FirstAscent
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 05:03:32 PM |
|
Here's the graph... Maybe now you can see what I meant when I said "the numbers are all over the place": ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fdl.dropbox.com%2Fu%2F146411%2Fgundeaths..jpg&t=665&c=Ao17LAJl8GnwgA) More permissive<----------------------------------------------------->Less permissive There's a trend... vaguely. But the highest data point is in the middle, and the least permissive has more gun deaths than the third most permissive, and the 9th. And all of the other top 5 least permissive states. Sure, I could slice and dice that data to show pretty much any result I wanted. But taken together, it doesn't show a very clear correlation. Especially when you consider that the data includes all gun deaths, accidental, violent, and suicide. I'd hardly call that conclusive. Tell you what: you find me data about gun crime, and I'll make another graph. If that one shows even this level of correlation, I'll eat my hat, switch positions, and start crying gun control from the rooftops. But I bet you can't. Oh! and while I was looking up info on that data, I found this lovely little nugget: "I am generally skeptical of gun laws," says Eugene Volokh, a law professor at UCLA. "The theory is that gun laws may prevent crimes of passion—domestic crimes, altercations over traffic incidents, or committed by someone who is otherwise law-abiding but has an anger problem… gun-control laws can potentially do something, but the kind of crime by which they can do the least is a mass shooting." That's from the very same article those data come from. The data just doesn't appear to be "all over the place" as you see it. Do you know how photography works? Please follow along. It's an excellent example of data sampling. The smaller the pixel on the sensor, and the less time the shutter is open, the more noise you get in the image. This is why small sensor cameras set at high ISO and shooting during nighttime will create a noisy image. Sampling theory explains all this. The image is there, but what happens is a reduction in the number of photons collected per pixel. Less photons per pixel will result in what appears to be randomness when you look at a small portion of the image - i.e. just a few pixels. But when you step back, a picture emerges. Very clearly, when you view the data as a whole, a picture emerges. Mathematically, this was demonstrated when a correlation coefficient of 0.7 resulted. Why don't you apply a running average to the data or perform some regression analysis on the data?
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 05:19:05 PM |
|
Very clearly, when you view the data as a whole, a picture emerges. Mathematically, this was demonstrated when a correlation coefficient of 0.7 resulted. There's a trend... vaguely.
But taken together, it doesn't show a very clear correlation. Especially when you consider that the data includes all gun deaths, accidental, violent, and suicide. I'd hardly call that conclusive.
You feel like addressing this part? Tell you what: you find me data about gun crime, and I'll make another graph. If that one shows even this level of correlation, I'll eat my hat, switch positions, and start crying gun control from the rooftops. But I bet you can't. Oh! and while I was looking up info on that data, I found this lovely little nugget: "I am generally skeptical of gun laws," says Eugene Volokh, a law professor at UCLA. "The theory is that gun laws may prevent crimes of passion—domestic crimes, altercations over traffic incidents, or committed by someone who is otherwise law-abiding but has an anger problem… gun-control laws can potentially do something, but the kind of crime by which they can do the least is a mass shooting." That's from the very same article those data come from.
|
|
|
|
FirstAscent
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 05:24:18 PM |
|
Very clearly, when you view the data as a whole, a picture emerges. Mathematically, this was demonstrated when a correlation coefficient of 0.7 resulted. There's a trend... vaguely.
But taken together, it doesn't show a very clear correlation. Especially when you consider that the data includes all gun deaths, accidental, violent, and suicide. I'd hardly call that conclusive.
You feel like addressing this part? I'll even go a step further: accidental, violent, suicide, and self defense. I don't have any desire to hide data, or mislead, as demonstrated by my statement above. Furthermore, if data existed to the left of the chart, assuming even more permissive gun laws existed (such as in NAP), we might even see the formation of a bell curve. The point is, data is data. The data does show a significant downward trend and does correlate. You can't argue that. What you can do is argue the underlying interpretation of it.
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 05:33:04 PM |
|
Oops. I'll take credit for that fuck-up. I should have been clearer that the following portion is what I wanted you to address, not the first part. The first part is done. .698 correlation, when you include all gun deaths, whether accidental, suicide, or as a result of violence (both in defense and as a result of attack, no less!), is simply not strong enough to show that restricting guns will reduce gun violence. Too much noise in the data, coming from all of those other deaths. Address this, please: Tell you what: you find me data about gun crime, and I'll make another graph. If that one shows even this level of correlation, I'll eat my hat, switch positions, and start crying gun control from the rooftops. But I bet you can't. Oh! and while I was looking up info on that data, I found this lovely little nugget: "I am generally skeptical of gun laws," says Eugene Volokh, a law professor at UCLA. "The theory is that gun laws may prevent crimes of passion—domestic crimes, altercations over traffic incidents, or committed by someone who is otherwise law-abiding but has an anger problem… gun-control laws can potentially do something, but the kind of crime by which they can do the least is a mass shooting." That's from the very same article those data come from.
|
|
|
|
FirstAscent
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 05:41:17 PM |
|
Oops. I'll take credit for that fuck-up. I should have been clearer that the following portion is what I wanted you to address, not the first part. The first part is done. .698 correlation, when you include all gun deaths, whether accidental, suicide, or as a result of violence (both in defense and as a result of attack, no less!), is simply not strong enough to show that restricting guns will reduce gun violence. Too much noise in the data, coming from all of those other deaths. Address this, please: Tell you what: you find me data about gun crime, and I'll make another graph. If that one shows even this level of correlation, I'll eat my hat, switch positions, and start crying gun control from the rooftops. But I bet you can't. Oh! and while I was looking up info on that data, I found this lovely little nugget: "I am generally skeptical of gun laws," says Eugene Volokh, a law professor at UCLA. "The theory is that gun laws may prevent crimes of passion—domestic crimes, altercations over traffic incidents, or committed by someone who is otherwise law-abiding but has an anger problem… gun-control laws can potentially do something, but the kind of crime by which they can do the least is a mass shooting." That's from the very same article those data come from. I don't feel inclined to bother until you accept that the correlation is strong (as per the wikipedia article, which states that anything above 0.5 is strong), and that the trend is clear despite noise. If you need to plot a running average or perform regression analysis to demonstrate this to yourself, then by all means, do so. If you can't concede a point based upon statistics because it disagrees with your ideology, then why should I expend effort in continuing the discussion in the direction you would prefer it to go. I believe my last post was more than fair with regard to discussion.
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 05:50:03 PM |
|
I don't feel inclined to bother until you accept that the correlation is strong (as per the wikipedia article, which states that anything above 0.5 is strong), and that the trend is clear despite noise. If you need to plot a running average or perform regression analysis to demonstrate this to yourself, then by all means, do so.
If you can't concede a point based upon statistics because it disagrees with your ideology, then why should I expend effort in continuing the discussion in the direction you would prefer it to go. I believe my last post was more than fair with regard to discussion.
Fucking troll. "Several authors have offered guidelines for the interpretation of a correlation coefficient. However, all such criteria are in some ways arbitrary and should not be observed too strictly. The interpretation of a correlation coefficient depends on the context and purposes. " .698 correlation, when you are looking at all gun deaths means nothing. So what if there are fewer gun suicides? That just means there are more pills or bathtub suicides. Find me data on gun crime. Then, if that data shows even a .5 correlation, I will eat my fucking hat. Respond to the quote from the article you linked to, at the very least.
|
|
|
|
FirstAscent
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 05:56:35 PM |
|
I don't feel inclined to bother until you accept that the correlation is strong (as per the wikipedia article, which states that anything above 0.5 is strong), and that the trend is clear despite noise. If you need to plot a running average or perform regression analysis to demonstrate this to yourself, then by all means, do so.
If you can't concede a point based upon statistics because it disagrees with your ideology, then why should I expend effort in continuing the discussion in the direction you would prefer it to go. I believe my last post was more than fair with regard to discussion.
Fucking troll. "Several authors have offered guidelines for the interpretation of a correlation coefficient. However, all such criteria are in some ways arbitrary and should not be observed too strictly. The interpretation of a correlation coefficient depends on the context and purposes. " .698 correlation, when you are looking at all gun deaths means nothing. So what if there are fewer gun suicides? That just means there are more pills or bathtub suicides. Find me data on gun crime. Then, if that data shows even a .5 correlation, I will eat my fucking hat. Respond to the quote from the article you linked to, at the very least. It's rather sad that you can't separate the concept of correlation from the interpretation of it. We can't move on until you understand the difference. I suggest you go focus on your Colorado conspiracy theory in the other thread ( https://bitcointalk.org/index.php?topic=94471.msg1046661#msg1046661 ) and reap the benefits of that fruitful exploration.
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 06:12:14 PM |
|
I don't feel inclined to bother until you accept that the correlation is strong (as per the wikipedia article, which states that anything above 0.5 is strong), and that the trend is clear despite noise. If you need to plot a running average or perform regression analysis to demonstrate this to yourself, then by all means, do so.
If you can't concede a point based upon statistics because it disagrees with your ideology, then why should I expend effort in continuing the discussion in the direction you would prefer it to go. I believe my last post was more than fair with regard to discussion.
Fucking troll. "Several authors have offered guidelines for the interpretation of a correlation coefficient. However, all such criteria are in some ways arbitrary and should not be observed too strictly. The interpretation of a correlation coefficient depends on the context and purposes. " .698 correlation, when you are looking at all gun deaths means nothing. So what if there are fewer gun suicides? That just means there are more pills or bathtub suicides. Find me data on gun crime. Then, if that data shows even a .5 correlation, I will eat my fucking hat. Respond to the quote from the article you linked to, at the very least. It's rather sad that you can't separate the concept of correlation from the interpretation of it. We can't move on until you understand the difference. I suggest you go focus on your Colorado conspiracy theory in the other thread ( https://bitcointalk.org/index.php?topic=94471.msg1046661#msg1046661 ) and reap the benefits of that fruitful exploration. I could put together data that shows correlation much stronger than .69 for anything I want, just by tossing in extra factors, or taking others out. You cannot separate the correlation from the context of the data and shout "It correlates!" Of fucking course making it harder to get a gun makes it less likely that someone who wants to kill themselves has a gun available to do it. Pills are easy to get a hold of. The data is too damn noisy to call .698 "strong". I'm going to repeat my request one last time: Show me data on gun crime vs. gun law, and if it correlates, I will heartily accept that gun control laws help. But I know you won't. You won't even respond to the article you fucking linked to. Whatever. Have a nice life, fuckwad. I'm done wasting my time with you. You are literally not worth my time. I don't know what you're trying to push, here, but I'm certain it's just more bullshit. Unless you have something productive to say, go take a flying fuck off a short pier.
|
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 06:18:51 PM |
|
Excellent data. They even offer it in a spreadsheet. It does not, unfortunately, include any measurement of "permissivity" of the gun laws in each state, I'll just use the rankings from the previous data. Graph is coming, but it will take some time to get everything collated.
|
|
|
|
Taz
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 06:20:07 PM |
|
Hmm.. Guess it's not over then.
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 22, 2012, 07:37:48 PM |
|
So, Here's the chart from that data... ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fdl.dropbox.com%2Fu%2F146411%2Fguncrime..jpg&t=665&c=TdqLfu-7hjGSrw) More Permissive<------------------------------------------->Less Permissive And, if anyone doubts me, here's the raw data that I used. You can verify it against the data Explodicle linked to. It's the same info. The original data (That FirstAscent posted) did not have information on DC's permissivity re: gun laws, but it's widely accepted as the strictest in the country, so I placed it at "51". If someone has data that contradicts this, I'd be glad to see it, and will revise my chart. Also, a note on Florida: There was no data for firearms murders for that state, though overall that does not look to have created much of an anomaly when you consider how much greater the other two numbers are in comparison, across the board. So, What's the correlation in that dataset, FirstAsshole?
|
|
|
|
nimda (OP)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 23, 2012, 01:37:11 AM |
|
So, What's the correlation in that dataset, FirstAsshole?
Well uh... I just watched a very nice performance of the Shakespeare play "Much Ado about Nothing," so I'm cracking up whenever someone says "ass" (no seriously, the guy who played Conrade was hilarious), and I don't wish to "written down AN ASS," but if you give me the raw data of that dataset I can find the correlation coefficient for you ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif)
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 23, 2012, 01:39:58 AM |
|
So, What's the correlation in that dataset, FirstAsshole?
Well uh... I just watched a very nice performance of the Shakespeare play "Much Ado about Nothing," so I'm cracking up whenever someone says "ass" (no seriously, the guy who played Conrade was hilarious), and I don't wish to "written down AN ASS," but if you give me the raw data of that dataset I can find the correlation coefficient for you ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) Ready and waiting for you. (click the words "raw data" in the sentence directly below the image in that post)
|
|
|
|
nimda (OP)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 23, 2012, 02:11:10 AM |
|
So, What's the correlation in that dataset, FirstAsshole?
Well uh... I just watched a very nice performance of the Shakespeare play "Much Ado about Nothing," so I'm cracking up whenever someone says "ass" (no seriously, the guy who played Conrade was hilarious), and I don't wish to "written down AN ASS," but if you give me the raw data of that dataset I can find the correlation coefficient for you ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) Ready and waiting for you. (click the words "raw data" in the sentence directly below the image in that post) Yes, well... uh... how about something parse-able? I'm a programmer, not a Microsoft Office wizard. Even Zamzar had nfi ![Tongue](https://bitcointalk.org/Smileys/default/tongue.gif) Luckily this poorly-built site worked for me. I hope my hacky Notepad++ regular expression produced the correct pared-down data of the " Total Gun Crime" statistic The result is that the data shows "only" a correlation coefficient of 0.1606
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 23, 2012, 02:18:06 AM |
|
The result is that the data shows "only" a correlation coefficient of 0.1606
For which dataset? I can split them off for you, if you'd like.
|
|
|
|
nimda (OP)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 23, 2012, 02:20:13 AM |
|
The result is that the data shows "only" a correlation coefficient of 0.1606
For which dataset? I can split them off for you, if you'd like. For the Total Gun Crime per 100,000 pop. I should have specified that. Would you like the data for the others? RegEx is a wonderful thing; it's not a big deal if there's interest. Oh, and what is this thread doing without a graphic from Yale? ![](https://ip.bitcointalk.org/?u=http%3A%2F%2Fwww.stat.yale.edu%2FCourses%2F1997-98%2F101%2Fcor.gif&t=665&c=gT_mBhX_bQpAWA)
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 23, 2012, 02:29:56 AM |
|
The result is that the data shows "only" a correlation coefficient of 0.1606
For which dataset? I can split them off for you, if you'd like. For the Total Gun Crime per 100,000 pop. I should have specified that. Would you like the data for the others? RegEx is a wonderful thing; it's not a big deal if there's interest. The other numbers would be nice. To clarify, In this case, does a high (positive) coefficient mean that high x->high y, ( / ) or does it mean that high x -> low y ( \ )?
|
|
|
|
FlipPro
Legendary
Offline
Activity: 1764
Merit: 1015
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 23, 2012, 02:34:18 AM Last edit: July 23, 2012, 06:52:45 AM by FlipPro |
|
Bitcoin geniuses who know more than law enforcement officials, compiling their own data in an attempt to disprove the obvious. ![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif) ... the amount of ego seen in this forum is astonishing. Do computer nerds really know EVERYTHING If you came to Bitcoin Talk it would sure seem that way... And you guys wonder why the Internet laughs at us... ![Undecided](https://bitcointalk.org/Smileys/default/undecided.gif)
|
|
|
|
myrkul
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
July 23, 2012, 02:35:40 AM |
|
Bitcoin geniuses who know more than law enforcement officials, compiling their own data in an attempt to disprove the obvious. ![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif) ... the amount of ego see in this forum is astonishing. I don't recall yanking your chain....
|
|
|
|
|