К оглавлению
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 
102 103 104 105 106 107 108 109 110 111 112 113 114 115 

Safety technologies have often been suggested as an alternative means

of preventing injury and crime. Locking technology might be used to limit

who can use a particular firearm. Protection technology might be used to

shield vulnerable persons or reduce the lethality of weaponry. Sensor and

tracking technology might be used to detect concealed weapons, provide

situational awareness for law enforcement, detect lost or stolen firearms,

limit when or where firearms can be discharged, or identify firearms that

have been discharged. To varying degrees, these different classes of techFIREARM


nologies are all being developed or considered by the National Institute of

Justice, the Office of Science and Technology, and other public and private


The potential of technology can be especially alluring. If widely adopted

and effective, safety technologies may alter the rates of gun ownership,

discharge, and mortality, as well as, more generally, the markets for weaponry

and injury. The actual effects of a particular safety device on violence

and injury, however, are difficult to predict. Even if perfectly reliable, technology

that serves to reduce injury among some groups may lead to increased

deviance or risk among others (Viscusi, 1992; Violence Policy Center,

1998; Leonardatos et al., 2001).

Many persuasive arguments have been made about the benefits and

costs of different firearms safety technologies. Despite the rhetoric, however,

there is almost no research that evaluates the efficacy and cost-effectiveness

of different interventions. The numerous arguments on the potential

benefits and costs of technology are largely speculative.

Locking Technology

To illustrate both the complexities of the issue and the lack of evidence,

it is useful to consider what is known about locking devices, perhaps the

most widely debated, studied, funded, and utilized firearms safety technology.

From simple trigger locks and gun safes to more sophisticated personalized

and “smart” guns, the promise of this technology is to reduce the

unauthorized transfer and use of firearms.2 Unauthorized transfers occur in

households, for example, from a parent to a child, in seizures from victims

to assailants, in thefts from residences, vehicles, and commercial places, and

in illicit transfers on the secondary market.

Much of the interest in locking technologies stems from the desire to

decrease the number of injuries and fatalities involving children. Children

under the age of 18 are not, in general, legally allowed to possess a handgun.

Yet each year, hundreds of children are fatally shot or injured in

firearms accidents and suicides. Juveniles also use handguns in criminal

activities, including the inner-city gang wars associated with the steep rise

in the juvenile homicide rate during the late 1980s and the highly publicized

1See the Office of Science and Technology web page,

sciencetech/welcome.html, for more details.

2Basic safety technologies have been around and widely used for over a century. Smith and

Wesson, for example, manufactured more than 500,000 guns with grip safeties between 1886

and 1940 (Teret and Culross, 2002). Mechanical locks are available commercially at negligible

cost. More sophisticated personalized guns, however, are either not yet developed or not

widely distributed.

mass school shootings in which, in many cases, the assailants obtained

firearms from their own homes (National Research Council and Institute of

Medicine, 2002).

While much of the attention and legislation regarding gun locks has

focused on reducing juvenile fatalities, these locking technologies may also

impact broader classes of unauthorized possession and discharge. The National

Institute of Justice has been particularly interested in the potential of

these technologies for reducing the handful of fatalities that occur each year

when police officers are fatally shot with their own firearm. More generally,

certain types of locking systems may decrease injuries that result from

firearms seizures, theft, and illegal transfers on the secondary market.3

While the specific numbers are unknown, the majority of criminals do not

obtain handguns via licensed dealers, and a large fraction of violent handgun

crimes are committed by proscribed users (see Chapter 5; Wright and

Rossi, 1986; Pacific Research Institute for Public Policy 1997; Cook and

Braga, 2001).

Locking technologies may also cause unintended injuries. In particular,

locking devices may compromise the ability of authorized users to defend

themselves. A lock may fail entirely or may take too much time for the

weapon to be of use. In fact, Wirsbinski (2001) and Weiss (1996), in

reviewing the engineering design of the different locks for the Sandia National

Laboratories, concluded that the existing personalized locking technologies

did not meet the reliability standards required for on-duty law

enforcement officers.4

The interaction between gun safety technology and the behavior of users

may also lessen the effectiveness of locking technologies. At the most basic

level, authorized users may not lock their guns and unauthorized users may

design ways to disable locks, access unlocked guns, or use different weaponry.

Safety technology may also lead to less cautious behavior around firearms:

authorized users may be careless in storing weapons, juveniles familiar

with locked guns may not be cautious around unlocked guns, and so forth.

Finally, these technologies may create new markets for firearms among consumers

who otherwise would not be inclined to own a gun.

3Presumably, for locks to deter illegal transfers in the secondary market, the key must be

maintained by a third party—for example, the authorized dealer—rather than the owner of

the gun (Cook and Leitzel, 2002). This may be possible with some of the automated biomechanical

technologies being developed (e.g., fingerprint technology) but may be more difficult

with many of the manual technologies.

4 Wirsbinski (2001) and Weiss (1996), and the New Jersey Institute of Technology (2001)

evaluated the reliability of different locking technologies in laboratory settings. A workshop

report of the National Academy of Engineering (2003) summarizes some of the key technological

and practical barriers to developing personalized handguns.

To evaluate the effect of locking technologies on injury, a number of

researchers have laid out conceptual models linking technology interventions

to injury. These models suggest that the efficacy of personalization

technology depends on the type and reliability of the technology, the extent

to which these technologies are integrated into the stock of firearms, and

the behavioral response of consumers and producers of firearms. Different

sets of assumptions about the nature of these factors lead to different

qualitative conclusions about the efficacy of safety technologies. Assuming

they are unreliable, not widely used, or result in unintended behavioral

responses, many conclude that locking devices may increase injury (see, for

example, Violence Policy Center, 1998; Leonardatos et al., 2001). Others,

under different sets of assumptions, conclude that these technologies may

decrease crime and injury (see, for example, Cook and Leitzel, 2002; Teret

and Culross, 2002). It is not known, however, which assumptions are

correct. Thus, without credible empirical evidence, the realized effects of

different safety technologies are impossible to assess.

In the absence of direct empirical evidence, a number of researchers

have appealed to the lessons learned from other product safety innovations

and legislation, especially automobile safety technologies. These analogies,

however, ultimately do not answer the question at hand—namely, how

firearms safety technologies impact injury. While a review of the product

safety literature is beyond the scope of this report, it seems clear that (1) the

efficacy of product safety innovations varies by product and (2) there are

ongoing and controversial debates on the effects of some of the most wellknown

innovations, including seat belts. In fact, scientists have long warned

that safety innovations can lead to offsetting behavioral responses. Auto

safety innovations may lead to increased recklessness (Peltzman, 1975);

child safety caps may lead to unsafe storage behaviors (Viscusi, 1984); and

low-tar cigarettes may lead to increased smoking (Benowitz et al., 1983;

Institute of Medicine, 2001). There is hardly consensus on the effects of

product safety innovations on injury. Furthermore, in contrast to most

other consumer products, firearms safety technology invariably reduces the

effectiveness of the weapon. Firearms, after all, are designed to injure.

Other safety devices do not generally impair the primary function of the

product. Seatbelts, for example, do not reduce the effectiveness of automobiles,

and safety caps do not reduce the effectiveness of medication.

Child Access Prevention Laws

Child access prevention (CAP) laws, sometimes referred to as “safe

storage” or “gun owner responsibility” laws, make owners liable if a child

uses an unlocked firearm. The first of these of laws was passed in Florida in

1989, and at least 17 other states and several cities have adopted similar

provisions (Brady Campaign, 2002). State laws differ in what age children

are covered, ranging from 12 to 18, in the penalty imposed, from civil to

criminal liability, and what it means to safely store a gun. Effectively,

however, CAP laws require gun owners with children to lock their firearms.

Two papers evaluate the effects of CAP laws on accidents, suicide, and

crime. Lott and Whitley (2002), using the same basic data and methods as

in the Lott and Mustard (1997) analysis of right-to-carry laws (see Chapter

6), conclude that CAP laws have no discernible effect on juvenile accidents

or suicide, but they do result in a substantial increase in violent and property

crime. In sharp contrast, Cummings et al. (1997) find that CAP laws

reduce accidents and may reduce suicide and homicide among youth as

well, although these are imprecisely estimated.5 They conclude that during

the five-year period from 1990 to 1994, these statutes prevented approximately

39 deaths of young children, and another 216 children might have

lived had these laws been in effect in all states.

It is difficult to explain the conflicting estimates. Using state-level injury

statistics, both analyses rely on interrupted-time-series designs that assume,

after controlling for observed factors, that CAP laws were the only notable

change in the environment. The formal models and specifications differ.

Cummings et al. (1997) estimates a negative binomial count model with

fixed state and time effects but an otherwise parsimonious specification of

control variables. Lott and Whitley (2002) use Tobit and log-linear models

with fixed state and time effects and a rich specification of 36 control

variables to account for variation in demographics (e.g., age, race, income,

education) and firearms laws. Lott and Whitley also evaluate different

outcomes and assess the sensitivity of their findings more generally.

In both studies, it is unreasonable to assume that CAP laws were the

only notable event that may have affected firearms related injury and crime.

Time-series variation in crime is thought to be a highly complex process

that depends on numerous economic, demographic, and social factors.

Moreover, CAP laws and other local firearms legislation may be adopted in

response to the local variation in the outcomes of interest. For example, a

sharp increase in accidental injuries and fatalities spurred a Florida legislature

that had previously turned down similar legislation to adopt the CAP

law in 1989 (Morgan, 1989). If the 1988-1989 wave of accidental injuries

would have naturally regressed back to some steady-state level, any observed

correlations between Florida’s CAP law and the injury rate would be

spurious. Even if all the other factors that may influence injury or crime are

time invariant, the dynamics that connect the law to the outcomes of inter-

5Webster and Starnes (2000), updating the Cummings et al. (1997) analysis, draw similar


est are likely to be complex. The impact of a CAP law adopted in a particular

place and time will almost certainly depend on how the law is enforced

and advertised over time, how this affects storage practices over time, and

how this in turn affects injury and crime over time.

The problems with this research are compounded due to the lack of

detailed data on the law, on ownership and storage, and on outcomes. The

data do not reveal information on the storage practices of particular households

or in the aggregate or how the laws are implemented and enforced.

The data do not link ownership to outcomes. Rather, we simply observe

aggregate correlations between injury and crime, and CAP law legislation

(see the discussion of ecological associations in Chapter 7). It is not known

whether the observed associations reflect changes in the behavior of firearms

owners, whether changes in accidents are associated with juvenile

shooters, or whether changes in victimization are associated with crimes

committed in households. A final data related concern is the possibility of

changes in reporting behavior. Webster and Starnes (2000) suggest that

whether a death is coded as an accident, a suicide, or a homicide is “likely

to vary across place and time.” If the coding behaviors change in response

to the legislation, for example, if after the law is passed accidental shootings

are more likely to be classified as suicides or homicides, then the observed

empirical results may be due to coding changes rather than the law.

Thus, in the committee’s view, until independent researchers can perform

an empirically based assessment of the potential statistical and data

related problems, the credibility of the existing research cannot be assessed.