Duration and restfulness, analysis revision 1, 2026-10-15.
Source: https://www.cdc.gov/nchs/nhis/documentation/2024-nhis.html
Public sample adult 2024 CSV; public codebook dated 2025-06-18; 2024 survey description.
Valid SLPHOURS_A=1..24; valid SLPREST_A=1..4. Rest codes 1/2=never/some days; 3/4=most days/every day. Duration groups <7, 7..9, 10..24. The 7..9 band is descriptive and not a sleep prescription. Refusal/not ascertained/don't know are excluded, not recoded.
Joint sample 31,463 of 32,629; excluded 1,166. Each cell denominator is the weighted valid joint sample, not its duration group.
Point p=sum(WTFA_A*I_cell)/sum(WTFA_A*I_validjoint).
Variance: e_i=WTFA_A_i*(I_cell_i-p*I_validjoint_i)/sum(WTFA_A*I_validjoint). Sum e_i within each (PSTRAT,PPSU). For each stratum h, add m_h/(m_h-1)*sum((PSUtotal-mean_h)^2). Retain all design rows and zero domain contributions outside the valid sample. There are 52 strata, 662 PSUs, 610 degrees of freedom; no singleton stratum. Standard error is square root of variance.
95% interval = inverse_logit(logit(p) +/- t(.975,610)*SE(p)/(p*(1-p))). Sampling uncertainty only, not all nonsampling errors. This documented logit interval need not duplicate every interval procedure used in a published brief.
Benchmark separate valid denominators: short sleep 30.476%=30.5% to one decimal, rested most/every day 54.835%=54.8%. Both reproduce Data Brief 559. Joint margins differ slightly because joint validity requires both answers. Minimum cell count=508; broad groupings retained, no detailed demographic cells published. No causal or diagnostic finding.
Public US government data reused with attribution, no respondent data in downloads, no agency endorsement. Next review 2027-01-15. Corrections: https://thebeddingsite.com/contact/.