To explore changes in the caregiving
environment, including maternal mental health
and contextual risks, in relation to the stability/
instability of attachment security from age 1 to
age 2.
METHOD
Participants
• 120 women followed from pregnancy through the
first few years of their infants’ life
• Mothers’ age range: 18-42, M = 26, SD 5.7
• 47% African–American, 36% Caucasian, 13%
Biracial, 5% Other
• 64% single/never married, 28% married, 4%
divorced, 4% separated
• 20% had a high school diploma or less, 44%
some college, 36% college degree
• Median monthly household income = $1500
• 88% received services from WIC, and 90% had
public health insurance
Measures
Attachment security was assessed at T3 and T4
using the observer-rated Attachment Q-Set (AQS;
Waters & Deane, 1985).
Scores were
dichotomized to form secure and insecure groups
using a cut-off at both time periods per
recommendation by the author (Waters, 1987).
Four groups were then created to capture patterns
of stability and instability (see Table 1).
Figure 1.
Figure 2.
Figure 3.
Changes in PTSD Symptoms Over
Time by Attachment Group
Changes in Negative Life Events
Over Time by Attachment Group
Changes in Monthly Income Over
Time by Attachment Group
Maternal and contextual risks measured at T1, T3,
and T4 included:
Posttraumatic Stress Disorder (PTSD). Maternal
symptoms of PTSD were assessed using the PTSD
Checklist (PCL; Weathers, Litz, Herman, Huska, &
Keane, 1993), a self-report measure assessing how
often participants have been bothered by PTSD
symptoms in the last month (T1 α = .87, T3 α = .
91, T4 α = .95).
Negative life events.
Negative life events
experienced by the mother were assessed using the
Life Events Scale (LES; Abdin, 1995), a self-report
measure of significant life events, positive and
negative, that have occurred during the past year.
The total number of negative life events was used
here.
Monthly income. The family’s monthly income
was assessed using a single item on a
demographics questionnaire: “Considering all
sources of income, approximately what is your
family income per month?”
Table 1.
Attachment Change Patterns
Age 1 Secure
Age 1 Insecure
Age 2
Secure
S-S
(n = 47)
I-S
(n = 13)
Age 2
Insecure
S-I
(n = 16)
I-I
(n = 10)
DISCUSSION
•  The results of the present study indicate that both
proximal and distal contextual factors have an impact on
the stability/instability of attachment security across time.
•  In regard to the effects of life events, it appears that
women initially experience a moderate level of negative
events during pregnancy, followed by an increase in the
first year postpartum and a decrease in the second year
postpartum. This makes sense because women tend to
experience several changes and strains in the first year
with a new baby such as economic and relationship
strain.
•  While the effects of monthly income are visually notable,
with clear patterns in expected directions, there were no
significant results. This may be due to overall low
sample size resulting in small groups for comparison.
Further, it is possible that a family’s income per se does
not have a significant impact on the stability/instability of
attachment, but rather, affects other important variables
that are more proximal to the infant.
•  For the insecure-secure group, the relationship between
concurrent maternal PTSD symptoms and attachment
security is counter-intuitive. While one would expect
higher levels of PTSD to be associated with insecure
attachment at the same time point, the results for this
group indicate otherwise. One possible explanation of
this finding is that mothers’ PTSD symptoms have a
lagged or delayed effect on infant attachment such that
PTSD symptoms during pregnancy are influencing
attachment at age 1, and PTSD symptoms at age 1 are
affecting attachment at age 2 for this group.
•  The results of this study provide implications for child
development as it is well known that secure attachment,
and the stability thereof, holds numerous benefits for
child development reaching from infancy into adulthood.
•  Future research will be benefited by the examination of
other factors, both proximal and distal, such as maternal
symptoms of depression and social support, for example,
on the stability/instability of attachment security.