Statistics, love them or hate them

[Octave]

As to statistics though, put me in the hater's camp. I think a lot of them are meaningless.

While I understand this sentiment (I think), they can also be incredibly useful when used properly. Statistics allow us as scientists to control for an endless amount of actually meaningless factors and find what matters in huge data sets. Remembering the difference between correlation and causation is what makes statistics fall into the camps of meaningful or meaningless, along with the presentation of what they data are trying to say. To write off statistics as being "meaningless" is essentially to throw away all scientific findings from the mid-18th century on..

[SteveP]

interesting if you watch sports...
Red Sox Xander Bogaerts doubled in the first 3 games of the season...

they go to the statistics trough of baseball ( surely the biggest in sports )

since 1908 it has only been done 1 other time by Adrian Gonzales in 2012.

They have a vast store for every thing.
the sport is full of arcane trivia

[j44ke]

I know I am forgetting some of the details, but when Virginia set up their lottery, there was a provision of some sort for electronic purchase of numbers. Evidently there are lottery "clubs" where each member pays into a kitty and when they find a lottery where the odds of winning compare favorably to the number of tickets the club can buy, they buy whatever that amount of tickets that is, each with a different number. I don't think they can completely eliminate odds but they can reduce them considerably. In the case of Virginia, I think the club that started winning the lottery originated in New Zealand, so Virginia instituted a rule that number purchases had to be done in person. That way the process of purchasing the numbers took too long to enter the numbers and print out the receipt, thus eliminating the interest of bulk number purchasers.

I spent some time going over probability in my writing class after several misconceptions appeared among the students. One was that buying multiples of the same number increased the chances of that number coming up in the final drawing. The other was that buying two tickets halved the odds over buying one ticket. They also struggled with the idea that the best "investment" of time and money was one ticket with one machine picked number. They really did not trust the machine picked number and felt that their number made up of birthdates or whatever was much more trustworthy. Of course, none of them figured how many times they had entered this number in the lottery and not won into their "trustworthiness" calculation. They really wanted to believe in that numbers had agency and weren't simply symbolic language based on talking about how many cows you have.

[monadnocky]

I spent some time going over probability in my writing class...

I think that an evolutionary argument would say that

WAIT! COME BACK! (yes I know that "evolutionary arguments" are completely un-falsifiable and I cringe whenever I hear the term) BUT I do think that there are basic worldviews and heuristics that have done a relatively decent job of not getting ourselves killed or maimed and allowed us to replace ourselves with mini-me's to repeat the whole cycle. As I said above, looking for connections is what we do, and that tends to work out pretty well in the wild. I think we're wired so that we tend to see patterns where none exist, or over-emphasize small effects, or can't wrap our heads around something happening due to chance.

Like the myth of the "hot hand" in basketball. Nonsense, but tell that to any basketball fan.

As for statistics in research, I'm a Cohen's d guy and not a p guy. But then again, we have this huge replication crisis thing going on right now (hey bud, I can assure you that you do in your field as well, it's just that nobody's really looked...)

[davids]

I have to admit I've been tempted by those drawings for convertibles where you pay $100/ticket and they only sell 500 tickets: "I like those odds! ...wait a minute. No, I don't."

And if you want to see an example of a horrible misuse of logic and statistics, here you go.

[j44ke]

A strong brain-based belief in agency is what allowed us to compensate for our relatively slow foot speed by using predictive triangulation and meeting the very fast (and tasty) animal as it came out of the bushes 100 feet to the right of where it went into the bushes while trying to escape.

[SteveP]

A strong brain-based belief in agency is what allowed us to compensate for our relatively slow foot speed by using predictive triangulation and meeting the very fast (and tasty) animal as it came out of the bushes 100 feet to the right of where it went into the bushes while trying to escape.

still only caught them 5% of the time*... cause they are faaaast.

* cavemanstatisticalanalysis.com

[monadnocky]

* cavemanstatisticalanalysis.com

oh, I see you have met some of my advisees....

[SteveP]

oh, I see you have met some of my advisees....

i had to pay them for the publishing rights to put this on the forum.
2 pieces of metal and a pointed rock

[TTX1]

I have to admit I've been tempted by those drawings for convertibles where you pay $100/ticket and they only sell 500 tickets: "I like those odds! ...wait a minute. No, I don't."

And if you want to see an example of a horrible misuse of logic and statistics, here you go.

I like the Firefly raffle at D2, because it feels like FLT wins no matter what.

Caught that WaPo article yesterday... felt like someone dropped a brick in my path. There is a 100% chance that I won't read it again.