Election Poll Aggregator — Bayesian Averaging Live (2D)

Prior belief

Prior mean vote share: 50.0% Prior std. dev: 10.0 pts

Add a poll

Sample size (n): 800 Observed share: 52.0%

Random poll generator

Simulated true share: 52.0%

Posterior

Polls combined0

Posterior mean50.0%

Posterior std. dev10.00 pts

95% credible interval—

Smallest single poll var.—

Posterior narrower than any poll?—

Each poll's sampling variance is p̂(1−p̂)/n, converted to precision (1/variance). The posterior combines the prior's precision with every poll's precision — larger, more decisive polls (n large, or p̂ far from 50%) carry more precision and move/narrow the posterior more. This 2D view stacks each step's posterior density as a row in a ridge (joyplot): row height = poll order (oldest at back, newest at front), curve shape = the density itself.

X = vote share % · rows stack newest (bottom, brightest) to oldest (top, dim)

Posterior mean (solid) and 95% credible interval (shaded) after each poll