LAB 04 / NORMAL-NORMAL UPDATE
The estimate of a mean moves with evidence strength
How does the estimate of a mean change with sample size and measurement variability?
← Seven labs / LAB 04About weighting means
More precise information gets more weight
In a normal–normal update, the prior mean and observed mean are weighted by precision, the inverse of variance. More observations or less variable measurements increase the precision of the observed mean.
The yellow curve's width shows uncertainty about the mean being estimated, not the spread of individual measurements.
Change evidence strength and compare means
Prior distribution
Dashed: means supported by the data (relative likelihood)
Posterior distribution
This model treats the observation SD as known. The likelihood curve is normalized to unit area here solely to compare shapes.
Posterior mean-
Posterior SD-
Weight of the observations-