TimelyScholarFree Tools › GOSH Plot Generator
Fit the meta-analysis to every combination of studies (or a large random sample of them) and plot each subset's pooled effect against its I², to see which studies drive heterogeneity.
FreeNo account neededYour data stays in your browser
20,000 randomly sampled subsets of the 32,767 possible are shown. Separate clusters of points, or a study whose inclusion (in red) shifts the estimate or I², point to influential studies or distinct subgroups. Fixed-effect results match metafor's gosh().
Copy the columns from Excel or Google Sheets, with names in the first row, or load the example.
Estimates, tests and the figure update instantly in your browser.
Download the figure (SVG or PNG) and copy a results sentence.
Each point is the result of one subset of studies. When the studies are consistent, the points form a single, roughly symmetrical cloud. Separate clusters suggest that one or a few studies shift the pooled effect or heterogeneity.
Highlighting subsets that contain a particular study shows whether that study is responsible for a cluster. Use the plot to explore, then report sensitivity analyses with and without the influential studies.
The number of subsets doubles with each added study, so larger meta-analyses are explored with a random sample of subsets.