Pre-registration of a test for statistical significance of a GALFA-HI -- WMAP ILC correlation
By Joshua Peek
In 2007 Gerrit Verschuur proposed that HI structures as seen in the Leiden-Argentina-Bonn 36' HI survey of the sky are correlated with structures in the WMAP sky, which are typically ascribed to the CMB.
He has recently proposed to see WMAP correlations in the higher-resolution GALFA-HI DR1 data. As the referee of this manuscript, I,(Joshua Peek) required a test of statistical significance to publish.
I proposed a test: I have fully reduced, but publicly unreleased GALFA-HI data that Verschuur has no access to. In the test, I select 34 WMAP peaks, to have WMAP 7 year ILC brightness greater than 0.1 mK, Galactic latitude > 20 degrees, and extract ~2x2 degree regions in the unreleased area. I extract the same regions from GALFA-HI, and give both to Dr. Verschuur, without coordinates. If he can match many of these pairs, one can easily calculate the p-value -- 6 matches would be convincing. While it might be possible to leverage the dust-gas correlation using the Planck 353 maps to 'cheat', I believe this would be very difficult.
Dr. Verschuur, the editor, Dieter Hartmann, and I have agreed to this test. The test expires in Jan 1 2017, or when GALFA-HI DR2 become public, whichever comes first.
Published: 16 Mar, 2016