{"id":468,"date":"2022-03-16T12:22:40","date_gmt":"2022-03-16T10:22:40","guid":{"rendered":"https:\/\/benediktehinger.de\/blog\/science\/?p=468"},"modified":"2022-03-16T12:22:40","modified_gmt":"2022-03-16T10:22:40","slug":"when-the-h0-distribution-of-tfce-is-not-uniform","status":"publish","type":"post","link":"https:\/\/benediktehinger.de\/blog\/science\/when-the-h0-distribution-of-tfce-is-not-uniform\/","title":{"rendered":"When the H0 distribution of TFCE is not uniform"},"content":{"rendered":"\n

I wrote about Threshold-Free-Cluster-Enhancement (TFCE) before<\/a>, this time I stumbled upon a weirdly looking H0 diagram. Let me explain: If you simulate data without any effect, you expect that the P(data|H0) distribution is uniform, that is, all p-values are equally likely. Here, I define the p-value as the minimal p-value over time that I get from one whole simulation (1000 permutations per simulated dataset) – I simulated only cluster in time not space (find the notebook here<\/a>, raw-jl here<\/a>). When I did this for 100 repetitions, each applying permutation TFCE and calculating the min-p, I got the following histogram of 100 p-values:
<\/p>\n\n\n\n

\"\"<\/a>
TFCE H0 distribution with integration step of 0.4<\/figcaption><\/figure>\n\n\n\n

That does not look uniform at all! What is going on? It turns out, that you can get this kind of “clustering” if your integration step-size is too large. Indeed, I change the integration step from 0.4 to 0.1<\/p>\n\n\n\n

\"\"<\/a>
TFCE H0 distribution with integration step 0.1<\/figcaption><\/figure>\n\n\n\n

Now it looks much more uniform; I should probably use more repetitions (indeed in full simulations I use 10x as many) – but this already took 500s and I am not prepared to wait longer \ud83d\ude09<\/p>\n\n\n\n

Thanks @Olivier Renauld for this explanation!<\/p>\n","protected":false},"excerpt":{"rendered":"

I wrote about Threshold-Free-Cluster-Enhancement (TFCE) before, this time I stumbled upon a weirdly looking H0 diagram. Let me explain: If you simulate data without any effect, you expect that the P(data|H0) distribution is uniform, that is, all p-values are equally likely. Here, I define the p-value as the minimal p-value over time that I get from one whole simulation (1000 permutations per simulated dataset) – I simulated only cluster in time not space (find the notebook here, raw-jl here). When I did this for 100 repetitions, each applying permutation TFCE and calculating the min-p, I got the following histogram of…<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5],"tags":[],"class_list":["post-468","post","type-post","status-publish","format-standard","hentry","category-blog"],"_links":{"self":[{"href":"https:\/\/benediktehinger.de\/blog\/science\/wp-json\/wp\/v2\/posts\/468","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/benediktehinger.de\/blog\/science\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/benediktehinger.de\/blog\/science\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/benediktehinger.de\/blog\/science\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/benediktehinger.de\/blog\/science\/wp-json\/wp\/v2\/comments?post=468"}],"version-history":[{"count":0,"href":"https:\/\/benediktehinger.de\/blog\/science\/wp-json\/wp\/v2\/posts\/468\/revisions"}],"wp:attachment":[{"href":"https:\/\/benediktehinger.de\/blog\/science\/wp-json\/wp\/v2\/media?parent=468"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/benediktehinger.de\/blog\/science\/wp-json\/wp\/v2\/categories?post=468"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/benediktehinger.de\/blog\/science\/wp-json\/wp\/v2\/tags?post=468"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}