A four-state publication-bias funnel plot in plain SVG — effect size against inverted standard error, shaded 95%/99% pseudo confidence contours around the pooled effect, flagged outliers and Egger's test.
Preview in your theme
Loading preview…
This component is part of Pro
The install command, prompt and concept map are included with a Pro license — one purchase unlocks every Pro component, forever.
status="ready" · 16 trials pooled at g = 0.28. Thirteen scatter inside the funnel; three small trials sit outside the 95% contour — all on the right, with the bottom-left corner bare — and Lindqvist 2018 is beyond 99%. Hover or walk the studies with the arrow keys; up/down jump between the flagged ones.
Caffeine and repeated-sprint power
16 randomized trials · standardized mean difference vs its standard error
Hedges' g · standard error ↓pooled 0.2816 studies3 studies outside 95%
Each mark is one study; the y axis is its standard error, inverted so the big, precise studies sit at the top and the small, noisy ones spread along the bottom. The wedge is where 95% and 99% of studies would land if sampling error were the only thing at work — a bottom corner that is populated on one side and bare on the other is the classic small-study signature. 3 studies fall outside the 95% funnel, all on the right — the one-sided pattern selective publication produces. Egger's test across 16 studies: intercept 1.1, t 2.1 — past the conventional |t| of 2, so the asymmetry is unlikely to be sampling error alone. Arrow keys walk the studies, up and down jump between flagged ones, Enter pins the readout and Escape clears it.
Caffeine and repeated-sprint power: every study with its effect, standard error, distance from the pooled effect of 0.28 in SE units, and which funnel region it lands in.
Study
Hedges' g
Standard error
SE from pooled
Region
Pfannerstill 2013
0.28
0.151
0.0
inside the 95% funnel
Torp 2015
0.19
0.112
−0.8
inside the 95% funnel
Green 2022
0.30
0.178
+0.1
inside the 95% funnel
Jones 2015
0.63
0.193
+1.8
inside the 95% funnel
Metz 2012
0.48
0.317
+0.6
inside the 95% funnel
McKenzie 2021
0.24
0.062
−0.6
inside the 95% funnel
Hyatt 2017
0.38
0.206
+0.5
inside the 95% funnel
Feeney 2012
0.11
0.371
−0.5
inside the 95% funnel
Christiansen 2023
0.32
0.134
+0.3
inside the 95% funnel
Wiza 2014
0.35
0.079
+0.9
inside the 95% funnel
Mueller 2012
0.32
0.073
+0.5
inside the 95% funnel
Bosco 2012
0.35
0.061
+1.1
inside the 95% funnel
Mertz 2021
0.10
0.157
−1.1
inside the 95% funnel
Marek 2016
0.95
0.32
+2.1
outside the 95% funnel, inside 99%
Osei 2021
1.10
0.38
+2.2
outside the 95% funnel, inside 99%
Lindqvist 2018
1.12
0.3
+2.8
outside the 99% funnel
status="empty" · a valid contract with no studies at all — kept distinct from "loading" and from a failed fetch.
Caffeine and repeated-sprint power
16 randomized trials · standardized mean difference vs its standard error
No studies yet
Each study lands here as an effect size and its standard error; the funnel needs at least one.
status="loading" · the skeleton keeps the funnel silhouette (apex, contours, pooled line) and stops pulsing under reduced motion.
Caffeine and repeated-sprint power
16 randomized trials · standardized mean difference vs its standard error
Loading studies
status="error" · the fetch failed; "Try again" only appears when onRetry is passed, and clicking it switches to ready.
Caffeine and repeated-sprint power
16 randomized trials · standardized mean difference vs its standard error
Zyeon UI