Statbitall

Hypothesis testing from scratch: the logic before the formula

Tue, 21 Apr 2026 00:00:00 GMT

Most courses start with p-values. That's the wrong place to start. Here's the logical framework that makes hypothesis testing coherent, before a single formula appears.

A/B testing under the hood: what the platform isn't telling you

Tue, 08 Jul 2025 00:00:00 GMT

Why peeking at results inflates your false positive rate, how multiple metrics break your significance threshold, and the pre-experiment checklist that makes experiments trustworthy.

Chi-square tests: how to make decisions from categories

Tue, 01 Jul 2025 00:00:00 GMT

When your data is counts, not measurements. The goodness-of-fit test, the test of independence, and why categorical data needs its own statistical tools.

Statistical power is why your A/B test found nothing

Tue, 24 Jun 2025 00:00:00 GMT

The false negative problem that most analysts ignore. What statistical power means, how to calculate required sample sizes before you run an experiment, and why 'no significant effect' often just means 'inconclusive test'.

ANOVA is not just multiple t-tests (and here's why)

Tue, 17 Jun 2025 00:00:00 GMT

Why running three t-tests on three groups gives you a 14% false positive rate instead of 5%. How ANOVA tests all groups simultaneously with one F-statistic, and when to use post-hoc comparisons.

p-values are not what you were taught

Tue, 10 Jun 2025 00:00:00 GMT

The most misused number in science. What a p-value actually measures, what it cannot tell you, why 0.05 is arbitrary, and how p-hacking turns null results into publications.

The bias-variance tradeoff controls every model you'll ever build

Tue, 20 May 2025 00:00:00 GMT

Why simple models miss patterns, complex models memorize noise, and the U-shaped curve that determines the sweet spot for every prediction problem.

Confidence intervals don't mean what you think they mean

Tue, 13 May 2025 00:00:00 GMT

The most misinterpreted concept in applied statistics. What a 95% confidence interval actually claims, what it doesn't, and why the distinction matters for every decision you make from data.

Variance is risk. Standard deviation is the language of risk.

Tue, 06 May 2025 00:00:00 GMT

Variance, standard deviation, standard error, and Bessel's correction. What each one measures, how they differ, and when high variance is a feature, not a bug.

Your sample is lying to you (and how to catch it)

Tue, 29 Apr 2025 00:00:00 GMT

Random sampling, sampling bias, stratified sampling, and the standard error. Why 1,000 observations can represent millions, and why 10 million observations can get it completely wrong.

Probability distributions are just rules for uncertainty

Tue, 22 Apr 2025 00:00:00 GMT

Normal, binomial, Poisson, exponential. What each one looks like, when data follows it, and what happens when you pick the wrong one.

Decision trees learn by asking the right questions

Tue, 15 Apr 2025 00:00:00 GMT

Information gain, Gini impurity, and why a greedy split strategy produces trees that are surprisingly good at finding structure in data.

The t-test: what it's really asking

Tue, 08 Apr 2025 00:00:00 GMT

Most people know to look for p < 0.05. Fewer know what the test statistic is actually measuring or why the t-distribution has fatter tails than the normal.

The central limit theorem is why statistics works

Tue, 01 Apr 2025 00:00:00 GMT

The CLT is the reason we can use normal distributions for nearly everything, even when the underlying data looks nothing like a bell curve.

Probability is not about luck. It's about measuring what you don't know.

Tue, 25 Mar 2025 00:00:00 GMT

Random variables, conditional probability, expectation, and the three axioms that make all of statistics possible. The true starting point.