For more than 40 years, psychologists John Gottman and Robert Levenson have conducted some of the most ambitious and rigorous research programs ever attempted in relationship science. One of their most enduring findings relates to how spouses move between emotional states during conflict in their marriage. What makes it so influential is that it allowed them to predict not only whether a couple will divorce, but also roughly when.
As explained in a 2017 review published in the Journal of Family Theory & Review, this insight stems from what was eventually nicknamed the “roach motel” model of unhappy marriage. With this framework, they captured one of the clearest, most empirically grounded explanations for the pathways that lead marriages toward divorce.
Here’s a breakdown of the model, why it works and what it reveals about the divorce trajectories Gottman and Levenson identified in their research.
Emotional States As A System In Marriage
The Gottman–Levenson framework serves to examine the pattern of transitions between emotional states in a relationship, as initially described in their famous 2002 Family Process study. To analyze 14 years of longitudinal marital data, coders were trained for longer than 200 hours in order to categorize every second of couples’ conflict as either positive, neutral or negative. Sequential analyses then calculated the probabilities that one state would lead to another.
They found that, for stable couples, emotional states were fluid. A moment of irritation would most be followed with a return back to neutrality. A disagreement might be softened by a brief joke or acknowledgment. In other words, they moved between these different emotional states both frequently and flexibly.
For distressed couples, however, emotional systems were significantly more “sticky.” Negativity would beget more negativity, rather than less. This means that once their conversations shifted into an even remotely negative register, it was likely to stay that way.
This is where Gottman and Levenson started to notice similarities between their data and pre-existing mathematical concepts. Specifically, they identified that for unhappy couples, negativity functioned as a “Markov absorbing state.” In the simplest of statistical terms, an absorbing state is one that is easy to enter and hard to exit. In the context of a relationship, this concept suggests that a negative interaction risks becoming self-sustaining for couples who are already distressed.
Notably, this wasn’t just a loose observation. They observed this absorbing state repeatedly across datasets, and it remained significant even after controlling for different demographic variables, personality traits and communication styles.
Stable couples did not fall victim to this absorbing state nearly as much as unstable ones. Although they weren’t immune to negativity altogether — because no one is — they didn’t remain stuck in it. In this sense, the distinction between recovery and absorption became a major predictor of marital outcomes.
Twelve years later, in his 2016 novel The Seven Principles for Making Marriage Work, Gottman eventually gave this model a memorable name: the “roach motel” model of unhappy marriage. The nickname was borrowed from an advertisement for a cockroach trap, which you might even remember yourself: “They check in, but they don’t check out.”
The metaphor captures the very essence of the Markov absorbing state of negativity in marriage: once a distressed couple enters a negative emotional state, they stay there. This prolonged negativity contributes to what Gottman and Levenson refer to as the “distance and isolation cascade.” Over repeated conflicts, partners begin withdrawing; their interactions become more guarded, less loving and more logistical. And in time, the absorbing dynamic becomes what structures the relationship.
Two Marriage Trajectories, Not One
This early-appearing negative trajectory wasn’t the only one the Gottman and Levenson discovered. With the same longitudinal data, the research duo were able to identify two distinct groups of couples who eventually divorced. Each couple category had its own emotional signature:
- Early Divorcers. These couples were married for an average of 5.6 years before divorcing, and experienced significant amounts of negativity. More specifically, they displayed high levels of hostility, rapid escalation during conflict, the classic negative absorbing state and frequent contempt (the strongest predictor of early divorce). The direction of the relationship became clear for these couples, and divorce came quickly within the first five to seven years.
- Later Divorcers. These couples were married for an average 16.2 years before divorcing. But instead of spending their days together in hostility, these couples didn’t show intense negativity. In fact, Gottman only described them as being “sad and mildly angry.” Instead, their marriage was characterized by a lack of positivity, rather than an abundance of negativity. Their conflicts were marked by sadness, irritability and emotional disengagement. Equally importantly, there was also minimal affection, humor, playfulness, curiosity or warmth.
These latter couples were stable enough to maintain the relationship through the early years. They typically stayed together long enough to raise children and manage daily responsibilities with efficiency. However, their emotional connection was dangerously thin.
This marital pattern has since been observed frequently — so much so that they’ve come to be labelled as “devitalized” marriages. The absence of love and affection, rather than the presence of hostility, is what predicts these couples’ eventual separation. It takes much longer to materialize than in contemptuous marriages, which may arguably be more painful to endure.
With these findings, Gottman and Levenson were able to do something incredibly rare in psychological research: predict not only whether a couple would divorce, but approximately when. Specifically, they proved that:
- High negative affect predicted earlier divorce
- Low positive affect predicted later divorce
In turn, they were able to completely reframe how relationship failure is understood. Although many people are still quick to assume that conflict is what kills a relationship, this framework argues otherwise. If anything, stagnation seems to be just as predictive; it just operates on a different timeline.
What Makes Happy Marriages Different
It’s worth repeating that the stable couples in Gottman and Levenson’s studies were not completely immune to conflict, negativity or frustration. Just like any other normal couple, they’d still interrupt each other, disagree or roll their eyes on occasion. In this sense, what differentiated them from the couples who didn’t survive were two core capabilities:
- They enter negative states less frequently. Their conflict discussions had measurable moments of humor, shared perspective or warmth — even when the conversation became noticeably tense. These intentional double backs toward positivity are what buffered them from the damage of negativity.
- They exit negative states more easily. Although negativity during conflict wasn’t common for them, it did still happen from time to time. But when it did, these couples’ repair attempts were both more frequent and more successful. Even the most modest apologies and attempts at affection were enough to shift the interaction out of negativity. This is because the attempts were sincere; these couples genuinely wanted to solve the problem together in order to go back to normal.
In other words, stable couples had emotional systems that were resilient. Negativity didn’t become sticky because they didn’t allow it to trap them.
The most striking takeaway from the Gottman–Levenson work is that most relationship outcomes aren’t mysterious, nor are they purely a matter of compatibility. They usually follow patterns that can be modeled, predicted and, in many ways, fixed.
This is because a relationship is not just a unit in which two separate individuals interact with one another. It is a dynamic, ever-evolving system with laws that govern the flow of emotion, including the roach motel model. The message for couples to take away is both straightforward and practical: reduce whatever conditions create absorbing negative states, and increase the frequency of positive states, even if it’s only briefly.
If you’d like to kick-start your marriage revitalization journey, take the Marital Satisfaction Scale to find out your baseline: Marital Satisfaction Scale