Stay-vs-go decision

A stay-vs-go decision is a structured choice about whether to remain in or leave a voluntary high-stakes commitment — a job, romantic relationship, graduate program, business partnership, or any domain where the person has the option to exit but exit is costly. The decision class is distinguished by three features: voluntariness (the person can leave), high stakes (the choice has substantial life consequences), and asymmetric uncertainty (the staying path has known characteristics while the leaving path is partly unknown). Stay-vs-go decisions recur across the adult life course and account for several of the most consequential individual decisions people make.

The decision class has been studied most extensively in voluntary turnover research (March & Simon 1958; Mobley 1977) and relationship dissolution research (Rusbult’s investment model, 1980; Levinger 1976). The decision-theoretic framing treats the choice as a comparison between two probability distributions over future outcomes; the behavioral-economic framing emphasizes the systematic biases that distort the comparison. Behavioral research suggests people may lean toward staying, though how far this departs from expected-value reasoning has rarely been measured; in one field study, people who made a major change such as quitting a job after a coin toss told them to reported being happier six months later (Levitt, 2021).

Three points are routinely missed in popular treatments. First, “should I quit?” is the wrong question; the right question is which distribution of future outcomes has higher expected value given honest probability assessments, which often differs from the immediate emotional reading. Second, sunk costs are particularly salient in stay-vs-go decisions and should be explicitly excluded from the comparison. Third, safeguards set up in advance, such as written exit criteria, outside review and time-bounded trials, may help more than effort at the moment of doubt, though they have not been tested on personal stay-or-go decisions.

Stay-vs-go decisions matter because they are among the highest-stakes choices most people make in adult life: whether to leave a job, end a relationship, exit a program, close a business, walk away from a project. The stakes are high not just for the person making the decision but for those connected to them; a single quit ripples through teams, families, organizations, and communities. The Great Resignation period of 2021-2022 and subsequent quiet quitting discussions surfaced the social-scale effects of stay-vs-go decisions when they aggregate across populations.

The research base has continued to develop. Hom and colleagues (2017) review a century of turnover research. Maertz and Campion's integrative framework extended the unfolding model with multiple "leaver types." This research describes who tends to leave and why; it does not tell any one person what to do.

For individuals, the practical value of the framework is that it prompts three questions, the first two drawn from turnover research: whether a shock has occurred, what embeds a person in their current situation, how asymmetric the costs of staying-when-wrong and leaving-when-wrong actually are. These factors are often visible to a person willing to examine them but get obscured by the emotional load that the decision itself produces.

The empirical research on voluntary departure decisions traces to early industrial-organizational psychology research on employee turnover. Mobley (1977) and Mobley, Griffeth, Hand, and Meglino (1979) framed quitting as the end-state of a sequential dissatisfaction process — first job dissatisfaction, then thoughts of quitting, then evaluation of alternatives, then turnover. This dissatisfaction-based account dominated the field for two decades and accurately describes some quitting trajectories, though it does not capture them all.

Lee and Mitchell's (1994) unfolding model reframed the field by identifying that many people leave not from gradual dissatisfaction but from shocks: discrete events prompting sudden re-evaluation. A merger, a missed promotion, a family medical event, a colleague's departure, an unsolicited offer — any can shift a stable calculation. The model proposed four decision paths: three set off by a shock and one driven by gradual dissatisfaction, which later work split into two. Replications by Lee et al. (1999) with accountants, Morrell et al. (2008) with UK NHS nurses, and Donnelly and Quirin (2006) with US accountants have generally supported the model.

Mitchell, Holtom, Lee, Sablynski, and Erez (2001) added embeddedness theory, in which people stay or leave based on links (connections), fit (compatibility), and sacrifices (what would be given up), and showed that embeddedness predicted leaving over and above job satisfaction and organizational commitment. Embeddedness research showed that people often stay in unhappy situations because links, fit, and sacrifices weigh against leaving. Steel and Ovalle's (1984) meta-analysis estimated the correlation between intent to leave a job and actually leaving at approximately r = 0.50; later reviews also find intentions among the best single predictors of leaving, with varying estimates.

Turnover research points to two factors that predict whether people leave a job, shocks and embeddedness; this entry adds a third, the costs of each possible error. Together they are a framework for reflection, not factors shown to separate good outcomes from poor ones.

• Whether a shock has occurred. Lee and Mitchell's unfolding model holds that many departures start with a shock rather than with slowly growing dissatisfaction. A shock can bring new information, or bring an old doubt to the surface, so it helps to ask which of the two it did.
• Embeddedness. Mitchell and colleagues' embeddedness theory frames staying through links (connections), fit (compatibility), and sacrifices (what would be lost). A person with high embeddedness will find staying more durable than dissatisfaction alone would predict; a person with low embeddedness should not assume more time will repair the situation. The framework explains why people sometimes leave situations they are happy with (low embeddedness plus shock) and stay in unhappy ones (high embeddedness plus tolerable dissatisfaction).
• Asymmetric costs of error. The cost of staying in a wrong situation differs from the cost of leaving a right one. Some commitments have high reversibility (a job is more reversible than a marriage); some have low reversibility (a doctorate, a major investment). The right decision depends partly on which error is more costly. A person facing a low-reversibility decision should weight staying-when-wrong more carefully; a person facing a high-reversibility decision can afford to leave with less certainty.

These three factors interact. A shock plus low embeddedness plus high reversibility makes leaving easy. Slow-drift dissatisfaction plus high embeddedness plus low reversibility makes the decision genuinely difficult, and whether structured reflection improves outcomes in this configuration has not been tested.

What it can do. The framework names three structural questions that the decision's emotional load can obscure. Whether a shock has occurred clarifies whether the impulse reflects new information or accumulated wear. Embeddedness clarifies whether staying is sustainable or held by inertia. Asymmetric costs clarify which error to weight more heavily. Writing down an answer to each can make them easier to weigh, though this has not been tested as a way to improve outcomes.

What it can't do. No framework can produce a single composite recommendation that overrides the person's own knowledge of their situation. Turnover research identifies factors that predict who leaves at population scale, but the application to any individual case requires the person's own knowledge of their situation, values, and possibilities. The framework also does not provide certainty; even decisions structured carefully against the empirical literature can produce poor outcomes when circumstances change unexpectedly. The point of structured reflection is to make better decisions on average, not to guarantee any particular outcome.

"Stay-vs-go decisions are about whether you're happy." Largely false. Embeddedness research shows that people often stay in situations they are unhappy with because their links, fit, and sacrifices weigh against leaving. Conversely, people sometimes leave situations they are happy with because of a shock or an unsolicited opportunity. Happiness is one input among several, not the determining factor in actual turnover behavior.

"If you're thinking about leaving, you should leave." Not consistently supported. The intent-to-leave-to-actual-turnover correlation is approximately r = 0.50 in Steel and Ovalle's (1984) meta-analysis — meaningful but not deterministic. Many people who think seriously about leaving actually stay, and the staying-after-considering-leaving outcomes are sometimes good (the consideration revealed information that improved the situation) and sometimes poor (the consideration was the symptom of a problem the staying did not solve). Structured reflection may help separate a passing impulse from a lasting problem, though it has not been tested as a way to improve these decisions.

"The decision logic is symmetric: cost of staying when wrong = cost of leaving when wrong." Almost never true. The two errors typically have very different costs depending on the reversibility of the commitment, the alternatives available, and the consequences of each choice. The framework's purpose is to make this asymmetry visible so it can inform the decision, rather than to produce a single composite recommendation that pretends the asymmetry doesn't matter.

"What others did in similar situations is a strong guide." Limited reliability. Other people's stay-vs-go decisions reflect their own embeddedness, shocks, and cost structures, often invisible from outside. Generic advice ("trust your gut," "give it more time") sometimes lands well but often misses the specific configuration of factors in any individual case. Turnover research describes what tends to happen across groups of people, which is useful context but cannot settle an individual case.

Consider someone seven years into a senior role at a stable company. The work has become routine. Pay is good, the team is good, there is no acute conflict. But the energy that once sustained the role has faded; mornings feel reluctant where they used to feel anticipatory. The intuition is "maybe time to leave" but no shock has occurred, embeddedness is high (long tenure, deep relationships, valuable benefits, real fit), and the alternatives are uncertain.

The framework analysis: no shock means the impulse reflects accumulated wear rather than new information; high embeddedness means staying is likely to feel comfortable for now; reversibility is moderate. The asymmetric cost analysis: staying when wrong loses years of potentially better fit elsewhere; leaving when wrong loses the embeddedness, certainty, and network. The two errors are not symmetric.

The framework does not produce "stay" or "leave" as output. It surfaces the structure: a high-embeddedness, no-shock, accumulated-wear configuration where slow-drift dissatisfaction has not been triggered by new information. No study shows how this configuration usually turns out. Embeddedness can keep staying comfortable, while research on people who changed jobs found that satisfaction rose after the move and then partly faded (Boswell, Boudreau & Tichy, 2005). Adding a shock (a missed promotion, a sudden opportunity, a family medical event) would change the configuration and the questions worth asking. The framework's value is in making the configuration visible, not in producing a single answer.

• Sunk cost — the cognitive bias that distorts stay-vs-go reasoning by making past investment feel like reason to continue. Distinguishing sunk cost from genuine embeddedness is one of the harder reasoning tasks in stay-vs-go decisions.
• Opportunity cost — the value of the alternative not chosen. Stay-vs-go decisions can underweight opportunity cost, because the unchosen alternative is often unknown and underspecified.
• Decision hygiene — the structural protections (writing the decision criteria in advance, considering the opposite, premortems) that aim to reduce reasoning errors. Particularly useful for stay-vs-go decisions where emotional load can swamp analysis.
• Decision support system — the broader category of tools designed to help structure complex choices. The Should I Quit? test is one example.
• Unfolding model of voluntary turnover — Lee and Mitchell's 1994 framework of decision paths to voluntary turnover, most of them set off by shocks. Foundational text for modern stay-vs-go research.
• Embeddedness theory — Mitchell and colleagues' 2001 framework explaining staying through links, fit, and sacrifices, over and above satisfaction. Explains why people often stay in unhappy situations and sometimes leave happy ones.
• Job embeddedness composite measures — validated multi-item scales used in turnover research to operationalize the embeddedness construct empirically.
• Risk aversion — Bernoulli 1738 / Pratt-Arrow framework; risk aversion underlies the asymmetric weighting of staying (lower variance) vs going (higher variance) in voluntary departure decisions.
• Bounded rationality — stay-vs-go decisions are typically made under bounded rationality with incomplete information about the alternatives.
• Nudge theory — choice-architecture features of the decision context (defaults, framing, salience) substantially affect stay-vs-go decisions.

The LifeByLogic Should I Quit? tool covers one kind of stay-vs-go decision: whether to leave a job. It scores burnout (exhaustion and disengagement), meaning at work and intent to leave with 26 LifeByLogic-written questions, then runs 10,000 simulated versions of staying and of leaving built from your own figures for pay, the salary range you expect, raises, expenses, search difficulty, the burnout relief and meaning boost you expect, and your time horizon. It reports the share of runs in which leaving came out ahead on a combined life-quality score, which summarizes your assumptions and is not a forecast or a probability, along with one of six patterns read from today's scores and an optional regret prompt; its report opens three next steps and the rest of the result. The questions have not been validated and have no norms, and the tool does not assess shocks, embeddedness or the costs of each error. The full methodology and its limits are documented on the tool methodology page.

Take the Should I Quit? test →

For the specific case where you have a defined pivot alternative in mind — not just "should I leave" but "should I leave for this" — the Career Pivot Decision Matrix structures a 6-domain weighted comparison between your current role and the pivot option, with a separate calibration check on how informed your evaluation actually is. The matrix complements Should I Quit: Should I Quit handles the leave-or-stay question; the Career Pivot Matrix handles the leave-or-stay-given-this-specific-alternative question.

Use the Career Pivot Decision Matrix →

This entry is intended as a citable scholarly reference. Choose the format that matches your context. The retrieval date should reflect when you accessed the page, which may differ from the entry's last-reviewed date shown above.

LifeByLogic. (2026). Should I Stay or Go? How to Make a Stay-or-Go Decision. https://lifebylogic.com/glossary/stay-vs-go-decision/

LifeByLogic. "Should I Stay or Go? How to Make a Stay-or-Go Decision." LifeByLogic, 31 Oct. 2026, https://lifebylogic.com/glossary/stay-vs-go-decision/.

LifeByLogic. 2026. "Should I Stay or Go? How to Make a Stay-or-Go Decision." October 31. https://lifebylogic.com/glossary/stay-vs-go-decision/.

@misc{lblstayvsgo2026,
  author = {{LifeByLogic}},
  title = {Should I Stay or Go? How to Make a Stay-or-Go Decision},
  year = {2026},
  month = {oct},
  publisher = {LifeByLogic},
  url = {https://lifebylogic.com/glossary/stay-vs-go-decision/},
  note = {Accessed: 2026-10-31}
}