Expected Value of Sample Statistics
You'll learn: When a sample statistic correctly targets the true population parameter on average.
What Does "Expected Value" Mean Here?
When we repeatedly draw samples from a population and calculate a statistic (like the mean or variance), we get slightly different values each time due to random sampling. The expected value is the long-run average of that statistic across all possible samples.
The key question: Does this average equal the true population parameter?
Unbiased vs Biased Estimators
An unbiased estimator is a sample statistic whose expected value equals the population parameter it's trying to estimate.
Think of it like a dartboard: if you throw many darts and they center around the bullseye (even if individual throws miss), your aim is unbiased. If they consistently land off-center, your aim has bias.
Examples You've Seen
- Sample Mean (x̄): The expected value of the sample mean equals the population mean (μ). It's unbiased.
- Sample Proportion (p̂): The expected value equals the true population proportion (p). Also unbiased.
- Sample Variance: When we divide by (n-1) instead of n, the expected value equals the population variance (σ²). This is why we use that adjustment — it makes it unbiased.
Why This Matters
If a statistic is unbiased, we know it's "aiming correctly" even if individual samples vary. This doesn't mean every sample will be spot-on, but on average, we're not systematically over- or underestimating.
Biased estimators might seem close but consistently miss in one direction, leading to wrong conclusions over time.
Key Takeaway: An unbiased estimator's expected value equals the true population parameter, meaning it targets the right value on average across all possible samples.