The authors investigated longitudinal stability of mother-infant attachment using an urban, poor population and compared their results with Waters (1978) who did a similar study using a middle-class sample.

The sample comprised 100 first-born infants (50 male, 50 female) who were amongst the first participants recruited for the Mionnesota Child Development study. Mother-infant dyads were examined using the Strange Situation procedure at 12 and 18 months and mothers (median age 19 years) were surveyed when the 18 month SSP was conducted to assess their exposure to stressful life events over the period of the study.

The authors found significant stability of attachment between 12 and 18 months (62% of the sample remained stable) but also found less stability than Waters (1978) found with his middle-class, unstable sample. Examination of maternal reports of stressful events indicated that: (1) Mothers of infants classed as insecure at 18 months reported significantly more stressful events than those of secure infants, and (2) Mothers whose infants changed from secure to anxious attachment reported higher levels of stressful life events over the intervening period than mothers of infants who remained securely attached.

These results indicate that family instability is associated with greater variability in infant/mother interaction and a consequent increased likelihood of insecure attachment. Increased maternal experience of stressful life events was associated with increased levels of insecurity in infants, in some cases leading from to a shift from secure to anxious attachment.

These results also show that the attachment bond continues to develop and be affected by external events beyond the initial 12 month period.

Note that this sample was used in a further study by Egeland & Sroufe (1981) to examine attachment in infants of neglecting and abusive mothers.