Trying to prove the sample mean $\hat{p} = \bar{X}$ is an efficient estimator for the Bernoulli parameter $p$
Consider a random sample $X_1, X_2, \dots, X_n$ from a Bernoulli distribution with parameter $p$. We want to show that the estimator $\hat{p} = \bar{X}$ is efficient.
#### **1) Unbiasedness**
First, we check if the estimator is unbiased.
$\hat{p} = \bar{X} \quad \text{is unbiased } \checkmark$
$E[\hat{p}] = E[\bar{X}] = E[X_1] = p$
#### **2) Cramer-Rao Inequality & Efficiency**
To show efficiency, we check if the variance of the estimator achieves the Cramer-Rao Lower Bound (CRLB). The Cramer-Rao inequality states:
$Var(\hat{p}) \ge \frac{1}{n \cdot I(p)}$
where $I(p)$ is the Fisher information for a single observation. For a Bernoulli distribution, the Fisher information is:
$I(p) = \frac{1}{p(1 - p)}$
**Actual Variance of the Estimator:**
$Var(\hat{p}) = Var(\bar{X}) = \frac{Var(X_1)}{n} = \frac{p(1 - p)}{n}$
Note: $MSE(\hat{p}, p) = E[(\hat{p} - p)^2] = E[(\bar{X} - p)^2]$
**Cramer-Rao Lower Bound (CRLB):**
$\frac{1}{n \cdot I(p)} = \frac{1}{n \cdot \frac{1}{p(1 - p)}} = \frac{p(1 - p)}{n}$
**Comparison:**
Comparing the actual variance to the lower bound:
$\frac{p(1 - p)}{n} = \frac{p(1 - p)}{n}$
#### **Conclusion**
Since the estimator is unbiased and its variance is equal to the Cramer-Rao Lower Bound:
$\Rightarrow (1), (2) \Rightarrow \hat{p} \text{ is an efficient estimator for } p$