# Gun prevalence vs homicide rates - correlation and causation

The discussion on gun ownership is often heated and detached from facts.

What does science say on correlation and more importantly on causality between gun prevalence and homicide rates?

There are academic studies that can be easily googled but they often contradict each other and many times their plausibility is questionable.

-
You may also want to take a look at suicide rates. And to Skeptics where we have addressed the same question. – Sklivvz Dec 27 '12 at 22:17
@Sklivvz unfortunately, it looks like that link only contains anedoctal stories like DVK's one in this post... – o0'. Dec 28 '12 at 16:52
@daniel.selacek Unfortunately, looking at those contradictory studies is all we have to go by. There is almost definitely no causality. As far as helping you make an informed decision, my opinion is that you can't just look at statistics. Depending on what nation you're in, gun ownership can be considered a civil right. When the question of rights comes up (even if they're minority rights), you have to set a very high bar to protect them. My opinion is that significant causality is that bar. But my answer below tries to give a view of the statistics none the less. – Javid Jamae Dec 28 '12 at 17:03
@Lohoris, anecdotal stories are not acceptable on skeptics. Which answers are anecdotal? – Sklivvz Dec 28 '12 at 18:08

There were absolutely no articles, blogs, or studies that I could find that suggested causation at all. Below is a list of articles that attempt to prove or disprove correlation.

This blog post attempts to show correlation, if any, on a scatter plot between gun ownership and murder / crime rates.

From the article:

It has the following data for all 50 states (except Florida, which does not submit the data for some reason):

• Brady Score (how much gun control a state has)
• Murder Rate per 100k population
• Firearm murder rate per 100k population
• Firearm Murder %

• Violent Crime Rate per 100k population

The following are the scatter plots that he generated, which (in the author's analysis and findings) show no statistical correlation between gun ownership and firearm murder or crime rate.

The original source for the article is another blog post that can be found here.

Looking at US data, another blogger did scatter plot analysis using source data from the About.com US Liberal Politics page and the Gun violence in the United States Wikipedia page. He concludes that:

There is literally 0 correlation between how many people own firearms, and the homicide rate with firearms, or overall homicide rate as below graphs and correlation coefficients illustrate.

His graphs are shown here:

In yet another scatterplot analysis of correlation between gun ownership and murder rate by country (again, using Wikipedia as the source), the author find a negative correlation between gun ownership and rate of homicide. He also makes a comment that these statistics aren't very helpful in the debate. From the article:

If all countries are included in the plot, there is a negative correlation between gun ownership and rate of homicide. Only when the plot is limited to OECD countries (with the outliers of US and Mexico included), does a very small positive correlation appear.

...

Because of the many uncontrolled variables involved in comparisons between countries, I don’t believe such statistics can help the case for or against gun control. I’m posting them here to counter the claims by advocates of increased governmental gun control that the statistics legitimise their preference.

A separate study entitled Would Banning Firearms Reduce Murder and Suicide? A Review of International and Some Domestic Evidence done by Don B. Kates (an American Criminologist and Attorney) and Gary Mauser (a Canadian Criminologist), was published in Volume 30, Number 2 of the Harvard Journal of Law & Public Policy (pp. 649-694). The authors in this article found that there is no correlation between gun ownership and murder.

From the article:

The non‐correlation between gun ownership and murder is reinforced by examination of statistics from larger numbers of nations across the developed world. Comparison of “homicide and suicide mortality data for thirty‐six nations (including the United States) for the period 1990–1995” to gun ownership levels showed “no significant (at the 5% level) association between gun ownership levels and the total homicide rate.” Consistent with this is a later European study of data from 21 nations in which “no significant correlations [of gun ownership levels] with total suicide or homicide rates were found.”

The authors then go on to say that they found a negative correlation between gun ownership and violent crime, meaning that populations with higher gun ownership had fewer incidents of violent crime. From the article:

The same pattern appears when comparisons of violence to gun ownership are made within nations. Indeed, “data on firearms ownership by constabulary area in England,” like data from the United States, show “a negative correlation,” that is, “where firearms are most dense violent crime rates are lowest, and where guns are least dense violent crime rates are highest.”

Another book entitled More Guns, Less Crime, written by John Lott, studies crime data for every county in the United States during 29 years from 1977 to 2005 and finds no correlation between gun ownership and murder and concludes that allowing concealed handguns decreases violent crime rates.

The wikipedia page (linked to above) for the book has a slew of papers that argue for and against his conclusions.

The Statistics section of the Gun Politics Wikipedia page references a slew of reports, studies, and anecdotes against gun control.

There are several sites that display or visualize data, but whose authors don't make any claims or assessments on correlation or causality whatsoever. The Guardian published data in an article entitled Gun homicies and gun ownership listed by country that can be downloaded an analyzed. They also have an interactive heat map that displays the following data:

• Average firearms per 100 people
• Homicide by firearm rate per 100,000 population
• % of homicides by firearm

This is a screenshot of the heatmap:

I took a (very) shallow dive into the data, looking at a 3-4 countries from each continent and there doesn't seem to be much correlation between ownership rates and murder rates. One commenter on the article makes an interesting point that we can't use the univariate solution of Gun Control to solve a multivariate problem of murder. He says:

If the rate of gun ownership is a determinant of the gun murder rate, how does one explain the gun ownership rates in Mexico (15 guns /100 and 9.97 murders/100k) and Jamaica (8.1 guns/100 and 39.4 murders /100K) and the very high gun murder rates? Guatemala has a gun ownership rate of 13.1/100 and a murder rate of 34/100k.

The US has 88.8 guns/100 and 2.97 murders/ 100k.

Univariate solutions for multivariate problems are unsatisfying.

On the other side of the debate, the article entitled Rates of Household Firearm Ownership and Homicide Across US Regions and States, 1988–1997 concludes as follows:

Although our study cannot determine causation, we found that in areas where household firearm ownership rates were higher, a disproportionately large number of people died from homicide.

FactCheck.org reviewed studies in 2008 and posted Violent Crimes and Handgun Ownership. They looked at a study by David Hemenway (director of the Harvard Injury Control Research Center and the Harvard Youth Violence Prevention Center), who reviewed commonly cited research from peer-reviewed journals. From the FactCheck.org article:

It found that studies of the United States or U.S. cities, states and regions "generally find a statistically significant gun prevalence-homicide association." The report said that the evidence from such "U.S. cross-sectional studies is quite consistent … where there are higher levels of gun prevalence, homicide rates are substantially higher, primarily due to higher firearm homicide rates."

With regard to a causal relationship, FactCheck.org says:

That’s not to say that taking the guns away from such areas would necessarily lead to a reduction in the murder rate. Such studies have shown a statistically significant relationship between guns and murder but not a causal one.

Disclaimer: I haven't personally reviewed the original sources or the analysis behind the sources I'm referencing. I'm not making an argument, one way or the other, as to whether or not I agree with the findings in these articles. I posted these because I think the format and nature of the findings in are specifically relevant to the original question.

Disclaimer 2: I listed many more articles that show little to no correlation, because it was more difficult to find ones that demonstrated a significant correlation. I am happy to add references to more articles that show positive correlation (or even causation) if you post them in the comments.

-
+1 - As I noted on DVK's answer below, correlation does not imply causation so we have to be careful there. The question is seeking information on causation, but I think you did a good job of laying out the data that is available without implying more than what is appropriate. – Michael Kingsmill Dec 27 '12 at 18:00

It appears that there have not been many (or any) peer reviewed studies that show a causal relationship between gun ownership and violent crime. There have been studies that show a statistically significant relationship, but causality as you know is a much higher burden to prove. There is a good discussion of the entire topic in scientific terms at this link.

In summation, it appears that higher gun ownership does imply higher rates of violent crime, but causation cannot be shown in the data available:

That’s not to say that taking the guns away from such areas would necessarily lead to a reduction in the murder rate. Such studies have shown a statistically significant relationship between guns and murder but not a causal one.

-

As you see, the answers and comments here also contradict each other. The statistics can be used and misused, it depends who publish them.

Take that article: http://www.landonswan.com/personal/gun-control-vs-murdercrime-rates-a-statistical-look/ which finds no relevant correlation between gun control and violent crime rate. But the problem is, it's based on internal USA statistics only. The author should be asked, how such data could be relevant in any way, because we have only one country that is analyzed. What's the difference what laws are in particular states, taken there's no border control between states?

Another cited article: http://www.guardian.co.uk/news/datablog/2012/jul/22/gun-homicides-ownership-world-list analyzes gun control and crime rate in various countries, but there are much more differences between those countries than gun control only. Having such data set we could try to find correlation between crime rate and language or religion and we could get significat results...

In the paralell discussion http://skeptics.stackexchange.com/questions/976/is-gun-control-effective one of the participants have posted in his comment the link: http://en.wikipedia.org/wiki/List_of_countries_by_firearm-related_death_rate with suggestion to compare Switzerland and Germany, which have similar culture and economic level, and the difference is gun control... But single example can't prove anything.

However, gun control is not 0-1 value. The ban is one thing, the control is the other. This is another factor that should be taken into account. How much arms there are per 100 people - this doesn't tell WHO has this arms, hunters, criminals or maybe collectors.

My personal suggestion: look carefully on the list http://en.wikipedia.org/wiki/List_of_countries_by_firearm-related_death_rate. Look which countries have the lowest crime rate. If you want to reduce violence, you should copy the concepts from those countries, but copying single concept such as gun control is too little, the whole system should be taken into account.

-
"If you want to reduce violence, you should copy the concepts from those countries, but copying single concept such as gun control is too little, the whole system should be taken into account." The problem there is that trying to do everything except gun control might not necessarily work either. – Joe Z. Dec 28 '12 at 22:52
-1: This is a non-answer. It doesn't attempt to answer the question, but rather addresses concerns with my post, which should be handled in chat or in the comments. – Javid Jamae Jan 2 '13 at 1:06