https://datatweets.com/courses/ab-testing/the-logic-of-experiments/Recent content in The Logic of Experiments on DATATWEETSHugoenCopyright (c) 2025 DatatweetsSun, 28 Jun 2026 09:00:00 +0200https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-1-why-we-run-experiments/Fri, 13 Feb 2026 09:00:00 +0200https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-1-why-we-run-experiments/The reason we run experiments is to claim cause and effect. This lesson shows, with a simulation you can run, how a hidden confounder makes a self-selected comparison overstate an effect four-fold — and how random assignment removes it and recovers the truth.https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-2-control-and-treatment-groups/Fri, 13 Feb 2026 09:00:00 +0200https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-2-control-and-treatment-groups/The control group stands in for the counterfactual — what would have happened without the change — while the treatment group carries exactly one change. This lesson shows why you change one thing at a time, why you randomize by user, and how a seeded Python split assigns thousands of users to A and B.https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-3-why-randomization-works/Fri, 13 Feb 2026 09:00:00 +0200https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-3-why-randomization-works/Randomization’s power is that it balances every trait across groups simultaneously, not just the ones you recorded. This lesson runs a balance check on a randomly split sample, shows the tiny differences that prove it worked, and explains why unmeasured confounders come along for free.https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-4-anatomy-of-an-ab-test/Fri, 13 Feb 2026 09:00:00 +0200https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-4-anatomy-of-an-ab-test/An A/B test is more than two groups: it’s a hypothesis, a randomization unit, a primary metric, guardrail metrics, and a decision rule you commit to in advance. This lesson names each part with Lumen’s signup-page test, and sets up the exact experiment you’ll analyze in the guided project next.https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-5-guided-project-your-first-experiment-on-lumen/Fri, 13 Feb 2026 09:00:00 +0200https://datatweets.com/courses/ab-testing/the-logic-of-experiments/lesson-5-guided-project-your-first-experiment-on-lumen/The Module 1 capstone: run Lumen’s signup experiment from a random split to a measured result. You’ll generate seeded data, compute observed conversion rates, calculate absolute and relative lift, and read the outcome descriptively — leaving the ‘is it real?’ question for Module 4.