Skip to main content

Methods in R: A topic reference for biomedical research and peer review

1 Find your method

Open the book here when you don’t know which chapter to read. Three routes are offered: a flowchart from research question to statistical test, a quick lookup table, and a click-through wizard hosted live.

  • Interactive wizard — three to five questions, one recommended test.
  • Static decision tree (below) — the same logic, single page, suitable for printing.
  • Quick lookup table — common scenarios mapped to a recommended chapter and section.

This chapter covers the core tree of classical tests (differences, associations, interdependence). Methods outside that core — survival, Bayesian, machine learning, time series, multivariate, meta-analysis, bioinformatics, study-design choices — are reached through their topic chapters or via the lookup table below.

1.1 The static decision tree

Colour encodes the scale level of the outcome; a dashed border marks a non-parametric test. Click a node in the rendered HTML to jump to the method page.

%%{init: {"securityLevel": "loose", "flowchart": {"htmlLabels": true, "useMaxWidth": true}}}%%
flowchart TD
    START["Research question"]:::root --> Q2{"Differences or<br/>associations?"}
    START --> Q_INT{"Exploratory<br/>structure?"}

    Q2 -->|Differences| QD1{"Kind of difference?"}
    Q2 -->|Associations| QA1{"Variable count?"}

    QD1 -->|Central tendency| QD2{"Outcome scale?"}
    QD1 -->|Variances| VAR["Variance tests<br/>(chi-sq, F, Levene)"]:::metric
    QD1 -->|Proportions| QDPROP{"Single variable vs. expected?"}

    QDPROP -->|Dichotomous| BIN["Binomial test"]:::nominal
    QDPROP -->|Categorical| CHI_GOF["Chi-sq goodness-of-fit"]:::nominal

    QD2 -->|Interval/ratio| QD3{"Number of groups?"}
    QD2 -->|Ordinal| QD6{"Groups &amp; pairing?"}
    QD2 -->|Nominal| QDPROP

    QD3 -->|2 groups| QD4{"Independent/paired?"}
    QD3 -->|3+ groups| QD5{"Factors &amp; design?"}

    QD4 -->|Indep. normal| TT["Independent t-test"]:::metric
    QD4 -->|Indep. non-normal| MWU["Mann-Whitney U"]:::metric_np
    QD4 -->|Paired normal| PT["Paired t-test"]:::metric
    QD4 -->|Paired non-normal| WSR["Wilcoxon signed-rank"]:::metric_np
    QD4 -->|Paired sign only| SGN["Sign test"]:::metric_np

    QD5 -->|1 between normal| OWA["One-way ANOVA"]:::metric
    QD5 -->|2+ between normal| FA["Factorial ANOVA"]:::metric
    QD5 -->|1 within normal| RM1["One-way rmANOVA"]:::metric
    QD5 -->|Mixed design| RMM["Factorial/mixed rmANOVA"]:::metric
    QD5 -->|Indep. non-normal| KW["Kruskal-Wallis"]:::metric_np
    QD5 -->|Paired non-normal| FR["Friedman"]:::metric_np

    QD6 -->|2 indep.| MWU
    QD6 -->|2 paired| WSR
    QD6 -->|&gt;2 indep.| KW
    QD6 -->|&gt;2 paired| FR

    QA1 -->|Two variables| QA2{"Scale levels?"}
    QA1 -->|More than two| QA5{"DV scale?"}

    QA2 -->|Both nominal| CHIC["Chi-sq contingency"]:::nominal
    QA2 -->|Both ordinal| SPE["Spearman"]:::ordinal_np
    QA2 -->|Ord + interval| SPE
    QA2 -->|Interval linear| PEA["Pearson"]:::metric
    QA2 -->|Interval directed| SLR["Simple linear regression"]:::metric
    QA2 -->|Interval non-norm| KEN["Kendall tau"]:::metric_np

    QA5 -->|Interval| MLR["Multiple linear regression"]:::metric
    QA5 -->|Dichotomous| LOG["Logistic regression"]:::nominal
    QA5 -->|Ordinal| ORL["Ordinal logistic regression"]:::ordinal
    QA5 -->|Nominal k+| MNL["Multinomial logistic regression"]:::nominal

    Q_INT -->|Reduce variables| FAC["Factor analysis"]:::interdep
    Q_INT -->|Group cases| QCL{"Scale/size?"}

    QCL -->|Mixed/small| HCL["Hierarchical clustering"]:::interdep
    QCL -->|Metric/large| KMC["K-means"]:::interdep
    QCL -->|Mixed/xlarge| TSC["Two-step clustering"]:::interdep

    classDef metric fill:#2A9D8F,stroke:#1d6e65,color:#fff
    classDef metric_np fill:#2A9D8F,stroke:#1d6e65,color:#fff,stroke-dasharray:6 4
    classDef nominal fill:#F4A261,stroke:#b4743f,color:#000
    classDef ordinal fill:#E9C46A,stroke:#a88a42,color:#000
    classDef ordinal_np fill:#E9C46A,stroke:#a88a42,color:#000,stroke-dasharray:6 4
    classDef interdep fill:#6A4C93,stroke:#4a3468,color:#fff
    classDef root fill:#ffffff,stroke:#333,color:#000,font-weight:bold

1.2 Quick lookup

A small table for the most common biomedical scenarios. The full tree above and the live wizard cover everything the table omits.

Scenario Recommended method Chapter
2 independent groups, metric, normal Independent / Welch t-test 5
2 independent groups, non-normal or ordinal Mann-Whitney U 5
2 paired groups, metric Paired t-test 5
3+ groups, between, normal One-way ANOVA 5
3+ groups, between, non-normal Kruskal-Wallis 5
Two continuous, linear Pearson correlation 5
Continuous outcome, 2+ predictors Multiple linear regression 8
Binary outcome, 1+ predictors Logistic regression 8
Time-to-event with censoring Kaplan-Meier, Cox 12
Diagnostic test accuracy Sensitivity/specificity, ROC 15
Aggregating effect sizes across studies Pairwise meta-analysis 16
Predictive model with cross-validation tidymodels workflow 14
Posterior inference with priors brms / Stan 11
Repeated measures within subjects Mixed-effects model 17
Two or more variances Levene’s test 5
Reduce variables to latent dimensions PCA / factor analysis 9

The static tree is mirrored from the decision-tree wizard at https://cttir.github.io/tutorials/decision-tree/. Structure inspired by the University of Zurich Methodenberatung (methodenberatung.uzh.ch).

This book was built by the bookdown R package.