researched argument draft essay

Regulating Guns among Young Adults

Gary Kleck1

Received: 19 January 2019 /Accepted: 1 April 2019 / Published online: 29 April 2019 # Southern Criminal Justice Association 2019

Abstract This paper reports the results of two studies of the impact of gun control measures on violent criminal behavior among persons age 18 to 20. The first study assessed the impact of state bans on gun carrying among persons age 18 to 20 on rates of violent crime committed by persons in that age group. The research used a state-level cross- sectional weighted least squares analysis of murder, robbery, and aggravated assault rates in 2000, controlling for possible confounding variables. The results indicate no significant effect of these carry bans on any of the three violent crime rates. The second study was a longitudinal analysis performed to evaluate the impact of a single previ- ously unstudied element of the federal Gun Control Act of 1968 – its ban on the purchase of handguns by persons aged 18 to 20. The analysis tested whether the share of arrests for three violent crime types trended downward, or less strongly upward, after the law went into effect, controlling for trends in the share of the population in this age group. Results indicate that there was no impact of this ban on the 18-to-20 year-old share of arrests for homicide, robbery, or aggravated assault.

Keywords Gun control . Young adults . Violent crime . Gun Control Act of 1968

Introduction

This paper performs two tests of the general hypothesis that gun control measures specific to young adults reduce violent crime within that group. This focus is especially important because criminal violence in America is at its highest in the young adult ages. For example, national arrest data for 2017 indicate that the single ages with the highest rates of arrest for murder were 18, 19, and 20 years old (U.S. FBI, 2019, Table 38). Society therefore has an especially strong interest in reducing violence among young

American Journal of Criminal Justice (2019) 44:689–704 https://doi.org/10.1007/s12103-019-09476-6

* Gary Kleck gkleck@fsu.edu

1 College of Criminology and Criminal Justice, Florida State University, Tallahassee, FL 32306-1127, USA

adults, and it would seem reasonable to focus violence-prevention efforts especially heavily on this group.

There are also good reasons to believe that gun control measures would be especially likely to be effective if they focused on young adults. First, precisely because violent criminal behavior is more common within this group, any one instance of denying a gun to such a person is more likely to prevent a violent crime with a gun. Second, young adults are more likely to use firearms when they commit violence crimes (Department of the Treasury and Department of Justice, 1999, pp. 7, 9). Consistent with this fact, there is some macro-level evidence that gun prevalence increases homicide rates among persons age 18 to 24 (Parker et al., 2011, p. 510), even though the technically strongest evidence does not show a net positive effect of gun prevalence on the overall (all-ages) homicide rate (Kovandzic, Schaffer, & Kleck, 2013). (Parker et al., however, failed to model the possible two-way relationship between gun prevalence and homicide, so the positive association they observed may reflect the effect of homicide on gun prevalence, rather than the reverse - Kovandzic et al., 2013). Third, it seems likely that, compared to older offenders, younger criminals are less likely to have acquired the personal contacts and knowledge that would enable them to evade restrictions on guns. Further, their higher rate of involvement in violence may make it especially easy to detect any crime- reducing effects of gun controls in this group. In contrast, among children and older people, there is little violent crime to be prevented, so even a large percentage decrease might be too small in absolute terms to be reliably detected.

There is little variation across jurisdictions in the regulation of firearms among persons under the age of 18. They have been categorically forbidden from purchasing firearms of any kind from licensed gun dealers under federal law since the Gun Control Act of 1968 (GCA) (18 USC Section 922(b)(1)), and are ineligible everywhere except in Vermont to get the state carry permits that are required in most states to lawfully carry a concealed firearm off of the possessor’s property (Giffords Law Center, 2019). This lack of variation makes it difficult to detect any effects of gun control measures applying to those under age 18. In contrast, the strictness of gun control laws pertaining to young adults age 18–20 has shown considerably more variation, both over time and across states. People in this age group can legally carry guns in public places in some states but not in others. This variation provides opportunities for researchers to estimate the effect of gun regulations on violence within this specific high-violence subpopulation.

The research reported here therefore focuses specifically on controls applying to this age group. In particular, it is aimed at two kinds of gun control measures: (1) state laws that forbid concealed carrying of firearms among 18–20 year-olds, and (2) a federal ban on the purchase of handguns by persons age 18–20, enacted as part of the 1968 GCA. We evaluate the first measure using a state-level cross-sectional design, and evaluate the second measure using a national-level longitudinal design.

Study 1 – The Effect of State Bans on Concealed Gun Carrying by Persons Age 18–20

Each of the 50 states has different regulations on the carrying of concealed weapons, and in particular the states differ regarding the minimum age at which persons become eligible to lawfully carry concealed weapons. Variation was especially high prior to the

690 American Journal of Criminal Justice (2019) 44:689–704

Supreme Court’s 2008 Heller and 2010 McDonald decisions, and subsequent court decisions interpreting them, which had the effect of striking down state laws that categorically forbade all concealed carrying. Prior to 2000, seven states forbade concealed carrying of firearms altogether, for all ages (see the Appendix for a listing of states and supporting statutory citations) a number of courts banned nearly all gun carrying among civilians. One state (Vermont) set 16 as the minimum age to carry concealed guns in public places, 14 states required a minimum age of 18, another 21 states set 21 as the minimum age, and one state (Oklahama) set the minimum age at 23.

The states also differ in how they implement age restrictions. Some require permits for concealed carrying, and specify a minimum age to get the permit, while others require no permit but nevertheless specify a minimum age for carrying. Still others do not specify an explicit age minimum, but rather grant discretion to authorities such as county sheriffs to assess the degree to which a prospective carrier is a Bsuitable person.^ As of 1999, 29 states forbade the carrying of concealed weapons by 18-to-20 year olds (including those that banned carrying by persons of any age) while the remaining 21 allowed it, either because (1) people this age could get a carry permit, or (2) the state neither required a carry permit nor stated a minimum age to carry (Appendix).

Study 1 Methods

Our strategy for testing for an effect of state carry laws on violent crime takes advantage of the age-related element of these laws. The provisions concerning minimum age either prohibit or allow carrying specifically among 18-to-20 year olds, so if they affect the frequency of violent crime, they should do so primarily by affecting violent crime rates among 18-to-20 year olds. There are no available data that directly measure the violent criminal behavior of Americans of specific age groups, but an approximation can be derived from data on persons arrested for crimes. Arrest data by age available from the Uniform Crime Reporting (UCR) program of the Federal Bureau of Investi- gation (FBI) are specific enough to establish national totals of arrests of 18-to-20 year olds. We analyzed three specific violent crimes that are the only crime types committed in significant numbers with firearms: (1) murder and nonnegligent manslaughter (hereafter denoted Bhomicide^ for brevity’s sake), (2) robbery, and (3) aggravated assault. About 73% of homicides, 41% of robberies, and 28% of aggravated assaults in 2017 were committed with firearms (U. S. Federal Bureau of Investigation, 2019).

We therefore estimated state rates of violent crime among 18-to-20 year olds by multiplying state crime rates concerning offenses by all ages by the fraction of persons arrested for a given crime type who were 18-to-20 years old. For example, we estimated the Alabama homicide rate among 18-to-20 year olds by multiplying the Alabama homicide rate in 2000 (7.4 homicides per 100,000 population) by the fraction of homicide arrests that 18-to-20 year olds accounted for in Alabama in 2000–2002 (0.1916), which yields an estimated homicide rate of 1.42 homicides committed by 18-to-20 year-olds per 100,000 resident population of all ages. Three years of arrest data, covering 2000–2002, were used to estimate the fraction of crimes committed by persons 18 to 20 because in smaller states there are too few arrests in any 1 year for any one type of violent crime (especially homicide) to provide a stable estimate of the 18- to-20 year old share of arrests.

American Journal of Criminal Justice (2019) 44:689–704 691

In the following regression analyses, three types of age-specific violent crime rates were analyzed: (1) the rate of homicide committed by 18-to-20 year-olds, (2) the rate of robbery committed by 18-to-20 year-olds, and (3) the rate of aggravated assault committed by 18-to- 20 year-olds. The analysis focused on the year 2000 rather than more recent years because there was far more variation in the strictness of state controls on carrying firearms than in later years. In more recent times there has been little meaningful variation in the strictness of carry laws. For example, in 1999 there were seven states that completely banned carrying guns in public (see Appendix), but by 2010, there were none (Gifford Law Center, 2019). State controls over gun carrying became lenient in all but a handful of states. By 2018, at least 30 states had adopted lenient shall-issue carry laws, which make it easy for noncrim- inal adults to get carry permits, while another 12 states had eliminated the requirement for a permit to carry concealed firearms altogether (Gifford Law Center, 2019). One might speculate that results pertaining to 2000 do not apply to more recent years. Certainly, the levels of the variables in our models, and which states have which laws, changes over time, but we are not aware of any evidence that the causal effects of gun laws have changed since 2000.

The present study focused on 2000 rather than any intercensal year (e.g., 2001– 2009) because the Census provides, for years when the dienniel Census is conducted, data on a wealth of crime-related variables that should be statistically controlled in order to isolate the effect of carry law provisions. A list of these other variables appears in Table 1, and the sources of data for the analysis are reported in the Appendix.

States were coded according to their carry laws as of 1999, rather than 2000, to make sure that they pertained to a time point prior to 2000–2002, the period to which the crime rates pertained. This makes it less likely that any relationship found between carry law provisions and crime rates is due to an effect of crime rates on the enactment or amending of carry law provisions, and more likely that the relationship reflects an effect of the carry law provisions on crime rates.

The statistical procedure used to estimate the relationship between carry law age provisions and crime rates was weighted least squares (WLS) regression. This procedure gives differing weights to each of the states, such that states with larger populations are given greater weight. This has the effect of reducing heteroscedasticity. The weight used in these analyses was the square root of the state’s resident population in 2000.

Only 48 of the 50 states could be included in the analysis because FBI arrest data by age were not available for all of the years from 2000 to 2002 for Florida or Wisconsin (they were missing for 2000 for Wisconsin and for all 3 years for Florida). It should be noted that it is not essential for present purposes that all arrests be reported to the FBI, since we make no use of the absolute frequency of arrests. Rather, it is only necessary that the 18-to-20 year old share of those arrests reported to the FBI be approximately the same as the 18-to-20 year old share of all arrests for a given type of violent crime, whether reported to the FBI or not.

Study 1 Findings

Table 2 displays the WLS estimates. Each column of numbers shows the estimated coefficients of a regression equation pertaining to one of the three dependent variables, and each row pertains to a particular independent variable that might affect these outcome variables. Each cell of the table shows three numbers. The topmost number is the WLS

692 American Journal of Criminal Justice (2019) 44:689–704

coefficient. The middle number is the ratio of the coefficient over its standard error, sometimes called a t-ratio, used to test whether the regression coefficient is significantly different from zero. The bottom number is the one-tailed statistical significance of the coefficient. A significance under .05 (5%) is generally considered to be statistically significant, i.e. not likely to be the product of random chance. Thus, if the bottom number in a given cell is lower than .05, it means that the associated independent variable shown in the row heading has a statistically significant association with the dependent or outcome variable shown in the column heading. Our primary interest is in the estimates shown in the first row of the table, those pertaining to the association of the 1999 state carry laws’ provisions regarding minimum age for concealed carrying with crime rates among 18–20 year olds on the estimated rate of violent crime among 18–20 year-olds. Note that CARRY1820 reflects whether persons age 18 to 20 are allowed to legally carry guns, so a positive coefficient for this variable means that violent crime rates are higher where these young adults may legally carry firearms.

Table 1 Variables in the analysis (as of 2000 unless otherwise indicated)*

Variable name

Variable description Mean Standard deviation

MURD1820 Murder, nonnegligent manslaughters among 18-to-20 year olds per 100,000 population

1.02 0.51

ROB1820 Robberies among 18-to-20 year olds per 100,000 population 117.66 61.76

ASLT1820 Aggravated assaults among 18-to-20 year olds per 100,000 population 35.84 14.27

CARRY1820 State law allows 18–20 carry, as of 1999 (1 = yes, 0 = no) 0.41 0.50

%POP18–20 % of resident population age 18–20 4.41 0.40

POVERTY % families under the poverty line, 1999 12.17 3.03

BLACK % African-American 11.55 9.19

HISPANIC % Hispanic

PRISONRS State, federal prisoners per 100,000 resident population 423.26 167.19

DIVORCE Divorces per 100,000 resident population 4.22 1.17

URBAN % population residing in urban areas 74.93 13.96

DENSITY Persons per square mile 0.21 0.25

FOREIGN % foreign-born 8.60 6.82

INSTATE % population born in same state 61.02 12.05

LIVLONE % of population that lives alone 9.73 1.04

MARRIED % of population married, living with spouse 51.81 2.60

MOVERS % of population age 5+ that changed residence, 1995–2000 45.61 5.16

OLDPCT % population age 65 or older 12.29 1.72

POLICE Sworn full-time police officers per 10,000 population 20.20 4.33

SOUTH State is in the South (former slave-owning state) 0.32 0.47

WEST State is in the West Census region 0.24 0.43

UNEMPLOY % civilian labor force unemployed 5.66 1.01

VETERAN Military veterans per 1000 population, 1999 89.16 9.76

* Means and standard deviations are based on weighted data, and cover the 48 nonmissing states used in the regression analyses – that is, they exclude Florida and Wisconsin

American Journal of Criminal Justice (2019) 44:689–704 693

Each regression equation also controlled for additional variables that affect crime rates and that might also be associated with carry law age provisions. Every equation controlled for the share of the state population that was in the 18–20 age range, regardless of whether this variable was significantly related to the dependent variable, because it was thought to be self-evident that the number of people in this age range would affect the number of crimes committed by persons in this age range. It turned out to make no difference to the key results whether or not this variable was included in the equations. The rest of the control variables were included in the equation only if they showed a significant relationship with the dependent variable. A generous standard of significance, 20%, was used in deciding whether to retain variables in the equation, to reduce the chances that a potentially important control variable was omitted. A much larger set of potential control variables were tested but found to have no significant association (even at a generous 20% level of significance) with any of the crime rates. All the variables shown in Table 1 that do not appear in Table 2 fall into this category.

Table 2 Weighted least-squares estimates – the effect of state bans on gun carrying by18-to-20 year-olds on rates of violence crime by that age group

Coefficient/ratio of coefficient over standard error/1-tailed significance

Independent variable Dependent variable

MURD1820 ROB1820 ASLT1820

CARRY1820 −0.116 1.006 −1.624 −1.112 0.342 −0.453 .136 .367 .326

%POP18–20 −0.052 −5.405 −0.496 −0.363 −1.361 −0.102 .360 .090 .460

POVERTY 0.044 1.269 1.434

2.301 2.366 2.191

.014 .012 .034

BLACK 0.033 0.611 0.665

5.542 3.699 3.298

.000 .000 .002

URBAN 0.006 0.427 0.357

1.561 3.987 2.715

.063 .000 .004

OLDPCT −0.069 −2.252 .015

Constant 0.767 −4.992 −13.158 Adjusted R2 .575 .447 .341

All variables shown in Table 1 that do not appear in Table 2 were found to have no significant (p < .20) association with any of the three violent crime rates studied and were therefore omitted from the crime rate models

694 American Journal of Criminal Justice (2019) 44:689–704

Collinearity among the independent variables was not a problem – all tolerances were over 0.7 in all equations.

The estimates shown in the first row (labeled CARRY1820) of the first three columns of Table 2 estimate the effect of allowing 18–20 year olds to legally carry concealed weapons. The estimates indicate that this policy is not significantly related to rates of any of the three violent crimes that are often committed with guns (homicide, robbery, aggravated assault) among persons age 18 to 20. Indeed, the associations were negative for two of the three violent crime types, indicating that, other things being equal, states allowing 18–20 year old carry have less homicide and aggravated assault among 18–20 year olds than states forbidding it, though not to a statistically significant degree. The homicide finding is consistent with the results of a recent study of the effect of state gun laws on homicide rates, which found no significant effect of a requirement that people be at least 21 years old to legally possess a firearm (Siegel, Pahn, Xuan, Fleegler, & Hemenway, 2019).

Study 1 Discussion and Conclusions

The analysis of state crime and arrest data indicates that provisions in state law prevailing in 1999 that allowed lawful concealed carrying of weapons among 18–20 year olds did not increase rates of murder, robbery, or aggravated assault within that age group. One partial explanation may be that states granting carry permits to persons under age 21 do so only for persons without criminal convictions or other predictors of violent crime, so legal carrying increased only among persons unlikely to commit violent crimes – even within a relatively high-violence age segment of the population. Another explanation, consistent with research on the frequency and prevalence of defensive gun use (Kleck, 2001a, b), is that the deterrent and defensive effects of gun carrying and defensive use among crime victims and prospective victims had crime-reducing effects that counterbalanced any crime-increasing effects of young adults carrying concealed guns. Alternatively, these restrictions may have failed to even achieve their proximate goal of reducing gun carrying in the target age group, because young adults carried guns illegally. In any case, the evidence indicates that allowing lawful concealed carrying of weapons among 18–20 year olds appears to have no net effect on rates of murder, robbery, or aggravated assault committed by people in this age group.

This finding fits well with complementary research done by Thomas Marvell (2001). Although he did not analyze age-specific crime rates as is done in the present paper, his fixed-effects panel analysis of state crime rates in the period 1968–1999 indicated that neither state laws banning juvenile gun possession nor a 1994 federal ban had any measurable net effect on violent crime rates (for all ages), including rates of homicide, robbery, and aggravated assault. This combination of studies suggests that banning juvenile gun possession or carrying does not, on net, reduce violent crime, and, conversely, that allowing lawful gun carrying does not, on net, increase violent crime.

Study 2 – The Effect of the 1968 Federal Ban on Handgun Purchases by Persons Age 18–20

The Gun Control Act of 1968 (hereafter GCA) was a major piece of federal gun control legislation that had many elements, one of which imposed a new restriction specifically

American Journal of Criminal Justice (2019) 44:689–704 695

applying to persons age 18 to 20. It provided that, while such persons could, like persons age 21 or over, lawfully purchase long guns (rifles and shotguns) from licensed gun dealers, those 18–20 were legally forbidden from buying handguns. Prior to this law, many states imposed age-based restrictions. We are not aware of any compilations of gun laws detailed enough to indicate how many states allowed handgun purchases by persons age 18–20 before 1968, but we do know that as of 2018, only 17 states and the District of Columbia required a minimum age of 21 to purchase handguns (Giffords Law Center, 2019). That is, as far of 2018 state law is concerned, people age 18 to 20 can lawfully buy handguns in 33 states. We assume that many states also allowed handgun purchases by 18–20 year-olds before 1968. Further, until 1968 no federal law prohibited handgun purchases by 18-to-20 year olds. Thus, the GCA introduced a new restriction on handgun buying by young adults.

The intent of the GCA was to reduce gun violence, partly by helping to restrict access to firearms among some subsets of the population. The introduction of this new restriction, applying specifically to 18-to-20 year olds, should directly affect only this age group, and should therefore have its largest effect on violent crime committed by persons age 18 to 20.

Study 2 Methods

The logic of the following analysis is simple. If the new handgun purchase restrictions of the GCA applying to 18-to-20 year olds was effective, the fraction of arrests for violent crimes that 18-to-20 year olds accounted for should have declined after the GCA went into effect late in 1968, since only this age group was newly subject to the age-based handgun purchase ban. Therefore, we analyzed arrest data for the United States for each year from 1963 (5 years before the GCA went into effect) to 1973 (5 years after the GCA), for trends in three measures: (1) the percent of arrests for murders and nonnegligent manslaughters (hereafter Bhomicides^) that 18-to-20 year olds accounted for, (2) the percent of arrests for robbery that 18-to-20 year olds accounted for, and (3) the percent of arrests for aggravated assault that 18-to-20 year olds accounted for. As with Study 1, these three offense types were studied because they are the only types of violent crime that involve handguns to any significant degree. Arrest data do not allow analysts to determine if the crime for which the person was arrested involved a gun, or more specifically a handgun, so it is not possible to directly measure trends in age patterns of persons committing, or arrested for, crimes in which handguns were used.

It should be emphasized that, unlike in Study 1, we did not analyze estimated per capita rates of crime by 18-to-20 year-olds in Study 2. Instead, we analyzed the estimated share of violent crimes (regardless of their number) committed by this subpopulation. This strategy strengthens our ability to isolate the effect of changes in age-specific gun control measures by reducing the need to control for numerous possible confounders. The alternative of analyzing per capita rates of violent crime in this age group would be afflicted by a far more severe need to control for possible confounding variables, since virtually any variable that could affect violent criminal behavior in the population as a whole could also affect violent criminal behavior among those age 18-to-20. Isolating the impact of the GCA’s new ban would require control- ling for all the other variables having causal effects among people of all ages. In

696 American Journal of Criminal Justice (2019) 44:689–704

contrast, only factors specific to persons 18–20 are likely to affect the percent of crimes committed by those age 18–20.

Violent crime (and arrests for violent crime) increased after 1968 but by itself this says nothing definitive about the impact of the GCA as a whole or of its new restrictions concerning 18-to-20 year olds, since violence increased in all age groups, and had already been increasing well before the GCAwent into effect. Thus, we need to introduce a feature into the research design that takes account of preexisting trends in violent crime, and thus in arrests for violent crime. We do this by measuring the average trend in the 18-to-20 year old share of arrests in the years immediately before the GCA went into effect in late 1968 (i.e., in 1963–1968) and comparing this with the average trend in the years immediately after the GCA (i.e., 1969–1973), excluding 1968 because part of it was before the GCA went into effect and part was after. If the relevant provisions of the GCA were effective, the post-1968 trends should be either downward or at least be less steeply upward than they had been before the GCA. Thus, the preexisting upward trend in the 18-to-20 year old share of violent crime arrests should have at least slowed, and possibly even reversed itself after 1968.

One other methodological feature was also essential to the analysis. The 18-to- 20 year old share of arrests could change even if the GCA’s relevant provisions had no effects, merely because the 18-to-20 year old share of the population changed. More specifically, the previously increasing 18-to-20 year old share of arrests could slow in its rate of increase if the previously increasing 18-to-20 year old share of the population slowed in its rate of increase. This is not a mere hypothetical – this is precisely what happened around 1968. Figure 1 shows that the 18-to-20 year-old share of the population (represented by the height of the line) was sharply increasing up until 1968, but then decreased in 1968 and thereafter increased much more slowly than it had before 1968. This means that if an analyst did not control for trends in the 18-to-

Fig. 1 Trends in the percent of the U. S. population that was age 18 to 20, 1963–1973

American Journal of Criminal Justice (2019) 44:689–704 697

20 year old share of the population, the effect of this shift in the age structure of the population would be confused with the effect of the GCA.

We therefore performed a linear regression analysis for each of our arrest outcome measures, in which the dependent variable was the 18-to-20 year old share of arrests for a given type of violent crime (murder, robbery, or aggravated assault), the year was the main independent variable, and the 18-to-20 year old share of the population was a control variable. The analysis was performed twice for each violent crime type, once for the years 1963–1968 (pre-GCA years) and once for the years 1969–1973 (post- GCA years). The coefficient for the YEAR variable represents the average annual change in the 18-to-20 year old share of arrests over the time period studied. If it is positive, it means that the 18-to-20 year old share of arrests for the target violent crime was increasing during that period, and if it is negative, it indicates that the share was decreasing. The larger the coefficient, the stronger the trend was. Thus, if the GCA was effective, the coefficient for YEAR in the post-GCA period should be either a smaller positive number than the same coefficient for the pre-GCA period (indicating a weakening of the pre-GCA upward trend), or the post-GCA coefficient might even be negative, indicating that the trend had reversed itself from an upward trend to a downward trend.

Study 2 Findings on the Impact of the GCA

Table 3 displays the key results from the regression analyses, showing the YEAR coefficients for the pre-GCA period and the post-GCA period, for each of the three arrest outcome measures. The upper panel (Panel A) shows the potentially misleading results obtained when there are no controls for the 18-to-20 year old share of the population, and the lower panel (Panel B) shows the corresponding results when a control for this variable was introduced.

Table 3 Changes in the trend in the 18-to-20 year old share of violent crime arrests, from the 1963–1967 (pre- GCA) period to the 1969–1973 (post-GCA) period

Panel A. No control for percent of the population age 18–20

Dependent variable

18-to-20 year-old share Coefficient for year Change from 1963 to 1968

of arrests for … 1963–1968 1969–1973 to 1969–1973

Murder .654 −.060 −0.714 Robbery .834 −.290 −1.124 Aggravated Assault .580 −.020 −0.600 Panel B. Percent of the population age 18–20 controlled

Dependent variable

18-to-20 year-old share Coefficient for year Change from 1963 to 1968

of arrests for … 1963–1968 1969–1973 to 1969–1973

Murder .410 3.417 +3.007

Robbery .840 2.651 +1.811

Aggravated Assault .525 1.649 +1.124

698 American Journal of Criminal Justice (2019) 44:689–704

The Panel A results initially indicated that trends in the 18-to-20 year old share of violent crime arrests moved in a desirable direction after 1968 – the previous upward trend prevailing prior to 1968 leveled off or even declined slightly after 1968. The Panel B results, however, indicate that this change in trends was at least partly due to the changing trends in the 18-to-20 year old share of the population documented in Fig. 1. Once that factor was statistically controlled, the positive coefficients for the YEAR variable for the pre-GCA period, which indicated an upward trend in the 18-to-20 year old share of violent crime arrests before 1968, became an even larger positive coefficient in the post-GCA period, indicating that the upward trend in the 18-to- 20 year old share of violent crime arrests became even stronger after 1968, despite the new federal ban on handgun purchases by persons age 18 to 20. That is, once one takes account of trends in the percent of the population age 18 to 20, it becomes apparent that upward trend in the share of arrests claimed by 18-to-20 year olds became stronger after 1968.

Study 2 Discussion and Conclusions

The federal ban on 18-to-20 year olds purchasing handguns from licensed dealers introduced in 1969 does not appear to have reduced the 18-to-20 year old share of violent crime. It is this specific age group whose violent behavior should have been most influenced if the purchase ban was effective, so the ban was apparently ineffective in reducing criminal violence. These results specific to young adults comport with research on the impact of the GCA on violent crime in the population as a whole. Magaddino and Medoff (1984) found no significant effect of the GCA on the U.S. homicide rate as a whole, i.e. for all ages. Zimring (1975) did not test the effect of the GCA on national crime rates, but did find that it failed to achieve the intermediate goals of reducing interstate movement of guns (pp. 181, 191) or decreasing the share of violent crimes committed with handguns (p. 172).

The focus on national-level data leaves open the possibility that the handgun purchase ban reduced young adult involvement in violence in some parts of the country, but not others. Since our results indicated no net effect for the nation as a whole, however, violence-reducing effects in some places could exist only if banning handgun purchases had violence-increasing effects in other places.

The apparent lack of impact of the handgun purchase ban on young adult involve- ment in violent crime does not mean that other elements of the GCA could not have had any beneficial effects. Nevertheless, the evidence indicates that the law as a whole apparently did not have any net effect on the homicide rate and did not reduce the share of the nation’s violent crimes committed with guns (Magaddino and Medoff 1984; Zimring 1975).

Why did the new ban on handgun buying not reduce violent crime among the young adults who should have been affected? One obvious explanation would be that they bought handguns from other sources besides licensed dealers, such as friends or relatives. They could have acquired handguns illegally, e.g., by stealing them or by using persons age 21 or older to act as Bstraw purchasers^ on their behalf (Wright & Rossi, 1986). Youth could also gain possession of handguns by borrowing their parents’ firearms, with or without their knowledge. It is even possible that the ban did reduce handgun acquisition in this group, but that this did not reduce the youths’

American Journal of Criminal Justice (2019) 44:689–704 699

involvement in violent crime. Criminally inclined young adults could substitute long guns like shotguns or rifles, or weapons other than firearms, to use in crimes, or commit offenses without using weapons of any kind - unsurprising possibilities given that most rapes, aggravated assaults, and robberies are in fact committed without firearms (U.S. Federal Bureau of Investigation, 2019, Table 19; U.S. Bureau of Justice Statistics., 2010, Table 66).

Overall Conclusions and Policy Implications

Neither study reported herein found any violent crime-reducing effects of gun control measures aimed specifically at young adults, notwithstanding the strong a priori reasons to expect that such effects would be strongest and easiest to detect within this population. These findings reduce any optimism one might have about the benefits of gun control restrictions directly specifically at young adults. Recent reviews of research on the impact of gun control laws as a whole likewise indicate that, with few exceptions, existing gun laws have no detectable net reducing effect on violent crime (Kleck, Kovandzic, & Bellows, 2016; Rand Corporation, 2018). For example, Rand searched for evidence bearing on the effects of 14 types of gun policies (excluding Stand-Your-Ground laws, which are not gun policies) with reference to eight violent crime outcomes, and found evidence bearing on 26 of the possible effects of policies on crime. Among these 26, the authors failed to find either Bstrong^ or Bmoderate^ support for a violence-reducing effect of any gun policy on any violent crime outcome, with a single exception – moderate support for the effect of prohibitions associated with mental illness. (The review also found moderate support for background checks on firearms homicides, but only Blimited^ support for an effect on total homicides - p. 304). For all the other possible effects, the authors rated the evidence as Binconclusive.^

Particularly relevant to the present report, the Rand authors found no strong or moderate support for an impact of age restrictions on purchasing or possessing guns - evidence regarding effects on homicide was rated as Binconclusive.^ Similarly, Kleck et al. (2016) failed to find even weak support for violence-reducing effects of bans on possession or purchase of guns by minors. Thus, in this research context, the present report’s null findings concerning restrictions on purchase or carrying of guns among persons age 18 to 20 are unsurprising.

This does not mean that no gun control measures can reduce violent crime. Even the generally unsupportive Rand review found some evidence that background checks and prohibitions associated with mental illness may reduce homicide. Likewise, Kleck et al. (2016, p. 508) found strong support for an effect of gun owner licensing (whose central element is background checks) on homicide, and weaker support for an assault- reducing effect of prohibition of gun possession by mentally ill persons. Further, there could be gun regulations that have never been implemented, or implemented but not evaluated, that might prove effective. Nevertheless, the full body of empirical evidence accumulated so far largely indicates that most gun control policies heretofore imple- mented, whether directed at young adults or the population as a whole, do not have any measurable net effect on violent crime.

700 American Journal of Criminal Justice (2019) 44:689–704

Appendix

Data Sources

State crime rate data

U.S. Dept. of Justice, Federal Bureau of Investigation. Crime in the United States - Uniform Crime Reports, 2000. Washington, D. C.: U. S. Government Printing Office.

State arrest data by age, 2000–2002 (available from ICPSR website at http://www. icpsr.umich.edu/icpsrweb/ICPSR/access/index.jsp):

U.S. Dept. of Justice, Federal Bureau of Investigation. Uniform Crime Reporting Program Data [United States]: Arrests By Age, Sex, And Race, 2000 [Computer file]. ICPSR03443-v2. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [producer and distributor], 2006-10-27.

U.S. Dept. of Justice, Federal Bureau of Investigation. Uniform Crime Reporting Program Data [United States]: Arrests By Age, Sex, And Race, 2001 [Computer file]. Compiled by the U.S. Dept. of Justice, Federal Bureau of Investigation. ICPSR03760- v2. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [producer and distributor], 2006-09-21.

U.S. Dept. of Justice, Federal Bureau of Investigation. Uniform Crime Reporting Program Data [United States]: Arrests By Age, Sex, And Race, 2002 [Computer file]. ICPSR04443-v2. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [producer and distributor], 2007-03-21.

State and Federal Prisoners

U.S. Bureau of Justice Statistics. 2001. Prisoners in 2000. BJS Bulletin NCJ 118207 at: http://bjs.ojp.usdoj.gov/content/pub/pdf/p00.pdf

Sworn police full-time employment

U.S. Department of Justice, Bureau of Justice Statistics. (1998). Directory of Law Enforcement Agencies, 1996: [UNITED STATES] [Computer file]. Con- ducted by U.S. Dept. of Commerce, Bureau of the Census. ICPSR ed. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [pro- ducer and distributor], 1998.

Divorces

U.S. National Center for Health Statistics. BProvisional Tables on Births, Mar- riages, Divorces, and Deaths 1998-2000^. National Vital Statistics Reports, Vol- ume 49, number 6, available at http://www.cdc.gov/nchs/data/nvsr/nvsr49 /49_06_02_03.pdf.

Population estimates by age: U.S. Bureau of the Census website at http://www. census.gov/popest/archives/pre-1980/.

http://www.census.gov/popest/states/asrh/files/SC-EST2009-AGESEX-RES.csv

American Journal of Criminal Justice (2019) 44:689–704 701

All other variables

U. S. Bureau of the Census. BCensus 2000 Briefs and Special Reports^ website at http://www.census.gov/population/www/cen2000/briefs.html.

Classification of the States as to Minimum Age for Lawful Carrying of Concealed Weapons, as of 1999.

No Concealed Carrying Allowed: IL, KS, MO, NE, NM, OH, WI. Minimum Age 23: OK. Minimum Age 21: AK, AZ, AR, FL, GA, HA, KY, LA, MA, MI, NV, NC, OR, PA,

RI, SC, TN, TX, UT, VA, WA. No express minimum age: AL, CA, CO, CT, NH, NY. Minimum age 18: DE, ID, IN, IA, ME, MD, MI, MN, MT, NJ, ND, SD, WV, WY. Minimum age 16: VT. State Law Provisions on Minimum Age for Concealed Weapons Carry as of 1999:

1. Alabama. Code of Ala. § 13A-11-75 (1999) 2. Alaska. Alaska Stat. § 18.65.705(1) (1999) 3. Arizona. Ariz. Rev. Stat. § 13–3112(E)(2) (1999) 4. Arkansas. Ark. Stat. Ann. § 5–73-309(1)(B) (1999) 5. California. Cal. Pen. Code § 12,050 (1999) 6. Colorado. Col. Rev. Stat. § 18–12-105.1 (1999) 7. Connecticut. Conn. Gen. Stat. § 29–28 (2000) 8. Delaware. 1 Del. Code § 701; 11 Del. Code § 1441(a) (1999) 9. Florida. Fla. Stat. § 790.06(2)(b) (1999) 10. Georgia. Official Code Ga. Ann. § 16–11-129(b)(1) (1999) 11. Hawaii. Haw. Rev. State § 134–9(a) (1999) 12. Idaho. Idaho Code §§ 3302(1)(l), (11) (1999) 13. Illinois. 720 Ill. Compiled Stat. Ann. § 5/24–1(4) (1999) 14. Indiana. Burns Ind. Code Ann. § 35–47–2-3(f)(2) (1999) 15. Iowa. Iowa Code § 724.8(1) (1999) 16. Kansas. Kansas Stat. Ann. § 21–4201(a)(4) (1999) 17. Kentucky. Kentucky Rev. Stat. § 237.110(2)(b) (2000) 18. Louisiana. La. Rev. Stat. § 40:1379.3(C)(4) (1999) 19. Maine. 25 Maine Rev. Stat. § 2003(1)(A) (1999) 20. Maryland. Md. Ann. Code art. 27, § 36E(a)(1) (1999) 21. Massachusetts. Mass. Ann. Laws ch. 140, § 131(d)(iv) (1999) 22. Michigan. Mich. Code Laws § 28.426 (1999) 23. Minnesota. Minn. Stat. §§ 624.713(1)(a), 624.714(5)(a) (1999) 24. Mississippi. Miss. Code Ann. § 45–9-101(2)(b) (1999) 25. Missouri. Rev. Stat. Mo. § 571.030 (1999) 26. Montana. Mont. Code Ann. § 45–8-321(1) (1999) 27. Nebraska. Rev. Stat. Neb. § 28–1202 (1999) 28. Nevada. Nev. Rev. Stat. Ann. § 202.3657(2)(b) (1999) 29. New Hampshire. N.H. Rev. Stat. Ann. § 159:6 (1999) 30. New Jersey. N.J. Stat. §§ 2C:58–3(c)(4), 2C:58–4(c) (1999) 31. New Mexico. N.M. Stat. § 30–7-2 (1999) 32. New York. N.Y. CLS Penal § 400.00(1) (1999)

702 American Journal of Criminal Justice (2019) 44:689–704

33. North Carolina. N.C. Gen. Stat. § 14–415.12(a)(2) (1999) 34. North Dakota. N.D. Cent. Code §§ 62.1–02-01(4), 62.1–04-03 (1999) 35. Ohio. Ohio Rev. Code Ann. § 2923.12(A) (1999) 36. Oklahoma. 21 Okl. St. § 1290.9(3) (1999) 37. Oregon. Oregon Rev. Stat. § 166.291(1)(b) (1999) 38. Pennsylvania. 18 Pa. C.S. § 6109(B) (1999) 39. Rhode Island. R.I. Gen. Laws § 11–47-11(a) (1999) 40. South Carolina. S.C. Code Ann. § 23–31-215(A) (1999) 41. South Dakota. S.D. Codified Laws § 23–7-7.1(1) (1999) 42. Tennessee. Tenn. Code Ann. § 39-17-1351(b) (1999) 43. Texas. Tex. Gov’t Code 411.172(a)(2) (1999) 44. Utah. Utah Code Ann. § 53–5-704(1) (1999) 45. Vermont. 13 Vermont Stat. Ann. § 4008 (1999) 46. Virginia. Va. Code Ann. § 18.2–308(D) (1999) 47. Washington. Rev. Code Wash. § 9.41.070(1)(c) (1999) 48. West Virginia. W. Va. Code § 61–7-4(a)(3) (1999) 49. Wisconsin. Wis. Stat. § 941.23 (1999) 50. Wyoming. Wyo. State. § 6–8-104(j) (1999)

References

Department of the Treasury and Department of Justice. 1999. Gun Crime in the Age Group 18–20. https://permanent-access-gpo-gov.proxy.lib.fsu.edu/lps20136/report.pdf. Accessed 1-9-19.

Giffords Law Center. 2019. BConcealed carry .̂ https://lawcenter.giffords.org/ gun-laws/policy-areas/guns-in- public/concealed-carry/. Accessed 10 Jan 2019.

Kleck, G. (2001a). The frequency of defensive gun use. In G. Kleck & D. B. Kates (Eds.), Armed (pp. 213– 283). Amherst, NY: Prometheus Books.

Kleck, G. (2001b). The nature and effectiveness of owning, carrying, and using guns for self-protection. In G. Kleck & D. B. Kates (Eds.), Armed (pp. 285–342). Amherst, NY: Prometheus Books.

Kleck, G., Kovandzic, T., & Bellows, J. (2016). Does gun control reduce violent crime? Criminal Justice Review, 41, 488–513.

Kovandzic, T., Schaffer, M. E., & Kleck, G. (2013). Estimating the causal effect of gun prevalence on homicide rates: A local average treatment effect approach. Journal of Quantitative Criminology, 28(4), 477–541.

Magaddino, J. P., & Medoff, M. H. (1984). An empirical analysis of federal and state firearm control laws. In D. B. Kates, Jr., (Ed.), Firearms and violence: issues of public policy (pp.225–258). Cambridge, Mass.: Ballinger.

Marvell, T. (2001). The impact of banning juvenile gun possession. Journal of Law and Economics, 44, 691– 713.

Parker, R. N., Williams, K. R., McCaffree, K. J., Acensio, E. K., Browne, A., Strom, K. J., & Barrick, K. (2011). Alcohol availability and youth homicide in the 91 largest US cities, 1984-2006. Drug and Alcohol Review, 30, 505–514.

Rand Corporation. 2018. The Science of Gun Policy: A Critical Synthesis of Research Evidence on the Effects of Gun Policies in the United States. https://www.rand.org/pubs/research_reports/RR2088.html. Accessed 10 Jan 2019.

Siegel, M., Pahn, M., Xuan, Z., Fleegler, E., and Hemenway, D. 2019. The impact of state firearm laws on homicide and suicide deaths in the USA, 1991-2016: A panel study. Journal of General Internal Medicine, Online First article.

U. S. Federal Bureau of Investigation. 2019. Uniform Crime Reports, 2017. Available online at https://ucr.fbi. gov/crime-in-the-u.s/2017/crime-in-the-u.s.-2017/topic-pages/tables/table-38. Accessed 1–9-19.

American Journal of Criminal Justice (2019) 44:689–704 703

U.S. Bureau of Justice Statistics. 2010. Criminal Victimization in the United States, 2008 –Statistical Tables, table 66. At https://www.bjs.gov/content/pub/pdf/cvus08.pdf. Accessed 10 Jan 2019.

Wright, J., & Rossi, P. (1986). Armed and Considered Dangerous. New York: Aldine de Gruyter. Zimring, F. E. (1975). Firearms and federal law: the Gun Control Act of 1968. Journal of Legal Studies, 4,

133–198.

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Gary Kleck is the Emeritus David J. Bordua Professor of Criminology and Criminal Justice at Florida State University. His research has focused on the topics of the impact of firearms and gun control on violence, deterrence, crime control, and violence. He is the author of Point Blank: Guns and Violence in America, which won the 1993 Michael J. Hindelang Award of the American Society of Criminology. He also wrote Targeting Guns (1997) and, with Don B. Kates, Jr., The Great American Gun Debate (1997) and Armed (2001), and, with Brion Sever, Punishment and Crime (2017). His articles have been published in the American Socio- logical Review, American Journal of Sociology, Social Forces, Criminology, Journal of Criminal Law and Criminology, Law & Society Review, Journal of Research in Crime and Delinquency, and many other journals. He has testified before Congress and state legislatures on gun control issues and his work has been cited by the U.S. Supreme Court. He has served as a consultant to the National Research Council, National Academy of Sciences Panel on the Understanding and Prevention of Violence, as a member of the U.S. Sentencing Commission’s Drugs-Violence Task Force, and a member of the National Research Council Committee on Priorities for a Public Health Research Agenda to Reduce the Threat of Firearm-related Violence. His most recent book is Punishment and Crime, with Brion Sever, and was published in October 2016.

704 American Journal of Criminal Justice (2019) 44:689–704

American Journal of Criminal Justice is a copyright of Springer, 2019. All Rights Reserved.