Undertook further investigations into the relationship between attachment style and adult romantic dyads. Subjects were 240 couples (mean age = 21.2 years) recruited from an undergraduate course who were currently in 'steady or serious dating relationship[s]' (p. 505). Telephone interviews were carried out on three occasions over a span of 3 to 4 years. On the first interview, the attachment measure used was the Attachment Style Prototypes from Hazan and Shaver's Adult Attachment Questionnaire (1987) but scored using a five point Likert-type scale. Scores from this were compared with scores from the Relationship Rating Form (RRF; 1995) which uses 68 items to assess 18 relationship characteristics.  Later telephone interviews enquired briefly about the current status of the relationship.

Sample Distribution

The authors noted that the distribution of attachment style was not typical. They undertook a second sampling (N=406) using the same sampling frame but without the requirement of being in a relationship.  Results from this second sample were similar to the expected distribution.  Significant differences were found between the 'undergraduates in a relationship' and 'undergraduates' sample for both avoidant and anxious attachment styles.  This was consistent with results reported by previous studies (1991; 1992) which suggested that avoidant and anxious individuals are less likely to be in a relationship than secure individuals.  

Pairings by Attachment Type

Significant results were found when looking at partner pairings by attachment style.  From the table below, it can be seen that the avoidant-anxious pairing occurred with a higher frequency than would have been expected from a random distribution. This was particularly true where the male was the avoidant and the female the anxious partner.  Non-avoidant males showed a preferance for non-anxious females and non-anxious females showed a preferance for non-avoidant males.

Secure males and avoidant females showed the next highest increase in frequency above random distribution, followed by secure with secure and then anxious males with secure females.  These results reinforce the undesirability of female anxiety and male avoidance, except in the anxious-avoidant pairing.

Apart from the secure male and anxious female dyads, the pairings of anxious with anxious and avoidant with avoidant showed the most negative differences in frequency, with the sample including no couples with either of these style pairings.  

Relationship Quality

Measures from the Relationship Rating Form (RRF) were compared with attachment style of self and of partner. Results indicated that an individual's own attachment style was a stronger predictor of perceived relationship quality that the partner's attachment style. Male avoidance was associated with low ratings on all dimensions by the subject, as was female anxiety.  In addition to this, where the female was classified as anxious, partners give low ratings on viability, satisfaction, commitment and care. Where the male was avoidant, however, this did not have a strong effect on the female partner's rating of the relationship, except that she rated it as high in passion. Male anxiety was associated with females rating the relationship as low in passion and high (reversed scores) in conflict.

Stability and Status

Relationship stability and status was measured at time 2 and time 3 of the study, over a period of around 3 years.  Two measures were employed.  Stability was measured as 'together' or 'no longer together'.  Status was measured against a scale with six categories, ranging from 'no longer dating' to 'married'.

For males, stability was found to be marginally correlated with attachment style.  Anxious men showed the lowest levels of relationship stability at both time 2 and time 3.  When scores were adjusted to factor out prior duration and commitment, the results became more significant.  Avoidant men displayed the highest rates of stability while anxious men displayed the lowest rates.  The pattern of attachment related findings was reversed for women.  Anxious women showed the highest scores for both stability and status, with avoidant women showing the lowest.  These results, in combination with the results on partner selection, add a further dimension to the picture.

Partnerships between avoidant males and anxious females showed the highest increase in frequency above random distribution.

This partnership showed the lowest overall levels of satisfaction, with lowest on commitment, care, satisfaction and viability. Yet the stability scores indicate that it is at least as stable as relationships between secure partners.  This is a stable, preferred and yet generally unsatisfying relationship.

Partnerships between anxious males and avoidant females showed the next highest increase in frequency above that expected from combining the attachment gender distributions. This dyad does not score particularly low or particularly high on any of the satisfaction metrics and yet is the least stable. This is an indifferent, frequent yet unstable partnership.

The authors theorise that these results can be accounted for by the fact that females are 'typically (although not necessarily) the maintainers and breakers of relationships'. Relationships involving avoidant women, who are unskilled in this area, would therefore show the highest breakup rates. Relationships involving anxious women, for whom relationships are very important, would show greatest stability.