← All articles
METHODOLOGY

Why "daily, zero-missing, 360-day" is rare: the value of longitudinal completeness

2026.06.25Methodology

Self-tracking data on health and behavior is everywhere. Yet most of it breaks off within days or weeks. The core value of our dataset (see the DATA LAB page) is not that "abstinence worked," but the completeness itself: two subjects measured 360+ days daily, with near-zero missing entries. This article explains why that completeness is statistically and scientifically rare and valuable.

01Dropout kills longitudinal studies

The greatest enemy of longitudinal research is dropout. When participants stop recording, the surviving data carries survivorship bias — "those who kept going = those who were already doing well" — inflating apparent effects. Much app data is distrusted precisely because dropout is uncontrolled. A near-zero-missing record structurally avoids this deepest bias.

typical self-tracking our data (near-zero) day0day365 dropout 365 days tracked
FIG 1Typical self-tracking drops out within weeks, creating survivorship bias (left). A near-zero-missing record avoids this structurally (right).

02Daily resolution captures the "waves"

Monthly measurement smooths away the short-term fluctuations — craving peaks, recovery, rebound — that abstinence produces. As the DATA LAB libido and EHS trajectories show, the real path is not a straight line but a non-linear wave of sharp drop, recovery, and renewed decline. This structure is observable only at daily resolution. Coarse intervals miss the most interesting phenomena.

WHY IT MATTERS
2 subjects × 360+ days × near-zero missing
To our knowledge, no other public dataset tracks abstinence daily across multiple indicators for 365 consecutive days without missing entries. The rarity comes not from scale but from the density of completion.

03Even n=2 matters if the record is near-complete

The "only two subjects" critique is fair. But as long as we make no claim of statistical generalization, a near-completely recorded pair can carry higher information density than 1000 coarsely measured people for certain questions. To observe intra-individual temporal dynamics precisely, you must first follow a single individual with near-continuous coverage. This is a design for the existence-proof and description of a phenomenon, not for population estimation.

KEY TAKEAWAY

The value of longitudinal data lies in completeness more than effect size. A daily, near-zero-missing, 360-day design avoids dropout bias and captures non-linear waves. n=2 is not a weakness but a starting point for precise description. See the actual trajectories on the DATA LAB page.

Note on scope. This article summarizes general findings from published research alongside observational patterns from our own project (n=2). It is not medical advice, and n=2 observations are illustrative, not statistically generalizable.

See the 360-day measured data

The records referenced in this article are published as actual trajectory charts on our DATA LAB page — daily records from two subjects over 360+ days with near-zero missing entries.

View DATA LAB →

© 2026 Reboot Data Lab Research Institute.

reboot-data-lab.com  |  contact@reboot-data-lab.com