Notes

Course-voice weekly lesson pages for the 15-week Intro to Statistics spine. Each weekly page is the main reading for that week — a short textbook-depth read that teaches the concepts end-to-end, with outbound links to OpenIntro IMS and ISLBS for an alternate voice.

Every page follows the same shape: where the week starts and why it matters beyond the exam, what you will be able to do by the end of it, a short table of the words worth owning, the concepts themselves, two fully worked examples, the misreading to avoid, and a set of practice questions for your own checking. Worked examples use small invented data sets chosen to make the reasoning visible; where that is the case, the page says so.

Available

Week 1 — Data, evidence, and statistics — cases, variables, and what one row of a table stands for; the four variable types and why the type decides every later move; what a mean keeps and what it discards; and separating a claim from the data behind it and the inference that joins the two.
Week 2 — Study design, bias, and causality — what happened before the data existed: populations and samples, sampling bias as a direction rather than a shortage, observational studies against experiments, what random assignment buys that random sampling does not, and the three explanations that fit any single association.
Week 3 — One-variable summaries — describing one variable through center, spread, and shape; naming skew by the tail rather than the peak; the median and IQR as the resistant pair; the boxplot read part by part; and what bin width does to the story a histogram tells.
Week 4 — Comparing groups — comparing two groups honestly: reading a group summary table, side-by-side boxplots, and the gap between averages set against the spread among individuals — so that an average difference is never mistaken for a personal one.
Week 5 — Association — scatterplots read as direction, form, and strength, in that order; correlation as one number for the straight-line part only; why a near-zero correlation can still hide a strong pattern; and how a single unusual point can move the number.
Week 6 — Confounding and multivariable thinking — the three conditions a third variable must meet before it can distort a comparison; comparing like with like by looking within strata; a worked reversal where the ranking flips; and what adjustment can and cannot repair.
Week 7 — First-half synthesis — the first six weeks assembled into one six-pass routine for reading any study, and a ladder matching each design to the strongest verb it can carry; worked end to end on one study, then on a claim that looks like the same study and is not.
Week 8 — Simple regression — fitting one line and reading it in the units of the study: slope and intercept, residuals and the residual plot, least squares as the rule for choosing a line, R² as descriptive strength, and the extrapolation caveat.
Week 9 — Multiple and logistic regression, by interpretation — adjustment written as a model: the holding the others fixed reading, why an adjusted coefficient moves and which way, and an interpretation-only look at logistic regression, odds, and the odds ratio for a yes-or-no outcome.
Week 10 — Probability as risk and diagnosis — probability as a count of people rather than a formula; conditional probability as among this subgroup; why P(A | B) is not P(B | A); sensitivity and specificity against predictive value; and why a positive test for a rare condition is often still wrong.
Week 11 — Simulation-based inference — could chance alone have produced this? Shuffling labels to build a null distribution and read a simulated p-value; then resampling with replacement to bootstrap an interval and say how precise an estimate is — and keeping the two procedures apart.
Week 12 — Classical hypothesis testing — the formula as a shortcut to the Week 11 simulation: null and alternative claims, the test statistic as (estimate − null) / SE, the p-value read honestly, the confidence interval, and what Type I and Type II errors, significance level, and power actually mean.
Week 13 — Categorical outcomes — reading a two-way table without swapping the denominator; the risk difference, relative risk, and odds ratio as three summaries of one gap; when an odds ratio exaggerates; and expected counts and the chi-square idea as a measure of discrepancy.
Week 14 — Meta-analysis and forest plots — how single studies become a body of evidence: reading a forest plot line by line, weighting by precision, heterogeneity as cross-study disagreement, and what the published record leaves out. Figures are course-built schematics using illustrative values, not results from real studies.
Week 15 — Final review — what claim can we responsibly make? The whole course as one argument, a chart for choosing the right comparison, the term’s recurring misreadings gathered in one place, and a checklist for writing a conclusion someone can act on. A review aid, not an exam preview.