Lab 13 / Uncertainty

Resample one sample

What can resampling one sample reveal about a median?

5 min synthetic data fixed seed

Draw one synthetic sample, then resample it with replacement many times to see how much the median can move. Change one control at a time, then open the values under the chart before interpreting the picture.

Interactive synthetic labruns in this browser

View the values behind this chart
Current controls, seed, and table

Loading the deterministic simulation...

SEED

01 / Question

What this display answers

Each histogram bar counts a median from a new bootstrap resample of the single observed sample. No new data from the source population is drawn.

The 2.5th and 97.5th percentiles of those resampled medians form a percentile interval for the median. It describes resampling uncertainty, not a guarantee about the true population value.

02 / Reading

What to notice

  1. Keep the observed sample small first. The resample spread stays visible.
  2. Increase the observed sample size. The bootstrap medians narrow even though the source population never changes.
  3. Use New draw. A new observed sample shifts the whole bootstrap distribution slightly.

03 / Mechanism

The model behind it

One synthetic sample is drawn once from the selected source model. Every bootstrap resample redraws values from that observed sample with replacement, keeping the observed sample size fixed.

The seed controls the initial sample and every resample, so the same controls and seed reproduce the same histogram exactly.

04 / Limits

Where the result stops

The percentile interval is a resampling summary of this one observed sample; it is not a formal confidence procedure with guaranteed coverage.

Small or heavily censored samples can make bootstrap intervals misleading. The experiment uses complete synthetic observations only.