Examples
Worked examples
- Is an instance
A diagram in a methods paper showing 32 plausible analytic pipelines branching from one dataset.
- Is an instance
An author's reflection on how many choices were latently available before any analysis ran.
Counter-examples
Looks similar, but isn't
- Not an instance
A pre-registered single-path analysis.
- Not an instance
A formal multiple-comparison correction (addresses but does not name the garden).
Editorial commentary
The “garden of forking paths,” a metaphor borrowed from Borges by Andrew Gelman and Eric Loken, describes the implicit, data-dependent multiplicity built into an empirical study even when the analyst never runs more than one analysis and is not engaged in deliberate p-hacking. Before any data are examined, an analyst faces many small decisions — how to code an ambiguous variable, which covariates to include, how to define the outcome, which observations count as outliers — and in practice many of these decisions are made, consciously or not, with half an eye on the emerging pattern in the data. The problem is that conventional inferential statistics (p-values, confidence intervals) assume a single, fixed analytical path decided independently of the data. When the actual path taken was one of many that could plausibly have been taken depending on what the data looked like, those statistics are miscalibrated, even though no individual step in the process looks improper.
The concept’s practical importance is that it motivates the entire family of transparency responses that have grown up around it: pre-registration and pre-analysis plans close off the garden in advance by fixing the path before data are seen; specification curve analysis and multiverse analysis map the garden after the fact by reporting outcomes across the plausible paths rather than just the one taken; a robustness check is a smaller, more selective probe of a few of those paths.
A note on terminology: “forking paths” and “garden of forking paths”
Researchers and journals use “forking paths” and “garden of forking paths” interchangeably to refer to this same Gelman & Loken concept — there is no meaningful distinction between the two phrasings in the methodological literature. This page uses the fuller, more precisely attributable form, matching the original paper's title.
References
- Gelman, Loken, 'The garden of forking paths' (Columbia University working paper, 2013/2014).
Also known as
Gelman-Loken garden of forking paths
Machine-readable encodings
Use in your systems
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="Garden of forking paths"
vocab-term-identifier="https://casrai.org/dictionary/term/garden-of-forking-paths" />{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://casrai.org/dictionary/term/garden-of-forking-paths",
"name": "Garden of forking paths",
"identifier": "https://casrai.org/dictionary/term/garden-of-forking-paths",
"description": "The Gelman-Loken metaphor for the implicit, data-dependent multiplicity of analytical choices made in the course of an empirical study, even by analysts not engaged in explicit p-hacking.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/reproducibility#set",
"url": "https://casrai.org/dictionary/term/garden-of-forking-paths",
"sameAs": [
"Gelman-Loken garden of forking paths"
],
"license": "https://creativecommons.org/licenses/by/4.0/",
"publisher": {
"@id": "https://casrai.org/#organization"
},
"author": {
"@id": "https://casrai.org/#editorial-team"
},
"datePublished": "2026-05-21T02:22:50",
"dateModified": "2026-08-22T15:41:30",
"inLanguage": "en-GB",
"isAccessibleForFree": true
}
