Gun Control – Answering The Issues Raised 3

“You start your latest posting by defending the use of the Small Arms Survey as the basis for the analysis that you quoted in your original piece. I’m sure that the survey is a useful source of information and I accept that it is quoted and used by reputable organizations. This does not equate to the survey being a suitable, or the most suitable, source of data for meaningful statistical analysis.”

Agreed, but nor does it preclude it from being such.

“You say that the survey seems, to you, to be the best source of data we have and challenge me to suggest a more reliable data source. I’d remind you that in your previous blog  you included a link to an academic article which argued that the proportion of firearm suicides to overall suicides is the most reliable proxy for establishing gun prevalence. This, of course, is the measure used in the peer reviewed article which I provided as evidence of a correlation between gun ownership and homicide.”

Any proxy must be validated against actual ownership. So what has it been validated against? To quote your preferred article

“FS/S, which measures the distribution of firearm vs nonfirearm methods used in suicide rather than the rate of suicide, has been validated against survey-based measures of household gun ownership. A recent study determined that FS/S is the best proxy for household firearm ownership rates of the half-dozen or more proxies that have been used in the literature.19 FS/S is highly correlated with the percentage of households reporting firearm ownership in studies of 16 developed nations (r = 0.91),22 the 9 US census regions (r = 0.93),23 21 US states (r = 0.90),23 170 US cities (r = 0.86),24 and 12 areas within a single state (r = 0.87).19

So  internationally it has been tested on just 16 developed nations. This would require the leap of faith that the correlation holds for all other nations, developed or not. I see no reason why this is intrinsically any more reliable than the small arms survey data. There are many cultural issues that influence the choice of suicide method which are liekly to invalidate any extrapolation. Perhaps we shall have to agree that there is no reliable international data on which to establish a correlation between gun ownership and homicide rates. Of course in the absence of such data we have no reason to reject the null hypothesis that there is no relationship.

However, all is not lost, the data I used for the US state analysis was a survey based measure of household gun ownership. If we have actual survey ownership data we don’t need to use a proxy, that correlates to it. We have actual survey data to work with, the gold standard and that confirms that there is no correlation!

“I didn’t question the US state gun ownership data. I confess that I haven’t given a moment’s thought to the reliability of this data – my time to expend on this debate is limited and my comments have been more than long enough as is. Whilst gun controls and gun ownership may well vary from state to state, there are no controls to prevent the transport of guns from one state to another (apart from in the case of Hawaii, which I believe has both the lowest gun ownership and the lowest homicide rate in the US). As such, state by state gun prevalence is less likely to be relevant than data for individual countries.”

Unfortunately, the “peer reviewed” article that you quote as your primary source of evidence for the correlation between gun ownership and homicide rates is a US state by state analysis!

Based on the above comment I assume you have changed your position and now assert that this is no longer compelling evidence?

I’m sticking with the evidence that I have supplied as the best direct analysis we’ve seen regarding the correlation.

Hmm, that will be the article that uses the second hand proxy of survey data instead of the available actual survey data and uses what you term unreliable state by state data instead of country by country data?

You state that I have ‘glossed over […] the strongest evidence for the lack of any significant correlation’ which you feel is provided by the British Journal of Criminology paper.

Your position is that a large group of criminologists – the bulk ofthose examining the inequality/homicide link over several decades – have all concluded that gun prevalence is irrelevant to crime (based upon evidence unknown), have decided that this is an issue which requires no further study, and haven’t bothered to inform the other large group of criminologists who have meanwhile been hotly arguing this very question. Furthermore, the many researchers funded by the NRA and other pro-gun groups haven’t spotted and pointed out that the weight of unbiased academic opinion is firmly set in their favour. I glossed over it because it is a ridiculous argument.

You have created a straw man argument. I make no comment on whether gun prevalence requires no further study.

It has been your position all along that my original analysis is invalid because it does not control for important variables. Your criticism was

“This analysis makes no attempt to adjust for socio-economic factors such as population density, poverty and inequality, levels of corruption within government and law enforcement etc.”

Your most recent comment abandons this line of attack altogether so perhaps that is no longer your view? However, not a single one of these criminologists felt it necessary to control for gun ownership levels in their studies of homicide rates.

Now you can’t have it both ways, either they didn’t bother to control for gun ownership because it is not worth controlling for. Which supports my position. Why control for an independent variable with zero correlation to the dependent variable?

Alternatively, it is not necessary to control for socio-economic factors and gun ownership in a study aimed at one or the other. Which undermines your criticism of my analysis.

Calling this point ridiculous and creating a straw man argument inferring incorrect views to my position may muddy the water, but does not answer the point.

If gun ownership was highly correlated to homicide rates, you would expect every study on homicide rates to need to control for it. (Most of them seem to control for poverty or inequality for example) They don’t control for it, why not?

“If what you infer from this paper was the case, then those same academics wouldn’t have bothered engaging in any subsequent research which relates to the correlation between gun prevalence and homicide. They’ve already decided, so
wouldn’t waste their time considering the question any further. With some difficulty (Google is my only tool), I’ve tracked down other work by a number of these researchers and can advise that this is not the case:… Long list of Gun Research Papers”

This is a straw man argument again, misrepresenting my view. There is no reason why these researchers would not examine the issues of gun control and violent crime, even homicide in specific circumstances.

I am not claiming, as you imply, that they made a decision to exclude gun ownership because, after extensive analysis they concluded it was not important. I am saying that academic criminologists do not consider gun ownership rates sufficiently highly correlated with homicides to control for it as part of their research into other areas. If it where they would have to, or it would invalidate all their research!

You have done a commendable job in tracking down the work of these authors, in and around the area of guns and gun control and violent crime. However the papers you quote  focus on specific situations e.g. New Mexico Juveniles, high rate offenders and violent crime, rather than any general investigation into homicide rates and gun ownership.

It is in no way inconsistent to accept no general correlation between homicide rates and gun ownership and still want to explore the impact of gun ownership on violent crime committed by persistent offenders or other specific narrower questions.

You provide a list of 20th century genocides and argue that
gun ownership is a defence against genocide.

Let’s start with this evidence that more Americans have died
from domestic gunfire since 1968 than in all wars in their history. That’s 1,384,171 domestic gunfire deaths. Even if we remove 2 thirds of this number as a generous estimate of the number of those deaths which were suicide that leaves
us with over 460,000 homicides and accidental deaths in just over 50 years. So your argument that ‘the smallest of these genocides’ outweighs the damage caused by domestic guns is demonstrably wrong.

Fair point, my error, only the top 6 genocides contained more murders. However, taken as a whole they account for over 67 Million murders, which equates to 7,287 years of US gun deaths! The point remains that it is orders of magnitude more.

By limiting your list to the 20th century, of course, you miss out the native Americans who were subject to a systematic genocide by the US government, despite being armed.

I am not sure that the rate of firearm ownership amongst native americans was very high?
This was also more of a case of invasion, rather than a domestic tyranny. (Which is, of course, in no way to excuse what was a horrendous genocide)

“To take one of the most striking examples which is on your list, the Nazi’s actually relaxed strict gun controls which had been put in place following World War I. Whilst they undoubtedly disarmed the Jews, ownership of guns by other Germans was regulated but not restricted. The Nazi genocide of the Jews was only possible with the assistance – or at the very least without the resistance – of the vast majority of the German people, and it is ridiculous to suggest that the Jewish community, even in possession of side arms, could have mounted a successful armed resistance that would have prevented this genocide. The same can be said of Stalinist Russia, Communist China and the rest.”

If someone came to take away my family in the middle of the night to send them to a concentration camp, rest assured it would be much more dangerous for them if I was armed. It is hard to get even storm troopers to break into a house with an armed citizenry. It only needs a few hundred of them to be shot, before the number of new recruits drys up.

Even with an armed population, there is little possibility that a
minority group within the US – say, Muslims, for example – could successfully defend themselves against a genocide conducted by the government which has gained the connivance of the majority of the population. Without having obtained this support, there is no possibility that the government could carry
out a widespread genocide, even if the population was unarmed.

You make two assertions without evidence.

Does Mugabe in Zimbabwe have the connivance of the majority of the population, or Assad in Syria, did Saddam Hussein in Iraq, Pol Pot in Cambodia? There are clearly counter examples to your position that mass murder of the civilian population only takes place with the general support of the population.

Even if you can find examples where an armed civilian population were subject to a genocide (I am not convinced by your native american example) it does not alter the fact that it is much more difficult and therefore at least some potential genocides would be avoided.

Where two armed groups within a country are sufficiently well matched, the resulting conflict between a government and a significant section of its population is a civil war. Whilst civil wars may be more palatable than genocides, they are often just as deadly.

Agreed, but in some cases neither will happen as both sides see the potential cost is too high. In the other cases the alternative is tyranny and genocide, so I am not sure you can be anything but better off by being armed.

It is worth noting the vast majority of countries with gun control where genocide has not occurred, and remember that the best defence against tyranny and genocide are democracy and international law.

This is logically very weak. Genocide is fortunately very rare. A more useful measure would be what % of countries suffering from genocides of their own citizens had gun control, versus the % of countries that did not.

International law didn’t do much for the 67 Million + murdered in 20th century genocides. Democracy was not much help for the Jews when the Germans democratically elected the Nazi Party to power.

This entry was posted in Uncategorized. Bookmark the permalink.
  • @simonr916

    OK. It seems that posting responses at 1am doesn’t ensure clarity of
    thought or expression. Thank you again for your further thoughtful
    responses to my comments.

    Taking the issues in order…

    Validating a proxy against other available data doesn’t mean that the
    data you validate against is more reliable – “the gold standard” – it
    is a sense check to help establish that your proxy is as reliable as
    possible. The point of using a proxy is that it is intended to be more
    reliable than other available sources of information, but wild
    variation from those other sources may indicate a problem with the

    I agree that establishing accurate and reliable figures of gun
    ownership is difficult, especially for international data where
    political and legal conditions in many countries may impact upon the
    accuracy of official statistics and survey data. However, I don’t
    accept your position that it is not possible to establish an accurate
    enough figure to allow you to test for correlation. Statistical
    analysis is possible but needs to take account of the reliability of
    those data (as part of confidence levels) when reporting upon its
    results – whether those results indicate a correlation or not.

    With regard to the study of state by state correlation between gun
    ownership and homicide, I didn’t say that the ease with which guns can
    be transported across state boundaries would render any analysis
    irrelevant. Instead I said that I hadn’t paid too much attention to
    your original data on this because it would seem to make sense that
    any correlation between ownership and homicide from state to state
    within a single country would likely be weaker than the correlation
    between separate countries with border controls. Any analysis, and in
    particular an unsophisticated one, is less likely to find a
    correlation. If I am right in this – and it is just my inference based
    upon logical consideration of the issue – then a valid study which
    demonstrates that there is still a correlation despite the ease of
    transport would actually be stronger evidence.

    I know that you’ll disagree with this, but my point is that I am not
    changing my position on the value of the article as evidence. Many
    other studies also point to a correlation – particularly where cross national
    analysis is limited to OECD countries – but we have agreed not to
    bombard each other with more articles.

    I still maintain that your original analysis is not useful in
    establishing whether or not there is a correlation between gun
    prevalence and homicide because it takes no account of other
    variables. I would remind you that the blogger who undertook the
    analysis makes this very point himself.

    So, I hadn’t ‘abandoned this line of attack’. I was answering new
    issues that you’d raised and had considered that the argument had
    moved on.

    You say that I created a straw man argument regarding your use of the
    British Journal of Criminology study. I don’t see that I do.

    Your interpretation of the study is that it shows that these dozens of
    researchers had decided that gun prevalence has a zero impact upon
    homicide so didn’t bother to control for it. I simply pointed out that
    you were making an inference from this that they considered that the
    issue is decided and settled, and I was pointing out how unlikely this

    I maintain that it is dangerous to extract conclusions from a study
    which are not relevant to the questions being asked by that study –
    particularly when these conclusions are to the knowledge or beliefs of
    another group of researchers whose data is the subject of

    It is interesting that the researchers didn’t control for gun
    prevalence. I’ve already said that I don’t know why. You have given a
    possible hypothesis, but that is a long step from proving anything. If
    you want to go away and provide further evidence to support your
    hypothesis, I’d be happy to hear it.

    Your interpretation is that this study is good evidence to support
    your position. My argument is that this study isn’t about guns and
    that, until and unless you can prove your hypothesis, it is completely
    irrelevant to the debate.

    On the question of genocide, I’ll just make a couple of additional points.

    The main weapon used to round Jews up into concentration camps was
    economic hardship. Jews were denied employment or assistance and, at
    least in the early years, were unaware of the genocide being committed
    in the concentration camps. They boarded trains to the camps on the
    promise of food and deportation.

    In Russia, it was the absence of gun control that allowed the
    Bolsheviks to arm themselves to overthrow the Czar, thus (eventually) allowing
    Stalin to come to power.

    Robert Mugabe had the general support of the Zimbabwean population,
    certainly in the early part of his leadership and arguably until
    relatively recently. Saddam Hussein initiated his genocide against the
    Kurds with a series of poison gas attacks against which small arms
    would have been no defence.

    The issue of genocide and the ‘defence against tyranny’ argument is a
    strong one. There is a clearly a debate to be had about what the best
    way is to prevent genocide. We’re clearly on different sides in this

    I think we’ve had an interesting and productive discussion on gun
    control. You’ve made some good and interesting points, and it is the
    first argument about gun control I’ve had online where I haven’t been
    called a ‘pussy’. Thank you for that.

    If you choose to answer these comments, I don’t propose to respond
    further because it has taken up more of my time than I can afford and
    I need to move on. It seems that I haven’t persuaded you that gun
    control works, and (whilst you have argued your case vigorously, and
    have presented some strong arguments) you haven’t persuaded me that it
    does not. I hope that you at least accept that it is possible to hold
    the view that gun ownership should be controlled based upon logic and
    reason, as this is the point in your first post which originally
    provoked me into response.

    • Murray Rothbard

      Thank you for the most interesting and educational debate I have had in a long time. We will agree to differ, but I think both appreciate that our positions are perhaps not quite so obvious and free from reasoned attack as we had thought.

  • Kokom Qomariyah

    gun ownership should be limited to avoid abuse and criminality.