從第一條盒鬚圖到 Cox 模型——13 章貫穿生物統計的核心思想、現代爭議與實作範例。每章皆有 Chart.js 互動模擬、R + Python 程式碼、決策樹、比較表與自我檢測題；內容對應 Rosner / Altman / Kleinbaum 教科書與 ASA 2016/2019、CONSORT、TRIPOD 等官方規範。
From your first boxplot to Cox models — 13 chapters traversing the core ideas, current controversies, and working code of biostatistics. Each chapter includes Chart.js simulations, paired R + Python code, decision trees, comparison tables, and self-checks; cross-referenced with Rosner / Altman / Kleinbaum textbooks and ASA 2016/2019, CONSORT, TRIPOD guidelines.
三條腿（位置 / 離散 / 形狀）+ 偏態下 mean vs median + 箱型/小提琴/dot plot 的選擇 + Anscombe 四重奏的故事 + Tukey 1977 EDA 哲學。
Location / spread / shape, mean vs median under skew, boxplot vs violin vs dotplot, Anscombe's quartet, and Tukey's 1977 EDA philosophy.
Normal / Binomial / Poisson / Negative Binomial / t / F / chi-square / log-normal——每個都有對應的生物學情境（突變數 / 細胞計數 / RNA-seq counts / 血液濃度）。
Normal / Binomial / Poisson / Negative Binomial / t / F / chi-square / log-normal — each tied to a biological scenario (mutations / counts / RNA-seq / biomarker concentrations).
SE vs SD 的混淆陷阱、CLT 的真實前提（獨立 + 有限變異數）、bootstrap 重抽樣與信賴區間建構。
The classic SE vs SD confusion, CLT's real assumptions (independence + finite variance), bootstrap resampling, and CI construction.
H₀ vs H₁ 的真正意涵、p-value 的 6 大誤解（ASA 2016）、p-hacking 與 garden of forking paths（Gelman 2014）、equivalence/non-inferiority 檢定、S-value 替代直覺。
H₀ vs H₁ properly explained, the 6 misuses of p-values (ASA 2016), p-hacking and the garden of forking paths (Gelman 2014), equivalence / non-inferiority tests, and the S-value intuition.
One/two-sample/paired t 完整流程，Welch 為預設（Delacre 2017）、Wilcoxon 非常態替代、CI 的 5 大誤讀、報告 effect size + CI 的正確姿勢。
One-/two-sample/paired t in full, Welch as default (Delacre 2017), Wilcoxon for non-normal data, the five common CI misreadings, and effect-size + CI reporting done right.
獨立性 / 適合度 / McNemar、Fisher's exact 何時用、Yates 連續性修正的爭議、2×2 表的 OR / RR / RD 三種效應量。
Independence / goodness-of-fit / McNemar; when to use Fisher's exact; the Yates correction debate; OR / RR / RD as three effect-size choices on a 2×2 table.
F 統計量幾何直覺、one-way / two-way / repeated-measures、post-hoc（Tukey HSD / Bonferroni / Dunnett）、Welch ANOVA、非常態 Kruskal-Wallis。
Geometric intuition for F, one-way / two-way / repeated-measures, post-hoc (Tukey HSD / Bonferroni / Dunnett), Welch ANOVA, and Kruskal-Wallis for non-normal data.
OLS 幾何意義、四圖殘差診斷、VIF / influence / Cook's distance、interaction 與 dummy variables、Table 2 fallacy（Westreich 2013）與 confounder 校正。
OLS as geometry, the four diagnostic plots, VIF / influence / Cook's distance, interactions and dummy variables, the Table 2 fallacy (Westreich 2013), and confounder adjustment done right.
Logit link、odds ratio 解讀、Wald vs LRT vs score、EPV ≥ 10 規則（Peduzzi 1996 / Riley 2020）、ROC + AUC + calibration + decision curve。
Logit link, OR interpretation, Wald vs LRT vs score, the EPV ≥ 10 rule (Peduzzi 1996 / Riley 2020), and ROC + AUC + calibration + decision-curve analysis.
Kaplan-Meier、log-rank、Cox PH 模型、Schoenfeld 殘差檢驗（Grambsch-Therneau 1994）、competing risks（Fine-Gray）、RMST 與時間相依變數。
Kaplan-Meier, log-rank, Cox PH, Schoenfeld residuals (Grambsch-Therneau 1994), competing risks (Fine-Gray), RMST, and time-varying covariates.
α / β / effect size / n 的四要素互動關係；G*Power、pwr (R)、statsmodels (Python)；絕不算 post-hoc power（Hoenig 2001）；SESOI、cluster RCT、survival sample size。
The four-way α / β / effect-size / n relationship; G*Power, pwr (R), statsmodels (Python); never compute post-hoc power (Hoenig 2001); SESOI, cluster RCT, and survival sample size.
FWER（Bonferroni / Holm / Hochberg） vs FDR（BH 1995 / BY 2001 / Storey q-value / 2dGBH 2024）、local FDR、scRNA-seq / GWAS / proteomics 場景應用。
FWER (Bonferroni / Holm / Hochberg) vs FDR (BH 1995 / BY 2001 / Storey q / 2dGBH 2024), local FDR, and applications in scRNA-seq, GWAS, and proteomics.
Fixed vs random effects、random intercept / slope、ICC、lme4 / nlme / statsmodels、GEE、避免 pseudo-replication（Hurlbert 1984）、Bayesian 混合模型導論。
Fixed vs random effects, random intercept / slope, ICC, lme4 / nlme / statsmodels, GEE, avoiding pseudo-replication (Hurlbert 1984), and a Bayesian mixed-model primer.
~90 個 DOI 連結：經典教科書（Rosner / Altman / Kleinbaum / Hosmer-Lemeshow / Pinheiro-Bates / Vittinghoff）、ASA p-value 聲明、CONSORT / STROBE / TRIPOD / ARRIVE 規範、Welch 預設、BH-FDR、Cox PH、Hoenig 等延伸討論。
~90 DOI links: classic textbooks (Rosner / Altman / Kleinbaum / Hosmer-Lemeshow / Pinheiro-Bates / Vittinghoff), ASA p-value statements, CONSORT / STROBE / TRIPOD / ARRIVE guidelines, the Welch-default debate, BH-FDR, Cox PH, Hoenig, and more.
涵蓋 13 章主題的單選、多選、是非題（每章 15+ 題），內建錯題複習、隨機練習、考試模式與進度紀錄。雙語對照。
Single-choice / multi-select / true-false questions across all 13 chapters (~15+ per chapter), with wrong-answer review, random practice, exam mode, and progress tracking. Fully bilingual.