Scientific Foundation

Framework: Schwartz’s Theory of Basic Human Values

Schwartz, S. H. (1992). Universals in the content and structure of values: Theoretical advances and empirical tests in 20 countries. Advances in Experimental Social Psychology, 25, 1-65.

Across 20 countries and over 10,000 participants using the Schwartz Value Survey, Schwartz identified 10 motivationally distinct value types that appear cross-culturally: Universalism, Benevolence, Tradition, Conformity, Security, Power, Achievement, Hedonism, Stimulation, and Self-Direction.

Critically, these 10 values are not independent — they form a circular structure (a “circumplex”) where adjacent values are compatible and opposite values conflict. This structure collapses into two bipolar dimensions: Self-Transcendence vs. Self-Enhancement and Openness to Change vs. Conservation.

How we use it

Our core quiz measures the two bipolar axes (values_self_transcendence and values_openness_to_change) as continuous 0–100 scales. These axes position the user on the circumplex. We then use 6 dedicated “value expansion” questions to separate the values within each pole — for example, distinguishing Universalism from Benevolence on the self-transcendence pole. This gives us a ranked list of all 10 values from a total of ~26 questions (20 core + 6 value), compared to Schwartz’s 57.

Does value similarity predict relationship satisfaction?

Gaunt, R. (2006). Couple similarity and marital satisfaction: Are similar spouses happier? Journal of Personality, 74, 1401-1420.

In a sample of 248 Israeli married couples, greater similarity between partners was associated with higher marital satisfaction and lower negative affect. Importantly, similarity on gendered personality traits and values was more strongly associated with relationship outcomes than similarity on attitudes or religiosity. Profile-based similarity (the overall pattern match) was a stronger predictor than simple difference scores.

How we use it — and why we’re honest about the limits

We weight value similarity as the strongest factor in Tier 1 matching (18% of Tier 1 weight for self-transcendence, 15% for openness-to-change). This is defensible because the research consistently shows value similarity predicts initial attraction and connection quality, which is exactly what our product facilitates — we’re matching people who haven’t met yet, not predicting long-term marital success.

We don’t claim “similar values = lasting relationship.” We claim “similar values = you’ll click when you meet.” That’s what the Montoya meta-analysis actually supports.

Framework: Five Factor Model (Costa & McCrae)

Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources.

John, O. P., & Srivastava, S. (1999). The Big Five trait taxonomy: History, measurement, and theoretical perspectives. In L. A. Pervin & O. P. John (Eds.), Handbook of Personality (pp. 102-138). New York: Guilford Press.

Five broad personality dimensions consistently replicated across cultures, languages, and measurement methods: Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism (OCEAN). Each dimension contains 6 subfacets. The Big Five is the dominant personality framework in academic psychology because of its predictive validity and replication record.

How we use it

We don’t use the Big Five as a complete framework — we selectively pull the dimensions most relevant to interpersonal compatibility:

• Openness:Split into two subfacets — intellectual (curiosity about ideas) and experiential (willingness to try new things). We dropped openness_aesthetic because it had zero reliable measurement in a 20-question quiz and correlates substantially (r ≈ .6) with the other two subfacets.
• Conscientiousness:Measured as a single dimension. Research consistently links conscientiousness similarity to relationship satisfaction, especially for cohabiting couples (Watson, Hubbard, & Wiese, 2000).
• Extraversion:Measured as a single dimension but used with complementarity matching rather than similarity matching (see Section 3).
• Neuroticism:We measure this as “emotional_volatility” (a subfacet of neuroticism focused on reactivity rather than chronic negativity) because reactivity is more relevant to interpersonal dynamics and less stigmatizing as a quiz result.
• Agreeableness:Not measured as a standalone dimension. Instead, it’s captured indirectly through warmth, nurture_vs_challenge, and values_self_transcendence, which together cover the interpersonal behaviors that agreeableness predicts.

How this shaped our design

We weight Openness dimensions heavily in Tier 1 (similarity matching) but place Conscientiousness and Neuroticism-related dimensions in Tier 3 (proximity matching with a tolerance band). This means we don’t require exact conscientiousness matches — just that people aren’t extremely far apart. This reflects the research: being somewhat similar on conscientiousness prevents friction, but exact matching doesn’t add predictive value.

Framework: The Interpersonal Circumplex (Wiggins, Kiesler, Sullivan)

Wiggins, J. S. (1979). A psychological taxonomy of trait-descriptive terms: The interpersonal domain. Journal of Personality and Social Psychology, 37, 395-412.

Kiesler, D. J. (1983). The 1982 Interpersonal Circle: A taxonomy for complementarity in human transactions. Psychological Review, 90, 185-214.

All interpersonal behavior can be mapped onto two axes: Affiliation (warmth–hostility) and Control (dominance–submission). The principle of complementarity states that on the affiliation axis, similarity is preferred (friendliness invites friendliness), while on the control axis, reciprocity is preferred (dominance invites submission, and vice versa).

The key empirical test: Dryer & Horowitz (1997)

Dryer, D. C., & Horowitz, L. M. (1997). When do opposites attract? Interpersonal complementarity versus similarity. Journal of Personality and Social Psychology, 72(3), 592-603.

Two experiments directly tested whether complementarity or similarity produces greater satisfaction in dyadic interactions. Participants in complementary pairings (dominant person + submissive confederate, or vice versa) reported significantly more satisfaction than those in similar pairings. Notably, satisfied participants also perceived their partners as similar to themselves — even when they were objectively complementary. This means complementarity feels like similarity from the inside.

How we use it

This directly shaped our Tier 2 matching rules. Dominance uses complementarity matching with an optimal gap of ~25 points on our 0–100 scale. Warmth uses similarity matching (corresponding to the affiliation axis). We set the penalty factor for dominance complementarity at 2.0 (the steepest penalty in our algorithm) because Dryer & Horowitz found it was the single strongest predictor of interaction satisfaction.

Framework: The ECR-R (Experiences in Close Relationships — Revised)

Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000). An item response theory analysis of self-report measures of adult attachment. Journal of Personality and Social Psychology, 78(2), 350-365.

The ECR-R measures adult attachment along two continuous dimensions: attachment anxiety (fear of rejection and abandonment) and attachment avoidance (discomfort with closeness and dependence). Low scores on both = secure attachment. The anxious-avoidant trap — where anxious individuals are drawn to avoidant partners — is one of the most documented dynamics in relationship psychology.

How we use it

Attachment dimensions are in Tier 3 (proximity matching) with the tightest tolerance band (±15 points) and the heaviest weight within Tier 3 (35% each for anxiety and avoidance, totaling 70% of Tier 3). For dating matches specifically, Tier 3 is weighted at 35% of the total score — meaning attachment compatibility alone accounts for roughly 25% of the dating match score.

Framework: The Humor Styles Questionnaire (HSQ)

Martin, R. A., Puhlik-Doris, P., Larsen, G., Gray, J., & Weir, K. (2003). Individual differences in uses of humor and their relation to psychological well-being. Journal of Research in Personality, 37(1), 48-75.

Humor is not a single trait — it operates through four distinct styles organized in a 2×2 matrix. The first axis is self vs. other orientation; the second is benign vs. detrimental. This produces: Affiliative (warm bonding humor), Self-Enhancing (coping humor), Aggressive (sarcasm, ridicule), and Self-Defeating (self-deprecation to gain acceptance).

The original study (N = 1,195) showed the four scales differentially related to Big Five traits and wellbeing measures. Internal consistency (Cronbach’s alpha) ranged from .77 to .81 across all four scales.

How we use it

We consolidated from 4 styles to 2 in the core quiz (humor_positive combining affiliative + self-enhancing; humor_edgy combining aggressive + self-defeating) because reliable 4-style measurement requires 32 total items which is impossible in a 20-question quiz. Two dedicated depth questions re-expand to 4 styles. Humor similarity is weighted at 28% of Tier 1 (15% for humor_positive + 13% for humor_edgy), reflecting its importance for initial social chemistry.

Framework: The Fast Friends Procedure

Aron, A., Melinat, E., Aron, E. N., Vallone, R. D., & Bator, R. J. (1997). The experimental generation of interpersonal closeness: A procedure and some preliminary findings. Personality and Social Psychology Bulletin, 23(4), 363-377.

Over a 45-minute period, unacquainted pairs who carried out structured self-disclosure tasks that gradually escalated in intensity reported significantly greater post-interaction closeness than pairs who engaged in comparable small-talk tasks. The closeness achieved in 45 minutes matched the average level participants reported feeling in their closest existing relationships.

Three additional findings shaped our design:

1. Matching pairs for attitudinal agreement did NOT increase closeness
2. Leading pairs to expect mutual liking did NOT increase closeness
3. Making closeness an explicit goal did NOT increase closeness

The only thing that mattered was the structure of escalating reciprocal self-disclosure.

How we use it in three ways

First, the quiz question ordering follows Aron’s escalation gradient: early questions are fun and preference-based, middle questions involve values and opinions, and late questions are personal. This isn’t just for engagement — gradual self-disclosure produces more accurate self-revelation than jumping straight to vulnerable questions.

Second, our conversation starters for matched users are designed as mini Fast Friends prompts — structured questions that invite escalating reciprocal disclosure based on the specific dimensions where users are compatible.

Third, Aron’s finding that attitudinal matching didn’t affect closeness generation is why we don’t over-index on similarity. The matching algorithm predicts who will have good chemistry, but the conversation design is what generates actual closeness — and Aron showed that structured disclosure works regardless of pre-existing similarity.

Framework: Reis’ Interpersonal Process Model of Intimacy

Reis, H. T., Clark, M. S., & Holmes, J. G. (2004). Perceived partner responsiveness as an organizing construct in the study of intimacy and closeness. In D. Mashek & A. Aron (Eds.), Handbook of Closeness and Intimacy (pp. 201-225). Mahwah, NJ: Erlbaum.

The single strongest predictor of relationship intimacy and satisfaction is not compatibility, personality, or shared interests — it’s perceived partner responsiveness (PPR): the extent to which someone feels understood, validated, and cared for by their partner. PPR predicts relationship satisfaction with medium-to-large effect sizes. It also predicts sexual desire, commitment, and even intellectual humility.

How we use it

This finding shapes the post-match experience rather than the matching algorithm. Our conversation starters are designed to create conditions for responsiveness — they’re prompts that invite one person to share something real and give the other person an opportunity to respond with understanding.

Why this matters for product strategy: The matching algorithm gets people in the door, but Reis’ research shows that what keeps them is the quality of their conversations. If our conversation design produces high-responsiveness interactions, users will feel closer to their matches faster than on any other platform. This is the real moat — not the quiz, not the algorithm, but the structured conversation experience.

Framework: The Third Dimension of a Good Life

Oishi, S., & Westgate, E. C. (2022). A psychologically rich life: Beyond happiness and meaning. Psychological Review, 129(4), 790-811.

Prior to this paper, wellbeing psychology recognized two dimensions of a good life: hedonic (happiness, pleasure) and eudaimonic (meaning, purpose). Oishi & Westgate proposed and empirically validated a third dimension: psychological richness — the extent to which one’s life involves diverse, interesting, perspective-changing experiences. In studies across 9 countries (N > 3,000), a significant minority chose a psychologically rich life over a happy or meaningful one when given the choice.

How we use it

We measure this through our life_meaning dimension (0 = experiential/hedonic, 100 = purpose/meaning), with psychological richness captured as a mid-range position plus high openness_experiential scores. Two people who both orient toward richness will understand each other’s life choices — why they took the weird job, moved to the unfamiliar city, left the stable relationship for uncertainty. Two people where one seeks richness and the other seeks security will fundamentally misunderstand each other’s motivations.

Framework: Intrinsic Motivation and the Three Basic Needs

Deci, E. L., & Ryan, R. M. (2000). The “what” and “why” of goal pursuits: Human needs and the self-determination of behavior. Psychological Inquiry, 11(4), 227-268.

Three universal psychological needs drive intrinsic motivation and wellbeing: autonomy (feeling volitional), competence (feeling effective), and relatedness (feeling connected). When all three needs are met, people are intrinsically motivated and thrive.

How we use it

SDT doesn’t directly inform the matching algorithm — it informs the product design to maximize engagement and retention.

• Competence: The quiz satisfies this — you learn something real about yourself, you feel understood.
• Relatedness: The sharing mechanic — seeing your match scores creates social connection.
• Autonomy: The optional depth and value modules — you choose how deep to go, at your own pace, with no gatekeeping.

Research consistently shows that autonomy-supportive framing increases both completion rates and data quality compared to coercive framing.

Why three modes, not one

Our algorithm uses three distinct matching rules depending on the dimension: similarity, complementarity, and proximity.

Context-dependent weighting

Why scenarios over Likert scales

Paulhus, D. L. (1991). Measurement and control of response bias. In J. P. Robinson et al. (Eds.), Measures of Personality and Social Psychological Attitudes (pp. 17-59). San Diego: Academic Press.

When you ask “I care about other people’s feelings — strongly agree to strongly disagree,” nearly everyone agrees. The question measures social performance, not actual behavior. Forced-choice scenarios where all options represent defensible behaviors force tradeoffs that reveal genuine preferences.

The Portrait Values Questionnaire methodology uses third-person framing to reduce self-enhancement bias. Our scenario format serves the same function — instead of asking “do you value honesty?” we describe a situation where honesty and kindness conflict, and your choice reveals which value actually wins.

Primary dimension architecture

Every question has a primary dimension that gets measured regardless of which option the user picks. This is inspired by Item Response Theory (IRT) principles: a well-designed item should discriminate along a single latent trait for all respondents. Options A, B, C, D all push the primary dimension to different positions on the scale, ensuring every user provides data on that dimension.

This architecture guarantees a minimum of 3 data points per dimension per user, regardless of the specific answer pattern. Standard psychometric practice recommends a minimum of 3 items per construct for internal consistency above α = .60.

Our scoring uses a diminishing returns function: as a dimension score approaches the extremes (0 or 100), additional pushes in the same direction have less effect. This is a deliberate psychometric choice, not just a mathematical convenience.

We display match scores as ranges (e.g., “78–92% match”) that narrow as users answer more questions. This is grounded in basic measurement uncertainty principles.

With only 20 data points (core quiz), each dimension score has substantial measurement error. The standard error of measurement decreases as the square root of the number of items. At 20 questions across 16 dimensions, each dimension averages ~4 data points — enough for a rough estimate but not for precision. At 36 (core + depth + values), confidence approaches acceptable levels.

We make this uncertainty visible to users rather than presenting false precision. This serves two purposes: (a) it’s honest, and (b) it creates a natural incentive to answer more questions without making current results feel worthless.