Another shooting in the USA. Remind me about the reason for having guns again. - by SubJeff
LarryG on 7/9/2012 at 20:52
Unfortunately, Chimpy's selection of just a subset of the data "proves" nothing. The best it does is indicate that there may be something else going on in those particular countries other than gun ownership which drives gun homicides when combined with gun ownership. But it offers no clues as to what. That's why I said that you need to look at "other operating factors (possibly poverty, possibly politics, possibly culture, possibly gun regulations, possibly, possibly, possibly, ...)" to see what may be the true necessary and sufficient conditions. Clearly without gun ownership no one could get shot. No guns, no shots. But just as clearly more guns does not imply more shots by itself. So what other conditions could be the contributing factors? That's where we need to focus our attention and stop fussing over the numbers of guns. That's a red herring. Personally, I like the idea that the state of a country's legal system and the level of political (and judicial) corruption is most likely to be directly related to the gun homicide rate. But I don't have the statistics to support that intuition.
Edit: Oh, if you use the same Y scale for both of Chimpy's graphs, you'll see that both regression lines are nearly flat, indicating no correlation. This is a way in which you can be fooled by "zooming in" on a graph. The slope of the line appears to be about 0.03 when you read the scales. (I used (0,.75) and (2.5,90) to get that). So while the slope is positive, it still is very likely that if you did a correlation analysis on that subset that you would see that there is no correlation relationship between the two variables. Sorry.
june gloom on 7/9/2012 at 21:22
Also, quoting the Bible should mean you lose the argument, same as if you pulled out Hitler/Nazis. Or socialism for that matter.
jay pettitt on 7/9/2012 at 21:26
Other factors are clearly important but don't necessarily exclude ownership as a factor.
Oh, and you don't engineer flatness by changing the Y axis. The Y axis should be chosen that best shows the data plots most clearly. You'd be fooling someone by trying to change the incline by altering the Y axis.
Eyeballing is always iffy, but I'd be very surprised if that data didn't show very strong correlation. It's clear as day. Especially if you dismiss NI as an outlier, which I don't think you'd have any trouble justifying,
LarryG on 7/9/2012 at 21:47
Quote Posted by jay pettitt
Oh, and you don't engineer flatness by changing the Y axis. The Y axis should be chosen that best shows the data plots most clearly. You'd be fooling someone by trying to change the incline by altering the Y axis.
Exactly my point. Changing the scale of the axis fools you into thinking the slope is significant, because you are used to looking at graphs where x and y are at the same scale. When you make them different scales, you have to be very careful to avoid being misleading. What can never be misleading is the actual slope value. And at 0.03, that is close enough to zero for me to say no correlation.
Quote Posted by dethtoll
Also, quoting the Bible should mean you lose the argument, same as if you pulled out Hitler/Nazis. Or socialism for that matter.
I was going to quote Jonathan Swift in his
Polite Conversation, where he quoted John Heywood (or was it Matthew Henry, the sources seem to disagree): "There are none so blind as those who will not see." but I thought I would elevate the quotations a bit to something that, perhaps, more would be familiar.
Chimpy Chompy on 7/9/2012 at 22:00
Quote Posted by jay pettitt
So the job now is to try and better understand why your analysis and Chimpy's give us different answers, and whether that might be instructive.
Just to be clear I'm not claiming to have done some rigorous statistical workout on this data. I just thought the plots could be better presented than the form Larry used.
The slopes on both plots are +or- 0.03ish, the rich-nations one just has lower y values overall. So... yeah I dunno what that means exactly.
LarryG on 7/9/2012 at 22:04
They probably could have been (better presented). I didn't use scatter graphs because I thought people would have more trouble interpreting them correctly. Silly me. It turns out that some don't know how to interpret either format.
SubJeff on 7/9/2012 at 23:31
The funny thing is it seems Larry G and I actually agree that Jason Moyer is wrong, but perhaps it's just for different reasons.
The first graph you posted, Larry G, shows nothing of interest at all to me. There is no correlation but given the nature of the graph it is meaningless, because of other issues.
jay pettitt on 8/9/2012 at 00:07
Quote Posted by LarryG
Exactly my point.
No it isn't.
You cogitate the gradient by LOOKING AT THE FUCKING NUMBERS. Artistic sensibility and the notion that perhaps it should look more flatter doesn't come into it. Not ever. The scope of the data points defines the axis.
Quote:
you are used to looking at graphs where x and y are at the same scale. When you make them different scales, you have to be very careful to avoid being misleading.
Larry, the X and Y axis are almost never on the same "scale" - if they are it's by accident. They're almost certainly not even of the same units for fucksake. The scope of the data points defines the axis.
Quote:
What can never be misleading is the actual slope value [\o/]. And at 0.03, that is close enough to zero for me to say no correlation [:(].
The slope is not a test for correlation. Correlation (of the linear sort) is how well the data points fit the slope.
Adam Nuhfer on 8/9/2012 at 00:13
Guns have been around and available for well over 100 years to the general populace. The shootings as of late seem to be more representative of changing morals as opposed to the ease of acquiring said firearms. Attempting to legislate morality indirectly by banning firearms is a political solution, not a morality change.
Look at various "areas" in Africa over the last 25 years. The carnage and death inflicted by various "people" upon other "people" with machetes has been well demonstrated. Sierra Leone, Congo, Uganda and others have shown how morality plays a role in what one is willing to inflict on ones neighbor regardless of the instrument used.
jay pettitt on 8/9/2012 at 00:22
Sure, but let's say (for the sake of argument) you can substantially reduce the number of deaths with gun controls within, say, 10 years. But doing all the things needed to change the underlying morality that leads to violent crime is maybe a 50 year project.
whatdayado?
