F is for Freedom
Shedding Light on the Brady Campaign's Claims
Our goal is to protect you, your family and your community from gun deaths and injuries. In America, we make it too easy for dangerous people to get dangerous weapons. There are only a few federal gun control laws on the books, and even those have loopholes. This leads to senseless gun violence affecting tens of thousands.
I think we all agree that laws are sensible if they provide a benefit for society and if the infringement on the individual's freedom is outweighed by the benefit. As such, any law is a balance that needs to be individually examined whether its purported benefits in fact justify the infringement of personal freedoms that is the inevitable consequence of any law. Conversely, I also think we can agree that a law that does not provide a significant benefit for society, yet infringes on individual freedoms, must be repealed, or - better still - never passed.
The Brady Campaign proposes "sensible" gun laws to reduce gun violence. Referring to the quote above, the Brady Campaign claims that "there are only a few federal gun control laws on the books", and even those have loopholes. Apart from the fact that there are about 20,000 gun laws on the books (check your knowledge of these laws with this quiz), I think it is also sensible to look at the laws and their efficacy.
I am grateful for the Brady Campaign's thorough analysis of the gun laws of the individual states. The Brady Campaign uses a score system where every law restricting the freedom to keep and bear arms is awarded points. Brady's state gun law list can be accessed here. In earlier years, a grading system was used with grades ranging from A+, A (achieved by no state), A- (California and some other states) and so on until F (for example, Alabama or Wyoming). The 2002 grading sheet, as one example, can be accessed here. More recently, the Brady campaign moved to using a score system with 100 being the highest score that indicates the maximum infringement on the rights of people to keep and bear arms. Many web links still refer to the old A-F grading scheme. As a first step, we will therefore attempt to build a translation table from scores to grades, although the table will have some uncertainties (Table 1).
Table 1: Approximate translation of grades to score points
|10 or less||11 - 20||21 - 35||36 - 50||More than 50|
Although scores and grades are related, their synthesis from individual criteria is very different. In general, an A grade in any category is only achieved if the word verboten is involved. Conversely, an F grade is associated with the absence of regulation, for example, the absence of secondary (private) sales background checks.
For this purpose, we have examined crime data by state, available from the FBI's CIUS for the year 2008 with Brady scores for the states, also for 2008.
Clearly, the Brady hypothesis is that a higher score - indicative of a larger number of laws regulating gun ownership, gun purchases, and gun transfers - will be associated with a lower incidence of violent crime. The null hypothesis is that the law density as quantified by the Brady score is not significantly related to violent crime incidence. The significance level to accept the hypothesis was set at the widely accepted value of P=0.05. For comprehensiveness, we analyzed three types of crime. First and most importantly violent crime, defined as murder and nonnegligent manslaughter, forcible rape, robbery, and aggravated assault. The Brady campaign's first and foremost claim is that "common-sense" gun laws reduce violence, that is, violent crimes. Second, the murder and nonnegligent manslaughter component of violent crime was examined separately, because in this category, the offenders used firearms in 69% of the crimes. Any trend would therefore be most prominent in this category. Third, property crime was examined. Property crime includes the offenses of burglary, larceny-theft, motor vehicle theft, and arson. Property crime is interesting because of its relatively low firearm involvement.
Figure 1 shows the correlation between the Brady score and the different types of crime. Each of the dots represents one state, and its position in the coordinate system is determined by the state's Brady score and its crime rate. The correlation between Brady score and crime rate was examined by linear regression. None of the regression lines were statistically significant from zero (violent crime: R2=0.00014, P=0.93; murder: R2=0.0013, P=0.80; property crime: R2=0.053, P=0.11). In other words, the Brady score and crime rate are not correlated. The complete absence of any correlation is clearly visible in the cases of murder and violent crime. However, a slight negative correlation of the Brady score with property crime can be seen. Do gun laws reduce property crime? First, the negative slope is not significantly different from zero, that is, from a statistical perspective, the correlation is not representative and is not considered to be a valid phenomenon, but rather a random effect of data scatter. Second, we included property crime because of its low firearm involvement (as opposed to murder). If the Brady scores had any predictive value for crime, it would show predominantly in the murder correlation. The opposite is true.
We must conclude that those laws that form the basis of the Brady score do not influence crime rate. Although individual states should not be used in individual state-to-state comparisons (see "caution against ranking"), some examples corroborate our observation. California ranks highest in the Brady scoring system with a score of 79, yet its violent crime rate is above the national average. Both Louisiana and North Dakota have F-level scores of 2 and 4 Brady points, respectively. Yet, North Dakota has the lowest murder rate in the nation. It becomes evident that the presence of gun laws that form the basis of the Brady score do not have any measurable effect on crime, notably on violent crime.
Even more prominently, the Brady campaign ranks the states by their law density (i.e., number of gun laws and law strength). In 2008, California ranked first, whereas the two examples above (Louisiana and North Dakota) shared rank 44. The lowest rank (2 Brady score points and rank 49) was given to Kentucky. The Brady campaign asserts that more gun laws (i.e., a higher rank like California) leads to less gun violence. Analogous to Figure 1, we can now test this hypothesis and correlate the rank to the state's crime rate.
Figure 2 exhibits weak apparent trends towards lower crime rate in states with a higher ranking (more rigid gun laws). However, none of these trends are statistically significant with P=0.79 (violent crime), P=0.36 (murder) and P=0.26 (property crime). The hypothesis that a higher rank (more rigid gun laws) reduces crime and violence must therefore be rejected, and the null hypothesis that the Brady rank and crime rate are unrelated must be accepted.
In Figure 1, a conspicuous gap between the B and C grades can be seen. This gap lies between Hawaii (score 43, rank 8) and Illinois (score 28, rank 9). There are only eight states with scores above the gap (HI, RI, NY, CT, MA, MS, NJ, CA). If the claims of the Brady campaign has even the least bit of validity, we can certainly expect a significant difference in crime rates between the "top eight" listed above and the "bottom eight" with F-level scores at the low end. These states are OK, LA, KY, WV, ND, UT, MO, and AK with the highest level of freedom for gun owners. To test this hypothesis, we have averaged crime rates of the two groups of states and tested them for a statistical difference with the t-test.
Figure 3 shows box-and-whisker plots of the crime rates within each group. A box-and-whisker plot contains key statistical parameters. the central box represents the central 50% with the line inside the box representing the median value. The whiskers extend to the maximum and minimum value of each group. Although the median value seems to be slightly lowered in the less free states, none of the difference shows statistical significance (P-values of 0.8, 0.4, and 0.3 for violent crime, murder, and property crime, respectively). In fact, the state with the lowest crime rate of the "bottom eight" (i.e., most free) group is lower than the lowest crime rate in the "top eight" (i.e., least free) group. Conversely, the state with the highest crime rate is higher in the "bottom eight" than in the "top eight". These observations clearly indicate that random scatter is responsible for any apparent difference seen, and that the hypothesis that the stricter gun laws in the "top eight" group reduce violence and crime in comparison to the "bottom eight" group must be rejected. From this analysis, we must conclude that the stricter gun laws in the "top eight" group do nothing to reduce crime and violence.
The main conclusion that needs to be drawn from these data is that a large number of gun laws, or a number of strict gun laws, do nothing to reduce any form of crime, both violent crime and property crime. Most importantly, murder rate (with its high firearm involvement) is not reduced with more gun laws.
The analysis of these data allows to draw a number of important conclusions with respect to those laws that the Brady campaign calls "common-sense" gun laws:
Clearly, the answer is NO. Self-defense is a fundamental individual right. The standard of restricting individual rights without societal benefit is characteristic of totalitarian states and must be vehemently rejected in free nations.
Why then, would the Brady Campaign continue its campaign for more gun laws, if the effect is negligible and unmeasurable? We submit that the goal of anti-freedom groups like the Brady Campaign is total disarmament of the population, irrespective of the use of guns as sporting, hunting or self- defense instruments. This goal aligns with many ideologues and totalitarians. To achieve this goal, the Brady Campaign uses one important instrument: the publication of half-truths. We will demonstrate this trick to sway less-informed parts of the public with half-truths by using several recent examples.
The most striking finding from the study is that individuals in possession of a gun were 4.5 times more likely to be shot in an assault than those not in possession. Among gun assaults where the victim had at least some chance to resist, the adjusted odds ratio increased to 5.5.
An interesting statement, considering that apparently a victim with "at least some chance to resist" was at higher risk of being shot than a victim that had no chance to resist. Common-sense question: How does a victim who does not have a chance to resist differ from an unarmed victim? The original study alludes to some higher-risk behavior of armed victims, but does not support this allusion statistically. The analysis of the study and the detection of its fundamental statistical flaws is quite difficult. This is likely the main reason why the Brady Campaign picked this study for their anti-gun pitch. However, thorough analysis of the original publication does reveal the flaws. Compare the above Brady quote to the original abstract (emphasis added):
Results. After adjustment, individuals in possession of a gun were 4.46 (P < 0.05) times more likely to be shot in an assault than those not in possession. Among gun assaults where the victim had at least some chance to resist, this adjusted odds ratio increased to 5.45 (P < 0.05).
Let us begin with the more obvious limitations of the study. Only gun assaults were considered, a situation in which the perpetrator already has a considerable advantage over the victim. Other forms of violent attacks are not considered. However, gun assaults are only 13% of all violent-crime offenses (see here for a detailed analysis). The defensive use of a gun to defend against a non-gun attack (83% of all cases) is not examined in this study, a fact that may severely bias the statistics. In fact, the authors of the study contend that "successful defensive gun uses can and do occur", but challenge that such occurrences are likely to happen (study, page 2037). Unfortunately, the authors do not provide an estimate to what extent the limitation to gun assaults biases the picture of overall defensive gun use.
The second fallacy is less obvious, but it is frequently found in less-scientific studies with a preconceived outcome. This fallacy is the confusion between correlation and causality. Let us take a look at Figure 4. Here, the state GDP growth between 1997 and 2007 (source: American Legislative Exchange Council) was compared between the eight states with the lowest gun law density ("most free") and the eight states with the most repressive gun laws ("least free"). Surprise! The "most free" states have a higher GDP growth by almost 4.5%, and this difference is statistically significant (P<0.05). Does that mean that fewer gun laws mean more GDP growth and more wealth? Not necessarily. This is the correlation/ causality confusion. With enough effort, some data can be found that significantly correlates with whatever criteria I want. For the record, I note that the linear correlation between GDP growth and Brady rank and score for all states is not significant (i.e., does not exist). We could, of course, decide not to publish the lack of correlation with 50 states and only show the top 8 vs. bottom 8 comparison (Figure 4), triumphantly announcing that fewer gun laws create wealth. And right here is my error: States with fewer gun laws may also be states with higher economic freedom, which in turn causes higher GDP growth, but this is a separate analysis. It is not acceptable to deduce a causality ("fewer gun laws cause higher GDP growth") from a statistically significant corelation. To provide a more extreme example, we could examine the incidence of diabetes and relate it to households that have insulin. We certainly get a very high correlation. Can we conclude that insulin causes diabetes? No! Yet this is exactly the fallacy of this study. There is a correlation between people carrying a defensive gun and people getting shot. Can we conclude the causality that people get shot because they carry? No! The causal relationship may be in the other direction, indirect, or nonexistent. The conclusions of this study ("On average, guns did not protect those who possessed them from being shot in an assault") are flawed beause of the fallacy we just explained.
The above considerations are still not likely to account for the surprisingly high risk that an armed victim is about five times more likely of getting shot than an unarmed victim. The third fallacy is arguably most difficult to detect, because a closer examination of the statistical methods employed in this study is required. The authors provide a meticulous break-down of the studied victims by situational and individual characteristics (Table 1 on Page 2036). They found that victims that got shot were younger, more likely of hispanic origin, more likely in a high-risk occupation (for example employees handling cash), had a higher number of prior arrests, and had a higher rate of alcohol or drug involvement. Furthermore, crimes where the victim got shot were almost twice as likely to occur in high drug-trafficking areas. Cases where the victim got shot were almost nine times more likely to occur outdoors than those cases where the victims successfully defended themselves. The authors consider these differences "confounding factors" and produced a statistical model that "adjusted" for those factors. Unfortunately, the authors provide only the comparison after "adjustment". Details of this "adjustment" are scant, and no information of the differences between groups before adjustment are given. It is therefore impossible to judge the effect of this "adjustment".
The challenge with any form of "adjustment", particularly any form of "adjustment" that is not well described, is that the adjustment can shift the data in either direction. In fact, adjustment methods are possible that allow to match the data to a desired outcome, and an intransparent presentation of the adjustment factors and methods effectively hides this bias. Could this have happened in the study in question?
Let us examine Table 2 (page 2037) of this study. As mentioned before, no un"adjusted" data are given, but there is partially adjusted data:
Table 2: Regression Results Showing the Association Between Gun Possession and Gun
Assault: Philadelphia, PA, 2003 to 2006
| Total Participants
(Cases and Controls), No.
| Full Models,
AOR (95% CI)
| Reduced Models,
AOR (95% CI)
|All gun assaults||1361||4.46 (1.16, 17.04)*||2.55 (1.00, 6.58)*|
|Fatal gun assaults||329||4.23 (1.19, 15.13)*||3.54 (1.18, 10.58)*|
|Gun assaults where the victim
had a chance to resist
|897||5.45 (1.01, 29.92)*||2.94 (1.01, 8.42)*|
In this context, AOR denotes the adjusted odds ratio and the values in parentheses are the 95% confidence interval. In the full model (middle column), data have been adjusted for all identified confounding factors. In the reduced model (right column), only adjustment for age, illicit drug involvement, being outdoors, and unemployment took place. Interesingly, the odds ratio is much lower (in fact, significantly lower with P≤0.05, denoted with a * symbol) with less adjustment in the reduced model than in the full model. The odds ratio for all gun assaults drops by a factor of 1.7 when less "adjustment" takes place. Without the detailed raw data and the detailed model, a conclusive proof is not possible, but Table 2 gives a strong indication that the high odds ratio is really an artefact created by the adjustment rather than a true observation. A sensitivity analysis (Table 3 on page 2038) shows an extremely high sensitivity (factors of more than 4) towards misrecorded gun possession. For example, if 10% of unarmed victims are randomly assigned to the group of armed victims, the adjusted odds ratio drops from 4.46 to 1.03. Clearly, this high sensitivity indicates a highly instable model. Assigning 10% of unarmed victims (who are claimed to have a 4.46-fold lower odds ratio of being shot) to the group of armed victims can - in a stable model - reduce the odds ratio by maximally 7% to a new odds ratio of 4.2. In the same sensitivity analysis, 10% of both armed and unarmed victims were randomly reassigned to the other group. In the "adjusted" model, the odds ratio fell from 4.46 to 1.78. This is another example of the hypersensitivity of the model to small variations (i.e., instability) that further corroborates the notion that the high odds ratio is more an artefact of the model than a real observation.
For those who are interested in data manipulation and data biasing by using, let's say, directed statistics, I strongly recommend the book "Damned Lies and Statistics: Untangling Numbers from the Media, Politicians, and Activists" by Joel Best (2001) and his more recent book "Stat Spotting - A Field Guide to Identifying Dubious Data" (2008). Those books help understand how statistics and statistical modeling can be used to give data a slant, and - most importantly - how to identify manipulative statistics.
This study itself is a good example of the half-truths that were mentioned above. Half of the truth, that is, the exact model and the adjustment parameters are not provided in this study. Because of this omission, the study cannot be validated by independent researchers and its validity must be doubted. It comes as no surprise that the Brady Campaign uses just one sentence of this study to support their claim that "gun ownership creates risks and having more guns correlates with more gun violence". Naturally, the Brady Campaign would not cite publications of the opposing view and attempt to find the reasons for the conflicting results. The half-truth seen at the Brady Campaign is the uncritical parroting of one study that supports its pre-conceived views without acknowledging that studies exist that support opposing views. Two examples for the latter are the books by Gray Kleck "Point Blank: Guns and Violence in America" (2005) and by John Lott "More Guns, Less Crime: Understanding Crime and Gun-Control Laws" (2000).
True - or, at least, half-true. This pistol uses a unique type of ammunition that more resembles tiny rifle rounds than typical pistol rounds. This type of ammunition is designated 5.7x28 mm. The projectile has a mass of 2.6 grams (SS195LF rounds) and a typical muzzle velocity of 576 meters per second. Its kinetic energy is 431 Joules. In fact, easy research (ref) reveals energies between 461 and 538 Joules, depending on the type of ammunition used. In comparison, the commonly-used .45ACP pistol round carries kinetic eneriges between 447 to over 700 Joules (ref). The kinetic energy is the primary determinant of a bullet's destructive force and therefore also its ability to penetrate body armor. Why is the higher-energy .45ACP never called cop-killer bullet, while the weaker 5.7x28 is?
The answer is fairly straightforward. FN Herstal offers the SS190AP round for the Five-seveN pistol. This ammuntion is designated by the BATFE as Armor Piercing (AP) ammunition, designed to penetrate body armor. Its sale to civilians is restricted, and BATFE mandates that FN Herstal stores the ammunition in a bonded warehouse and releases only upon sale to law-enforcement or military purchase order.
Cop-killer pistol? Only if you have cop-killer ammunition. And the SS190AP cop-killer ammunition is restricted to law-enforcement and military. Civilians can legally purchase the SS192 and SS196 sporting rounds which were found to not penetrate body armor. This finding has been published on the official ATF web site (local copy, accessed 11-26-2009).
Cop-killer pistol? Certainly not for civilians. However, this half-truth where a crucial piece of information (that is, the existence of armor-piercing rounds available to law-enforcement and military only) is deliberately omitted, serves to stoke unwarranted fears and brand an excellent target and sporting pistol with the ugly moniker "cop-killer pistol". A confusing, carefully worded muddle of true statements, half-truths and outright lies as found, for example, in the Huffington Post is frequently published by anti-freedom groups to convince the public of the need to introduce yet another element of prohibition. See here for more detailed information about the effects of the Five-seveN in the Fort Hood terrorist attack. Of primary interest is the alleged propensity of the high-velocity 5.7x28 round to over-penetrate and cause less traumatic wounds than, for example, the .40S&W or .45ACP hollow point cartridges that are favored by American law enforcement.
So why bother prohibiting a handgun that is too expensive for most civilians, poses less danger to victims than other handgun types, and is owned only by a small fraction of gun owners?
The strategy is simple. Anti-freedom groups have long since realized that the desired "big coup" of general civilian disarmament is difficult to achieve. Therefore, these groups revert to a long-term strategy where in successive small steps more and more freedom is eroded away - ideally in steps so small that the general public does not realize the erosion of their rights. Every time, anti-freedom groups successfully completed one step, they are ready for the next step in the direction of the ultimate goal: a defenseless, easily victimized populace.
True - or, at least, half-true. The 90% number comes from the statement of ATF Special Agent William Newell ( link to original document and local copy) who submitted
Firearms are now routinely being transported from the U.S. into Mexico in violation of both U.S. and Mexican law. In fact, 90% of the firearms recovered in Mexico, and which are then successfully traced, were determined to have originated from various sources within the continental U.S.
In the same document, we find that in Fiscal Year 2008, 2514 firearms seized in Mexico were traced to sources in Texas, Arizona and California. The remaining 47 States accounted for 1,053 traces in FY 2007. Let us assume that the differences between FY 2007 and 2008 are small. In this case, we can assume that approximately 3600 firearms were successfully traced in one year.
In a report by USA Today (local copy), a total of 29000 seized firearms was reported for the last two years, that is likely the years of 2007 and 2008. A rough estimate gives 3600 successfully traced firearms out of 14500 (in one year), which is about 25%. It can be assumed hat the other 75% are not even submitted for tracing in the USA.
A number of conflicting reports exist. Fox News reported:
In 2007-2008, according to ATF Special Agent William Newell, Mexico submitted 11,000 guns to the ATF for tracing. Close to 6,000 were successfully traced -- and of those, 90 percent -- 5,114 to be exact, according to testimony in Congress by William Hoover -- were found to have come from the U.S.
Factcheck.org estimates this number to be closer to 36%, but admits that the percentage of firearms traced to US origins is an elusive number. The Mexican newspaper El Universál (local copy) presents completely different numbers. According to this article, 31000 firearms were recovered between December 2005 and January 2009 from the cartel "Zetas" alone. This number corresponds to an estimated 10000 firearms confiscated per year between 2005 and 2009 from the Zetas. Further following this article, ATF estimates about 7 million firearms sold in the United States and smuggled into Mexico, contrasted to 5 million firearms that were recovered by the authorities in 2006 and 2007 combined. This ratio is again close to 25%. However, the disparity between the sources does not enhance their trustworthiness.
It appears to be extraordinarily difficult to present a valid number of firearms of US origin that the Mexican drug cartels use. There exist too many conflicting and contradictory statistics to obtain a percentage value of firearms confiscated in Mexico that can be traced to legal sales in the United States.
However, an inference can be made that this number must be fairly small. The El Universál article lists some of the firearms: M72 anti-tank missiles, RPG-7 rocket-propelled grenades, hand grenades (actually, about 750 hand grenades confiscated per year), MGL 37mm grenade launchers, 37mm and 40mm grenades, among other military arms. A related article in the Wall Street Journal, Activities of drug gangs at the Mexico/US border (local copy) confirms the use of heavy military weaponry. In particular, fully automatic weapons are mentioned. None of the military weapons listed above are freely available in the United States. Fully automatic weapons are restricted in the United States and cannot be simply purchased from the local gun store. Rifles with mounted grenade launchers are illegal, as are the grenades. It would appear that civilian-legal US firearms are of little interest to the Mexican drug cartels who actually focus on high-power military arms starting at the level of a fully automatic assault rifle (again: illegal in the US).
How convenient to be able to grab a number (those 90%) out of context and present it as shocking fact in a different context. Obviously, the news outlets did extremely poor research or used the value unchecked because it supported their bias (see commentary here). And anti-freedom groups like the Brady Campaign are all too happy to parrot these numbers in order to mislead the public and advance the anti-freedom agenda.
Laws are necessary that prevent firearm possession and firearm use by criminals. Law enforcement must have the ability to prosecute illegal firearm ownership and use in a similar manner that law enforcement must have the ability to prosecute criminals who do not use firearms.
We have seen that a large number of laws does not have any measurable effect on crime rate. We can conclude that the number of laws is sufficient. Even more, many states have too many laws that infringe on personal freedoms without reducing the crime rate. Many laws need to be re-evaluated, and many laws need to be repealed.
Anti-freedom groups, foremost the Brady Campaign, purport to help reduce gun violence and support measures that reduce gun violence. A close analysis reveals that the reduction of gun violence is not their primary goal, but rather the complete disarmament of the population. To pursue their real agenda, anti-freedom groups do not hesitate to resort to propaganda, half-truths and outright lies. To study the network of interconnected anti-freedom groups and their freedom-eroding agenda, I recommend the web site www.discoverthenetworks.org.
Based on the statistics, a large number of laws do not reduce violent crime. Analysis of the laws shows that states with very few gun laws do not have higher crime rates than those with excessive number of gun laws. Therefore, a good target for the allowable number of laws are those states that are F-rated by the Brady Campaign. Because F is for freedom.