Statistics for Data Science Interviews — From Descriptive Stats to Inference

Statistics is the foundation that makes everything else in data science credible. You don't need to be a statistician — you need to understand key concepts well enough to apply them correctly and explain them clearly to a non-technical stakeholder.

This guide covers every statistical concept that comes up in product DS technical screens:

✅ Descriptive statistics — mean vs median, variance, IQR, percentiles, when to use each

✅ Probability fundamentals — conditional probability, Bayes intuition, independence, base-rate fallacy

✅ Distributions — Bernoulli, binomial, normal, Poisson, and when to use each in a product context

✅ Sampling, CLT, and standard error — including the critical distinction most candidates get wrong

✅ p-values, alpha, confidence intervals, Type I/II errors, power — with correct and incorrect definitions

✅ One-tailed vs two-tailed tests — when to use each and why

✅ The p = 0.051 edge case — the model answer that doesn't p-hack

✅ Statistical vs practical significance — with the closing sentence pattern for every answer

✅ Multiple testing and Bonferroni correction

✅ Correlation vs causation, confounding, selection bias, survivorship bias, Simpson's Paradox

✅ Performance metrics vs statistical tests — the distinction senior DS candidates need

✅ Regression basics — linear, logistic, multicollinearity, key assumptions

17 pages. Concept-first, interview-ready throughout.