What Actually Predicts Relationship Success Better Than MBTI Compatibility

Why MBTI compatibility is a weak predictor

MBTI is a useful self-understanding tool. It's a poor outcome predictor. The reason isn't mysterious. Most popular compatibility charts are built on theoretical pairings (cognitive function stacks that "balance" each other) rather than on longitudinal studies of actual couples. When researchers have looked at personality similarity and dissimilarity using the more rigorously validated Big Five model, the effect sizes on marital satisfaction are real but small — much smaller than the effect of how partners handle conflict, how securely attached they are, or whether they share core values.

There's another structural problem. MBTI sorts people into 16 boxes, but the boundaries between boxes are softer than the labels suggest. A person who scores 51% Introvert is grouped with someone scoring 95% Introvert, and treated as fundamentally different from someone scoring 49%. Compatibility charts then multiply these noisy categorizations across four dimensions, which compounds the noise rather than canceling it. The output looks precise. It mostly isn't.

MBTI tells you how two people think and communicate. It doesn't tell you whether they'll be kind to each other under pressure. The second question is the one that actually decides relationships.

Predictor 1: Attachment style compatibility

Attachment style is the closest thing relationship psychology has to a master variable. The framework comes from John Bowlby's attachment theory, adapted to adult romantic relationships by Cindy Hazan and Phillip Shaver in their landmark 1987 paper. Decades of follow-up research have refined the picture, but the core finding has held up unusually well: how you behave when you feel threatened, disconnected, or abandoned shapes the quality of your closest relationships more than almost any other trait.

There are roughly four adult attachment patterns — secure, anxious-preoccupied, dismissive-avoidant, and fearful-avoidant. They aren't personality types, exactly. They're defaults under relational stress. A secure person tends to assume good faith, ask for what they need, and tolerate brief disconnections. An anxious person tends to feel disconnection as threat and escalate to reestablish closeness. An avoidant person tends to feel closeness as threat and pull back to reestablish autonomy. Fearful-avoidant carries elements of both, often arising from chaotic early environments.

The pairing data is striking. Secure-secure couples consistently report the highest relationship satisfaction and the lowest conflict, in part because they're both good at the same things: clear communication, emotional regulation, and repair. Secure-anxious and secure-avoidant pairings often work well, because the secure partner can model regulation. Anxious-avoidant is the worst combination by a meaningful margin: one partner's pursuit triggers the other's withdrawal, which triggers more pursuit, which triggers more withdrawal. The pattern is so well-documented it has a name (the anxious-avoidant trap), and it's self-reinforcing in a way personality differences rarely are.

Knowing your own attachment style, and your partner's, is the single most useful piece of relational self-knowledge most people can get. If you haven't taken one, the attachment style test is a reasonable starting point. It's not deterministic — attachment can shift over time, especially with secure partners or good therapy — but it's a better compatibility signal than anything MBTI produces.

Predictor 2: Gottman's Four Horsemen

John Gottman's work at the University of Washington across multiple decades produced one of the most-cited findings in relationship research: by observing how couples argue, his team could predict divorce with high accuracy — often above 90% in his original samples, though independent replications have settled in around the 80% range, still a remarkable result.

The strongest predictors were four specific conflict behaviors he called the Four Horsemen:

• Criticism — attacking your partner's character rather than addressing a specific behavior. "You never…" and "You always…" are the common forms.
• Contempt — sneering, mockery, eye-rolling, name-calling, sarcasm aimed at humiliation. Gottman called this the single best predictor of divorce, period.
• Defensiveness — deflecting any criticism rather than considering it, often by counter-attacking. The underlying message is "I am not responsible for anything."
• Stonewalling — shutting down completely, refusing to engage, going silent or leaving the room without repair. Often a sign of physiological flooding, but corrosive over time regardless of cause.

None of these are personality features in the MBTI sense. They're learned conflict behaviors. Two ENFJs can have a contempt-saturated marriage. Two ISTPs can have a marriage with none. The presence or absence of the Horsemen does more to predict whether a couple will still be together in ten years than any combination of personality traits.

The corollary matters as much: Gottman's research also identified positive patterns that protect relationships — turning toward bids for connection rather than turning away, maintaining a roughly 5:1 ratio of positive to negative interactions during conflict, and what he called "repair attempts," the small moves (humor, a touch, an apology, a shift in tone) that de-escalate before things get bad. Couples who repair well stay together. Couples who can't mostly don't.

Predictor 3: Shared values and life direction

Personality differences can be navigated. Value differences on load-bearing questions usually can't. The hard ones are predictable: whether to have children, how to raise them, how to handle money, how religion fits in, whose career takes priority and when, how to relate to extended family, what "a good life" even looks like. Couples who align on these — or who genuinely respect each other's differences without trying to convert each other — tend to do well across a wide range of personality combinations.

Couples who don't usually run into trouble at predictable inflection points: marriage, kids, career moves, illness, aging parents, retirement. The argument feels like it's about whose turn it is to do school pickup, but it's actually about whose career is presumed to matter more. The argument feels like it's about the in-laws, but it's actually about how much obligation each partner thinks family deserves. Personality style colors how these arguments unfold. Value alignment determines whether they're resolvable in principle at all.

The practical implication for dating: spend less time matching personality types and more time finding out what the other person actually wants their life to look like in ten years, and whether your answers can coexist. That conversation isn't romantic. It's also one of the most useful filters available.

Predictor 4: Conflict repair skill

All long-term couples fight. The question isn't whether you fight; it's whether you can recover from fighting. Gottman's longitudinal work and parallel research from labs like Susan Johnson's on Emotionally Focused Therapy converge on the same finding: the ability to repair after rupture is one of the strongest predictors of relational survival.

Repair looks unremarkable from the outside. It's an apology that names the specific thing. It's a partner who, in the middle of an argument, says "Wait, I'm getting defensive — let me try again." It's the move from "you're being ridiculous" to "I see why this lands the way it does." These micro-skills aren't innate. They can be learned, and couples who learn them do measurably better over time, regardless of personality match.

This is one place MBTI does have something useful to offer: knowing your own and your partner's communication defaults can speed up repair. An NT partner who instinctively reaches for logical reframes can learn that an NF partner needs the emotional acknowledgment first. A J partner who wants resolution by bedtime can learn that a P partner sometimes needs to sit with a question overnight. The information isn't a compatibility verdict; it's a manual for repair.

Predictor 5: Big Five Agreeableness

Of the standard Big Five traits, Agreeableness has the most consistent positive relationship with marital satisfaction, both for the individual high in it and for their partner. People high in Agreeableness are more cooperative, more forgiving, less prone to contempt, and more willing to do the small daily maintenance work that keeps closeness alive. Neuroticism is the mirror image — higher trait neuroticism in either partner predicts more conflict and lower satisfaction, an effect documented across many cultures.

What about similarity? The pop-psych idea that "similar personalities make better couples" is at best half-true. Some traits — Conscientiousness, shared values around order and effort — benefit from similarity. Others, like Extraversion, tolerate quite a bit of mismatch. Across the board, similarity effects are smaller than people expect. Two emotionally regulated, kind, conscientious people tend to do well together regardless of whether their Big Five profiles look alike.

The takeaway: if you're going to use a personality framework to inform partner choice, Big Five is the more rigorous one, and the most useful single number to look at is Agreeableness — your own and the other person's. It correlates with the actual behaviors that make day-to-day life with someone good or bad.

How to use personality tests intelligently

Given all of this, the question isn't whether to use personality tests in relationships. It's how to use them without overreaching. Three reasonable uses, in roughly increasing order of value:

As conversation starters."I'm an INFJ, you're an ESTP — what does that mean for how we make decisions together?" is a better date conversation than what most people are running. It surfaces real differences in low-stakes language. Whether the typology is metaphysically true matters less than whether it gives you a useful vocabulary for the differences that are already there.

As blind-spot detectors. A profile that says "people with your pattern tend to underweight their partner's need for X" is genuinely useful self-knowledge, even if the underlying typology is approximate. Same for love languages: knowing that you default to acts of service while your partner registers care primarily through words of affirmation predicts a class of common misunderstandings, and lets you preempt them. The love language test is best used this way — not as compatibility, but as translation.

As a layered self-portrait. Take MBTI to understand cognitive and communication style. Take attachment to understand how you behave under relational stress. Take love languages to understand how you give and receive care. Look at the three together. Share the results with a partner. The combination is much more informative than any single result, and much more honest about its own limitations than a single "compatibility score" ever will be.

Use them as instruments, not verdicts. Treat MBTI compatibility charts as starting points for self-knowledge rather than scoring rubrics. The 16-type test is a fine entry point; attachment is the more important follow-up if you want a single test that actually correlates with relational outcomes.

What this looks like in practice

If you're assessing a current or potential relationship, a more honest checklist than "are our types compatible?" might be:

• Can we have a hard conversation and come out the other side feeling closer rather than further apart?
• When we're at our worst, do contempt, stonewalling, or character attacks show up — and when they do, do we repair?
• Do we want roughly the same kind of life ten years from now, including the parts that involve compromise?
• Do we both feel safe asking for what we need, and is the answer usually some version of yes?
• When one of us is dysregulated, does the other person co-regulate or escalate?

None of those questions are answered by a four-letter type. They're answered by paying close attention over time. Personality tests can support that attention by giving you better language for what you notice. They can't replace it.

Frequently asked questions