Tag Archives: statistics

Polar Vortex, Climate Change, Red Herring?

Wow, it was unusually cold last week. In a small fraction of the globe. For a couple of days. And what does that cold snap, that big wiggle in the Polar Vortex that carries high-atmospheric winds around the North Pole, imply about “climate change”, also known as “global warming”, also known as “global weirding”?

The answer is very simple. Nothing.

If you heard anyone suggest otherwise — whether they said that the extreme cold implies that there is no global warming going on, or they said that the extreme cold implies that global warming is happening — you should seriously question anything that person says when it comes to climate change. Because that person does not respect (or perhaps even understand) the difference between anecdote and evidence; between weather and climate; between a large fluctuation and a small but long-term trend. Or between media hoopla and science.

In the interest of an imperfect analogy: Let me ask you this. Are you generally happier, or less happy, than you were five years ago? Answer this as best you can.

Now let me ask you another question. Did you, within the last month, have a really, really bad day, or a really, really good one?

Does the answer to the second question have much to do with the answer to the first one?

Barring an exceptional recent disaster in your personal or professional life, the fact that, say, last Thursday your car broke down, you locked yourself out of your house, your dog vomited on the carpet and you got caught in the rain without your umbrella does not have anything to do with whether you are a happier person than you were five years ago. Being a happier person has more to do with whether you have a better job, a happier family, a better sense of self-esteem, and things like that. And even if you love your job, you know there are going to be really bad days in the office sometimes. That’s just the way it goes. We all know that.

It’s the same with daily and monthly and yearly fluctuations in the stock market compared to the slow but fairly steady century-long growth of the U.S. economy (both curves corrected for inflation.)

So why, when there’s a big fluctuation in the daily, monthly or even seasonal weather, do people jump up and down about what the implications are for the long-term trends in climate? Continue reading

Two Higgs Bosons? No Evidence for That

[Note Added: this afternoon the author of the Scientific American article made a few corrections.  I leave it to you to judge for yourself whether he addressed the issues raised here.]

There’s been a little silliness floating around (sadly, in Scientific American, whose article contains at least two factual errors) unscientifically speculating that ATLAS’s new results on the Higgs-like particle, from data collected at the Large Hadron Collider [LHC], suggest there are two such particles rather than one. The mass measurement of this particle using the data when it decays to two photons, 126.6±0.3±0.7 GeV/c², is different, by 2.7 standard deviations, from the mass measurement obtained from its decays to two lepton/anti-lepton pairs, 123.5±0.9+0.4-0.2 GeV/c².  So… huh… gee… maybe there are two Higgs-like particles, a lighter one which rarely decays to two photons and a heavier one which rarely decays to two lepton/anti-lepton pairs?

[Note Added: I should emphasize, lest anyone blame ATLAS for this implausible line of speculation, that in the ATLAS presentation last week, which was one of several presentations that morning, these two mass measurements were presented simply and responsibly, as results from data.  Not a single speculative word was said about there being a hint of two Higgs particles.  I don’t know who got the ball rolling on that idea, but it wasn’t ATLAS.  And it’s not a plausible idea: see below.]

Take a deep breath. For not only would the two types of particles somehow have to be magically and implausibly arranged to mimic, at first glance and to a rough extent, a single Standard Model Higgs particle (the simplest possible type of Higgs particle), there’s another experiment, which unfortunately the writer of the Scientific American article neglected to consult.

ATLAS’s mass measurement from the events with two lepton/anti-lepton pairs also disagrees with CMS’s mass measurement obtained from the same type of events: 126.2±0.6±0.2 GeV/c².  Two similar experimental detectors, same measurement, moderate disagreement. Nature is nature; there’s no way that ATLAS can be making one type of particle all the time, while CMS is making a different one all the time.  So there is no evidence here, taking ATLAS and CMS together, favoring the existence of a separate particle with a mass of about 123.5 GeV/c² that decays to two lepton/anti-lepton pairs.

What is behind these discrepancies, then?  ATLAS and CMS each have scarcely a dozen of these two lepton/anti-lepton events, and their extraction of the Higgs particle’s mass from each event is somewhat uncertain, which is why many events are required for a good mass measurement.  When you still have small amounts of data, funny statistical fluctuations will often occur.  We’ve seen this before; back in 1989, when the Stanford Linear Collider (SLC) produced its first few Z particles at the Stanford Linear Accelerator Center, the plot of the Z particle’s mass gave a double resonance peak, instead of the single peak that was expected.   A brief moment of speculation occurred, but with more data the anticipated single peak structure emerged.  I’ve heard at least one other similar story from an earlier decade.  In fact ATLAS and CMS had a 2 GeV mass discrepancy when the first Higgs hints came in; that was just an effect of statistics.  Combine a fluctuation of this form with a minor detector calibration problem, and you’ll get discrepancies like this.

Multiple types of Higgs particles are certainly possible; people have considered this scenario for decades, and I’ve written about it here, for instance. Efforts to search for a second type of Higgs particle have been going on since the discovery of the first one.  But let’s not manufacture one out of thin air by looking selectively at the data; that’s not how reliable science gets done.