Great stuff. "Correlation doesn't equal causation" is taught in basic science classes everywhere. :up:

IMO it's useful to reflect on and critically examine one's own thinking with regard to any pursuit. This is but one important example of how that can be done.

 

wrong forum. :shifty:




















:p

 

Ugh, how about this:

"Correlation matched with observation of events leads to logical conclusions"?

This is one of the many many problems with a stats based approached to sports, "we" see things, stats say otherwise, what is more accurate?

When in doubt, trust what you see, not what the stats say, your eyes do not lie to you, however you may not know what you are seeing..that is where stats enter the picture.

 

This is what you're after here (consilience):

http://en.wikipedia.org/wiki/Consilience

 

I'm going back to sleep.

 

Good for you. That'll make you not unlike the good number of people here who appear to do something very similar when confronted with objective information inconsistent with their beliefs. ;)

 

I don't care what everyone else says, you're aight with me.

 

Another reason this thread is relevant:

"There is an inverse relationship between ice cream sales and car accidents in New Jersey."

FALSE.

The compounding factor is the weather. People buy more ice cream when the weather is nice and hot outside (few car accidents). On the other hand, people buy less ice cream when it it stormy, cold, icy, and generally bad weather out (more accidents). Therefore, correlation does not mean causation!

 

I'd hardly call your statiscal analyses objective, let alone relevant. What I've been reading is each analysis has a modicum of truth that you then stretch to your conclusion (or implied conclusion).

 

Hmm, not neccessarily, when the weather is warm, people drive more miles, which raises the risk of car accidents.

This is why insurance companies cut discounts if one lives within 10 miles of where one works or goes to school, stats wise the risk of accident is quite low.

 

I think you completely missed his point :tongue2:

 

Oh well. Of course you're entitled to your opinion. :)

 

It's a very valid point and worth noting.

But it does NOT mean it's not worthwhile to see which factors correlate highly with winning. Especially when you're able to do so with the entire sample (all 32 NFL teams.) That makes a world of difference when you're talking about a sample of the population vs the entire sample.

 

If a stat doesn't accurately reflect what happened on the field then it's probably the result of the person watching the game whose responsible for recording the stat making an error. Most sports stats are recording an observation. The more advanced stats combine the observed stats with statistical formulas and models. Those are subject to error, but again, mostly on the part of the person or those interpreting the results.

The problem with relying on what we see is that our memories are not perfect. How many of Brady's INTs on Sunday were tipped passes? How many of Ryan Tannehill's INTs this season were on tipped passes? I remember my eyes telling me that some of their INTs were a result of tipped passes. But I trust whoever was watching the game along with me and was responsible for recording that data more than my memory.

 

Correlation not equalling causation is always a useful thing to keep in mind. But I think it's less a worry in a controlled environment like football, and sports in general. Since the game is bounded by rules and a time restriction it's easier to find causal links, as opposed to something like explaining why and how the ACA passed (my background is in political science, that's why I used that as an example). In other words, since there are only so many people on the field at once, doing only so many things at once, there are only so many reasons why a result occurred.

Football is difficult to analyze quantitatively because we don't have good ways to measure everything, yet. That's not an inherent flaw of stats or the idea of correlation. Even people who don't like the analyses that someone like shouright does use those ideas. When you watch the game and make observations and draw conclusions based on that you are seeing and use the relationship between what you saw and what you know about the nature of the game to draw conclusions you are making a correlation. It's just a less quantitatively rigorous one than what something like shouright does.

 

Shou is attempting to draw conclusions using incomplete data. You can draw your own conclusions for his intent. I feel he does it to intentionally bait you all into arguments solely for his own sadistic amusement.

 

A good example is Donald Thomas, he was cut quickly here, but you can see him move really well.

PFF's #2 rated Guard this year, it is a case of the stats saying he was not reallly very good, the eye ball test said otherwise.

 

Statistical discussion is very welcome in this forum, as it has been for years now. Just as much as a QBs arm angle or a lineman's ability to anchor.

 

This is getting more annoying. If you can't argue statistics, just say so or walk away. Seriously some of you are starting to sound like anti-global warming nut cases. Well if you don't like what it says, just pretend it's not science!!

 

The Earth is cooling off and I have stastical data that shows it! :tongue2:

 

EVERYTHING in life can be boiled down with mathematics.

A great quote:

-The gods did not reveal all things to men at the start; but as time goes on, by searching, they discover more and more.- Xenophanes (570-475 B. C. E)


You're wrong. let's move on.

 

because looking at a sheet of paper gives you incomplete information, very much like only seeing a guy play in person.

 

there are certain metrics, that when factored together can predict to damn near 100% accuracy the standings in several sports. Football among them.

Those are just facts.

 

Not enough data recorded.

 

Ah, but here stats are telling, namely his completion percentage.

THill's issue was not Hartline and Co, it was flat out bad passes to often, that was a good example of stats and eyeballs agreeing.

 

This kind of disagreement between two people who saw the "same" thing (i.e., the same games) is precisely why "seeing" alone is insufficient and objective analyses are needed.

If "seeing" alone were sufficient, there would never be any disagreement about anything in this forum.

 

There are also many stats that show that you are most likely to get into an accident within 10 miles of your home. This of course depends on where you live, the route you live on, the visibility when pulling out of your driveway, are you backing out... the speed limit on the road... etc. So many variables.

 

But not all of those are predictive of an accident. You can bet your insurance company has the predictive ones figured out and is using them, however. ;)

 

I feel he does it to get people to step back and reassess what they are seeing and to take emotion out of the equation. Sure he could go about it another way, but sometimes it is best to let others try to figure it out on their own. In the long run they will be better for it.

 

More miles driven with each extra foot exposing the insured to more risk of getting into an accident.

 

I agree.

 

Entirely accurate. :up:

The only other way to do it would be to apologize for saying what lots of people don't want to hear, but I'm not going to do that. That isn't my problem, I'm not a politician, and this isn't a popularity contest. :)

 

Tiredness in the morning... feeling comfortable about you surroundings... getting into a habit because you know that when you leave there is "USUALLY" no one coming down your road. Going about your usual routine (we are creatures of habit). All lead to more accidents close to home than when you are out on the road and paying more attention. Just my opinion.

 

That's the tough love part, hence the reason I said the rest in that sentence. :up: After you have spelled it out and the point still hasn't been taken you have to leave up to them to figure it out.

 

for single games...yes. for a full season...no.

You can absolutely predict who will be good with a pretty high degree of certainty.

The Playoffs are a series of single events that make for a lot of uncertainty and randomness. That's when numbers tend to not matter as much.

 

Really it's an exercise in appreciating the inherent boundaries among individuals, the freedom each person has to think and do as they please, and the development of the ability to offer information without needing it to have any influence whatsoever on anyone else. :up:

 

for example...

Matt Ryan.

for his career..when he equals or surpasses his career QB Rating, the Falcons were 38-3 going into that SF game. Now they are 38-4. (he had a 114 rating and lost) There is your randomness of a one game sample size.