The Science ofMarriage Success
Four decades of watching couples argue produced a short list of what predicts divorce and what predicts lasting. Most of it shows up in a text thread.
John Gottman's "Love Lab" at the University of Washington spent decades recording couples: talking, arguing, wired to sensors, then followed for years. The output is the most-cited body of marriage research there is, and the most misquoted. Here is what it found, how confident you should be in it, and what it looks like on your phone.
Key takeaways
- Stable couples keep roughly five positive interactions for every negative one during conflict; couples heading for divorce sit near 0.8 to 1 (Gottman, 1994).
- Four behaviours predict divorce: criticism, contempt, defensiveness and stonewalling. Contempt is the strongest single predictor.
- The famous "over 90% accuracy" figures are real but in-sample; on new couples, accuracy drops sharply (Heyman & Slep, 2001). Treat the patterns as risk factors, not prophecy.
- Small "bids" for attention, and whether a partner turns toward them, matter more than grand gestures. Texting is made of bids.
What the Love Lab did
Couples came to an apartment-style lab and spent a day or a weekend doing ordinary things while being recorded; they also had structured conversations about an ongoing disagreement. Observers coded every turn (affect, tone, who repaired, who withdrew) while sensors tracked heart rate, skin conductance and movement (Gottman & Levenson, 1992). Then the lab followed the couples for years to see who stayed together and how satisfied they were.
The design is why the findings are concrete. Gottman was not asking couples what they thought went wrong; he was watching what they did and checking it against what happened next.
The prediction claims, and the caveat
Carrère and Gottman (1999) coded the first three minutes of a conflict discussion in 124 newlywed couples and could predict who would divorce over the following six years from those three minutes alone. Read the study. Gottman and Levenson (2000) predicted divorce timing over 14 years with 93% accuracy using marital satisfaction, thoughts of separation and the emotional pattern of two conversations. Read the study.
These numbers are true and easy to over-read. The models were fitted to the same couples they were scored on. Heyman and Slep (2001) showed that when such divorce equations are cross-validated on fresh couples, accuracy and predictive value fall precipitously. Read the critique. Gottman's patterns are well-replicated as risk factors; the "we can predict your divorce in three minutes" version is not something anyone can do to you.
The Four Horsemen, and their text versions
Gottman named four conflict behaviours that predict dissolution. They tend to escalate in this order.
Criticism
Attacking the person, not the problem. "You didn't text when you landed" is a complaint. "You never think about anyone but yourself" is criticism. Antidote: a soft start-up ("I got worried when I didn't hear from you").
Contempt
Disgust, mockery, superiority. In text: sarcasm with an edge, "wow. ok.", an eye-roll emoji at a serious message, screenshots sent to friends. The strongest single predictor of divorce. Antidote: say what you appreciate, regularly and specifically.
Defensiveness
Counter-attack or innocent victim. "I only did that because you..." Antidote: take responsibility for your part, even a small part.
Stonewalling
Shutting down. In text it is the unanswered paragraph, "k", or the read receipt with nothing after it. Usually a sign of being flooded. Antidote: say you need a break, name when you will be back, then come back.
Our guides to passive-aggressive texting and toxic text patterns go through the horsemen message by message.
The 5:1 ratio
In Gottman's observations, couples who stayed together and satisfied maintained about five positive interactions (humour, affection, interest, agreement) for every negative one during disagreements, while couples heading for divorce ran close to 0.8 to 1 (Gottman, 1994). The ratio is not a quota; it is a description of what a repairable argument looks like. Most of the positives are tiny: a joke mid-fight, "fair point", a heart on a message you disagree with.
"In emotionally intelligent marriages, a dynamic is established where negative thoughts and feelings are kept from overwhelming the positive ones."— Dr. John Gottman, The Seven Principles for Making Marriage Work
Texting makes this measurable in a way conversation never was. A week-by-week tone timeline, of the kind MosaicChats builds from an exported chat, is a crude version of the same count, and it shows the thing couples cannot see from inside: whether the balance has shifted, and when.
Bids and turning toward
A bid is any small reach for attention: a link, a photo of the dog, "you up?" Partners can turn toward (respond), turn away (ignore), or turn against (snap). In Gottman's newlywed observations, couples who were still married six years later had turned toward each other's bids 86% of the time; couples who had divorced, 33% (Gottman & DeClaire, 2001).
A text thread is a bid log. Unanswered links, questions that never get a reply, and photos met with silence are turning away, at scale, with timestamps.
The Seven Principles, with a texting example each
- Enhance your love maps. Know your partner's inner world. In text: you know what the meeting they are nervous about is called, and you ask how it went.
- Nurture fondness and admiration. Say the specific thing you admire, unprompted, more often than you send logistics.
- Turn toward each other. Answer the bid, even with an emoji, even late.
- Let your partner influence you. "You're right, let's do it your way" appears in the thread, from both of you.
- Solve your solvable problems. Soft start-ups, and repair attempts ("that came out wrong") that get accepted rather than ignored.
- Overcome gridlock. Perpetual problems get talked about with curiosity instead of re-fought; ask what the dream behind the position is.
- Create shared meaning. The in-jokes, rituals and plans that make the thread yours.
Count your ratio
Upload a chat and MosaicChats shows the tone week by week, who initiates, and how reply patterns have shifted: the closest thing to a Love Lab you can run from your phone.
Analyze your chatFrequently asked questions
Can Gottman's research really predict divorce?
It identifies strong risk factors, and the original studies reported high accuracy (for example 93% over 14 years). But those models were fitted to the couples they were tested on; Heyman and Slep (2001) showed accuracy drops sharply on new couples. Use the patterns as warning signs, not as a forecast.
What are the Four Horsemen?
Criticism (attacking character rather than behaviour), contempt (mockery, disgust, superiority), defensiveness (counter-attacking or playing the victim) and stonewalling (shutting down). Contempt is the strongest single predictor of divorce. Each has an antidote: soft start-ups, expressing appreciation, taking responsibility, and self-soothing breaks.
What is the 5:1 ratio?
During conflict, stable couples in Gottman's observations had about five positive interactions for every negative one; couples heading for divorce were near 0.8 to 1. The positives are mostly small: humour, agreement, affection, interest.
Does Gottman's research apply to texting?
The behaviours do. Criticism, contempt, stonewalling, bids and repair attempts all appear in messages, often more visibly than in person because there is a record. The exact ratios were measured on face-to-face conflict, so treat text counts as an indicator, not a replication.
Related reading
References & Sources
- Carrère, S., & Gottman, J. M. (1999). Predicting divorce among newlyweds from the first three minutes of a marital conflict discussion. Family Process, 38(3), 293–301. Source
- Gottman, J. M. (1994). Why marriages succeed or fail: And how you can make yours last. Simon & Schuster.
- Gottman, J. M., & DeClaire, J. (2001). The relationship cure: A five step guide to strengthening your marriage, family, and friendships. Crown.
- Gottman, J. M., & Levenson, R. W. (1992). Marital processes predictive of later dissolution: Behavior, physiology, and health. Journal of Personality and Social Psychology, 63(2), 221–233. Source
- Gottman, J. M., & Levenson, R. W. (2000). The timing of divorce: Predicting when a couple will divorce over a 14-year period. Journal of Marriage and Family, 62(3), 737–745. Source
- Gottman, J. M., & Silver, N. (1999). The seven principles for making marriage work. Crown.
- Heyman, R. E., & Slep, A. M. S. (2001). The hazards of predicting divorce without crossvalidation. Journal of Marriage and Family, 63(2), 473–479. Source