What the data actually shows
The most influential framework is Caryl Rusbult's investment model of commitment. It finds that commitment — and therefore whether people stay — rests on three things: satisfaction with the relationship, the quality of perceived alternatives, and the size of one's investment (time, shared history, finances, children, mutual friends). Crucially, satisfaction is only one of the three. High investment and few perceived alternatives can keep people committed even when satisfaction is low.
That helps explain a pattern that otherwise looks puzzling. Someone can be unhappy yet stay because they have poured years into the relationship and cannot easily picture a better alternative — both of which the model identifies as independent forces pulling toward staying. The relationship not 'working' in the satisfaction sense does not, on its own, predict leaving.
Other well-documented factors layer on top. Sunk-cost effects — the human tendency to keep investing in something because of what has already been put in — have been studied in relationships specifically. And research on the fear of being single, associated with work by Spielmann and colleagues, finds that people who are more afraid of being alone tend to settle for and stay in less satisfying relationships. None of these reflect weakness; they reflect ordinary, broadly shared psychology.
Why this feels different from how it actually is
From the outside, staying in an unhappy relationship can look like a simple failure of will, which is why the question is so often asked with a note of judgment. The research points the other way: it is the predictable result of investment, limited perceived alternatives, and entanglement — forces that act on almost everyone, not signs of a personal flaw.
It also feels different from the inside because the costs of leaving are concrete and immediate while the benefits are uncertain and distant. A shared home, finances, children, and a long history are vivid and present; a better future alone or with someone else is hypothetical. Human decision-making weights the concrete and the already-invested heavily, which tilts the felt calculation toward staying.
And hope plays a quiet, powerful role. People remember how the relationship once was and expect it could return, so the present difficulty can read as a phase rather than a settled state. That hope is not irrational — relationships do go through phases — but combined with sunk-cost thinking it can keep someone waiting longer than they might otherwise.
What the research says to do about it
The research is descriptive here — it explains the forces at work rather than prescribing what anyone should do, and this page is context, not advice about any specific relationship. What the investment model offers is a clearer way to see the question: separating satisfaction from investment and from perceived alternatives can make an otherwise foggy situation more legible.
Naming sunk cost for what it is can be clarifying. The honest version of the question is not 'how much have I already put in?' — that is gone either way — but 'how do I feel about what comes next?' Studies of sunk-cost effects suggest that simply recognizing the bias reduces its grip, in relationships as elsewhere.
Where alternatives feel non-existent, the research on the fear of being single is worth knowing. Because that fear can make people overstate the costs of being alone and understate their other options, awareness of it can restore a more accurate sense of the choices available. None of this points toward staying or leaving as the right answer — both can be valid — only toward seeing the decision more clearly.
What the research says does not help
Framing it as weakness or a lack of self-respect does not help and does not fit the evidence. The research shows people stay because of investment, few perceived alternatives, and entanglement — ordinary forces acting on almost everyone — so the moralizing version of the question misreads what is actually happening.
Treating past investment as a reason to stay is the sunk-cost trap. Time, effort, and shared history already spent cannot be recovered by staying, so basing the decision on them rather than on how you feel about what comes next is exactly the bias the research identifies. The honest question is forward-looking, not backward-looking.
Assuming there is a single correct answer that applies to everyone does not help either. The evidence describes why people stay; it does not declare that staying or leaving is right. Depending on the situation, either can be a valid choice, and a one-size-fits-all verdict ignores the real specifics that any particular person is weighing.
Real numbers in context
Caryl Rusbult's investment model identifies three drivers of whether people stay: satisfaction, the quality of perceived alternatives, and the size of one's investment. Because two of the three are independent of satisfaction, high investment and few perceived alternatives can keep someone committed even when the relationship is unsatisfying — which is the core of why people stay.
Layered on top are well-documented forces: sunk-cost thinking, which keeps people investing because of what they have already put in, and the fear of being single — associated with research by Spielmann and colleagues — which is linked to settling for and remaining in less satisfying relationships. These are understandable, broadly shared patterns, presented as neutral context rather than advice; staying or leaving can each be valid.