Anisfeld, Casper, Nozyce & Cunningham, 1990. “Does Infant Carrying Promote Attachment? An Experimental Study of the Effects of Increased Physical Contact on the Development of Attachment”.


On to our third research study on carrying/babywearing. This is also one of a few who are done quite specifically on carrying babies in “soft baby carriers” and would therefor be easy to transfer over on babywearing. It is quite commonly referred to in the babywearing world as scientific evidence for “babywearing helps to build secure attachment”. Continue reading and I’ll try to clarify on why that might not be the case.

The study sought to investigate if “increased physical contact would promote attachment”. In the study included mother-baby dyads were divided into two groups; one handed a soft baby carrier and the other group a infant seat. The mothers responsiveness and sensitivity to the child and the attachment between the dyads were then measured, as well as product use, and later compared.

The study was an experimental study set in a low socioeconomic sample (low-SES), a sample that was expected to hold several social risk factors that could influence the attachment process as well as it being a sample that could benefit from an intervention such as this one (if successful, one might ad).

“The main hypotheses of the study were that:
(1) mothers who carry their infants in soft baby carriers during the first months of the infants’ lives will be more responsive and sensitive to their infants at 3,5 months than will mothers who use infant seats
(2) at 13 months, the carried infants will be more securely attached to their mothers than will the control infants
(3) maternal responsiveness at 3,5 months will be related to security of attachment at 13 months. In addition, it was hypothesized that infant behaviours such as vocalizing and looking will be influenced by the experience of being carried in a soft baby carrier.


Who were included in this study? Well, the women gave birth at a large inner-city hospital, were “… from a low-income clinic population that is predominantly Hispanic and black.” . The women had to be between 18 to 37 years of age, not given birth more than 4 times before, conversational English speaking and accessible by phone. The phone part stands out to me, in this particular setting and at this time it’s pointed out that of the eligible population only 70% actually had phones, which in turn then means that the sample in the study wasn’t fully transferable to the population as well as that the 30% that didn’t owned a phone aren’t represented (possibly, I’m guessing here, this could have been the 30% who simply couldn’t afford a phone and therefor might have benefitted this intervention the most). The women had to have a vaginal birth at term, and the baby had to be healthy, they had to have had prenatal care, had to been enrolled in a “hospital-based pediatric practice for medical follow-up of the infant” and also the women had to planned to stay at home with the child for at least the first three months.

Women who were eligible for this study were, the day after giving birth, read a list of items for infants and asked if they would use the items if the item was gifted to them. Both the soft baby carrier as well as the infant seat was on the list. 60% of the women who got the list read for them was eliminated from this study, 14% because they had already planned to use a soft baby carrier and 46% (which to me is a quite surprisingly high percentage, at least I think it would be today) because they would NOT consider using a soft structured carrier. The study was explained for the remaining women and most of these women (95%) agreed to participate. The women were instructed to use the product they would be gifted, and NOT use the other type. The researchers meant that they by this procedure “… tried to ensure that women in both groups would initially have similar attitudes to soft baby carriers and infant seats.”.

The seats and carriers were handed out before the dyad left the hospital and was instructed to be used every day. Pedometers were sewn in inside of the baby carriers to obtain an “… objective estimate… “ of its use.

Infant seats were not considered to constitute an intervention; they are ubiquitous in this culture and are not thought to exert a powerful influence on attachment.”, THIS little sentence I find fantastically interesting! The seats weren’t considered to be an intervention, but according to the instructions the mothers were instructed to use them every day, and to me if you’re asked to do or use something in a special way or to a certain extent that is an intervention (but that’s me, I’m picky). “… ubiquitous in this culture…”, if that won’t make your eyebrows touch your hairline I suggest you give it another read. Which culture? The low-income? The low-SES? Or another not named culture?

The dyads were randomly placed in the two groups depending on the sex of the infant. The researchers manged to gather a total sample of 60 mother-baby dyads of which 11 weren’t able to finish all of the assessments in the study protocol, this gives us roughly 18%non-completion mainly due to the families moving out of the area (8), one infant died, one mother had brain surgery and one dyad couldn’t be located. The final experimental group (soft baby carrier) consisted of 23 dyads, and the control group (infant seat) of 26.

I love charts and tables and since the original article had a slight lack of them, I did some, I find them helpful when comparing numbers, as one might want to when it comes to numbers within the groups when researching experimental interventions. You’re welcome! 😛

Mothers Experimental
group
Control
group
Mean age in years: 23,7 24,5
Mean education level:* 5.0 5.2
Mean Hollingshead score: 1.8 2.0
Received public assistance: 74% 70%
Married or living with the infants’ father: 52% 61%
Living in extended families: 35% 31%
Primiparous: 30% 58%
Black: 35% 62%
Hispanic: 65% 38%
Total number in the group: 23 26

(* an educational level of 5 = completed high school)

And for the infants the numbers look like this in a table:

Infants Experimental group Control group
Male sex: 57% 50%
Mean birthweight: 3333,5 g 3255,7 g
Breastfed: 48% 38%


Okay, so what was the result and conclusion of the study?

Using a transitional probability analysis of a play session at 3,5 months, it was demonstrated that mothers in the experimental group were more contingently responsive than control mothers to their infants’ vocalizations. When the infants were 13 months old, the Ainsworth Strange Situation was administered. Significantly more experimental than control infants were securely attached to their mothers. We infer from these results that for low-income, inner-city mothers, there may be a causal relation between increased physical contact, achieved through early carrying in a soft baby carrier, and subsequent security of attachment between infant and mother.”

And here is a picture of their table for the attachment measurements:

Table 1 from Anisfeld, Casper, Nozyce & Cunningham, 1990.

I’m a nerd, I will happily admit that, and to me this study and article took quite some time to digest and sort out. A certain amount of uncertainty or blurriness is quite common and that’s when you need to be very very clear with the limitations, possible confounders/covariates, differences in groups etc. When that is the case it’s most commonly described thoroughly in the discussion part of the article, and hopefully also pointed out in relation to the results and the conclusion. Here are a few points that I would like to highlight:

Because of differences between this study and previous studies in the nature of the rating scale (5point vs. 9-point, respectively), the basis for assigning the ratings (10-min video of play vs. extensive home observation, respectively), and the age at which the mother’s sensitivity was rated in relation to the age at measurement of attachment (prospective vs. concurrent), no conclusions can be drawn from the present study about the relation of sensitivity to security of attachment.”.

Due to “mismatching” in the design of this study the results were not comparable to other studies, which makes results based on or including maternal sensitivity hard to draw any confident conclusions from.

“There were some mother-infant dyads for whom it was not possible to compute this measure
(Maternal Responsivity, [Editor´s note]) because they did not have enough intervals of infant alone (2% or more) to be included.”.

Of an already small sample some dyads, unknown how many, weren’t included in the measurement of maternal responsivity. Which makes results based on or including maternal responsivity hard to draw any confident conclusions from.

“To assess the association between maternal responsivity and sensitivity in the early months and subsequent security of attachment, the mean scores on the 3,5-month measures (Maternal Responsivity and Maternal Sensitivity) for the secure and insecure dyads within each group were calculated. Because there were only four insecurely attached infants in the experimental group, no further analyses were done.”.

“Five variables (parity, ethnicity, household composition, Maternal Attitude Scale Factor 1 and Factor 3) were related to one or more of the dependent variables {p < .25). They were therefore entered as covariates in two ANCOVAs (for Matemal Responsivity and Sensitivity) and in a logistic regression analysis (for attachment classification) to test the relation between treatment condition and the dependent variables.”

“There were significantly more securely attached (B) infants in the experimental group (83%) than in the control group (38%), according to logistic regression analysis. The regression coefficient was significantly different from zero (p = 1.93, SE = .82, Z = 2.35, p = .019) in a model that included the five covariates. Because of small numbers in some of the cells, the two types of insecure attachment, insecure avoidant (A) and insecure resistant (C), were combined (Table 1).”

“To determine whether the differences in attachment between the groups were due solely to differences in Matemal Responsivity, a second logistic regression analysis was done, adding Matemal Responsivity to the model. The regression coefficient remained significantly different from zero (|3 = 2.02, SE = 1.01, Z = 2.00, p = .045), suggesting that the experimental treatment had an effect above and beyond that attributable to Maternal Responsivity.”

“In conclusion, the present experimental study indicates a causal relation between physical contact, achieved through carrying an infant in a soft baby carrier, and security of attachment between mother and infant within this present sample of low-SES dyads.”


“Confirmation of these findings with other SES populations and with larger samples will be necessary before we can conclude that increased physical contact between mother and infant is a major factor in creating secure attachment.”.

Quite a lot to read, right? And it’s quite tough reading. But what I want to try to point out is that this is a study that simply doesn’t provide rock solid, crystal clear evidence that carrying/babywearing promotes secure attachment. It does indicate it in this sample, 30 years ago, but cant deliver a clear reason on why. The study has flaws, in many parts and places, and the authors do point that out but sometimes you really have to do some tough reading to find out where those flaws are and what they might mean for the result and how/to whom the results might apply.

Do I still think that newborns should be carried? Well, yes! But not due to this article 😊 What I do think this article proves in a very clear manner is the difference helpful gifts can have, especially for those who may not be able to get them for themselves!

And as always, I’m not doing this to prove anyone wrong (or right), this is just me nerding out on some of my favourite things!

Lämna ett svar

E-postadressen publiceras inte. Obligatoriska fält är märkta *