If you have enough random distributions based on the same underlying probability distribution, you will find 3-σ "bumps" (or dips!) in one or more distributions--that is, a change big enough to make you think "There's something there!" So to understand the true significance of a "bump" in the (eg) mass distribution, you have to do some statistical analysis that takes into account how big the region is that might have a random bump, and de-weight the significance of your distribution accordingly. The broader the range, the more likely you are to find a meaningless statistical fluctuation that gives you a nice-looking peak, and therefore the less meaningful your bump is. This is the Look-Elsewhere-Effect.
We saw a "more significant" peak than this melt away already this year, and so aren't claiming anything yet. And the fact that both CMS and Atlas, after each excluding a large swath of possible masses, see a small bump in about the same spot lends a lot of weight to it--it reduces the look-elsewhere issue. But...
The only channel that shows it is the gamma-gamma decay mode--there's nothing much in b-bbar. That's not unexpected; the background is higher for b-bbar, making the signal muddier and harder to see. But there could be something else happening in gamma-gamma that we haven't accounted for correctly, that gives both groups a bump in the mass plot. Of course that would be something new also, and good to learn about; perhaps new physics and perhaps better models of the old. So we'll keep at it, waiting for both a better gamma-gamma peak and some verification in another channel.
2 comments:
Astronomy and physics are so fascinating right now. Every discovery makes scientists shout, "What?"
Yeah, the choke-collar of the arithmetic of random numbers is a bitch, ain't it? In my field we keep forgetting - or ignoring - how much pharma data must? be random.
But a great study that tells you exactly what you want to hear is just sooo gratifying.
Post a Comment