5-sigma signals come and go!

December 23, 2007 at 6:39 pm 3 comments

Slowly I am managing to return to the land of physics blogging, and much of what I see now is great!

One item which strikes me is a post written in Charm, etc. in which the author describes how a very nice 5-sigma pentaquark signal came and went! This may well serve as a cautionary tale for all of us at high-energy colliding beam experiments. If this peak had been the Higgs boson, the collaboration(s) might have been delighted to announce discovery. (Some would argue that the original W boson and top quark discoveries were on shakier ground, statistically speaking.)

I think it is great that the CLAS Collaboration wrote a article demonstrating the results from their first set of data, and the second, which is fives times larger than the first. This plot sums up the situation:

CLAS data sets, with and without peak

The points represent the first data sample, on the basis of which a 5-sigma significance was claimed. The solid line represents the second data sample, which is more than five times larger than the first, which contradicts that claim. Both data samples were taken with the same apparatus under the same conditions, by the same people. The collaboration wrote a paper exploring the statistical issues involved (see arXiv:0709.3154). See the clear and concise article for details, or or Charm’s original peak-finding post.


Entry filed under: Particle Physics.

Return After the budget debacle

3 Comments Add your own

  • 1. Luis Sanchez  |  January 26, 2008 at 5:49 pm

    Which is the current status of the pentaquark, can we safely say that the theta+ was an illusion? This is a quite interesting case, because I even saw an 8-sigma detection!!!

    And by the way, which is the current status of X(3872)?

    Luis Alberto

  • 2. Hur säker kan man vara? « Stjärnstoft och kugghjul  |  June 17, 2010 at 1:57 pm

    […] Man måste vara så noga, för även osannolika saker händer ibland. Det finns exempel på 5 sigma-signaler som dyker upp och försvinner igen. […]

  • 3. Cees Timmerman  |  November 28, 2011 at 2:59 am

    “5-sigma”, that’s 233 bad results per million tests, right? http://money.howstuffworks.com/six-sigma4.htm

    So for normal people, it’s 100 * (1 – 233 / 10**6) = 99.9767 or 99.98%.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Trackback this post  |  Subscribe to the comments via RSS Feed

December 2007
« Jun   Jan »

%d bloggers like this: