Why Developers Need Statistics

Statistics underpins every aspect of modern software — from A/B testing and feature flags to model evaluation and capacity planning.

Descriptive Statistics

  • Central tendency — mean, median, mode
  • Spread — variance, standard deviation, IQR, range
  • Shape — skewness, kurtosis

Key Distributions

  • Normal — bell curve, 68-95-99.7 rule
  • Binomial — coin flips, binary outcomes
  • Poisson — events per time interval (requests/second)
  • Exponential — time between events (latency modeling)

Hypothesis Testing

The core workflow: define H₀, choose α (typically 0.05), collect data, compute p-value, make decision. Common tests: t-test, chi-square, Mann-Whitney U.

"p < 0.05 doesn't mean your feature works — it means your data would be unlikely if it didn't."

Common Pitfalls

  • p-hacking — running tests until significance appears
  • Simpson's Paradox — aggregated trends reversing in subgroups
  • Survivorship Bias — only analysing successes