to-do list for all

[itchyshin]

Things to be finished before submission

Stage 1

Figures

• [x] Figure 1 (SE, V, 1/SE, 1/V) = @mlagisz
• [x] Figure 2 (Survey) = @roseodea
• [x] Figure 3 (Funnel) = @itchyshin
• [x] Figure 4 (Egger) = @itchyshin
• [x] Figure 5 (P + Selection) = @itchyshin
• [x] Figure 6 (Mulitlevel) = @itchyshin
• [x] Figure 7 (Summary + Recommendation) = @roseodea + @itchyshin
• [x] Checking related code = @roseodea + @daniel1noble

Appnedices

• [x] Appnedix S1 (Survey) = @roseodea
• [x] Appnedix S2 (Egger to meta-regression) = @itchyshin
• [x] Appnedix S3 (Fail-safe N) = @itchyshin
• [x] Appnedix S4 (main worked examples) = @ASanchez-Tojar + @Yefeng0920
• [x] Appnedix S5 (averaging and sampling) = @ASanchez-Tojar + @Yefeng0920
• [x] Checking related code = @roseodea + @daniel1noble

Stage 2

• [x] getting approvals from coauthors group 1 (people on this repo)
• [x] emailing coauthor group 2 (Tim, Julia, Micahel)

[daniel1noble]

Sweet!

A great plan @itchyshin. I’m happy to also help @ASanchez-Tojarhttps://github.com/ASanchez-Tojar + @Yefeng0920https://github.com/Yefeng0920 on appendices. Did you want this to be markdown?

Dan

[Yefeng0920]

@daniel1noble Markdown is great! I really like it. Thanks

[Yefeng0920]

@ASanchez-Tojar There may exist big differences between us: how we show the worked example, (2) the techniques of scripts. I think it is fine for our first version. I tried an example for Appendix S4:

• first I estimate the overall mean;
• next, use the conventional Egger's test (SE as the moderator) to detect publication bias;
• when I find publication bias using Egger's test I turn to use Shinichi's method (eq. 27 but with moderator) to detect the publication bias;
• then I find the publication bias disappears (slope is not statistically significant) - so the publication bias is something "pseudo";
• last, I use eq. 28 to obtain ‘potentially’ bias-adjusted overall estimate and compare it with the native estimate (i.e., original overall mean) to quantify the degree of bias

I have not done Appendix S5 - averaging and sampling. Because my R currently can not update to the latest metafor_version. I do not know why. I think we will have several questions to discuss after we have a look at each's pilot Questions like should we do parallelly for time lag bias test? Or we just conflate the variable year with other moderators (i.e., eq. 29). After we achieve consensus I think we can consult Shinichi to meet his requirement and let Dan check scripts