Stress, context and the Vulnerability–Stress–Adaptation model

In 1995 Benjamin Karney and Thomas Bradbury published a review in Psychological Bulletin covering more than a hundred longitudinal studies of marriage. Alongside the review they proposed an organising framework, the Vulnerability–Stress–Adaptation model, which has shaped a great deal of the research that followed.1

The model has three components. Enduring vulnerabilities are the stable characteristics each person brings — family history, personality, mental and physical health. Stressful events are the circumstances a couple encounters, from job loss and illness to housing instability and caregiving. Adaptive processes are how the pair handles what arrives: the conversations, the support, the negotiation.

The shift it introduced

The argument is that these three components interact rather than operate separately, and in particular that adaptive processes are partly downstream of the other two. The same couple, with the same repertoire, handles a disagreement differently when under heavy external load than when not.

That reframing matters because a great deal of earlier work treated interaction quality as a fixed property of a couple, something they possessed in greater or lesser measure. The model reads it instead as a performance that varies with conditions.

Lisa Neff and Karney pursued this empirically, examining how stress originating outside a relationship appears inside it. Their work reported that periods of higher external stress were associated with poorer problem-solving behaviour in the same couples, and with a weaker link between what a partner did and how satisfied the other felt.2

The sampling problem, stated by the authors

The more consequential limitation concerns who was studied. The longitudinal literature reviewed in 1995 was built largely on volunteer samples that were disproportionately white, educated and middle class, recruited near universities in the United States.

Karney and Bradbury have written about this at length in the years since, and the point they press is not only methodological. Conclusions drawn from those samples fed into publicly funded relationship-education programmes directed at low-income populations whose circumstances differed substantially from the samples that generated the findings.3

Large federal evaluations of those programmes in the United States, including the Building Strong Families and Supporting Healthy Marriage projects, reported effects that were small or absent on the outcomes they set out to move.4 Several readings of that result are available. The one consistent with the model is that programmes concentrating on adaptive processes, while leaving stressful conditions untouched, are working on the component least able to move on its own.

What follows, and what does not

What the model supports is a claim about levels of explanation: that circumstance belongs in the account of relationship outcomes rather than being treated as background noise around interaction. Research on household income and satisfaction trajectories has continued in this direction.5

What it does not support is a claim in the other direction — that interaction does not matter, or that circumstances determine outcomes. The model is a framework for organising evidence, not a finding in itself, and the studies testing it remain correlational.