TigerMatch

TigerMatch Research Study

Help us understand romantic compatibility through a comprehensive couples questionnaire

About This Study

This study examines how personality traits, values, and lifestyle preferences relate to relationship compatibility. You and your partner will each independently complete a 50‑question survey that takes approximately 20–25 Minutes.

20–25 Minutes
Complete at your own pace
Confidential
Your data is secure
For Couples
Both partners participate
  • Answer 50 questions about your personality, values, and preferences
  • For each question, indicate your answer, what you’d accept from a partner, and how important it is
  • Navigate easily with keyboard shortcuts (arrow keys)
  • Your progress is automatically saved as you go
Important: you and your partner will complete the survey separately. You’ll share a “Couple Code” to link your responses.
Begin QuestionnaireMethods & MathPrivacy Policy

Conducted by Ammaar Alam (COS 398) under the supervision of Prof. Marcel Dall'Agnol.

Research lens

What we're modeling

We're probing whether an interpretable scoring functional can consistently recover true partners from a massive synthetic negative sample. Each respondent yields a directed acceptability window (what you'd tolerate) plus an importance gradient (how costly a mismatch is), which we treat as a utility surface. Those surfaces feed logistic, calibrated-margin, and boosted heuristics to see which rule best separates held-out couples from randomized pairings.

Assemble mutuality tensors from answer / tolerance / importance vectors to capture asymmetric fit.
Apply finite-sample regularization and light fairness constraints so the score stays interpretable.
Track ROC/AUC lift, calibration, and stability to test whether a single algorithm can generalize across cohorts.

Bottom line: we're asking if an “optimal” couple-matching algorithm exists in practice and what guarantees it can honestly make.

Participation & outputs

What you commit & receive

Requirements

  • Both partners complete the 50-question instrument separately but with the same Couple Code.
  • Responses stay private; only aggregated statistics show up in COS 398 deliverables.
  • After both submissions, you can request your compatibility score plus a short interpretation.

The score is a research artifact—it measures how strongly the current model distinguishes your pair from synthetic matches. Think of it as a benchmark against the learned algorithm, not a deterministic forecast for your relationship.

Data collection is still underway; once enough datapoints are collected and the findings are finalized, the paper will be published for COS 398.
Questions? Contact Ammaar Alam. Participation is voluntary; you may withdraw at any time.