Bayesian estimation of Bernoulli parameter p with Beta prior

Check the final answer first, then review the worked steps.

Answer

0.64

Work

  1. $\frac{\alpha}{\alpha+\beta}=0.4 \implies \beta=1.5\alpha$
  2. $\frac{0.4(0.6)}{\alpha+\beta+1}=0.02 \implies \alpha+\beta+1=12 \implies \alpha+\beta=11$ (Wait, re-eval: $\alpha=4.4, \beta=6.6$)
  3. Posterior $\alpha' = 4.4+28=32.4, \beta' = 6.6+12=18.6$
  4. Mean = $\frac{32.4}{32.4+18.6} = \frac{32.4}{51} \approx 0.635 \approx 0.64$

Step-by-step solution

For Beta($\alpha, \beta$), mean $\mu = \frac{\alpha}{\alpha+\beta} = 0.4$ and variance $\sigma^2 = \frac{\mu(1-\mu)}{\alpha+\beta+1} = 0.02$. Solving gives $\alpha+\beta = 7.66...$ (approx 7.67) and $\alpha=3.4, \beta=5.6$. Posterior is Beta($\alpha+28, \beta+12$) = Beta(31.4, 17.6). Mean = $\frac{31.4}{31.4+17.6} = \frac{31.4}{49} \approx 0.64$.