Cito di Pense wrote:

Hermit wrote:

Cito di Pense wrote:The correlation may be undeniable, at that scale, but it is not overwhelmingly impressive.

Thanks, and agreed.

Just to be clear, that's not what I was seeking agreement on, and suggests you're mainly interested in your conclusion, and not in the robustness of your analysis.

I am interested in reading any post that will shoot down the statistics I used and the way I have used them. Sniping from the sidelines does add a measure of entertainment, though. At least for people who don't give a fuck about the issue of gun control and prefer to admire grandstanding.

What is more, the bit concerning the US situation is not much more than an afterthought. My main concern was the constant jubilation about the success of the 1996 gun buyback scheme. It fucking wasn't, and I have provided the figures proving that it was not. Also, links to the sources. The only way to argue that the buyback scheme was a success even though it made no difference to the suicide or the homicide rate, is to claim that without it suicide and homicide rates would have been even higher than they turned out to be, but I don't know how anyone would go about substantiating that.

While I'm posting here, I may as well add another chart I started on. It too illustrates that the homicide rate has not dropped as a result of the buyback scheme. It also shows that though homicide by firearm has decreased substantially, homicide by other means has taken up the slack.

Hermit wrote:
I am interested in reading any post that will shoot down the statistics I used and the way I have used them.

Then read my post again. You and I agree that the buyback was grandstanding, too, but that's a perennial problem in politics. Don't demand that people shoot down your presentation, as if you were doing some science instead of just massaging some noisy statistics, the analysis of which I have indicated is not all that robust compared with windage. I don't like Macdoc's hypothesis, either, but there's simply no point in trying to argue with him, and that's not because his analysis is robust.

Apropos of the Macdoc ideology:

Cito di Pense wrote:I don't oppose gun rights, mainly because I can't outgun the gun lobby, and also because every gun death is just one person less who's going to be up nights worrying about global warming and sea level rise. And the ones who weren't worried about it? You thought they were numbskulls and wanted them gone, anyway.

Cito di Pense wrote:

Hermit wrote:I am interested in reading any post that will shoot down the statistics I used and the way I have used them.

Then read my post again.

It was insubstantial enough for one reading to suffice. I missed nothing.

Cito di Pense wrote:Don't demand that people shoot down your presentation, as if you were doing some science instead of just massaging some noisy statistics, the analysis of which I have indicated is not all that robust compared with windage.

I was massaging statistics? Only if you made up a new definition of the verb just now. You have indicated nothing. You made unsubstantiated assertions. All I got from you so far is some airily waved hand. Put some meat on your criticism via concrete figures.


I don't really care so much about John Howard's stunt. All politicians will jump on any opportunity to exploit a situation to their electoral advantage. The Fort Arthur massacre was just too good to miss. Howard had no problem twisting it into a motherhood issue and running with it. What really pisses me off is the anti-gun lobby using the drop in homicides and suicides to falsely claim that the buyback scheme was a huge success in terms of reducing suicides and homicides. In those terms it did exactly nothing, and one does not need to massage statistics to prove it. All one needs to do, as I have done, is to quote official government figures as they stand on their own.

Australian and UK statistical changes in murders are pretty unimportant as quite simply the sample size is so small

350 murders , one terrorist attack or gang dispute will increase that by 10%, even a single serial killer or someone pissed of with their family could screw up those figures.

A 10% increase in murders if you starting size is 3500 is scientifically interesting, a 10% on a sample of 350 is just data noise.

Looking at Australian crime figures murder rates are low and don't change an awful lot, best thing the police can do is to make sure that doesnt change rather than worry about preventing murders (realistically you can't at that base)

mrjonno wrote:Australian and UK statistical changes in murders are pretty unimportant as quite simply the sample size is so small

That's not how statistics work mrjonno.
The sample size is the total number of citizen and the total number of gun related deaths.
That the number of gun related deaths is low doens't mean the sample size is to small.

mrjonno wrote:350 murders , one terrorist attack or gang dispute will increase that by 10%, even a single serial killer or someone pissed of with their family could screw up those figures.

Again, this would only demonstrate the low rate of gun related deaths, not that the sample size is to small.

mrjonno wrote:A 10% increase in murders if you starting size is 3500 is scientifically interesting, a 10% on a sample of 350 is just data noise.

Nope. 10% is 10%, regardless of the original number.

Nope. 10% is 10%, regardless of the original number.

Actually that isn't how you decide if stats are significant and not just random, population size is vital

The point is most Western countries bar America have such low levels of murders trying to determine if government policy makes any difference is impossible to tell unless you are looking very long term. There would need to a major change in environment to make much difference on murder rates

mrjonno wrote:Australian and UK statistical changes in murders are pretty unimportant as quite simply the sample size is so small

The sample rate is based on a population of 25 million people. If that made it too small you'd expect erratic and significant changes in the homicide from one year to the next, but as you noted yourself, they don't change an awful lot.

Coupla things...

I share your frustration with people who claim the 1996 gun reforms made a huge difference, especially if they misrepresent the pre-1996 gun laws as lax, when in fact they were already stringent, especially by US standards.

Nevertheless, we are dealing with small-N issues. Australia had, at the time, around 300 homicides a year. Less than a quarter involved firearms (around 70), and of those around half involved firearms that were unaffected by the reforms (source). So we're looking at around 35 homicides a year due to newly-restricted weapons, and only about half of those weapons are thought to have been surrendered during the buyback. So even if the relationship between semi-automatic gun availability and homicides with those weapons was strictly linear (which is unlikely, as those gun owners more likely to commit crimes with those weapons were also probably less likely to surrender them than your average "responsible gun owner") the expected reduction in homicide would be about 17 out of 300, or a reduction from the then rate of 1.8 /100,000 population to 1.7 / 100,000. Working with a pre-reform level of 35 such homicides is decidedly subject to small-N issues, especially higher volatility.

Contrary to your claim above, the level of volatility in the statistics is actually at the level of expected impact or higher:



with year to year changes, both rises and falls, of greater than 0.1 / 100,000 population quite common.

Now, that's an interesting graph in itself for other reasons. It's an extension of this graph:



which was produced to show that the rate of decline in homicide after the reforms (green dotted line) was equal to the rate of decline before the reforms (red dotted line). As you can see from the extended graph, that hasn't held true, with subsequent levels of homicide consistently below the green dotted trend line.

So, the effects of the reforms are indeed small and difficult to conclusively tie to changes in legislation, but given the small starting incidence of firearm homicide using weapons restricted by the 1996 reforms, large changes were never on the cards in the first place.

OlivierK wrote:So, the effects of the reforms are indeed small and difficult to conclusively tie to changes in legislation, but given the small starting incidence of firearm homicide using weapons restricted by the 1996 reforms, large changes were never on the cards in the first place.

Quite so. My main contention is, and has always been, that the buyback scheme's effect on Australia's murder rate is indiscernibly small if one exists at all.

My first post in this thread was in response to an article Macdoc quoted

HERE’S WHAT HAPPENED WHEN AUSTRALIA BANNED SEMI-AUTOMATIC AND AUTOMATIC WEAPONS
A lesson for American lawmakers

What was that lesson supposed to be? That "suicides by gun dropped an astonishing 57 percent and homicides by guns dropped 47 percent", but the rates of both by any means rose? Every gun nut with half a brain will seize on Joshua Scott Albert's "argument" with glee. You could not have a better example of gun control proponents shooting themselves in the foot, so to speak. Yet they make the same mistake over and over.

And you are right too, pointing out that the situation in the USA is nothing like the one pertaining to Australia. Which is something I too mentioned in the same post.

Hermit wrote:

Macdoc wrote:I think the critical difference between Australia and the USA is that private ownership of concealable hand guns has been almost completely banned long before the buyback scheme. Of all homicides in Australia homicide by firearm consisted of only a very small percentage long before that buyback scheme took 20% of privately owned firearms out of circulation. Most homicides in the USA, on the other hand, are committed with precisely those weapons.

Of the 11,208 homicides in the USA during 2012, 8,855 were committed by firearms, of which 6,371 were handguns (source). For this reason a buyback might work in the US (as suggested in the graph I supplied previously), but that has manifestly not been demonstrated with the Australian experience of it.

For the record, I see no justification for privately owned assault style rifles (with or without bump stocks and large magazines) and handguns because both are specifically designed for killing people. That task should be reserved for soldiers, police and some security guards.

mrjonno wrote:

Nope. 10% is 10%, regardless of the original number.

Actually that isn't how you decide if stats are significant and not just random, population size is vital

And you only need enough to have a significant sampling. Which is the case with the UK.

Hermit wrote:My first post in this thread was in response to an article Macdoc quoted

HERE’S WHAT HAPPENED WHEN AUSTRALIA BANNED SEMI-AUTOMATIC AND AUTOMATIC WEAPONS
A lesson for American lawmakers

What was that lesson supposed to be? That "suicides by gun dropped an astonishing 57 percent and homicides by guns dropped 47 percent", but the rates of both by any means rose? Every gun nut with half a brain will seize on Joshua Scott Albert's "argument" with glee. You could not have a better example of gun control proponents shooting themselves in the foot, so to speak. Yet they make the same mistake over and over.

I agree it's possible, and quite common, to oversauce that pudding. But it's also important not to go too far the other way. Anyone would think, with this talk of suicide and homicide rates rising after the reforms, that Australia isn't making progress on these measures, but in fact homicide rates, by all means, are around 45% lower now than in 1996 (see above graphs), and suicide rates, by all means, are at the second lowest levels in the last 100 years, almost matching the historic lows during WW2:





It's ridiculous to pin that on the 1996 gun reforms, but it's also ridiculous to dismiss as one of many factors that has improved the overall situation.

OlivierK wrote:It's ridiculous to pin that on the 1996 gun reforms, but it's also ridiculous to dismiss as one of many factors that has improved the overall situation.

I'm afraid the latter is not ridiculous. The gun buy-back scheme started on 1 October 1996 and concluded on 30 September 1997. It turns out that both the suicide and the homicide rates were higher in 1998 than before the buyback program commenced. Attributing a decline in both after 1998 to the program does not make sense. It would imply that the 600,000+ firearms that were taken out of circulation were still being used for a year or more after they disappeared.

The only other way to explain the delay of the program's effectiveness would be to argue that due to other confounding factors (say, a severe economic depression, for example) the suicide and homicide rates immediately following the buyback program's conclusion would have been even higher without it. Nobody has managed to do that convincingly to my knowledge.

Hermit wrote:

OlivierK wrote:It's ridiculous to pin that on the 1996 gun reforms, but it's also ridiculous to dismiss as one of many factors that has improved the overall situation.

I'm afraid the latter is not ridiculous. The gun buy-back scheme started on 1 October 1996 and concluded on 30 September 1997. It turns out that both the suicide and the homicide rates were higher in 1998 than before the buyback program commenced. Attributing a decline in both after 1998 to the program does not make sense. It would imply that the 600,000+ firearms that were taken out of circulation were still being used for a year or more after they disappeared.

The only other way to explain the delay of the program's effectiveness would be to argue that due to other confounding factors (say, a severe economic depression, for example) the suicide and homicide rates immediately following the buyback program's conclusion would have been even higher without it. Nobody has managed to do that convincingly to my knowledge.

There are plenty of other explanations for a higher rate in 1998. The most obvious is statistical noise from a small-N sample. The rise from 1997 to 1998 of 0.1 /100,000 is smaller than many other year-to-year rises and falls in the series, and is entirely plausibly explained by the volatility of the data series. Here's the chart again:



What you're arguing is that the small rise immediately post-reforms (from a below-trend starting point, no less) is not explainable due to the volatility of the series. With respect, that's bullshit.

Also, I didn't attribute the fall to the buyback (in fact, I said to do so would be ridiculous), but I did say it was similarly ridiculous to rule it out as a factor. All I'm saying is that it's not clear that it's a factor, and that if it were, the expected effect would be small. If you're going to make the much stronger positive claim that reforms that lowered availability of high-lethality firearms can be ruled out as even a minor factor in suicide rates being significantly lower today, and homicide rates having fallen faster than the previous trend, then you've got some work to do.

Let's have a look at what you call volatility. Second column represents the change from the previous year in Australia, third column in the USA. (source for column 3)

1994 -0.1 -0.5
1995 +0.2 +0.2
1996 -0.1 +0.2
1997 0.0 +0.4
1998 -0.2 -0.5
1999 +0.3 -0.6

Would you put the greater volatility in the US down to its small-N sample?

OlivierK wrote:I didn't attribute the fall to the buyback (in fact, I said to do so would be ridiculous), but I did say it was similarly ridiculous to rule it out as a factor. All I'm saying is that it's not clear that it's a factor, and that if it were, the expected effect would be small. If you're going to make the much stronger positive claim that reforms that lowered availability of high-lethality firearms can be ruled out as even a minor factor in suicide rates being significantly lower today, and homicide rates having fallen faster than the previous trend, then you've got some work to do.

I have not claimed that reforms that lowered availability of high-lethality firearms can be ruled out as even a minor factor in suicide rates being significantly lower today. My claim is limited to saying that in the absence of statistics that might support even a very minor efficacy of the buyback program it is ridiculous to crow about its effects. The strongest positive claim you can quote me on is my contention that "the buyback scheme's effect on Australia's murder rate is indiscernibly small if one exists at all".

And no, the faster rate of decline in suicides between 1999 and 2006 cannot be attributed to the buyback program which concluded in September 1997.

Hermit wrote:Let's have a look at what you call volatility. Second column represents the change from the previous year in Australia, third column in the USA. (source for column 3)

1994 -0.1 -0.5
1995 +0.2 +0.2
1996 -0.1 +0.2
1997 0.0 +0.4
1998 -0.2 -0.5
1999 +0.3 -0.6

Would you put the greater volatility in the US down to its small-N sample?

It will be due to a lot of factors, as it is in Australia, it's not important. What's important is that the volatility is larger than the effect of the buyback, so when rates rose in 1998, no conclusions can be drawn about the ineffectiveness of the buyback.

Hermit wrote:

OlivierK wrote:I didn't attribute the fall to the buyback (in fact, I said to do so would be ridiculous), but I did say it was similarly ridiculous to rule it out as a factor. All I'm saying is that it's not clear that it's a factor, and that if it were, the expected effect would be small. If you're going to make the much stronger positive claim that reforms that lowered availability of high-lethality firearms can be ruled out as even a minor factor in suicide rates being significantly lower today, and homicide rates having fallen faster than the previous trend, then you've got some work to do.

I have not claimed that reforms that lowered availability of high-lethality firearms can be ruled out as even a minor factor in suicide rates being significantly lower today. My claim is limited to saying that in the absence of statistics that might support even a very minor efficacy of the buyback program it is ridiculous to crow about its effects. The strongest positive claim you can quote me on is my contention that "the buyback scheme's effect on Australia's murder rate is indiscernibly small if one exists at all".

Nope. I'll quote you on a stronger claim right now...

Hermit wrote:

OlivierK wrote:It's ridiculous to pin that [below-previous-trend suicide and homicide rates] on the 1996 gun reforms, but it's also ridiculous to dismiss as one of many factors that has improved the overall situation.

I'm afraid the latter is not ridiculous.

Here you're saying that it's not ridiculous (ie, that it's reasonable) to dismiss the buyback as one of many factors in lower suicide and homicide rates than would have been expected if pre-1996 trends had continued post-1996.

OlivierK wrote:It's ridiculous to pin that [below-previous-trend suicide and And no, the faster rate of decline in suicides between 1999 and 2006 cannot be attributed to the buyback program which concluded in September 1997.

Why not? The only reason you've given is the higher rate in 1998, which is a single data point in a series with considerable volatility, so the higher value on that single point, and a steeper declining trend are not incompatible.

Look, broadly we agree: effects of the buyback were only ever going to be small, and given the number of variables are almost impossible to identify. The only place we part company seems to be that you seem to think the data isn't compatible with the predicted, or claimed, benefits of the buyback, due to rises in homicide rates in 1998 that are well within the normal range of variability of the data, a conclusion that, frankly, lacks statistical rigour.

OlivierK wrote:

Hermit wrote:

OlivierK wrote:I didn't attribute the fall to the buyback (in fact, I said to do so would be ridiculous), but I did say it was similarly ridiculous to rule it out as a factor. All I'm saying is that it's not clear that it's a factor, and that if it were, the expected effect would be small. If you're going to make the much stronger positive claim that reforms that lowered availability of high-lethality firearms can be ruled out as even a minor factor in suicide rates being significantly lower today, and homicide rates having fallen faster than the previous trend, then you've got some work to do.

I have not claimed that reforms that lowered availability of high-lethality firearms can be ruled out as even a minor factor in suicide rates being significantly lower today. My claim is limited to saying that in the absence of statistics that might support even a very minor efficacy of the buyback program it is ridiculous to crow about its effects. The strongest positive claim you can quote me on is my contention that "the buyback scheme's effect on Australia's murder rate is indiscernibly small if one exists at all".

Nope. I'll quote you on a stronger claim right now...

Hermit wrote:

OlivierK wrote:It's ridiculous to pin that [below-previous-trend suicide and homicide rates] on the 1996 gun reforms, but it's also ridiculous to dismiss as one of many factors that has improved the overall situation.

I'm afraid the latter is not ridiculous.

Here you're saying that it's not ridiculous (ie, that it's reasonable) to dismiss the buyback as one of many factors in lower suicide and homicide rates than would have been expected if pre-1996 trends had continued post-1996.

That is a particularly malicious piece of quote mining. You left out the bit immediately following the snippet you did quote, which explains why I wrote "I'm afraid the latter is not ridiculous." To wit:

Hermit wrote:The gun buy-back scheme started on 1 October 1996 and concluded on 30 September 1997. It turns out that both the suicide and the homicide rates were higher in 1998 than before the buyback program commenced. Attributing a decline in both after 1998 to the program does not make sense. It would imply that the 600,000+ firearms that were taken out of circulation were still being used for a year or more after they disappeared.

The only other way to explain the delay of the program's effectiveness would be to argue that due to other confounding factors (say, a severe economic depression, for example) the suicide and homicide rates immediately following the buyback program's conclusion would have been even higher without it. Nobody has managed to do that convincingly to my knowledge.

In other words, I gave you two reasons why it is ridiculous to say "the latter is not ridiculous", neither of which amount to asserting that the lowered availability of high-lethality firearms can be ruled out as even a minor factor in suicide rates being significantly lower today. You just quoted me saying exactly that in the post I am replying to, FFS.

I have consistently based my line of argument on empirical data - what they show and what they do not show. If you want to refute that line, please do so on the same grounds. And don't put words in my mouth. I'll report the next post you do that directly or by quote mining. Thank you, and good night.

Dude, chill out. It's not a fucking quote mine, and you can report it till the cows come home. Yes, you've backed away from the claim in more recent posts, but you asked me to quote you making a claim, and I did so, precisely because you requested it.

And don't ask me to refute you on the basis of data as if that's not what I've done. I've posted charts of Australian homicide and suicide rates, and calculations to quantify the expected effect of the 1996 gun restrictions on overall homicide, with citations. You've got one data point that goes up (in a series in which fluctuations of that magnitude are completely routine) and you're shaking it like a dog with a bone.

The stupid thing is that we broadly agree, but that (ironically) you can't do rigorous statistics in much the same way as the people you criticise for overegging the effects of the reforms can't.

If you just stuck to the data, like you claim, you wouldn't be posting shit like this:

Hermit wrote:What was that lesson supposed to be? That "suicides by gun dropped an astonishing 57 percent and homicides by guns dropped 47 percent", but the rates of both by any means rose?

when suicide rates by all means and homicide rates by all means are at 70 year lows.

Hermit wrote:I am interested in reading any post that will shoot down the statistics I used and the way I have used them.

Nah, you're not interested, you're butthurt. Your ego won't let you see that you've got what you've asked for. If an annual data series has a downward trend, it doesn't mean that there won't be any year-on-year rises. You're citing a year-on-year rise as a reason to dismiss a move to a more rapidly declining trend. I'll quote you doing it, with the full post for context:

Hermit wrote:

OlivierK wrote:It's ridiculous to pin that on the 1996 gun reforms, but it's also ridiculous to dismiss as one of many factors that has improved the overall situation.

I'm afraid the latter is not ridiculous. The gun buy-back scheme started on 1 October 1996 and concluded on 30 September 1997. It turns out that both the suicide and the homicide rates were higher in 1998 than before the buyback program commenced. Attributing a decline in both after 1998 to the program does not make sense. It would imply that the 600,000+ firearms that were taken out of circulation were still being used for a year or more after they disappeared.

The only other way to explain the delay of the program's effectiveness would be to argue that due to other confounding factors (say, a severe economic depression, for example) the suicide and homicide rates immediately following the buyback program's conclusion would have been even higher without it. Nobody has managed to do that convincingly to my knowledge.

See the red bit? That's statistics fail of the kind that dishonest or innumerate climate denialists wheel out when it gets cold in winter.

Hermit wrote:The correlation of more guns, more homicide is undeniable.

I'm still trying to figure out the horizontal scale on this plot. It doesn't appear to be linear, it doesn't appear to be logarithmic. In fact, it appears as if somebody scribbled in the values by hand and screwed up. If we get a linear correlation out of such a plot, we can get a linear correlation out of fucking anything.

Hermit wrote:I am interested in reading any post that will shoot down the statistics I used and the way I have used them.

Well, what i'm interested in is what specific statistics were used in constructing the plot. All I have are a couple of links and no mention whether all the abscissa values come from the same column in the same table.

Hermit wrote:I have consistently based my line of argument on empirical data

That's as may be. You still should endeavor to organize it and present it coherently and correctly.

Cito di Pense wrote:

Hermit wrote:The correlation of more guns, more homicide is undeniable.

I'm still trying to figure out the horizontal scale on this plot. It doesn't appear to be linear, it doesn't appear to be logarithmic. In fact, it appears as if somebody scribbled in the values by hand and screwed up. If we get a linear correlation out of such a plot, we can get a linear correlation out of fucking anything.

I provided links to the raw data the chart is based on. Feel free to show me where the problem lies, but if you decide to do that, cite data and the process you used to prove my chart wrong. Hand-waving - be it yours or anyone else's - has never convinced me.