Habit Dynamics
Independent research · In progress
A SINGLE-PARTICIPANT STUDY
Every day leaves a pattern.
Exploring how daily habits, internal states and sleep relate over time—and what the models still cannot explain.
Preliminary analyses. One person’s daily records. No causal claims.
A FRAMEWORK FOR QUESTIONS
Time & history
Study framework · Schematic, not a results chart.
Connections represent research questions, not causal results.
01 / THE STUDY
Small moments. A longer view.
A single person’s repeated daily records can reveal questions that a one-time snapshot misses. This study connects everyday activities and self-reported states with wearable sleep records to examine how they relate across dates.
The aim is to describe patterns carefully, show uncertainty and test what remains unexplained. It is an observational study, so a relationship in the records does not establish cause and effect.
Daily records
Activities, timing and context from Lognote.
State ratings
Anchored self-reports collected through Google Forms.
Wearable sleep
Sleep records from WHOOP, aligned to the study timeline.
1
participant
4
preliminary analysis packages
9
research groupings
The nine groupings organize the planned portfolio; they are not nine completed analyses.
From daily records to better questions.
01
Collect
Daily records + wearable sleep
02
Align
Dates, ratings + missingness
03
Ask
Nine research groupings
04
Estimate
Models + uncertainty
05
Check
Diagnostics + future validation
Association asks what was related in these records. Forecasting asks what predicts later data—and still needs its own validation.
02 / THE RESEARCH MAP
Nine ways to ask about a day.
A plain-language map of the planned model portfolio. Each area contains specific questions; only some have preliminary results.
01
Daily routines & sessions
How are activities spread across a day: one long session or several shorter ones? These models ask whether session count, duration and timing accompany different outcomes.
Planned
02
Habits & context → states
How do recorded behaviors and circumstances relate to daily state ratings? F-ES1 has preliminary school-context models, while the broader habit questions remain open.
Partially analyzed · F-ES1
03
States → behaviors
Do earlier states accompany a later recorded behavior? R-ES1 has preliminary results for one outcome; links between other behaviors remain a separate planned branch.
Partially analyzed · R-ES1
04
The routine map
Which activities occur together, in what amounts and in what recorded order? This descriptive atlas provides context for the fitted models without claiming causal clusters.
Planned
05
Nutrition & recovery
How do recorded nutrition and recovery activities relate to states such as energy and focus? These planned questions include calories, protein and meditation.
Planned
06
Time, history & context
Does a relationship depend on timing, recent history or the work environment? These branches explore lagged, nonlinear and interaction questions, with selected sensitivity checks available.
Selected checks available
07
State-to-state relationships
Does one state relate to another on the next date after accounting for prior state history? CURRENT_CS also examines whether a state carries over, with preliminary estimates and diagnostic follow-up.
Preliminary results · CURRENT_CS
08
Joint states & networks
What happens when states occur together, such as achievement alongside low psychological load? These models examine combinations and residual co-variation, where an edge does not establish causation.
Planned
09
Wearables & forecasting
How do sleep and physiology relate to daily functioning? WP04 supplies one exploratory sleep baseline; forecasting is a separate future step that needs validation on later data.
One baseline available · WP04
03 / PRELIMINARY RESULTS
What we can say so far.
Four analysis packages have preliminary results. Their scope is narrower than the full research plan, and the direction of the primary comparisons remains uncertain.
F-ES1
Preliminary
School context & daily state
QUESTION
How does prior-date school context relate to the next day’s state?
CURRENT RESULT
Ten ordinal models have preliminary estimates. Selected initiation and persistence comparisons remain uncertain; their intervals include zero.
NEXT STEP
Complete the broader habit questions and assess model adequacy.
R-ES1
Preliminary
States before a recorded behavior
QUESTION
Do prior ratings relate to later engagement in one recorded behavior?
CURRENT RESULT
One outcome has been modeled. The reported contrasts do not establish a clear direction or a general state-to-behavior effect.
NEXT STEP
Assess remaining outcomes and sensitivity checks.
CURRENT_CS
Preliminary
How daily states relate
QUESTION
Do state-to-state relationships extend beyond prior state history?
CURRENT RESULT
Preliminary estimates and follow-up checks are available. Primary probability contrasts remain uncertain, and residual patterns raise questions about model adequacy.
NEXT STEP
Investigate unexplained time structure and check calibration.
WP04
Preliminary
Sleep duration & getting started
QUESTION
Does recorded sleep duration relate to ease of task initiation?
CURRENT RESULT
One exploratory baseline produced nearly flat median curves with broad uncertainty. This does not establish no association or an optimal sleep duration.
NEXT STEP
Check calibration, date alignment and additional planned models.
An interval spanning zero means the direction remains unresolved; it does not prove there is no relationship. Many comparisons from the same records are not independent confirmations.
AN OPEN QUESTION
What might the models be missing?
Follow-up checks found structure in residuals across multiple day gaps. That is a reason to investigate how the models represent time—not proof of behavioral cycles.
These checks summarize uncentered residual products selected after fitting. They are not correlations, independent discoveries or multiplicity-adjusted tests.
Useful questions. Clear limits.
One participant
The results describe this record and may not generalize to others.
Observational data
Self-report, missingness and shared context limit interpretation; causation is not established.
Models in progress
Estimates depend on assumptions. Passing sampling checks does not establish scientific adequacy, and forecasting has not been validated.
04 / RESEARCH LIBRARY
Start here. Go deeper when you need to.
GOOGLE DOC · AVAILABLE
The technical overview; interpretation and study counts remain under review.
Short methods guide
In preparation
Preliminary family reports
Links to be added
Full comparison sets
Curated release pending