Plank Hold: NNYFS Microdata Derivation
These percentiles are a FitnessNorms derivation from public CDC National Youth Fitness Survey microdata. McDonough et al. (2026) independently analyzed the same survey and provide the marginal values used to validate the filtering and quantile implementation.
Files, variables, and eligibility
- Measurement file:
Y_PLX.xpt; plank time isMPXPLANKand component status isPLSTS. - Demographic file:
Y_DEMO.xpt; examination age isRIDEXAGY, screening age isRIDAGEYR, sex isRIAGENDR, and the full-sample MEC examination weight isWTMEC. - Eligibility: aged 6 or older at the examination, male or female, complete plank status, and a recorded hold time within the CDC-published 1-450 second range.
- Age bands: 6-11 and 12-15, matching the age groups printed by McDonough et al. FitnessNorms crosses those bands with sex rather than inventing narrower cells.
Examination age, not screening age
NNYFS carries two age variables: RIDAGEYR, the age at screening, and RIDEXAGY, the age on the day of the examination. The distinction decides the sample. Filtering on examination age at 6 and above returns 1,209 plank scores with a weighted mean of 69.6 seconds, matching the count and mean McDonough et al. report, and reproducing all 25 of their published cut-points to the printed decimal. Filtering on screening age between 6 and 15 instead returns 1,201 scores and leaves residuals of up to 0.6 seconds. The eight-record gap is six participants who were 15 at screening and had turned 16 by their examination, plus two who crossed the lower boundary. No upper bound is applied because the survey itself stops there. The analytic distribution contains 352 boys and 373 girls aged 6-11, plus 248 boys and 236 girls aged 12-15.
Weights and percentiles
For each sex and age cell, the script orders scores and accumulates their MEC weights into an estimated population CDF. The requested quantile is linearly interpolated between adjacent distinct hold times using the definition documented for SAS PROC SURVEYMEANS. P5, P25, P50, P75, and P95 are rounded to 0.1 second. All four cells have five distinct, increasing cut-points.
Published-data validation
Table 2 of McDonough et al. reports P10, P25, P50, P75, and P90 for five margins: overall, male, female, ages 6-11, and ages 12-15. The derivation recomputes those 25 values before writing the crossed table and requires every one to match the published figure exactly; a single deviation aborts the run. It also checks the weighted mean against the paper's Table 1, and pins the analytic n, the CDC file SHA-256 hashes, cell sizes, and strict percentile ordering. An earlier tolerance of 1.0 second passed on the wrong age variable, which is why the gate is now an exact match.
Interpretation and uncertainty
The examination weight supports nationally representative point estimates for US youth in 2012. FitnessNorms does not calculate design-based standard errors or confidence intervals, and the broad age bands still combine different stages of growth and maturation. A percentile describes position in this historical distribution; it is not a health threshold.
Protocol differences can move a score. NNYFS used a forearm plank, one practice attempt, one form correction, and no stated time cap. High planks, side planks, weighted holds, and capped school tests should be treated as different measurements. The complete derivation is retained in scripts/derive_nnyfs_plank_percentiles.py, and its accepted artifact is scripts/nnyfs_plank_percentiles.json.