Who contributes at each visit?
Subject by scheduled visit, including explicit missing rows. Start with aggregate completeness, then locate individual gaps in the observation matrix. Missing assessments are not automatically withdrawals.
Clinical Figure Library / By Jaime Yan
From a clinical question
to an inspectable analysis.
Five executed R/Python teaching cases connect distributions, missingness and uncertainty. All inputs are fictional. Choose a path below, then inspect results, assumptions and evidence together.
Choose by question and data structure
What does the mean conceal?
Subject by baseline/follow-up outcome. Inspect the complete-pair change distribution. For multiple domains, compare descriptive profiles with within-arm changes and intervals.
What is the between-arm contrast?
Subject, treatment, baseline and follow-up per domain. Move from within-arm summaries to baseline-adjusted contrasts. Complete-case adjustment does not resolve missing-data bias.
Review a case in four steps
- Results: read the clinical question and inspect the full population before switching strata or language.
- Methods: check the target, population, units, missingness, interval and interpretation limits.
- Values: reconcile counts and estimates with the current-selection CSV. Full-strata downloads contain overlapping populations.
- Code & QC: download the case bundle, run the minimal example, and inspect numerical checks and source hashes.
Population switching uses separately computed results. Downloads follow the current language and population unless labeled otherwise. Each case links its own methods and QC; numerical agreement is not external clinical validation.
Explore the five analysis cases
Actual executed previews, using Python and all fictional participants. Open a case to switch to R or a population stratum.
Beyond the mean
Distribution of symptom change
Is the observed improvement broadly distributed, or concentrated in a few participants?
Descriptive analysis · Independent R + Python · Synthetic data
Review results, methods & evidence →Denominators over time
Scheduled measurement completeness
How much of the scheduled population contributes at each visit?
Descriptive analysis · Independent R + Python · Synthetic data
Review results, methods & evidence →A quantitative companion to radar
Symptom changes with uncertainty
How large are the observed changes in each domain, and how precisely are their means estimated?
Descriptive analysis · Independent R + Python · Synthetic data
Review results, methods & evidence →Locate missing observations
Subject-by-visit observation matrix
Are missing assessments isolated gaps or trailing runs within individual participants?
Descriptive analysis · Independent R + Python · Synthetic data
Review results, methods & evidence →Model-based comparison
Baseline-adjusted domain contrasts
What is the fitted between-arm Week-12 difference after adjusting for baseline within complete cases?
Model-based complete-case contrast · Independent R + Python · Synthetic data
Review results, methods & evidence →Research, reuse & attribution
Quality framework & clinical coverage · Research directory · Comparisons and known disadvantages
Upstream rights · Cite Jaime Yan and upstream work · License
Original work: personal noncommercial use with required attribution. Upstream software retains its own licenses. Formal screen-reader and different-engine evaluation remain incomplete. No regulatory-validation claim.