How much of Science is based on Statistics

[StatsClue]

Very random and possibly silly question, but then this is random chat. We'll add silly to it.

Does Physics use statistics and probability as heavily as Medicine does? My guess is No.

In Medicine, we're limited in that we can't experiment in most situations and therefore end up with conclusions drawn from analysing outcome measures that may have more predictor variables that we can conceive, yet we can only work with what we know, until we know more. We use large p-values of 5 percent. And we often revise our truths. Medicine changes rapidly.

How heavily is Physics invested in Statistics?

 

[Dason]

I'll give you a search term that might suggest an area that uses probability and statistics heavily: probability in quantum mechanics.

 

[spunky]

Another important search term for your googling adventure: 5 sigma

A big chunk of evidence for the discovery of the Higgs Boson by CERN relies on that and when you read the description you realize it really is just a p-value.

 

[hlsmith]

Well a difference is physics can be performed in a vacuum, so the underlying relationships can be teased out. Medicine is based on physiology and each person is different, even twins. So it is hard to find signals and make generalizations - thus the use of stats.

Monte carlo simulations came from the Manhattan project, which would be another day example of stats used in physics. Also the search for exoplanets uses stats, if you want to call that astrophysics or not.

Also, i feel like a have heard of many people who were in physics that transitioned over to data science. So the jump must not be that far between the fields.

 

[noetsi]

I think science is primarily theory, but depending on your definition of it you would probably answer your question different ways.

 

[ondansetron]

Another important search term for your googling adventure: 5 sigma

A big chunk of evidence for the discovery of the Higgs Boson by CERN relies on that and when you read the description you realize it really is just a p-value.

An that the p-value they note is often misinterpreted (I've seen it as .00005 chance [or whatever number] that it's a fluke, which we know is a bogus interpretation).

 

[StatsClue]

I vaguely remember the Higgs Boson experiment at CERN from the news actually and the p-values quoted were 0.---many many zeros and then a one and I remember thinking that most of Medicine allows for a huge probability of a chance result by comparison. Anyway, we all understand that for obvious reasons, Medicine can't experiment like Physics can.

Ondansetron - I didn't really get the bogus interpretation comment. Are you talking about the general interpretation of p-values or some interpretation of the Higgs Boson, in the context of the associated p-value?

Thanks all for your comments.

 

[ondansetron]

Ondansetron - I didn't really get the bogus interpretation comment. Are you talking about the general interpretation of p-values or some interpretation of the Higgs Boson, in the context of the associated p-value?

Thanks all for your comments.

I'm referring to the TV shows and news articles where I have seen people treat the p-value as an error probability/probability that higgs boson was a fluke. Those are totally incorrect interpretations of p-values.

https://www.nytimes.com/2012/07/05/...vered-higgs-boson-particle.html?pagewanted=all
"Both groups said that the likelihood that their signal was a result of a chance fluctuation was less than one chance in 3.5 million, “five sigma,” which is the gold standard in physics for a discovery."

http://www.independent.co.uk/news/s...very-of-higgs-boson-god-particle-7907677.html
"meaning that there is less than a one in a million chance that their results are a statistical fluke."

https://www.newscientist.com/article/dn22014-celebrations-as-higgs-boson-is-finally-discovered/
"There’s a 5-in-10 million chance that this is a fluke. That was enough for physicists to declare that the Higgs boson – the world’s most-wanted particle – has been discovered."

Just several out of a ton of misinterpretations of the p-value.

 

[hlsmith]

There is also the flip, the use of physics in statistics - Hamiltonian Monte Carlo.

 

[hlsmith]

Since I got a like from @ondansetron - I thought I would also mention that I read last night Yule was able to grasp and explain autocorrelation in time series using physics (e.g., pendulum swinging).

Also, since you all were referencing extreme p-values and collider research, I reviewed an abstract yesterday on Mendelian randomization and it presented a pvalue to E-88. Hmm.

 

[StatsClue]

Can't say I understood most of your replies but thanks.