4 Florida: Learning About an Unknown Proportion from Survey Data

In this example, beliefs about an unknown proportion are updated from new survey data. The particular example is using survey update beliefs about support for Bush in Florida in the 2000 presidential election campaign (Jackman 2004).2

  data {
  real y;
  real y_sd;
  real mu_mean;
  real mu_sd;
}
parameters {
  real mu;
}
model {
  mu ~ normal(mu_mean, mu_sd);
  y ~ normal(mu, y_sd);
}

The prior polls had a mean of 49.1% in support for Bush, with a standard deviation of 2.2%. The new poll shows 55% support for Bush, with a standard deviation of 2.2%.

After observing the new poll, the mean for the posterior is 52, with a 95% credible interval of 48.9–55.1.

References

Jackman, Simon. 2004. “Bayesian Analysis for Political Research.” Annual Review of Political Science 7 (1). Annual Reviews: 483–505. https://doi.org/10.1146/annurev.polisci.7.012003.104706.


  1. This example is derived from Simon Jackman, “Florida,” BUGS Examples, 2007-07-24, URL.