← Bank of America Interview Insights
This looked like a warmup and turned into a 15-minute deep dive.
Start by clearly defining the mean and variance of a random variable, then transition to the sample estimators. Explain unbiasedness of the sample mean, then show the algebra for why the sample variance with n is biased and how (n-1) corrects it, emphasizing the loss of a degree of freedom due to estimating the mean.
Pro tip: Mention that the bias arises because deviations are taken from the sample mean, which is itself estimated from the data, leading to underestimated variability. This shows deep understanding and connects to practical implications in A/B testing where unbiased variance estimates are crucial.
Define the mean (expected value) and variance of a random variable X: μ = E[X], σ² = E[(X-μ)²] = E[X²] - (E[X])².
Define the sample mean x̄ = (1/n)Σx_i and sample variance with denominator n: s²_n = (1/n)Σ(x_i - x̄)².
Compute E[x̄] = (1/n)ΣE[x_i] = μ, proving unbiasedness.
Show E[s²_n] = σ² - σ²/n = σ²(n-1)/n, hence biased downward. Use algebra: Σ(x_i - x̄)² = Σ(x_i - μ)² - n(x̄ - μ)², take expectations.
Define s² = (1/(n-1))Σ(x_i - x̄)². Since E[s²] = σ², it's unbiased. Explain that dividing by n-1 corrects for the fact that deviations are from the sample mean, which underestimates variability.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.