Methodology.

01 — Source material

Consulted, not copied.

We drew on peer-reviewed journal articles, textbook chapters, and public research summaries in couples and close-relationships psychology, to think through which behaviors matter and why. We don't reproduce or license any specific researcher's instrument.

02 — Four dimensions

Chosen on purpose.

Every question maps to one of four things the research keeps coming back to:

• Warmth & friendship — the ordinary, day-to-day fondness that predicts stability over time.
• Corrosive conflict — patterns like criticism, contempt, defensiveness, and shutdown.
• Small bids for connection — the tiny moments of turning toward or away that quietly add up.
• Repair — the ability to de-escalate after a rupture and reconnect.

03 — Item design

Why the questions sound the way they do.

• About moments, not judgments. Questions ask what happened, not how you'd rate yourselves overall.
• Past week only. A short recall window gives a snapshot of now, not a colored memory.
• Mixed phrasing. Some items are worded positively, some negatively, so you can't just agree your way through.
• Five-point scale. Enough range to catch change over time without pretending to false precision.
• Perspective is labeled. Each result is tagged self-report, partner-perceived, or mixed, so you know what kind of answer you're looking at.

04 — Pattern question

A categorical layer on top of the scales.

The "who withdrew first" question isn't scored. It's used to help tell apart patterns the scales alone can't: one-sided withdrawal, mutual quiet, ongoing friction without repair, and general avoidance.

05 — Scoring & stages

From items to a stage of drift.

• Equal-weight dimensions. Each of the four dimensions is scored 0–100% and averaged, then mapped to a 0–90 scale. No single dimension can dominate the result.
• Skipped items don't count against you. If you skip a question, that dimension is scored on what you did answer.
• Stage cutoffs are a judgment call. The five stages come from thresholds we chose for clarity, not from population norms. Read them as a direction, not a verdict.
• How long, and how hopeful, are shown separately. They sit alongside your stage but don't lower your score. Feeling discouraged shouldn't make your reading worse.
• A consistency check. A few items are worded so the healthy answer is the less common one, and one question repeats in different words. If your answers are inconsistent, we show a gentle "take this lightly" note.

06 — From stage to program

Why each stage gets different scripts.

The toolkit, missions, and scripts change with your stage. Earlier stages focus on noticing bids and small fondness rituals. Middle stages add repair and softer openings. Later stages focus on de-escalation, and often point you toward a licensed clinician alongside the program.

07 — Honest limits

What this instrument is not.

This is a short reflection screen, not a validated clinical instrument.

• No published numbers yet. We haven't published reliability or validity statistics. Once we have enough responses to do that responsibly, we plan to, including anything unflattering.
• One person's view. You answer alone, at one moment, and much of what's measured is your read on your partner. Partner Perspective is a companion path for the other side of that.
• Cutoffs are a judgment call. Stage boundaries were chosen for clarity, not drawn from a normative sample.
• Some assumptions built in. Questions assume an ongoing partnership, shared time, and familiar affection norms. It fits less well for long-distance, neurodivergent, or non-monogamous relationships.
• Out of scope. This can't detect abuse, infidelity, untreated mental health conditions, or substance use. Those need a clinician.

08 — Independence

Affiliation disclosure.

Undrift Us is independently developed. It is not affiliated with, endorsed by, or certified by any therapy institute, training program, or research group.