Final adult CRIB sedentary-time analysis

The active workspace reproduces the adult NHANES 2011–2014 CRIB report: 9,308 participants and 55,722 eligible days. The non-wear screen retains runs bounded within the day by known Wake or Sleep states, including matching states such as Wake–Non-wear–Wake.

• Full report
• Report source
• Primary analysis specification
• Parameter sensitivity methods and reproduction

Render and check

From the project root, with the project R environment and local inputs available:

Rscript R/render_primary_report.R
Rscript tests/check_primary_day_rule.R
Rscript tests/check_hourly_summaries.R
Rscript tests/check_paired_sensitivity.R

Use renv::restore() when setting up the project R environment on another machine. The wristsed package supplies the survey metadata and hourly comparison inputs. The large minute records and original CHAP/SWaN inputs are maintained on Unity.

Active layout

Directory	Contents
R/	Final-report renderer, primary day selection, daily summaries, method comparison, sensitivity analysis, and figure generation
docs/	Analysis specifications, correction notes, and data-source documentation
reports/source/	The single final CRIB report Rmd
reports/rendered/	The rendered final CRIB HTML report
data/nhanes/	Public NHANES demographic, body-measure, and self-reported sleep source files
data/derived/	Final day-level and participant-level analytic datasets used by the report
data/figure1/	Raw reference signal used to construct Figure 1
data/upstream/	Local or Unity-only minute-level inputs; ignored because of file size
outputs/final_analysis_adult/	Final tables, sensitivity results, and required source summaries/audits
outputs/waking_time_method_comparison/	Current four-method comparison inputs and results
outputs/final_analysis/results/	Shared CRIB minute-source files retained at the upstream pipeline's existing path
figures/final_analysis_adult/	Figures used in the final report
tests/	Primary day-rule, hourly-summary, and paired-inference checks
old/	Superseded analyses, reports, presentations, diagnostics, and other workspace material

Data provenance

Public NHANES source files

The files under data/nhanes/ were obtained from the CDC/NCHS NHANES public-use releases. They are stored with descriptive local names and are not modified by the analysis.

Files	How obtained	Use in this report
nhanes-demo-2011-2012.xpt, nhanes-demo-2013-2014.xpt	Downloaded from the NHANES 2011–2012 DEMO_G and 2013–2014 DEMO_H releases	Supplies age; the wristsed participant frame supplies the examination weight, masked stratum, and masked PSU
nhanes-bmx-2011-2012.xpt, nhanes-bmx-2013-2014.xpt	Downloaded from the NHANES 2011–2012 BMX_G and 2013–2014 BMX_H releases	Supplies measured BMI used to define underweight, normal-weight, overweight, and obesity categories
self-sleep-2011-12.xpt, self-sleep-2013-14.xpt	Downloaded from the corresponding NHANES sleep-questionnaire releases	Supplies self-reported sleep duration for the supplementary waking-time method comparison; it is not used to define the primary CRIB estimate

The download URLs and SHA-256 checksums are recorded in the data-source documentation.

Derived analytic datasets

The report creates the files under data/derived/ from the verified CRIB and CHAP daily summaries. They are the main analysis datasets committed to this repository.

File	How derived	Use in this report
primary_daily_sitting.csv	Starts with complete 1,440-minute participant-days. CRIB is applied to the full minute sequence, waking sedentary minutes are calculated by summing CHAP sitting fractions over CRIB-wake minutes, and the adopted within-day non-wear rule is applied. Only eligible adult participant-days are retained.	Primary day-level dataset; 55,722 days from 9,308 adult participant-cycles. It is also the fixed day set for the CRIB parameter sensitivity analysis.
primary_analytic_participants.csv	Daily waking sedentary and waking-time values are averaged within participant. The result is joined to age, measured BMI, and the NHANES survey-design metadata, then restricted to adults with complete design information.	One record per analytic participant-cycle; used for the overall, cycle-specific, and subgroup survey estimates.

The report rewrites these two files when it is rendered, so their contents must agree with the current eligibility rule and report code. Cohort counts and an MD5 fingerprint guard the downstream sensitivity analyses against stale data.

Figure 1 reference data

data/figure1/raw80HZ_83724_data/2000-01-12.csv is a reference extract from the raw 80-Hz wrist-acceleration input maintained on Unity. It is used only to show the signal-processing workflow in Figure 1 and does not contribute to any population estimate.

Large upstream data kept outside GitHub

The minute-level CHAP and CRIB files are maintained on Unity because they are too large for the repository. The local merged input, when present, is stored as data/upstream/minute_level_classification.parquet and is ignored by Git. R/sitting_time_est_batch.R summarizes the minute records minute records by participant-day and clock hour, and R/sitting_time_est_merge.R combines the batch summaries. The final report reads the merged daily_sitting.csv and hourly_sitting_daily.csv files from the ignored outputs/final_analysis_adult/results/ directory. Participant survey metadata and supporting hourly comparison inputs are obtained from the wristsed R package.

Some required non-wear inputs retain historical audit directory names. They are used by the primary day-selection code and remain active. The local shared minute parquet predates the current Unity source; use Unity's repaired merged minute records for upstream recalculation. Local report rendering uses the verified daily/hourly summaries and reference-minute extract instead.

Archive

Archived files retain their original relative paths beneath old/. The move manifest records original and archived paths, file sizes, and SHA-256 checksums. Files were moved without changing their contents; the archive is not a separate runnable project.

This organization applies to the local workspace. Existing Unity paths and jobs are unchanged.

The superseded different-flanking-state analysis is preserved in old/nonwear_rule_correction_20260922/. See the correction audit for the rule change and its effect on the analytic sample.

Paper

Sedentary Behavior in the United States Measured by Accelerometer, NHANES 2011–2014. Christopher Odoom, John Staudenmayer, and additional coauthors to be added. Department of Mathematics and Statistics, University of Massachusetts, Amherst. Funded by NIH 5R01HL1685355-02. Manuscript in preparation.