View Single Post
  #215  
Old 10-17-2018, 06:31 PM
Damuri Ajashi is offline
Guest
 
Join Date: Jul 2006
Posts: 20,317
Quote:
Originally Posted by iiandyiiii View Post
They didn't data dredge. They were specifically looking for data about shootings of black and white people by police. And they presented that data. They're not saying, and I'm not saying, "this is conclusive evidence of racism". They're just saying "this is interesting and should be studied further", which is what I'm saying as well.
Did they analyze the data or did they present a small subset of the dataset that presents a gross disparity when you look at only that small dataset? They are reaching a conclusion from about a 3 dozen killings out of 3000 over 3 years. They didn't go into wondering what the murder rate was for 15-19 year olds. Based on Pro-Publica's agenda, I suspect they went into this looking for large disparities because 3::1 isn't shocking enough. Especially when you have studies saying that there are all these factors that makes the disparity disappear. That is data dredging.

Quote:
You appear to have made a conclusion based on other studies, and no other data presented can affect your conclusion. IMO, that's ridiculous and unscientific, but you're free to make such a conclusion if you wish.
Because those other studies were ACTUAL STUDIES. My mind cannot be changed by bullshit statistical tricks and sleight of hand. And neither should yours. There is NO value to the pro-publica "study" None. All it does is a calculation. No analysis just "HEY LOOK BLACK KIDS GET KILLED 21 TIMES AS OFTEN AS WHITE KIDS. AREN'T YOU OUTRAGED!!!!! that is the sole purpose of that "study" Its not trying to uncover some truth. it is playing statistical games.


Quote:
As to whether such numbers are "noise" or not, further investigation might reveal this. We certainly don't have enough information to make such a conclusion at this point.
No further investigation necessary. We have enough information to make a pretty good guess that the Pro-Publica "study" is bullshit. This is not a "teach the controveersy sort of situation. This is like flipping a coin 100 times, picking out a series where you got 5 heads in a row and concluding that the coin only comes mostly heads.

Quote:
Your certainty is unconvincing to me.
Do you know any statisticians? Please. Go to them. Ask them about data dredging and then ask them if pulling a subset of 3 dozen out of 3 THOUSAND datapoints is generally good statistical science.

The actual peer reviewed studies (and ask your statistician friends if the pro-publica calculation qualifies as a study) say that cops do not shoot blacks more frequently than they shoot whites. Until you provide a peer reviewed study that legitimizes what pro-publica presents, you are comparing science with factoids.

You are effectively the guy who thinks vaccines cause autism because you know some kid that got autism, that thinks that global warming is a myth because the weather got cooler over the last few weeks, that thinks women always lie about rape because of Tawana Brawley and the Rolling Stone article.

Like I said, one the one hand two peer reviewed studies that contradict your point of view and one back of the napkin calculation that the statistician that did the calculation won't stand behind but at least it is consistent with what you believe. At least global warming deniers have actual scientists willing to stand behind the notion that global warming is a myth.