3 ms·

> We can conclude, from the fact that we did not find a positive correlation, that the hypothesis is false. We could if it was a lab experiment where the only

> We can conclude, from the fact that we did not find a positive correlation, that the hypothesis is false. We could if it was a lab experiment where the only factor that varies would be the popularity of violent video games. But it's not. It's entirely plausible the the increase of home entertainment has indeed reduced the rate of juvenile crimes. (kids don't stay in the street any more). Yet it doesn't say much about how violent games induce violent behaviours.

sergiosgc 13y ago
We can. If we didn't have to account for other factors, we wouldn't even require tools like correlation and linear regression. These are statistical tools needed to deal with signals affected by both error and multiple inputs. These characteristics do not prevent analysis, though care must be taken. Your explanation is obviously possible, but it does not preclude my conclusion. If your explanation for the signal correlation is true, then video games account for comparatively little youngster aggression. I am terribly sorry for the ad-hominem, but I think, at this moment, you need to brush up on statistics 101. The arguments used are typical of pre-statistical studies, and HN is really not the place to be taught fundamentals.
- bnegreve 13y ago
  > I am terribly sorry for the ad-hominem, That was a bit cheap but ok. Let me add more. Here is the same data that I have reploted http://imgur.com/fdtc0tH http://imgur.com/fdtc0tH This chart is less biased (y axis starts at 0) and the correlation is less obvious. The correlation is actually -0.6557, which is significant but probably not as high as you would expect from the first chart. But ok the correlation between game software revenue and juvenile violence is probably here (we're talking about ~20 data points...). To go further and conclude anything about violent games, you would need at least the proportion of violent games among games, we don't have it here. Here's the data that I've extracted from the pixel position. It's not 100% accurate, but it should be good enough. 1990 412.999995 1991 430.999995 1992 448.999995 1993 469.999995 1994 495.49999995 1995 487.99999995 3.2 1996 432.49999995 3.67407407407407 1997 384.49999995 3.74179894179894 1998 355.99999995 4.55449735449735 1999 322.99999995 5.43492063492064 2000 298.99999995 5.5026455026455 2001 291.49999995 5.90899470899471 2002 276.49999995 6.92486772486772 2003 285.49999995 7.06031746031746 2004 270.49999995 7.33121693121693 2005 279.49999995 6.92486772486772 2006 297.49999995 7.26349206349206 2007 286.99999995 9.43068783068783 2008 288.49999995 11.6656084656085 2009 261.49999995 16

sergiosgc 13y ago

We can. If we didn't have to account for other factors, we wouldn't even require tools like correlation and linear regression. These are statistical tools needed to deal with signals affected by both error and multiple inputs. These characteristics do not prevent analysis, though care must be taken. Your explanation is obviously possible, but it does not preclude my conclusion. If your explanation for the signal correlation is true, then video games account for comparatively little youngster aggression. I am terribly sorry for the ad-hominem, but I think, at this moment, you need to brush up on statistics 101. The arguments used are typical of pre-statistical studies, and HN is really not the place to be taught fundamentals.

bnegreve 13y ago
> I am terribly sorry for the ad-hominem, That was a bit cheap but ok. Let me add more. Here is the same data that I have reploted http://imgur.com/fdtc0tH http://imgur.com/fdtc0tH This chart is less biased (y axis starts at 0) and the correlation is less obvious. The correlation is actually -0.6557, which is significant but probably not as high as you would expect from the first chart. But ok the correlation between game software revenue and juvenile violence is probably here (we're talking about ~20 data points...). To go further and conclude anything about violent games, you would need at least the proportion of violent games among games, we don't have it here. Here's the data that I've extracted from the pixel position. It's not 100% accurate, but it should be good enough. 1990 412.999995 1991 430.999995 1992 448.999995 1993 469.999995 1994 495.49999995 1995 487.99999995 3.2 1996 432.49999995 3.67407407407407 1997 384.49999995 3.74179894179894 1998 355.99999995 4.55449735449735 1999 322.99999995 5.43492063492064 2000 298.99999995 5.5026455026455 2001 291.49999995 5.90899470899471 2002 276.49999995 6.92486772486772 2003 285.49999995 7.06031746031746 2004 270.49999995 7.33121693121693 2005 279.49999995 6.92486772486772 2006 297.49999995 7.26349206349206 2007 286.99999995 9.43068783068783 2008 288.49999995 11.6656084656085 2009 261.49999995 16