You are on page 1of 12

Does Infant Carrying Promote Attachment?

An Experimental Study of the Effects of

Increased Physical Contact on the Development of Attachment
Author(s): Elizabeth Anisfeld, Virginia Casper, Molly Nozyce, Nicholas Cunningham
Source: Child Development, Vol. 61, No. 5 (Oct., 1990), pp. 1617-1627
Published by: Blackwell Publishing on behalf of the Society for Research in Child
Stable URL:
Accessed: 15/11/2009 19:29

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact

Blackwell Publishing and Society for Research in Child Development are collaborating with JSTOR to digitize,
preserve and extend access to Child Development.
Does Infant Carrying Promote Attachment?
An Experimental Study of the Effects of
Increased Physical Contact on the
Development of Attachment

Elizabeth Anisfeld, Virginia Casper, Molly Nozyce,

and Nicholas Cunningham
Columbia University College of Physicians and Surgeons

Infant CarryingPromoteAttachment?An ExperimentalStudy of the Effects of IncreasedPhysical
Contact on the Development of Attachment. CHILDDEVELOPMENT,
1990, 61, 1617-1627. This study
was designed to test the hypothesis that increasedphysical contact,experimentallyinduced, would
promote greatermaternalresponsiveness and more secure attachmentbetween infantand mother.
Low-SES mothers of newborn infants were randomlyassigned to an experimentalgroup (n = 23)
that received soft baby carriers(more physical contact)or to a controlgroup(n = 26) that received
infant seats (less contact).Using a transitionalprobabilityanalysisof a play session at 3'/2 months,it
was demonstratedthat mothersin the experimentalgroupwere more contingentlyresponsive than
controlmothersto their infants'vocalizations.When the infantswere 13 monthsold, the Ainsworth
Strange Situationwas administered.Significantlymore experimentalthan controlinfantswere se-
curely attachedto their mothers.We infer fromthese resultsthatforlow-income,inner-citymothers,
there may be a causal relationbetween increasedphysical contact,achieved throughearly carrying
in a soft baby carrier,and subsequent security of attachmentbetween infantand mother.

This article reports an experimental in- imity between mother and infant during the
vestigation of the effects of physical contact first months of the infant's life. Bowlby postu-
on the development of attachment between lated that infants have in their repertoire cer-
mother and infant. The amount of physical tain instinctive (genetically programmed) be-
contact was experimentally regulated by ran- haviors (e.g., crying and smiling) that have as
domly assigning mother-infant pairs right af- their goal the promotion of proximity to and
ter the infant's birth to one of two groups: a close physical contact with the caregiver
group that received soft baby carriers that af- (Ainsworth, Bell, & Stayton, 1974). Through
ford more physical contact, or a group that the exercising of these behaviors and the
received infant seats that afford less physical proximity thus achieved, infants gradually de-
contact. The two groups were later compared velop an attachment to their caregivers. Re-
on measures of maternal responsiveness and cent theoretical formulations (Main, Kaplan,
attachment. Previous correlational and obser- & Cassidy, 1985) suggest that the manner in
vational studies have identified close physical which the caregiver responds to the infant's
contact as an antecedent to attachment (Ains- attempts to seek and maintain proximity
worth, 1967; Ainsworth, Blehar, Waters, & largely determines the nature of the attach-
Wall, 1978; Egeland & Farber, 1984; Gross- ment relationship formed.
mann, Grossmann, Spangler, Suess, & Unz- In her classic studies in Uganda and the
ner, 1985). United States, Ainsworth found that the
Bowlby's attachment theory (Bowlby, amount of time mothers held their infants was
1969) places central importance on close prox- related to the security-of-attachment rating

Portionsof this study were submittedto YeshivaUniversityby the second and thirdauthorsin
partialfulfillmentof the requirementsfor the Doctorateof Philosophy.The researchwas supported
by funds from Snugli Cottage Industries, Ross Laboratories,and the generous contributionsof
several private donors. We gratefully acknowledge the help of Everett Waters in the scoring of the
Strange Situation tapes. We also thank the following, who contributed at various times throughout
the study: Moshe Anisfeld, Hetty Cunningham, Carol Feingold, Karen Fleiss, Gabriela Garcia,
Oriana Linares, Pat Shrout, and Deborah Wagner. Virginia Casper is currently affiliated with Bank
Street College of Education, New York, NY. Molly Nozyce is currently affiliated with the Bronx-
Lebanon Hospital Center, Bronx, NY. Requests for reprints should be addressed to: Elizabeth
Anisfeld, Department of Pediatrics, Columbia University, 630 West 168th St., New York, NY 10032.
[Child Development, 1990, 61, 1617-1627. ? 1990 by the Society for Researchin Child Development,Inc.
All rightsreserved.0009-3920/90/6105-0013$01.00]
1618 Child Development
that the infants received (Ainsworth, 1967). seems likely that close physical contact in the
Mothers who in the first months of life held early months of life plays a major role in the
their infants for relatively long periods and formation of healthy, responsive relationships
were tender and affectionate during the hold- between mothers and infants. This article re-
ing had infants who, at 12 months of age, had ports an experimental study of that concept in
developed secure relationships with them a low socioeconomic (SES) sample, which can
(Ainsworth et al., 1978). In contrast, if be expected to have a range of social risk fac-
mothers were inept in handling their infants tors influencing the development of attach-
and provided them with unpleasant experi- ment (Spieker & Booth, 1988) and to benefit
ences during holding, the infants developed from attachment-bolstering experiences. The
an anxious-ambivalent pattern of attachment. design took into account other factors known
Several studies (Egeland & Farber, 1984; to affect the development of attachment.
Main, 1977; Main & Stadtman, 1981) have These factors include neonatal irritability and
found that mothers of avoidant infants had re- infant temperament (Crockenberg, 1981;
jected or sought to minimize physical contact Grossmann et al., 1985; Waters, Vaughn, &
with their infants. Thus, there is evidence that Egeland, 1980) as well as the mother's social
the amount and quality of physical contact be- support and the household composition
tween mother and infant is related to security (Chisholm, 1983; Crittenden, 1985; Konner,
of attachment. By increasing the quantity of 1977; Munroe & Munroe, 1971).
physical contact, the experimental treatment The main hypotheses of the study were
"mayafford the mother opportunities to show
affectionate and tender behavior, thus affect- that: (1) mothers who carry their infants in soft
baby carriers during the first months of the
ing the quality of interaction. infants' lives will be more responsive and
A variety of factors could explain how sensitive to their infants at 31/2 months than
close body contact might have an effect on the will mothers who use infant seats; (2) at 13
infant, the mother, and the mother-infant re- months, the carried infants will be more se-
lationship. Investigators with diverse ap- curely attached to their mothers than will the
proaches have underscored the key role of control infants; and (3) maternal respon-
contact comfort (Harlow & Mears, 1979), early siveness at 3/2 months will be related to secu-
tactile experiences (Frank, 1957; Montagu, rity of attachment at 13 months. In addition, it
1971), vestibular-proprioceptive stimulation was hypothesized that infant behaviors such
(Gregg, Haffner, & Korner, 1976; Korner & as vocalizing and looking will be influenced
Thoman, 1972), and kinesthetic stimulation by the experience of being carried in a soft
(Yarrow, Rubenstein, & Pedersen, 1975) in baby carrier.
promoting healthy social and emotional infant
development. Infants carried in soft baby car- Method
riers are exposed to all these types of stimula- Subjects
tion. The increased and more varied time Women who delivered on the ward ser-
spent by the carried infant in close contact vice of a large inner-city hospital were poten-
with the mother may affect certain infant be- tially eligible for participation in the study.
haviors, such as crying, smiling, and vocaliz- These women come from a low-income clinic
ing, that were identified by Bowlby as related population that is predominantly Hispanic
to the development of attachment. and black. Approximately 30% do not have
The actual physical presence or proxim- telephones.
ity of the infant may have effects on the To participate in the study, women had
mother, making her more aware of and thus to be 18 to 37 years of age, with a parity from 1
more responsive to her infant's needs and through 4. They had to be accessible by tele-
states. A mother can more easily recognize phone and able to speak conversational En-
prodromal signs of hunger or discomfort in a glish. Other criteria were that the women had
carried infant than in an infant in a crib or received prenatal care, had vaginally deliv-
stroller at some distance from her. Several ered a full-term, healthy infant, were enrolled
previous studies have identified maternal re- in a hospital-based pediatric practice for med-
sponsiveness or sensitivity as an antecedent ical follow-up of the infant, and were plan-
to secure attachment (Ainsworth et al., 1978; ning not to return to work or school for at least
Ainsworth, Bell, & Stayton, 1972; Clarke- 3 months after delivery.
Stewart, 1973; Egeland & Farber, 1984;
Grossmann et al., 1985). The final sample consisted of23 women
in the experimental group (out of 30 who en-
From a theoretical point of view, and on rolled) and 26 women in the control group
the basis of observational studies, it thus (out of 30 who enrolled). These women had
Anisfeld et al. 1619
completed all assessments during the 13 to participate. It was stressed that if they
months of the study. Of those lost to follow- agreed to be in the study, they would be ex-
up, one infant died, one mother had brain pected to use whichever product they re-
surgery, one mother-infant pair could not be ceived by chance and not use the other. In
located, and the remaining families moved this way, we tried to ensure that women in
out of the area. both groups would initially have similar at-
For the purposes of the study, extended titudes to soft baby carriers and infant seats.
families were considered to be ones in which On the second day after giving birth,
a relative other than the infant's father lived women who signed the Informed Consent
in the household. Nuclear families were were officially enrolled in the study. Back-
defined as families composed of a mother, ground information was obtained from the
father, and their children, or as mothers living mother, and she completed (if necessary, with
alone with their children. Mothers' educa- the help of the research staff) the short form of
tional level was rated on a scale in which 4 = the Maternal Attitude Scale (Cohler, Weiss, &
some high school, 5 = completed high Grunebaum, 1970). On the second or third
school, and 6 = some college. Hollingshead day of the infant's life, one of the authors who
scores are based on a 7-point scale in which a had received training in Boston administered
higher score indicates a higher SES level the Brazelton Neonatal Behavioral Assess-
(Hollingshead, 1975). ment Scale (NBAS) (Brazelton, 1973) accord-
Of the mothers in the experimental ing to standard procedures.
group, 30% were primiparous; 35% were The mother was randomly assigned
black, 65% Hispanic; 52% were married or (stratified by sex of infant) to the experimental
living with the infant's father and 35% were (soft baby carrier) group or the control (infant
living in extended families (as determined by seat) group. The product was given to the
the 2-month questionnaire); 74% were receiv- mother while she was still in the hospital and
ing public assistance. The mean age of the manner of its use demonstrated. Mothers
women in this group was 23.7 years. Their in each group were asked to use the product
mean educational level was 5.0, and their every day. In order to obtain an objective es-
mean Hollingshead score was 1.8. timate of the amount the soft baby carriers
were used, pedometers were sewn inside
Of the women in the control group, 58%
were primiparous; 62% were black, 38% His-
panic; 61% were married or living with the The soft baby carriers used were "Snug-
infant's father, while 31% were living in ex- lis," commercial versions of cloth carriers
tended families; 70% were receiving public derived from those used in many parts of Af-
assistance. The mean age for this group was rica. Infants are carried in an inner pouch that
24.5 years. The mean educational level was is supported by the adult's upper torso. Adult
5.2, and the mean Hollingshead score was 2.0. and infant are chest to chest, creating a secure
and protective environment that allows the
Fifty-seven percent of the infants in the baby to be with the parent in a wide variety of
experimental group were males. The infants activities.
in this group had a mean birthweight of
3,333.5 grams; 48% were breast fed. In the Plastic infant seats can be used for carry-
control group, 50% of the infants were males. ing babies from place to place or for posi-
The group had a mean birthweight of 3,255.7 tioning an infant near its mother. They do not,
grams and 38% of the infants were breast fed. of course, involve close physical contact. In-
fant seats were not considered to constitute an
Intake Procedures
On the day after giving birth, women intervention; they are ubiquitous in this cul-
ture and are not thought to exert a powerful
who met the recruitment criteria were read a
list of different baby items and asked whether influence on attachment.
or not they would use each of the items if it Follow-up Assessments: Introduction
were given to them as a gift. Embedded in To insure objectivity of measurement,
the list were a soft baby carrier and an infant two teams of researchers were employed.
seat. Those who had already decided to use a Team A did the recruiting and follow-up
soft carrier (14%) or who would not consider interviews and knew which product each
using one (46%) were eliminated from consid- mother received. Team B administered all the
eration. The study was explained in detail to standardized tests, did the videotaping and
those women who indicated that they were coding, and were "blind" as to group mem-
willing to use either a soft baby carrier or an bership of the dyads. Team A kept in touch
infant seat. Only 5% of these women refused with the mothers and gathered information on
1620 Child Development
product use via phone calls and through short 60 = mother's and infant's faces lined up, eyes in
interviews with the mothers when they re- contact,
turned for well-baby checkups. 50 = mother'sand infant'sfaces lined up, eyes not
in contact,
Follow-up at 2 Months 40 = infantlooks at mother'shand or at some object
At the clinic visit nearest to 2 months af- she is holding,
ter each infant's birth, two questionnaires 30 = infant partiallyaverts his head frommother's
were administered to the mother. Lifestyle face,
20 = infant totally averts his head from mother's
Questionnaire I documented the exposure of face,
the infant to different situations and people, 0 = uncodable.
the infant's biorhythms and habits, the age of
onset of social smiling, the amount of the in- Each second of the 10 min of tape was as-
fant's crying, and the relative amount of time
signed a position code, producing a maximum
(rank ordered) that the infant spent in various of 600. The percentage of time the infant
locations during the day (crib, stroller, infant
spent in each position was calculated. Scores
seat, sofa, bed, arms or lap, soft baby carrier). on Looking At (60) and Looking Away (20)
The lowest number (i.e., 1) was assigned to were compared.
the location where the infant spent the most
time. Questions were also included about Vocalization.-Analyses of the vocaliza-
sleeping arrangements and household compo- tion data were based on the methods of Bake-
sition at 2 months. man and Brown (1977). As the 10 min of tape
were played, an event recorder was used to
The Product Use Questionnaire asked record simultaneously whenever the mother
the mother under what circumstances she or infant vocalized. One coder recorded ma-
used the product and who else used the prod- ternal vocalizations and one recorded infant
uct with the baby. Daily and weekly patterns vocalizations. Each 2-sec interval was then as-
of product use were ascertained. Pedometer
signed a code according to what occurred dur-
mileage readings were also recorded on this ing that interval: M-mother vocalized alone,
form by a Team A member. I-infant vocalized alone, C-both mother
Assessment at 31/2 Months and infant vocalized, Q-neither mother nor
This assessment consisted of four parts: infant vocalized. For each mother-infant pair,
the percentage of time in each dyadic state
Bayley Scales (Mental and Psychomotor) of was calculated to determine the amount of
Infant Development (Bayley, 1969), the time that the mother vocalized alone (%M),
videotaping of a mother-infant play session, the infant vocalized alone (%I), etc. As a reli-
the Carey Infant Temperament Scale (Carey,
1973), and the Product Use Questionnaire. ability check, a systematic subsample of 10
tapes was recoded by two other coders. The
Videotaping was done with the mother mean percent agreement between the two
sitting on a mattress on the floor with her in- sets of coders was 79 overall (M-92%, I-
fant propped up on a pillow facing her. A 73%, C-75%, Q-75%).
large mirror placed behind the infant's head A second analysis of the vocalization
allowed for a clear view of both the infant's
data was done to determine whether or not
face and the mother's face. Mothers were
the experimental intervention affected the
asked to play with their infants the way they
mothers' tendency to respond contingently to
would at home for about 15 min. They were
their infants' vocalizations. Through sequen-
asked not to use toys. In order to minimize
tial analysis of the 2-sec intervals already
distractions for the infant, the testing room
walls were unadorned and the room was fur- coded, transitional probabilities were gener-
ated that denote the likelihood of one specific
nished with only a small testing table and
vocal event following another. These were
averaged for each group separately to see
Three different analyses of the video- whether the two groups differed on the transi-
tapes were undertaken. In each case, the first tional probabilities following an infant-alone
minute of the tape was dropped to allow the (I) vocalization.
mother to adapt to the taping situation. The
next 10 min were analyzed. A measure of maternal vocal respon-
siveness, labeled Maternal Responsivity, was
Looking behavior.-The tapes were devised by adding together the proportion of
coded for the direction of the infant's gaze in mother-alone (M) following infant-alone (I)
relation to the mother's face, according to the vocalizations, and co-acting (C) following in-
following system (Beebe & Gerstman, 1980): fant-alone (I) vocalizations. There were some
Anisfeld et al. 1621
mother-infant dyads for whom it was not pos- household composition and the infant's sleep
sible to compute this measure because they patterns.
did not have enough intervals of infant alone
(2% or more) to be included. The Product Evaluation Questionnaire
asked mothers in each group to indicate the
Global ratings.-Cmic's system (Cmic, age of the infant when they stopped using the
Ragozin, Greenberg, Robinson, & Basham, product and how frequently they had used it
1983), which involves rating maternal and in- during the period when it was most useful. In
fant affect on three 5-point scales, was used addition, control mothers were asked whether
for the third analysis. Specifically, the scale they had used soft baby carriers (of any sort)
that rates the mother's sensitivity to the and how much.
baby's cues, state, and rhythm was used. For
the purposes of the study, this was labeled the Product Use
Maternal Sensitivity Scale. A graduate stu- Experimental group.-At 2 months, 86%
dent in psychology who was blind to the of the experimental group mothers reported
mothers' group assignment rated all the tapes. that their infants spent some time in soft baby
In addition, two undergraduates rated 50% of carriers. At 13 months, in response to the
the tapes. Intercoder reliability, using Co- question about how frequently the mothers
hen's weighted kappa, was .55 for agreements had used the soft carrier during the time
within 1 scale point. when it was most useful, the following results
were obtained: daily, 48%; two to three times
Assessment at 13 Months a week, 48%; less than two to three times a
The final assessment consisted of three week, 4%. The average length of use was 8.5
parts: the Ainsworth Strange Situation (Ains- months. The shortest period of use was 4
worth et al., 1978), the Lifestyle Question- months. At 13 months, four mothers were still
naire II, and the Product Evaluation Ques- using the carrier.
tionnaire. The first of these, the Strange
Situation procedure, was administered ac- Information from the pedometer read-
cording to the guidelines in Patterns of At- ings, the Product Use Questionnaire (at sev-
tachment (Ainsworth et al., 1978). This situa- eral time points), and the Product Evaluation
tion activates the infants' attachment systems Questionnaire (at 13 months) was combined
by putting them through a series of separa- to categorize the experimental group mothers
tions from and reunions with their mothers as high users (16) or moderate/low users (7).
and a stranger. The procedure was video- All the high users used the carrier every day
taped in a laboratory in the hospital. The aver- during the time it was most useful to them or
age age of the infants at testing was 13.5 had a pedometer reading of over 9 miles at 3
months. months. One moderate user used the carrier
every day (pedometer reading 3); the others
The videotapes were scored using Ains- used it only two or three times a week.
worth's global scoring technique by one of
the authors (V.C.), who was blind to the group Control group.-At 2 months, 96% of the
assignment of the mother-infant pairs. The control infants spent some time during a typi-
scoring system assigns infants to one of three cal day in the infant seat. At 13 months, the
basic patterns of attachment based on their control mothers reported the following
behavior throughout the Strange Situation, amounts of use during the time when the in-
with particular emphasis on their behavior fant seat was most useful to them: daily, 72%;
during the reunion episodes. The three pat- two to three times a week, 20%; less than two
terns are: securely attached (labeled B), to three times a week, 8%. The average length
insecurely attached-avoidant (labeled A), of use was 5.3 months.
and insecurely attached-resistant/ambivalent
Another consideration concerning use of
(labeled C). A reliability check with an expe-
rienced Strange Situation rater (Dr. E. Wa- the product was the issue of cross-contami-
nation. Our primary concern was the use of
ters) on 10 of the tapes revealed a 90% agree-
ment rate on major group (i.e., A, B, or C) soft baby carriers by control mothers. Four
control group mothers used a soft baby car-
rier. Most started to use it later than the ex-
The Lifestyle Questionnaire II elicited perimental group mothers (e.g., at 2 months).
information on various aspects of the dyad's Because of the randomized trial design, it was
life 1 year after the infant's birth. The points necessary to leave these mothers in the con-
covered included the mother's work status, trol group for the statistical analyses. (Three
relation to significant other, perceived social of the four infants were securely attached to
support, current welfare status, as well as their mothers.)
1622 Child Development
Results model that included the five covariates. Be-
To confirm that the random assignment cause of small numbers in some of the cells,
had resulted in similar groups, we compared the two types of insecure attachment, inse-
the experimental and control groups on rele- cure avoidant (A) and insecure resistant (C),
vant variables. They were similar on all dem- were combined (Table 1).
ographic characteristics except parity and To determine whether differences in
ethnicity (p < .10 by x2 analysis), the control product use within the experimental group
group having more primiparous and more were related to security of attachment,
black dyads. No differences were found (us- we compared high and moderate/low users.
ing t tests and p < .25) between the experi- Fifteen of the 16 high users had infants
mental and control groups in two analyses of who were securely attached. Of the seven
the NBAS data, one a cluster analysis as per moderate/low users, four were securely at-
Lester, Als, and Brazelton (1982) and the tached, X2(1, N = 23) = 2.35, N.S.
other an analysis of the number of nonoptimal
scores (Waters et al., 1980). No differences Whether insecurity of attachment might
were found between the two groups on any be related to use of the infant seats was as-
factor of the Maternal Attitude Scale (Casper, sessed by examination of the mean rank order
1985). assigned to the use of infant seats by control
mothers whose infants became securely at-
Major Outcome Variables tached (mean = 2.9), insecure avoidant (mean
A series of analyses was carried out to
test for possible associations between each = 3), and insecure resistant (mean = 4). This
analysis indicates that less frequent infant
dependent/outcome measure and each of the seat use (i.e., higher rank) was associated with
following independent variables: sex, parity, more insecure attachment.
ethnicity, household composition, mother's
education, perceived social support, method Maternal responsivity.-The ANCOVA
of feeding, Maternal Attitude Scale, neonatal on the Maternal Responsivity score showed
characteristics such as irritability (measured that the experimental group mean (.61) was
by the NBAS), and infant temperament at 31/2 significantly higher, F(1,31) = 6.57, p < .02,
months. The subjects' scores on the indepen- than the control group mean (.44). Experi-
dent variables were categorized for these mental mothers responded significantly more
analyses. Chi-square analysis was used for often within 2 sec of an infant-alone vocaliza-
comparisons involving security of attachment, tion, either by vocalizing alone or by joining
and t tests were used for Maternal Responsiv- with the infant in vocalizing (i.e., co-acting),
ity and Sensitivity. (All chi-squares were cal- than did control mothers. For the control
culated with Yates's correction for continuity, group, an infant vocalization was more likely
when necessary.) Method of feeding (i.e., to be followed by another infant vocalization
breast or bottle) was not related to Maternal or by quiet. It is known that infants interpret a
Sensitivity or attachment (Cunningham, Anis- response that follows within 2 sec of their
feld, Casper, & Nozyce, 1987), but showed a own behavior as being contingent on their be-
slight association with Maternal Responsivity havior (Millar, 1972). For this reason, it is ar-
(.20 > p > .10). gued that the experimental mothers were
Five variables (parity, ethnicity, house- more contingently responsive than the con-
hold composition, Maternal Attitude Scale trol mothers to their infants' vocalizations
Factor 1 and Factor 3) were related to one or (Nozyce, 1985).
more of the dependent variables (p < .25). In addition, experimental mothers were
They were therefore entered as covariates in more likely to vocalize following quiet (Q)
two ANCOVAs (for Maternal Responsivity than were control mothers. In the control
and Sensitivity) and in a logistic regression group, an interval of quiet was more likely to
analysis (for attachment classification) to test be followed by another interval of quiet.
the relation between treatment condition and Thus, experimental mothers were more likely
the dependent variables. to initiate a vocal interchange following quiet
and to respond to the infant's vocalization by
Security of attachment.-There were
significantly more securely attached (B) in- joining in during the succeeding interval.
fants in the experimental group (83%) than in Maternal sensitivity.-Experimental moth-
the control group (38%), according to logistic ers received higher ratings on the Maternal
regression analysis. The regression coefficient Sensitivity Scale at 31/2months than did con-
was significantly different from zero (13 = trol group mothers. However, this difference
1.93, SE = .82, Z = 2.35, p = .019) in a fell short of significance in the covariance
Anisfeld et al. 1623


Insecure- Insecure-
Avoidant Resistant
Secure (B) (A) (C)
GROUP n % n % n %
Experimental ................. 19 83 3 13 1 4
Control ...................... 10 38.5 10 38.5 6 23
Total ....................... 29 59 13 27 7 14

Secure Insecure
(B) (A & C)
n % n %
Experimental ................. 19 83 4 17
Control ...................... 10 38 16 62
"aZ= 2.35, p = .019. Z is based on logistic regression analysis that included five variables (parity,ethnicity,
household composition, MaternalAttitude Scale factors 1 and 3) in the model.

analysis: experimental group mean = 3.43, In the control group, a significant differ-
control group mean = 2.81, F(1,42) = 2.96, ence was found between the mean responsiv-
p = .09. ity scores of the mothers whose infants later
became securely or insecurely attached: se-
Of the various independent variables that
cure group mean = .56, insecure group mean
were tested, only household composition was = .33, t(20) = 2.58, p < .02. Infants in the
found to be significantly related to Maternal
control group whose mothers were more vo-
Responsivity and Maternal Sensitivity. Fur-
ther analysis revealed that the overall effect cally responsive to them at 31/2 months were
was due to a significant relation within the more likely to be securely attached to the
mothers at 13 months. On Maternal Sensitiv-
control group: Maternal Responsivity, t(20) =
ity, there was no significant difference be-
2.25, p < .05, and Maternal Sensitivity, t(24) tween the control mothers whose infants
= 2.73, p < .02. Within the experimental
were securely attached and insecurely at-
group, the association did not reach signifi- tached: secure group mean = 3.00, insecure
cance. Control group mothers who lived in
extended families were more likely to re- group mean = 2.50, t(24) = .93, N.S. Because
of differences between this study and previ-
spond contingently to their infants' vocaliza- ous studies in the nature of the rating scale (5-
tions and to react sensitively to their infants
point vs. 9-point, respectively), the basis for
during the 31/2-month video than were
mothers living in nuclear families, confirming assigning the ratings (10-min video of play vs.
extensive home observation, respectively),
previous findings (Chisholm, 1983; Konner, and the age at which the mother's sensitivity
1977; Munroe & Munroe, 1971). was rated in relation to the age at measure-
Relation between 31/2-Month and 13-Month ment of attachment (prospective vs. concur-
Measures rent), no conclusions can be drawn from the
To assess the association between mater- present study about the relation of sensitivity
nal responsivity and sensitivity in the early to security of attachment.
months and subsequent security of attach-
ment, the mean scores on the 31/2-month mea- The analyses up to this point indicate
sures (Maternal Responsivity and Maternal that the experimental mothers were more
Sensitivity) for the secure and insecure dyads contingently responsive and their infants
within each group were calculated. Because were more likely to be securely attached. Pre-
there were only four insecurely attached in- vious studies and results from the control
fants in the experimental group, no further group presented in the preceding paragraph
analyses were done. indicate that maternal responsiveness is asso-
1624 Child Development
ciated with more secure attachment. To deter- tors was controlled for in the analysis. Thus,
mine whether the differences in attachment the experimental condition, by promoting
between the groups were due solely to differ- physical contact, appears to have made these
ences in Maternal Responsivity, a second lo- mothers more contingently responsive to
gistic regression analysis was done, adding their infants.
Maternal Responsivity to the model. The re-
There were many fewer avoidant (A)
gression coefficient remained significantly
different from zero (P = 2.02, SE = 1.01, Z = babies in the experimental group than in the
control group (13% vs. 38.5%). Main and her
2.00, p = .045), suggesting that the experi-
mental treatment had an effect above and colleagues (Ainsworth et al., 1978; Main,
beyond that attributable to Maternal Respon- 1977; Main & Stadtman, 1981; Main & Wes-
sivity. ton, 1982) have shown that mothers of
avoidant babies often have an aversion to
Effects of Treatment on Other Outcome physical contact and may even reject their in-
Measures fants' approaches. Based on their studies with
Analyses of the play sessions videotaped motherless monkey mothers, Harlow and
at 3'/2 months showed that infants in the Suomi (Harlow & Suomi, 1971; Suomi, 1973)
experimental group vocalized alone (I) sig- suggest that potentially negative, rejecting
nificantly less than infants in the control maternal behaviors can be modified by close
group: experimental mean = 4.57%, control body contact. In the present study, it appears
mean = 9.73%; t(47) = 2.95, p < .01. Experi- that the process of carrying a baby in a soft
mental infants scored above the median on carrier enabled some mothers who might
Looking at Mother more frequently than did have had an initial aversion to physical con-
control infants, X2(1,N = 49) = 4.65, p < .05. tact to overcome it and form healthy relation-
The ages of onset of social smiling as re- ships with their babies. The physical contact
inherent in carrying seems to have brought
ported by the mothers at 2 months were com-
out latent nurturing behavior in the low-
pared. The experimental infants reportedly
started to smile later (mean, 4.86 weeks) than income mothers in our sample. Whether the
the control infants (mean, 3.68 weeks), t(47) same effect would be achieved in a lower-risk
= 2.31, p < .05. In addition, 52% of the con- sample is a question for future studies.
trol infants reportedly had a regular daily pe-
Infant Behavior
riod of crying, whereas only 21% of experi- The major consequence of the carrying
mental infants did so, yielding a X2(1,N = 49) intervention was that, in comparison with the
of 5.23, p < .05. control infants, significantly more of the car-
Carrying did not appear to have any ef- ried infants developed secure attachment re-
fect on infant temperament as measured by lationships with their mothers. The distribu-
the Carey Scale at 31/2 months: there were tion of secure versus insecure experimental
four "difficult" infants in each group. In addi- group infants did not differ significantly, X2(1,
tion, there were no differences in early devel- N = 46) = 1.80, N.S., from that found by
opment between the two groups as measured Ainsworth (Ainsworth et al., 1978) among a
by the Bayley Scale at 31/2 months. middle-class sample. Thus, the experimental
condition appears to have dramatically im-
Discussion proved the chances of infants in this study
population becoming securely attached to
The results of this investigation support their mothers.
the hypothesis that increased physical contact
achieved through the use of a soft baby carrier The relatively high percentage of
makes mothers more responsive to their in- avoidant classifications within the control
fants and promotes the formation of more se- group resembles data from other inner-city,
cure attachment between infant and mother at low-SES populations (Crittenden, 1985; Ege-
13 months of age in a sample of low-income, land & Sroufe, 1981; Gaensbauer & Harmon,
inner-city mothers. 1982; Lyons-Ruth, Connell, Grunebaum, Bo-
tein, & Zoll, 1984; Schneider-Rosen, Braun-
Maternal Behavior wald, Carlson, & Cicchetti, 1985; Waters et
The finding that more experimental al., 1980). In such populations, neglectful care
group mothers than control group mothers re- has recently been associated with avoidant at-
sponded contingently to their infants' vocali- tachments (Spieker & Booth, 1988). In our
zations at 31/ months cannot be explained by clinic population, we see neglectful care rang-
pre-existing differences between the groups. ing from insensitive caregiving to maltreat-
The effect of other possible contributing fac- ment. The predominant problem seems to be
Anisfeld et al. 1625
a lack of appropriate stimulation and a lack of the experimental intervention affected the be-
attention to the infant's needs and wants. haviors of both mother and infant. The
However, there were no reported cases of ne- mothers became more contingently respon-
glect or abuse in the study sample, which was sive to their infants, and the infants were
followed closely by a hospital pediatric con- more likely to develop secure attachments to
tinuity practice. their mothers. The process of being carried
close to the mother seems to have had an ef-
The differences in attachment behavior fect on the infant's security of attachment
shown by the experimental and control in- above and beyond that attributable to in-
fants at 13 months can be traced back to ear- creased maternal responsivity as measured in
lier differences in behavior patterns in- this study.
fluenced by the close body contact provided
by the soft carrier. Bowlby suggested that By intervening at a sensitive time in a
vocalizing, smiling, and crying are social re- mother's emotional life, right after she has
leasers that infants use in early months to pro- given birth, and encouraging her to carry her
mote proximity to their mothers. The carried infant in a soft baby carrier, it may be possible
infants in this study vocalized alone less and to influence the pattern of her parenting and
started to smile socially somewhat later than to promote a healthier mother-infant relation-
the control infants. Moreover, fewer of the ship. The process by which the physical con-
carried infants had regular periods of crying at tact leads to healthier relationships requires
2 months. It may be that the carried infants elucidation by further research. This might
had less need to activate these behaviors than include exploration of the possible effects on
control infants because they were already in mothers' working models of attachment and
close proximity to their mothers. Several stud- exploration of the possible channels through
ies provide support for this interpretation. which carrying affects the infant directly.
Anderson, Vietze, and Dorechi (1978), Jones Confirmation of these findings with other SES
and Moss (1971), and Lewis and Freedle populations and with larger samples will be
(1973) reported that at 3 months of age infants necessary before we can conclude that in-
vocalized more when they were further away creased physical contact between mother and
from the mother and less when they were infant is a major factor in creating secure at-
close (e.g., on the mother's lap). Anderson et tachment.
al. (1978) concluded that "from Bowlby's per-
spective a decrease in infant vocalization
would be expected when the young infant has References
the closest possible physical contact with the Ainsworth, M. D. S. (1967). Infancy in Uganda:
mother" (p. 388). Hunziker and Barr (1986), in
Infant care and the growth of love. Baltimore:
a randomized trial of soft baby carriers, re-
Johns Hopkins University Press.
ported that carried infants cried significantly Ainsworth, M. D. S., Bell, S. M., & Stayton,D. J.
less at 6 weeks than did control infants. (1972). Individual differences in the develop-
Our finding that control infants spent less ment of some attachment behaviors. Merrill-
time looking at their mothers' faces than did Palmer Quarterly, 18, 123-143.
experimental infants could be related to the Ainsworth, M. D. S., Bell, S. M., & Stayton,D. J.
fact that their mothers were generally less re- (1974). Infant-motherattachmentand social de-
sponsive. This interpretation is supported by velopment: Socializationas a productof recip-
the studies of Brackbill (1958) and Watson rocal responsiveness to signals. In M. P.
Richards(Ed.), The integration of a child into
(1972), which showed that unresponsiveness
to infant face-to-face behaviors can lead to a social world (pp. 99-135). Cambridge:Cam-
fussing and gaze aversion on the part of the bridge University Press.
infant. An alternative interpretation is that the Ainsworth, M. D. S., Blehar, M. C., Waters,E., &
carried infants may have had less opportunity Wall, S. (1978). Patterns of attachment:A psy-
for direct en face interaction and therefore chological study of the Strange Situation.
now sought it out and found it rewarding. Hillsdale, NJ: Erlbaum.
Anderson,B. J., Vietze, P., & Dorechi, P. R. (1978).
In conclusion, the present experimental Interpersonal distance and vocal behavior in
study indicates a causal relation between the mother-infantdyad. Infant Behavior and
physical contact, achieved through carrying Development, 1, 381-391.
an infant in a soft baby carrier, and security of Bakeman,R., & Brown,J. V. (1977).Behavioraldia-
attachment between mother and infant within logues: An approach to the assessment of
the present sample of low-SES dyads. The mother-infantinteraction.Child Development,
different ecology of interaction promoted by 48, 195-203.
1626 Child Development
Bayley, N. (1969).Bayley Scales of Infant Develop- Gaensbauer, T. J., & Harmon, R. J. (1982). Attach-
ment. New York: Psychological Corp. ment behavior in abused/neglected and prema-
Beebe, B., & Gerstman, L. J. (1980). The "packag- ture infants: Implications for the concept of at-
ing" of maternal stimulation in relation to in- tachment. In R. N. Emde & R. J. Harmon
fant facial-visual engagement: A case study at (Eds.), Attachment and affiliative systems (pp.
four months. Merrill-Palmer Quarterly, 26, 245-279). New York: Plenum.
321-339. Gregg, C. L., Haffner, M. E., & Korner, A. F. (1976).
Bowlby, J. (1969).Attachment and loss: Vol. 1. At- The relative efficacy of vestibular-proprio-
tachment. New York: Basic. ceptive stimulation and the upright position in
Brackbill, Y. (1958). Extinction of the smiling re- enhancing visual pursuit in neonates. Child
sponse in infants as a function of reinforcement Development, 47, 309-314.
schedule. Child Development, 29, 115-124. Grossmann, K., Grossmann, K. E., Spangler, G.,
Brazelton, T. B. (1973). Neonatal Behavioral As- Suess, G., & Unzner, L. (1985). Maternal sen-
sessment Scale. London: Spastics International sitivity and newborns' orientation responses as
Medical Publications. related to quality of attachment in Northern
Carey, W. B. (1973). Measurement of infant temper- Germany. In I. Bretherton & E. Waters (Eds.),
ament in pediatric practice. In J. C. Westman Growing points of attachment theory and re-
(Ed.), Individual differences in children (pp. search (pp. 233-256). Monographs of the Soci-
298-306). New York: Wiley. ety for Research in Child Development, 50(1-
Casper, V. (1985). The effects of early carrying on 2, Serial No. 209).
infant-mother attachment: The role of mater- Harlow, H. F., & Mears, C. (1979). The human
nal attitudes. Unpublished doctoral disserta- model: Primate perspectives. New York:
tion, Yeshiva University, New York. Wiley.
Chisholm, J. S. (1983).Navajo infancy:An etholog- Harlow, H. F., & Suomi, S. J. (1971). Social recov-
ical study of child development.New York:Al- ery by isolation-reared monkeys. Proceedings
dine. of the National Academy of Sciences, 68,
Clarke-Stewart, K. A. (1973). Interactions between 1534-1538.
mothers and their young children: Characteris- Hollingshead, A. B. (1975). Four-Factor Index of
tics and consequences. Monographs of the So- Social Status. Unpublished manuscript, New
ciety for Research in Child Development, Haven, CT.
38(6-7, Serial No. 153). Hunziker, U. A., & Barr, R. G. (1986). Increased
Cohler, B., Weiss, J., & Grunebaum, H. (1970). carrying reduces infant crying: A randomized
Child-care attitudes and emotional disturbance controlled trial. Pediatrics, 77, 641-648.
among mothers of young children. Genetic Psy- Jones, S. J., & Moss, H. A. (1971). Age, state and
chology Monographs,82, 3-47. maternal behavior associated with infant vocal-
Crittenden, P. M. (1985). Social networks, quality of izations. Child Development, 42, 1039-1051.
child rearing, and child development. Child Konner, M. J. (1977). Infancy among the Kalahari
Development, 56, 1299-1313. Desert San. In P. H. Leiderman, S. R. Tulkin,
Cmic, K. A., Ragozin, A. S., Greenberg, M. T., & A. Rosenfeld (Eds.), Culture and infancy:
Robinson, N. M., & Basham, R. B. (1983). So- Variations in the human experience (pp. 287-
cial interaction and developmental compe- 328). New York: Academic Press.
tence of preterm and full-term infants during Korner, A. F., & Thoman, E. B. (1972). Relative
the first year of life. Child Development, 54, efficacy of contact and vestibular-proprio-
1199-1210. ceptive stimulation in soothing neonates. Child
Crockenberg, S. B. (1981). Infant irritability, mother Development, 43, 443-453.
responsiveness, and social support influences Lester, B. M., Als, H., & Brazelton, T. B. (1982).
on the security of mother-infant attachment. Regional obstetric anesthesia and newborn be-
Child Development, 52, 857-865. havior: A reanalysis toward synergistic effects.
Cunningham, N., Anisfeld, E., Casper, V., & Child Development, 53, 687-692.
Nozyce, M. (1987, Feb. 14). Infant carrying, Lewis, M., & Freedle, R. (1973). Mother-infant
breast feeding and mother-infant relations. dyad: The cradle of meaning. In P. Pliner, L.
Lancet, p. 379. Krames, & T. Alloway (Eds.), Communication
Egeland, B., & Farber, E. A. (1984). Infant-mother and affect: Language and thought (pp. 127-
attachment: Factors related to its development 155). New York: Academic Press.
and changes over time. Child Development, Lyons-Ruth, K., Connell, D., Grunebaum, H., Bo-
55, 753-771. tein, S., & Zoll, D. (1984). Maternal family his-
Egeland, B., & Sroufe, L. A. (1981). Attachment tory, maternal caretaking and infant attachment
and early maltreatment. Child Development, in multiproblem families. Preventive Psychia-
52, 44-52. try, 2, 403-425.
Frank, L. K. (1957). Tactile communication. Ge- Main, M. (1977). Analysis of a peculiar form of
netic Psychology Monographs,56, 209-255. reunion behavior seen in some day-care chil-
Anisfeld et al. 1627
dren: Its history and sequelae in children who ternal vocal responsivity as measured at 31/2
are home-reared. In R. A. Webb (Ed.), Social months. Unpublished doctoral dissertation,
developmentin childhood:Day-careprograms Yeshiva University, New York.
and research (pp. 33-78). Baltimore: Johns Schneider-Rosen, K., Braunwald, K. G., Carlson, V.,
Hopkins University Press. & Cicchetti, D. (1985). Current perspectives in
Main, M., Kaplan, N., & Cassidy, J. (1985). Security attachment theory: Illustration from the study
in infancy, childhood, and adulthood: A move of maltreated infants. In I. Bretherton & E. Wa-
to the level of representation. In I. Bretherton ters (Eds.), Growing points of attachment the-
& E. Waters (Eds.), Growing points of attach- ory and research (pp. 194-210). Monographs of
ment theory and research (pp. 66-104). Mono- the Society for Research in Child Develop-
graphs of the Society for Research in Child ment, 50(1-2, Serial No. 209).
Development, 50(1-2, Serial No. 209). Spieker, S. J., & Booth, C. L. (1988). Maternal an-
Main, M., & Stadtman, J. (1981). Infant response to tecedents of attachment quality. In J. Belsky &
rejection of physical contact by the mother. T. Nezworski (Eds.), Clinical implications of
American Academy of Child Psychiatry, 20, attachment (pp. 95-135). Hillsdale, NJ: Erl-
292-307. baum.
Main, M., & Weston, D. R. (1982). Avoidance of the Suomi, S. J. (1973). Surrogate rehabilitation of mon-
attachment figure in infancy: Descriptions and keys reared in total social isolation. Journal of
interpretations. In C. M. Parkes & J. Steven- Child Psychology & Psychiatry, 14, 71-77.
son-Hinde (Eds.), The place of attachment in Waters, E., Vaughn, B., & Egeland, B. (1980). Indi-
human behavior (pp. 31-59). New York: Basic. vidual differences in infant-motherrelation-
Millar, W. S. (1972). A study of operant condition- ships at age one: Antecedents in neonatal
ing under delayed reinforcement in early in- behavior in an urban, economically dis-
fancy. Monographsof the Societyfor Research advantaged sample. Child Development, 51,
in Child Development, 37(2, Serial No. 147). 208-216.
Montagu,A. (1971). Touching. The human signifi- Watson, J. S. (1972). Smiling, cooing, and "the
cance of the skin. New York:Harper& Row. game."Merrill-PalmerQuarterly,18, 323-340.
Munroe, R. H., & Munroe, R. L. (1971). Household Yarrow, L. J., Rubenstein, J. L., & Pedersen, F. A.
density and infant care in an East African soci- (1975). Infant and environment. Early cogni-
ety. Journal of Social Psychology,83, 3-13. tive and motivational development. New
Nozyce, M. (1985). Soft baby carrier use and ma- York:Wiley.