Handgun Control Measures: How Much and from Whom?

Home , Guns (essay), Gun ownership, Long gun, Sporting Gun

GUNSIN AMERICA

Resultsof a Comprehensive NationalSurvey on FirearmsOwnership andUse

SummaryReport

PhilipJ. Cook JensLudwig

Police Foundation Washington,DC The Police Foundation is a privately funded, independent, nonprofit organization established by The Ford Foundation in 1970 and dedicated to supporting innovation and improvement in policing. Tl~ePolice Foundation's research findings are published as an information service. The research findings in this publication were supported under award number 93-IJ-CX-0017 from the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of view in this document are those of the authors and do not necessarily represent the official position of the U.S. Department of Justice or the Police Foundation. This is a summary report of The National Survey of Private Ownership of Firearms in the United States. The full technical report is available from the Police Foundation, 1001 22nd Street, N.W., Washington, D.C. 20037, telephone (202) 833-1460.

01996 by the Police Foundation All rights, including translation into other languages, reserved uncter the Universal Copyright Con- vention, the Berne Convention for the Protection of Literary and Artistic Works and the Interna- tional and Pan American Copyright Conventions. Permission to quote readily granted. ISBN 1-88461414-0

Library of Congress Catalog Card Number: 9746573 List of Tables ...... vii Acknowledgments ...... x Foreword ...... xi

I . Introduction ...... 1

11 . Methodology ...... 5 Introduction ...... 5 Survey Method ...... 5 Response Rates ...... 5 Characteristics of the NSPOF Population ...... 8 Gun Ownership Rates ...... 9

111 . Stocks and Flows of Guns in Private Hands ...... 13 Introduction ...... 13 Distribution and Characteristics of the Private Gun Stock ...... 13 Storage ...... 19 Acquisitions ...... 23 Disposals ...... 28

IV . Gun Ownership ...... 31 Introduction ...... 31 Patterns of Gun Ownership ...... 32 Reasons for Ownership ...... 36 Feelings of Safety with a Gun in the Home ...... 43 V. Gun Uses and Misuses ...... 45 Introduction ...... 45 Firearms Training ...... 45 Recreational Uses of Firearms ...... 46 Gun Accidents ...... 49 Gun Carrying ...... 53

VI . Defensive Gun Uses ...... 57 Introduction ...... 57 How Many Defensive Gun Uses? Previous Findings ...... 58 How Many Defensive Gun Uses? NSPOF-Based Estimates ...... 61 Resolving the NCVS-NSPOF Discrepancy ...... 68 Interpretation of Defensive Gun Use Estimates ...... 75

VII . Attitudes Toward Gun Control ...... 77 Support for Handgun Control Measures: How Much and from Whom? ...... 77 Gun Involvement and Support for Handgun Control ...... 80 Support for Handgun Control and Concern about Crime ...... 82

VIII . Conclusion ...... 85 References ...... 89 Table 2.1 Final NSPOF Sample Disposition ...... 7 Table 2.2 Two Definitions of Response Rate ...... 8

Table 2.3 Comparison of Personal Gun-Ownership Rates by Survey ...... 12

Table 3.1 Privately Owned Firearms ...... 14 Table 3.2 Gun Ownership Across Individuals and Households for Those Who Report Owning at Least One Gun ...... 15 Table 3.3 Individual Gun Collections ...... 16 Table 3.4 Magazine Capacity of Private Gun Stock ...... 17 Table 3.5 Size of Handguns in Private Gun Stock ...... 18 Table 3.6 Years Since Acquisition of Private Gun Stock ...... 19

Table 3.7 Storage Methods of Private Gun Stock ...... 20

Table 3.8 Storage Location of Private Gun Stock ...... 21 Table 3.9 Training Received by Gun Owners ...... 22 Table 3.10 Training and Storage Methods ...... 23 Table 3.11 Gun Acquisition Within the Past Two Years ...... 25

Table 3.12 Annual Average Gun Acquisitions. 1993-1994 ...... 26 Table 3.13 Retail Price of Guns Acquired Within the Past Two Years ...... 27 Table 3.14 Primary Versus Secondary Markets of Guns Acquired Within the Past Two Years ...... 28

Table 3.15 Gun Ownership at Time of Most Recent Acquisition ...... 29

Table 4.1 Gun Ownership Patterns: Demographic Characteristics ...... 33

Table 4.2 Gun Ownership Patterns: Socioeconomic Characteristics ...... 34

Table 4.3 Gun Ownership Patterns: Other Variables ...... 35

Table 4.4 Gun Ownership Patterns and Experiences with Crime ...... 37

Table 4.5 Gun Ownership by Cohort ...... 38

Table 4.6 Primary Reasons for Owning a Gun ...... 39

Table 4.7 Reasons for Not Owning a Gun ...... 40

Table 4.8 Current Ownership and Plans for Acquiring a Gun for Protection: Individual Characteristics ...... 41

Table 4.9 Current Ownership and Plans for Acquiring a Gun for Protection: Experience with Crime ...... 42

Table 4.10 Perceptions of Safety Among Unarmed Residents of Armed Households ...... 43

Table 5.1 Firearms Training ...... 46

Table 5.2 Population Estimates for Firearms Training ...... 47

Table 5.3 Gun Ownership and Training ...... 48

Table 5.4 Prevalence of Recreational Gun Use ...... 48

Table 5.5 Type of Gun Used by Recreational Gun Users ...... 49

Table 5.6 Characteristics of Recreational Gun Users Versus Others ...... 50

Table 5.7 Number of Households with Gun Accidents ...... 51

Table 5.8 Gun Accidents Reported in NSPOF ...... 52

Table 5.9 Number of Guns Carried Regularly ...... 54

Table 5.10 Prevalence of Gun Carrying ...... 55

Table 5.11 Frequency of Gun Carrying Not Related to Work ...... 56 Table 6.1 Defensive Gun Uses Reported for Preceding 12 Months ...... 62

Table 6.2 Defensive Gun Use, One- and Five-Year-Recall Period ...... 63 Table 6.3 Circurnstances of Defensive Gun Use Against Humans, Previous Five Years ...... 64

Table 6.4 Characteristics of Perpetrator in Defensive Gun Use Against Humans, Previous Five Years ...... 64

Table 6.5 Characteristics of Gun in Defensive Gun Use Against Humans, Previous Five Years ...... 65

Table 6.6 Circumstances and Outcomes of Defensive Gun Use Against Humans, Previous 5 Years ...... 66

Table 6.7 Demographic Characteristics of Gun Owners, Defensive Gun Users, and Non-Owners ...... 69

Table 6.8 Defensive Gun Use Compared to Total Crime Counts ...... 70

Table 6.9 Defensive Gun Use Reports: Lives Saved ...... 71

Table 7.1 Views on Gun Control by Sociodemographic Characteristics ...... 79 Views on Gun Control by Extent of Involvement with Guns ...... 81

Table 7.3 Views on Gun Control in Relation to Concerns about Crime ...... 83 Acknowledgments The authors thank Lois Mock at the National Institute of Justice and Earl Hamilton at the Police Foundation for their advice and support in completing this project, and to President Hubert Williams for providing Police Foundation support for publication of this report. Garen Wintemute provided helpful comments on an earlier draft. We also thank David Hemenway and the peer reviewers for their helpful comments on an earlier draft. We especially want to acknowledge the skilled assistance of Karen Price and Robin Turner at Duke, and Brian Komar and Ruth Almeida at Georgetown.

Guns in America The reality of policing in America in- against criminal attackers; and attitudes cludes dealing with citizens who pos- toward gun control regulation. sess firearms: there are about 200 million guns in private hands, according Among other things, the survey found to this survey and others. So huge is that handgun owners most often gave the domestic arsenal that American self-protection as their primary reason police must be aware that a firearm may for gun ownership, whereas owners of be at hand in any situation they en- long guns cited hunting or target shoot- counter. Tragically, in thousands of situ- ing as their main reason for owning a ations each year, the potential for in- gun. Furthermore, handguns are much jury or death by firearms is realized. more likely than long guns to be carried in public, and to be kept unlocked The National Survey of Private Owner- and loaded in households. ship of Firearms in the United States (NSPOF) provides the most compre- While there are enough guns in pri- hensive information to-date on vate hands to provide every adult in America's private stock of firearms. America with one, only 25 percent ac- Topics covered in the NSPOF include: tually own one and those who do usu- the size, composition, and ownership ally own several. Middle-aged, college- of the gun stock; how and why fire- educated residents of rural areas and arms are acquired; gun storage and small towns are most likely to own carrying; the defensive use of firearms guns, according to this survey. -- Police Foundation By the year 2003, according to the For the nation's police, the nexus of Centers for Disease Control (CDC), drugs and guns creates daily and the leading cause of death by injury deadly challenges to tlieir ability to con- in the U.S. will be from gunshots. The trol crime and ensure public safety. prevalence of firearm ownership and Civil debate and rational policy about use is of concern to law enforcement guns require that we arm ourselves personnel, health officials, educators, with the facts about the extent and na- policymakers, families, and commu- ture of gun ownership and use in nities all across America. The impact America. As with all of the work which that guns have on our lives contin- the Police Foundation has conducted for ues to generate passionate debate. over a quarter century, the results of Americans are ambivalent about The National Survey of Private Owner- guns: they fear them and at the same ship of Firearms presented in this report time they feel safer possessing them, are an effort toward informing the as reflected by the growing number debate. of states that have or are considering Hubert Williams concealed weapons-"right-to- President carryn-laws. February 1997

Guns in America The United States is unique among these concerns in mind. This nationally wealthy nations in its vast private arse- representative telephone survey was nal of firearms. The 200 million guns in conducted by Chilton Research Services ... we now private hands provide a sense of secu- during November and December 1994 have a rity for many of their owners. Yet they for the Police Foundation under the statistical also make a lethal contribution to an- sponsorship of the National Institute of other uniquely lavish feature of Ameri- Justice (NIJ). It consisted of telephone description of can life, criminal violence. In the debate interviews with a probability sample of the private over the proper regulation of gun com- over 2,500 adults, who were asked ques- gun stock by merce and use, it is concern about crime tions about gun ownership, training, that holds center stage. One side touts transactions, uses, and related attitudes. size, caliber, guns as an important deterrent and To an extent the NSPOF visits familiar value, bow source of protection; the other de- territory, well mapped by previous sur- the firearms nounces them for the damage they do. veys; this redundancy helps provide a It is clear that America's vast stock of check on its validity. But the NSPOF sur- were obtained, firearms has impact on our lives in a passes other surveys in its comprehen- and when and variety of ways and that it behooves us siveness. As a result, we now have a sta- to learn what we can about it. bow they are tistical description of the private gun stored and The 1994 National Survey of the Pri- stock by size, caliber, value, how the fire- vate Ownership of Firearms (NSPOF) arms were obtained, and when and how used. was designed and implemented with they are stored and used. The NSPOF does less well in providing ies about each respondent's gun collec- reliable information on two vital events tion, this survey provides the basis for a involving guns: defense against crime detailed portrait of the U.S. gun stock. and accidental wounding. Both of these This report includes information not just are rare occurrences, at least proportion- on the types of guns, but also on how ately, but figure prominently in the pub- they were acquired. Previous specula- lic debate over the costs and benefits of tion on the importance of the "second- gun ownership. As we shall discuss in ary" market in gun tsansactions may have this report, the estimated incidence of been exaggerated; we find that a major- these occurrences from this and similar ity of transactions involve federally li- surveys is subject to a large positive bias censed dealers. Our analysis of how guns and should not be taken serio~~sly. are stored documents an all-too-preva- ... handgun This summary report is presented in lent practice of keeping at least one gun ownersbip chapters, beginning with an assessment loaded and unlocked. Strangely, train- of the quality of the survey. The NSPOF ing does not appear to promote safer follows pretty was conducted in a fashion that meets storage. However, when we disaggre- much the the norms for an academic-quality, ran- gate formal training by source we find same pattern dom-digit-dial telephone survey. Initially that some training programs (such as we were concerned that the NSPOF- those offered by the National Safety as long gun based estimate of household gun preva- Council) are effective in reducing the ownersbip lence was just 35 percent, when the prevalence of unsafe storage practices. despite the conventional wisdom suggests that some- Chapter 4 switches focus from the guns weapons' very thing closer to 50 percent is correct. But to their owners, documenting the pat- the conventional wisdom may be out of terns and motivations for gun owner- different uses: date; our estimate is not far out of line ship. The most evident division is be- handguns are with other recent surveys. And the tween the sexes: men own most of the predominantly NSPOF estimate of 192 million guns in guns. About two-thirds of adult men private hands is close to other careful have owned a gun at some time in their kept for self- estimates. In our investigation, a new lives, and over 40 percent do so cur- defense, and version of the gender gap became evi- rently. The corresponding numbers for long guns for dent: inat-ried women are considerably woinen are far smaller. In other re- less likely to report a gun in the house- spects as well, gun ownership follows sport. hold than married men. Apparently, patterns of military service and rural wives are either ignorant of the firearins' sporting traditions. One interesting fea- presence or uncomfortable in discuss- ture is that handgun ownership follows ing them. We solved this problem by re- pretty much the same pattern as long stricting most of our analysis to the data gun ownership despite the weapons' on self-ownership. very different uses: handguns are pre- Chapter 3 presents results that are unique dominantly kept for self-defense, and to the NSPOF. Thanks to detailed inquir- long guns for sport. The woinen who

Guns in America do own guns are far more likely than lent crimes in the United States. These male owners to report protection as comparisons reveal gross inconsistencies. their primary motive. Adults who have The filndamental problem here is inher- been arrested for nontraffic offenses are ent to the task of estimating a rare event; more likely to own firearms than others. false positives tend to outnumber false negatives, producing a positive bias. Chapter 5 analyzes the uses of guns, Consistent with this explanation, a sub- beginning with the extent to which the stantial proportion of the defensive gun public (gun owning or not) has been use reports are in some way internally trained in the proper handling of guns. inconsistent, or otherwise don't make The primary sporting uses of guns are sense. We also note that respondents' hunting and target shooting, with about descriptions of their "defensive" gun uses 15 million people involved in each. We do not allow us to properly determine ... the estimate that 4 million people carry whether the incident, if accurately re- American guns for protection on tlie job and an- ported, was legal or appropriate, nor can other 10 million people carry guns, at we determine the respondent's culpa- public would least occasionally, for protection out- bility, if any, in the incident. like to see side of their work. All of this activity produces some accidental shootings, Finally, in chapter 7 we explore the views somewhat but so rarely as to preclude accurate of the respondents with respect to hand- more stringent estimation in this type of survey. gun regulation. As has been the case since pollsters started asking the ques- controls The dangers of relying upon telephone tion, the great majority of the American on gun surveys to estimate rare events are also public would like to see somewhat more highlighted in our attempts to estimate stringent controls on gun commerce and commerce the number of defensive gun uses use than currently exist in most jurisdic- and use than (DGUs), reported at some length in tions. On the other hand, only a minor- chapter 6 The NSPOF data offer the ity is willing to express support for actu- currently first opportunity to replicate the recent ally banning handgun possession. Tlie exist in most well-publicized survey finding that 2.5 gender gap is again much in evidence jurisdictions. million citizens use a gun defensively on these matters and so is the respon- each year. Since many defensive gun dents' concern about crime. uses may go unreported to law enforcement, these estimates cannot be veri- This collection of findings does not add fied by official records. up to any specific conclusion about appropriate regulation of guns, but However, the DGU estimate from the policy concerns influence our choice NSPOF-which is not inconsistent with of topics throughout. The result, we the 2.5 million figure--can be compared hope, is better understanding of the to well-known facts about the number causes and consequences of America's of firearms injuries, homicides, and vio- extensive involvement with guns.

Police Foundation Gunsin America Introduction are eligible under this method, and The NSPOF is a nationally representa- each telephone household in the U.S. tive telephone survey. The methods essentially has equal likelihood of be- employed by Chilton generally follow ing selected. Each selected household the norms of good practice for tele- was scheduled for an original call and phone surveys (Frey 1989). The main up to five follow-ups (Chilton 1995, 2- cause for concern is the low response 4). When a call was completed, the rate-more than 40 percent of the con- Chilton interviewer asked to speak with tacts with "eligible" telephone numbers the adult in the household who had could not be completed, or resulted in the most recent birthday. Since this a refusal. Hence the completed sample method randomizes the choice of re- may be somewhat unrepresentative of spondent from the adults living in the the U.S. population. However, for bet- household, the NSPOF is able to pro- ter or worse this response rate is not duce a probability sample of English- unusually low for surveys of this sort. or Spanish-speaking adults in the United States (Waksberg 1978). Survey Method The NSPOF employs a list-assisted ran- Response Rates dom-digit-dial sampling method, as The survey response rate1 is relevant discussed in Brick et al. (1995). House- to judging the accuracy of survey esti- holds with unlisted telephone numbers mates. The sample of completed interviews will be somewhat unrepresenta- about the representativeness of the tive if those who refuse to cooperate completed sample) is the number of tend to be different in relevant ways households that provided interviews or than those who are successfully inter- were willing to do so-that is, com- viewed. While we do not know pleted interviews plus those terminated whether those who refused to partici- because the sample quota had been pate in the NSPOF are more or less filled. Response-rate measures differ likely than the national average to, say, with respect to what is included in the own a gun, we cannot rule out that denominator. At a minimum, the de- possibility. The larger the group of nominator includes, in addition to tlie refusers in comparison with coopera- count of willing participants (the nu- tors, the larger is the likely magnitude merator), the count of refusals (3,618 mile we of "nonresponse bias." The response in the case of NSPOF). We believe it do not know rate was quite low in the NSPOF, and appropriate to include also the count hence a matter of concern. of telephone numbers in which a call whether those was never completed (4,724), since this The final sample disposition is pre- form of nonresponse may also produce refused sented in Table 2.1. Of the 29,917 tele- an unrepresentative sample. to participate phone numbers tliat were randomly selected, 32 percent were ineligible (not in the NSPOF Less clear-cut is what to do about the working or not residential). Of the other type of nonresponse-those cases are more or 20,302 telephone numbers in the sam- (3,059 in all) in which some member of pling frame, 6,333 contacts were ter- less likely than the ho~lsehold had been contacted and minated by Chilton before conducting the national been cooperative, but no interview had the interview because tlie initial re- actually been completed by the time the average to sponses indicated that the household survey ended despite one or more fol- was not needed to complete pre-es- ... own a gun, low up calls. These cases can be viewed tablished sampling quotas. These quo- we cannot as "cooperators" (because the fact that tas were defined for the NSPOF with there was a successful initial contact sug- rule out that respect to race and gun-ownership sta- gests at least the willingness to cooper- tus (Chilton 1995, 16). What remains possibility. ate), in which case they would be in- after netting out these cases is 13,969 cluded in both the numerator and tlie telephone numbers of households that denominator. Or they could be viewed are either known to be eligible for in- as non-respondents, since the chosen clusion or at least not known to be in- adult in the household proved some- eligible. Of these, 2,568 interviews were what difficult to contact. in which case completed. they should be included in the denomi- There is no single definition for "re- nator but not the numerator. We des- sponse rate" (Frey 1989). The appro- ignate them as "initial cooperators," priate numerator (given our concern and calculate the response rate in two

Gunsin America FinalNSPOF SampleDisposition

Total phonenumbers in sample

Phone numberseliminated from sample: Non-operational Not a household Language barrier (~on'tspeakEnglish or Spanish) Other ineligible Total

Phonenumbers insampling frame 20,302

Phone numberseliminated fromsampling sample Interviewsnot completed: Terminated(quota filled) 6,333 One or more callbacks,no interview 3,059 Unable to completecall 4,724 Refusals 3,618 Total 17,734

Total numberseliminated 27,349

Total completedinterviews 2,568

Source:Chilton Research Services (June 1995), The 1994 NationalStudy of Privote Ownership of Firearmsin the United States:Methodology Report. (p, 7). Note: "Unable to completecall" includes"no answerlbusy"and "answeringmachine." Of the callbacks that did not leadto an interview, 173were determinedto be eligiblefor the study, while eligibilitystatus was not knownfor 2,886. Of the refusals,527 weredetermined to be eligible,

ways, one with them included as "co- refuse to be interviewed or who are operators" and one with them unavailable to be interviewed may be included as nonrespondents. different from the population as a whole in relevant ways. Hence we urge If we include the "initial cooperators" caution in the interpretation of results in the numerator, the response rate is based on NSPOF data. On the other 59 percent; if we exclude them, the hand, there is no reason to believe that response rate is just 44 percent. In ei- this survey has a less representative ther case, there is clearly a possibility sa~lnple than other commercial tele- of nonresponse bias in estimates of phone surveys. For example, Heck and population parameters. Those who Gertz (1995) report a response rate of

Police Foundation TwoDefinitions of ResponseRate

(# Cooperators)+ (# lnitialCooperators) 1. (# Cooperators+ # lnitialCooperators +# Refusers+ # Nonrespondents)

(# Cooperators) *' (# Cooperators+# InitialCooperators + # Refusers+ # Nonrespondents)

where:

# Cooperators count of completed interviews + countof interviewsterminated by Chilton # lnitialCooperators countof initialcooperators # Refusers count of thosewho refused togive an interview # Nonrespondents count of those telephone numberswhere a call was never completed + countof initialcooperators

61 percent for their national survey of lated population projection weights2 for gun ownership and use, defined as the both households and persons. House- number of households willing to par- hold weights reflect the sampling quo- ticipate divided by the number of com- tas for gun ownersliip and race, and pleted calls. If we followed this prorp- include household income as a con- dure, our response rate would be at trol variable. Person weights use age, least as high as theirs. race, sex, education, number of adults in the household, and household income as control variables. Characteristics of the NSPOF Population While Chilton's weights did not account For various reasons, some households for the number of working telephone were more likely than others to be se- lines in each contacted household, lected into the final sample of com- NSPOF includes data on that item. In pleted interviews. For example, the the NSPOF sample, 351 respondents sampling procedure had the effect of (13.7 percent) reported the existence including a disproportionate number of of more than one telephone line. (Re- gun-owning households. To account sponses to this item were missing for for differences in the probability of an additional 80 survey participants.) selection in the NSPOF. Chilton calcu- In order to account for the greater like-

Gunsin America lihood that a household with two or cent during the 199Os, several points more telephone lines would be selected lower than during the 1970s and much to participate, we adjusted the Chilton lower than the Gallup Poll estimate. In weights by dividing through by the particular, the GSS estimate for 1994 number of working telephone lines in was just 41 percent. The Kleck-Gertz the respondent's household. Thus, a survey, conducted in 1993, produced household with two working telephone a still-lower estimate of gun ownership, numbers would receive one-half the 38 percent of households. Our best weight of an equivalent household that guess, then, is that the NSPOF estimate had one telephone line, as suggested understates the "truth," but only by by Sud~lnan(1976). We made an addi- about five percentage points, rather tional modification to ensure that the than the 15-point deficit implied by the sample projects to the entire U.S. adult conventional wisdom. NSPOF data population. What is the cause of this apparent un- indicate Gun Ownership Rates derestimate? One possibility is that the that just net effect of the various distortions in 35 percent. .. Estimates for Household the NSPOF sample is to underrepresent Gun Ownership gun owners. Alternatively, we offer of bousebolds The December 1993 Gallup Poll esti- three observations concerning the spe- owned a... mated that 49 percent of households cifics of how gun ownership is deter- gun in 1994. possess a gun, a result that affirms one mined in the NSPOF, any one of which of the seeming constants in American may lead to a small negative bias in life: the fraction of American house- the gun ownership estimate: holds owning a gun has remained at The NSPOF focuses on guns owned about half since polling on the subject by household members aged 18 began in 1959 (Kleck 1991). Given this and older. A l~ousehold that in- conventional wisdom, it is of consid- cludes a gun-owning adolescent erable concern that the NSPOF data in- but no gun-owning adult is counted dicate that just 35 percent (plus or mi- as a "no gun" household. However, nus l .3 percent) of households owned such households are apparently a (working) gun in 1994. We believe very rare. An NSPOF follow-up of that this estimate may be somewhat off 200 households in which the first- the mark, but not by much. The con- round respondent had indicated ventional wisdom appears out of date. that no adult owned a gun found The best of the available series on gun no cases in which a minor in the ownership is the General Social Sur- household owned a gun. Hence we vey (GSS), conducted by the National doubt that the focus on adults ac- Opinion Research Center. Its estimates counts for the low estimate of gun have been in the range of 40-43 per- ownership.

Police Foundation II The NSPOF was conducted by tele- in the household is chosen as the re- phone, whereas in the GSS respon- spondent. The individual who actually dents are interviewed face-to-face. owns the gun appears more willing (or The 6 percent of U.S. households able) than other adults in the house- that lack telephone service will be hold to report that the household excluded from the NSPOF sampling includes a gun. We document that in- frame; that only matters if teresting phenomenon below, and pro- phoneless households are more vide evidence that using the data on likely to own a gun, which we individual, rather than household, doubt. More important may be the ownership has the effect of reducing effect of the interview mode (tele- the negative bias. phone versus face-to-face) on the tendency for the respondent to re- me Self-Ownership Versus Household port gun ownership. While there Ownership indZvldual might be some difference in this - Surveys typically interview only one who actually regard, we can only speculate as randomly selected adult from each to what it would be. owns the sample household in order to minimize gun appears The gun ownership question is surveying costs. As Smith (1985, 2) rnore willing more complex in the NSPOF than notes: in the GSS. The GSS asks, "Do vou Interviewing each adult in a house- (or able) happen to have in your home (IF hold is usually considered unnec- HOUSE: or garage) any guns or than other essary when the information being revolvers?" The NSPOF asks re- adults in the sought is either 1) household level spondents, "Do you or any mern- information accessible to any adult household to bers of your household 18 years of Family member.. .2) a joint behavior age or older currently have any report- that the shared with the informant ...or 3) a firearms in your home, car, or else- household basic demographic or observed be- where around vour home? Do not havior that is commonly known by includes include airguns, toys, models, or household members. a gum starter pistols." Some NSPOF respondents may become confused But gun ownership is typically not a by the complexity of this question, "joint behavior" and in some house- and end up giving the wrong holds it may not even be "commonly answer. known" or "accessible." We have rnore direct evidence on yet Evidence on this issue comes from another possible explanation for the comparing responses by men and NSPOF "undercount." Whether a women. Given the NSPOF's method for household is reported to have a work- selecting respondents, husbands and ing gun depends in part on which adult wives should be equally likely to re-

Gunsin America port a gun in the home because they Differences BetweenSpouses in are both describing the same event Repoflinga Gunin the Household (gun ownership in households with a Percent married couple). Yet for households 49 50 headed by a married couple, 49 per- - A cent of the husbands report a gun in the home, compared with just 36 per- 40 - 36 cent of the wives. Since this difference is far larger than can be explained by 30 - chance, it appears that many wives either do not know about their 20 - husband's gu~zor are reluctant to discuss it with a stranger. Smith (1985) 10 - found the same pattern in the GSS, though in that case the difference was only three percentage points (57.2 Husbands Wives percent of the husbands versus 54.5 percent of the wives). that is, they can be fired?" and "Does the gun/do any of these guns belong This result suggests that reports of self- to you personally?" ownership may be more complete in the NSPOF than reports on household Table 2.3 presents self-ownership rates ownership. Population estimates based from three surveys: the NSPOF, Kleck on self-ownership are much larger than and Gertz (1995), and the GSS. The self- estimates based on household ques- ownership estimates from the NSPOF tions. The NSPOF estimates based on are quite similar to those found by a respondent's report of all guns in the Kleck and Gertz, but about four per- household is 107.2 million working fire- centage points lower than in the GSS. arms. The NSPOF estimate based on a That difference could be due to the respondent's report of his or her own focus on working firearms in the firearms is 192.1 inillion working fire- NSPOF: three percent of respondents arms. who report a firearm in the household indicate that none of these guns are in Estimates for Individual working order. Overall, we conclude Gun Ownership that the NSPOF estimates of individual gun ownership are credible. Respondents who indicated that a firearm was present in the home were In what follows in subsequent chap- asked the follow-up questions, "How ters, we make far greater use of data many of the firearms in your house- on self-ownership than on household hold are currently in working order- ownership.

Police Foundation Comparisonof Personal Gun-OwnershipRates by Survey

Kleck and General Social NSPOF Gertz Surveyfor 1994

Percent

Total 25.5 28.7 24.6 Male NIA 47 .O 41.8 Female NIA 12.7 9.0

NA = Not available. Notes:GSS figures taken from Smithand Smith(1 995, 147), from 1994General Social Survey data. The surveyby Kleckand Gertz (1 995) was conductedin 1993.

Endnotes

1. For a fuller explanation of how the response rate was calculated, please refer to the technical report of the study available from the Police Foundation. 2. For a thorough discussion of the population projection weights employed, please refer to the technical report.

Guns in America Introduction (about one third) are handguns, 70 The NSPOF was designed to gather million are rifles, and 49 million are detailed information on the stock of shotguns (Table 3.1). About 40 percent privately held guns in the United States, of handguns and long guns are semi- including: automatics. These estimates are reason- ably close to other survey-based NSPOF what types of guns are in estimates and only about 15 percent estimates circulation, less than the cumulative total of known who owns them, suggest that additions (imports and domestic manu- how they were acquired, and factures minus exports) since 1899 80 percent of how they are stored. (Zawitz 1995). While that is not much all firearms This information helps fill a major gap attrition over nearly a century, the ex- in private in our knowledge. planation is not so much the dura- bility of guns as in the great upsurge bands were Distribution and of gun sales in recent decades: NSPOF acquired Characteristics of the estimates suggest that 80 percent of all since 1974. Private Gun Stock firearms in private hands were acquired since 1974. Concentration of Ownership The NSPOF-based estimate for the to- While there are enough guns to pro- tal number of privately owned firearms vide every adult in the United States is 192 million, of which 65 million with one, only one-quarter of adults PrivatelyOwned Firearms

NSPOF Guns Entering Estimate Circulation, 1899-1 993*

Millions

Handguns Revolvers Semi-automatics Other

Rifles Semi-automatics Other

Shotguns

*FromZawitz (1 995) ** Includes"other longgun" and 'other gun"categories not ~ncludedabove

... the owners actually own a gun. Those who have Gun ownership is quite concentrated, of four or one gun usually have several. Sev- but not more so than for other durable enty-four percent have two or more. goods. In marketing circles, the "80/ guns As shown in Table 3.2, the owners 20 rule" suggests that the top fifth of (about 10 of four or more guns (about 10 per- all consumers of a product typically ac- percent of cent of the nation's adults) are in count for four-fifths of all purchases possession of 77 percent of the total by value (Clotfelter and Cook 1989, 93, the nation's stock of firearms. 279). NSPOF data indicate that the top are in It is also of interest to consider the 20 percent of gun owners have just 55 possession of concentration of ownership among Percent of the tota1 quantity (It is Pos- 77percent of households, since some households sible that if we could adjust for the include more than one gun owner. value of guns, the degree of concen- the total stock Accordinn- to the NSPOF results, 35 tration would be still greater and would of firearms. percent of households include at least better fit the rule.) The bottom line is one gun, and the relative distribution this: 9.7 million individuals own 105.5 of guns across households is very million guns, while the remaining 86.6 similar to the distribution among million guns are dispersed among 34.4 individuals. million individuals.

Gunsin America GunOwnership Across Individuals andHouseholds for ThoseWho Report Owning at LeastOne Gun

Individuals Households Gun Stock: GunStock: WhoOwn Guns That Own Guns Individuals Households (N=789) (N=1,158) (N=789) (N=1,158)

PercentDistribution

1 gun 23.5 30.1 2 guns 19.6 22.3 3 guns 12.4 13.4 4 or moreguns 42.1 34.3

Note:35 percentof NSPOFhouseholds and 25 percent ofindividuals report owning at least onegun.

Gun owners who have several guns Design Characteristics ... two-thirds tend to have a varied collection, The 200 million guns in private hands (68 percent) including rifles, shotguns, and hand- include everything from cheap .22- guns. Fifty-five percent of individu- caliber "snubbies" to finely made of handgun als who own four guns or more have high-powered rifles worth thousands owners also at least one of each of these three of dollars. The variety of firearm de- own at least categories. But not everyone who signs reflects the multiplicity of uses buys a gun is so deeply involved. Of for which they are intended. For ex- one ripe or particular concern is the novice who, ample, a gun designed to be carried shotgun, sug- without having any experience or in a pocket or handbag is usually light gesting some know-how with guns, buys a pistol and short-barreled, while a gun de- for self-protection. While there are signed with the primary purpose of experience such people, we note that two-thirds shooting accurately over long dis- and interest in (68 percent) of handgun owners also tances will be larger and heavier. The the sporting own at least one rifle or shotgun, design characteristics of firearms have suggesting some experience and in- figured importantly in legislation uses of guns. terest in the sporting uses of guns. regulating commerce in firearms. Table 3.3 provides the details. Guns capable of firing continuously

Police Foundation IndividualGun Collections

Compositionof Millions Percent Distribution GunCollection

1 handgun 1 long gun 2+ guns, hand only 2+ guns, long only 10.0 2+ guns, both handand long 19.2

Total 44.3*

*Total differs fromsum of columnentries dueto rounding,

me NS. with one pull of the trigger are Respondents who reported personally i1~3ludes...design banned from commerce, as are owning a firearm were asked a series sawed-off shotguns and other "weap- of questions about the characteristics characteristics ons of mass destruction." Handguns of one gun that was randomly selected of respcmdmzs' are regulated more closely than rifles from the owner's gun collection. The guns. With and shotguns. Small, cheaply made estimates presented in this chapter are handguns are banned from importa- computed by use of the respondent these data tion, as are some foreign-made "assault weights multiplied by the number of we develop a weapons." guns in the respondents' collection.' These estimates should provide a valid desm.ptiun of The NSPOF includes several items on representation of the national stock of the design characteristics of respon- the p'vate guns in private hands. In the discus- dents' guns. With these data we de- sion below, we focus on the barrel gun stock: velop a description of the private gun length, caliber, and the number of stock. Among other things, these data rounds of ammunition included when may be relevant for evaluating the im- the gun is fully loaded. pact of proposed regulations that distinguish among different types of Table 3.4 provides details on maga- guns. zine capacity. Owners were asked to

Gunsin America MagazineCapacity of PrivateGun Stock

AllGuns Handguns Long Guns (N=789) (N=347) (N=442)

PercentDistribution

Numberof rounds: 1-5 rounds 45.9 12.9 63.1 6 rounds 20.2 38.4 10.6 7-9 rounds 15.6 27.7 9.2 10-1 5 rounds 13.0 14.6 12.1 16 or more rounds 5.4 6.3 4.9

Average magazine capacity 6.6 rounds 8.1 rounds 5.8 rounds

indicate how many rounds of ammu- 10 rounds. According to NSPOF re- Almost nition were included when the gun was sults, such magazines were used with fully loaded. (Some guns were 25 million guns in that year. halfof all equipped with more than one maga- handguns Questions concerning barrel length zine, in which case the owner was hold seven asked to provide information for the and caliber were asked only of those magazine he used most commonly.) respondents for whom a handgun was or more The modal handgun holds six rounds, randomly selected. The most common rounds. By caliber is .38 (28 percent) and another reflecting the prevalence of revolvers comparison, in the handgun stock. But these days 19 percent of guns are close at 9 or 10 most handguns sold are pistols (Zawitz millimeters. At the smaller end of the only about 1995), which typically have larger ca- spectrum, the .22 caliber handgun one-quarter pacity. Almost half of all handguns hold constitutes 23 percent of the stock, and seven or more rounds. By comparison, 11.5 percent are .25 or .32. At the larger of long guns only about one-quarter of long guns end, only 8 percent are .44 caliber or bold that hold that many rounds. larger. many rounds. In 1994 Congress enacted a ban on the Table 3.5 reports the distribution of sale of magazines holding more than handguns by length of barrel. This

PoliceFoundation Sizeof Handguns in PrivateGun Stock

Length Percentwith PercentDistribution of Barrel SmallCaliber of HandgunStock

1-3 inches (N=86) 36.9 4-5 inches (N=129) 30.7 6 or more inches (N=132) 37.5

Note:"Small Caliber"is definedhere as .32 or less.

...the characteristic of handguns has been legislative efforts to extend the ban particularly important in the debate to domestic manufacture. majority over appropriate regulation, because As shown in Table 3.5, only about one of short short-barreled handguns are well in six handguns have a barrel length adapted to street crime in the sense handguns of three inches or less-1 1 million guns that they are conveniently carried in all. The median length of the hand- concealed. This, however, is the only gun stock is five inches. In this table la~ecaliber, advantage of a short barrel; since a we also report the percentage of hand- short barrel reduces the accuracy and in excess guns in each of the length categories power of the gun, someone who did that have small caliber (32 or less). of 32. not want to carry it concealed would ordinarily prefer a longer gun. (In The important news here is that the most states carrying a concealed fire- majority of short handguns have fairly arm is banned outright or is restricted large caliber, in excess of .32. One to those who have a special license.) implication is that definitions of "Sat- The 1968 Gun Control Act restricted urday night specials" specifying small the importation of short-barreled caliber do not apply to the majority of guns, and there have been recurrent short-barreled handguns in the nation's

Guns inAmerica Years SinceAcquisition of PrivateGun Stock

All Guns Handguns Long Guns (N=789) (N=352) (N=437)

PercentDistribution

0-1 year 2 years 3-5 years 6 or moreyears

Averageyears since gunwas acquired 12.8 10.9 13.9

me average gun was acquired about 13 years ago firearms stock. It should be kept in were acquired by their current owners by its current mind that caliber has no effect on one within the past two years. owner. Twenty- of the problematic dimensions of these guns, their concealability (Cook 1981, two percent of Storage 1740)' Indeed' it appears In 1993. there were 1.740 accidental all guns, and studies that have a preference deaths causedby fire=,rms, including 27percent of for handguns with and 170 involving children fourteen years handguns, Zawitz 1995) Based On the NSPoF fatal "gunshot accident there are a num- were acquired data, we that this preference her of accidental shootings that cause for larger caliber is shared by nonfelons by their serious injury, Guns were also the current owners who own small handguns. means of destruction in 19,590 suicides, Finally, Table 3.6 lists the years in which 2 10 involving a child fourteen or within the past respondents' randomly selected guns younger (National Center for Health two years. were acquired. The average gun was Statistics 1994). For these reasons, the acquired about 13 years ago by its cur- safe handling and storage of firearms has rent owner. Twenty-two percent of all attmcted the attention of the public health guns, and 27 percent of handguns, community.

Police Foundation StorageMethods of PrivateGun Stock

AllGuns Handguns LongGuns (N=789) (N=352) (N=437)

Percent

Gun lockedup, secured with trigger locketc. 45.6 Gun loaded 26.2 Gun loaded and unlocked 16.4

... 20 percent Despite exhortations by almost every gun in your home currently loaded and leading gun-related organization that unlocked?" (if there is just one gun) or of all firearms be stored unloaded and un- "How many of the guns in your home gun-owning der lock and key, a considerable mi- currently are loaded and unlocked?" (if households nority of gun owners are in violation there are several). Second, as part of of this stricture (Weil and Hemenway the sequence of questions concerning had a loaded 1992, 3035; Cook and Moore 1995, the randomly selected firearm, each and unlocked 269). The obvious solution, training, respondent was asked "Where do you gun in the does not appear effective. One recent usually keep this gun?" followed by "Is study found that those who have re- this gun currently locked up or secured home. .. ceived formal training in the use of fire- with a trigger lock or some other kind arms are no more likely than other gun of locking device?" and "Is this gun owners to store firearms safely (Weil currently loaded?" and Hemenway 1992). We found that 20 percent of all gun- The NSPOF presents two separate op- owning households had a loaded and portunities for examining how firearms unlocked gun in the home at the time are stored. First, each respondent who of the NSPOF. This figure was signifi- reported having an operable firearm in cantly higher among households that the household was asked either "Is the had a handgun than among households

Guns in America Storage Locationof Private Gun Stock

All Guns Handguns LongGuns (N=789) (N=352) (N=437)

Percent Distribution

Bedroom Gun closet 44.1 26.5 53.3 Other closet 16.6 11.5 19.2 With or on respondent 3.3 8.7 0.5 In car or truck 2.6 7.7 0.1 Other 9.7 8.5 10.2

with long guns only (30 versus 7 percent are kept in some other type of ... one closet. In contrast, just over one-quar- cent). In Table 3.7 we present the rein three sults for the other questionnaire items ter of all handguns are kept in gun clos- on storage, those concerning how the ets, while 37 percent are kept in the handguns, randomly selected firearm was stored. owner's bedroom. Nine percent of and one in handguns are regularly stored on the We found that 53 percent of long guns six long guns, and 57 percent of handguns are usu- owner's person, with an additional 8 ally kept unlocked. Further, 26 percent percent regularly stored in a motor is kept of all guns, and 55 percent of all hand- vehicle. loaded and guns, are usually kept loaded. Loaded unlocked. guns of either type are a bit more likely Storage and Training to be unlocked than are unloaded guns. Each gun owner in the NSPOF was The result is that one in three hand- asked "Have you ever had any instruc- guns, and one in six long guns, is kept tion or training on how to use guns?" loaded and unlocked. and, if so, "From whom have you re- As shown in Table 3.8, over half (53 ceived training?" Up to three different percent) of all long guns are kept in a sources of training were recorded for gun closet, while 17 percent are kept respondents. Table 3.9 reports the rein the owner's bedroom and 19 per- sults. The five categories-military, law

Police Foundation TrainingReceived by GunOwners

All GunOwners HandgunOwners

PercentDistribution

One or more sources of formal training Informal sourcesof training only No training

Sourcesof formaltraining: Military 25.0 Lawenforcement 11.9 NationalRifle Association 6.6 Gun club 1.O NationalSafety Council 26.0 and othersources

... adults in enforcement, National Rifle Association, esis" that formal training in the use of gun-owning gun club, and National Safety Council firearms does not affect the likelihood and other sources-constitute what we of unsafe gun storage. homes who designate to be "formal training." In all, have received 58 percent of gun owners have received Analysis for Storage formal formal training from at least one of the and Trahhg sources listed. The rest have received To further explore this surprising re- training are either ''informal" training (from friends, sult, we conducted a multivariate logit 2 as likely as Or the person from the analysis of data on storage practices purchased the firearm) Or by gun owners. The results of this and those without no training at all. training to related analyses confirm that adults in In Table 3.10, we present results on how gun-owning homes who have received keep a gun training relates to the likelihood of un- formal training are as likely as those unlocked and safe gun storage. As it turns out, there is without training to keep a gun unloaded. little difference among the various train- locked and loaded. Other control vari- ing categories in the percentages of re- ables made a difference. Not surpris- spondents who keep a gun unlocked and ingly, ownership of a handgun, or any loaded. Given the standard errors on our sort of gun for protection, was associ- estimates, we accept the "null hypoth- ated with an increased likelihood of

Guns in America Trainingand Storage Methods

Training Oneor More Unlocked and LoadedGuns inHousehold

Percent

Allgun owners: R has formal training(N=442) R has informal training(N=245) R has no training(N=102)

Handgunowners: R has formal training(N=307) R has informal training(N=160) R has no training (N=67)

Note: Figuresabove are for storageof randomlyselected gun in NSPOF;figures are calculatedusing respondent weights. R = Resgondent. keeping an unlocked loaded gun in the Acquisitions TIje training home. The likelihood was lower when In 1983 Wight, Rossi, and Daly (315) provided by there were children in the home. wrote: the National We then explored the null result on A complicated set of marketing training programs in more detail; in- mechanisms holds the key to the Safety stead of lumping all forms of "formal circulation of weapons, including Council was training" into a single category, we trade, sale, and barter among in- associated separated them into the five categories dividual gun owners, black-mar- described above. While most forms of ket sales of stolen guns, sales by with a formal training had no discernible ef- recognized dealers in firearms, signzjkant fect, there was one exception. The sales or other dispositions by po- reduction in training provided by the National Safety lice departments and the military-, Council was associated with a signifi- and so on. Although some of the likelihood cant reduction in the likelihood of these market mechanisms are of keeping an keeping an unlocked, loaded gun. This known fairly well, what is not result either speaks well of that train- known is tl~eshare of the total unlocked ing program or of the people who circulation of weapons obtained loaded gun. choose to take it, or perhaps both. by each.

PoliceFoundation Not much has been learned about gun Sources and Means of Acquisition flows in the 13 years since the publica- Our objective is to characterize the flow tion of that report. The potential im- of guns for a defined period of time. portance of this information is in evalu- The choice of how long a period to ating regulations on firearms corn- use entailed a trade-off. A longer pe- merce. For example, some regulations riod, such as the five years preceding apply only to transfers that involve a the survey, would give us a larger federally licensed firearms dealer (FFL) sample size (more acquisitions) than a (Cook, Molliconi, and Cole 1995, 68- shorter one, but at the cost of greater 91, so it would be useful to know the recall error and a greater under- volume of informal transfers that do representation of those who were most not involve FFLs. But heretofore there active in the market. In what follows Almost has been no survey or other data-gath- we focus on the results for the two- all guns ering mechanism designed to measure year period before the survey. Table the relative importance of primary and 3.11 provides some results. acquired in secondary flows of guns. Almost all guns acquired in the previ- the previous The NSPOF provides a sound basis for ous two years were either purchased two years making such estimates. The results, by the respondent (73 percent) or re- were either developed below, suggest that the sec- ceived as a gift (19 percent). The re- ondary market is somewhat less impor- maining 8 percent were obtained purchased by tant than previously thought. through inheritance, a swap of some the respondent kind, or other means. The predomi- In the NSPOF, the sequence of questions nant source of guns, not surprisingly, ... or received concerning the respondent's randomly was a store (60 percent). Other impor- selected gun concluded with the ques- as a gz3 .... tant sources included family members tion "About how many years ago did JT~I1 (17 percent) and acquaintances (12 per- acquire this gun?" followed by the ques- cent). The 3 percent of respondents tion you acquired any other gun who indicated that they obtained the since you obtained the one we just fin- gun "through the mail" (which is ille- ished talking about?" If yes, the inter- gal for all but FFLs) may have mis- viewer established some basic charac- remembered or may be referring to a teristics of this more recently acquired mail-order purchase arranged through gun (type and how long ago it was ac- an FFL. quired) and then proceeded to a sequence of questions for all gun owners The volume of gun acquisitions im- concerning details of the most recent fire- plied by the 251 NSPOF cases (Table arm acquisition. Thus, we have the date 3.12) is 13.7 million; 6.5 million of these of acquisition for one or two guns per gun transactions involved a handgun. Note owner, as well as detailed data on each that 60 percent of the long guns and gun owner's most recent acquisition. 68 percent of the handguns were new

Guns in America GunAcquisition Within the PastTwo Years

AllGuns Handguns Long Guns (N=251) (N=128) (N=121)

PercentDistribution

What best describeshow you obtainedyour gun? Boughtit Receivedit as a gift Traded somethingfor it Inheritedit Other me

Fromwhat source predominant did you obtain thisgun? source of Gun store Pawnshop guns, not Hardware, department, otherstore surprisingly, At a gun show or flea market Throughthe mail was a store.... Memberof the family Friendor acquaintance Other

Gun was new when respondent first acquiredit 64.0 68.8 60.0 at the time of acquisition. The implica- and 3.6 million long guns in 1994 tion is that 4.5 million handguns and (Table 3.12). 4.3 million long guns were added to the U.S. gun stock over the two-year Respondents were also asked, "At the period. That estimate is somewhat less time you purchased your (gun/most re- than the actual sales volume of new cent gun), about how much would it guns in the United States during 1993 have cost to purchase in a store?" Table and 1994, although the 50-50 split be- 3.13 summarizes the responses. The aver- tween handguns and long guns is age gun was worth $392 at the time of about right. Bureau of Alcohol, To- transfer, with little difference between hand- bacco, and Firearms (ATF) tabulations guns and long guns. Fewer than one in indicate sales of 3.6 million handguns twenty guns were valued at less than $100.

Police Foundation Annual AverageGun Acquisitions, 1993-1 994

All Guns NewGuns U.S. Manufactures* (NSPOFData) (NSPOFData) Imports-Exports

Thousands

Handguns 3,240 2,229 3,602 Riflesand shotguns 3,620 2,173 3,598 Allguns 6,860 4,402 7,200

* Source:U. S. Bureauof Alcohol,Tobacco and Firearms. Note: Nomore than oneacquisition isincluded for eachNSPOF respondent.

Theaverage Gun Sources and Gun Regulation or the source was an acquaintance or a member of the family. Giving all such gun was Estimating the proportion of firearms cases the benefit of the doubt brings the transfers that involve an FFL, and hence worth $39.2 at total up to 64 percent. are subject to regulation, is somewhat the time of difficult due to various ambiguities in An additional 10 percent of all acquisi- transfer, the answers. In the simplest case the tions were gifts (or inheritances) for with little respondent reported purchasing a gun which the respondent indicated an FFL from a store, pawnshop, or other as source. Presumably the FFL was not difference source that (according to the respon- the gift giver in these cases. The refer- between dent) was licensed. Of guns acquired ence is probably to cases in which the handguns within the past two years, 57 percent FFL was the source but some third party fall into this category (Table 3.14). The was paying. In some of these cases the and long guns. ambiguity arises in cases where either paperwork in connection with the sale the respondent was not sure whether the was in the name of the gift giver rather seller was an FFL or the respondent said than the ultimate recipient. Those cases that the source was an FFL but then gave would be best considered part of the other details of the transaction that raised secondary market, since the regulatory some doubt-for example, the transac- apparatus does not directly affect the tion was a trade rather than a cash sale, final transfer.

Guns in America RetailPrice of GunsAcquired Withinthe PastTwo Years

AllGuns Handguns LongGuns (N=234) (N=116) (N=117)

PercentDistribution

From this discussion and the statistics it does not apply to secondary sales. But ... 60 to 70 in Table 3.14, we conclude that ap- some buyers with criminal records still proximately 60 to 70 percent of gun attempt to buy from an FFL, and each percent of gun acquisitions occur in the primary mar- year tens of thousands are denied ac- acquisitions ket. A somewhat higher percentage of cess to a handgun because of the results occur in the handgun than long gun acquisitions are of the background check. Do these de- from the primary market. The remain- nials accomplish anything? Wright (1995, primary ing acquisitions, amounting to about 63) suggests that waiting periods and market. ... The two million per year, are off-the-books background checks for handguns are remaining transfers in the secondary market. unlikely to l~ztvemuch effect in discour- aging gun crime, since "many (and con- ... are off- Gun Ownership, Gun ceivably nearly all) of the new guns corn- the-books Acquisitions, and Gun Control ing into circulation are being purchased transfers in by people who already own guns." The The 1993 Brady Law instituted a nation- NSPOF data allow us to determine the the secondary wide requirement that handgun buyers accuracy of this assertion. submit to a background check (usually market. including a waiting period) prior to ac- As seen in Table 3.15, we find that a quiring a handgun. One loophole in this substantial minority (23 percent) of requirement, as discussed above, is that handgun buyers did not own a gun at

PoliceFoundation PrimaryVersus Secondary Markets of Guns Acquired Withinthe PastTwo Years

PrimaryMarket Definition AllGuns Handguns LongGuns (N=248) (N=126) (N=121)

Percent

(1 ) Cash purchase fromgun, hardwareor departmentstore, from pawnshop,or from seller at gun show, flea marketor military, or throughmail that respondentsays "yes" was FFL

(2) Add cash purchasefrom seller ... when we at gun show, flea marketor restrict the military, or throughmail, that respondentsays "probably analysis just waslthinkso"

to cheaper (3) Add non-cashtransactions 60.1 handguns (trades) with sourcesin ... (1 and (2) which are (4) Add cash purchases,trades 64.3 used dispro- with family, friendslacquaintance that respondentsays are or portionately probablyare FFCs

in crime, we (5) Add gifts, inheritances,prizes 73.6 84.2 64.7 find that fully from sourcesin (1) through(4) 56percent of buyers did not the time of their purchase. Thirty-five Disposals percent of handgun buyers did not have a own a handgun at the time of purchase. Thefts handgun at Moreover, when we restrict the analy- One major theme highlighted by Wright the time of the sis just to cheaper handguns (those and Rossi's survey (1986, 183) of in- with a price less than $130 in 1994 carcerated felons was that theft is an transaction. terms), which are used disproportion- important source of firearms for those So it appears ately in crime (Cook 1981), we find with criminal intentions: 32 percent of that waiting that fully 56 percent of buyers did not surveyed felons had stolen the most have a handgun at the time of the trans- recent handgun that they had acquired. [periods are] action. So it appears that the waiting Estimates from the National Criminal not futile. period is not futile. Victimization Survey (NCVS) suggest

Guns in America GunOwnership at Time ofMost RecentAcquisition

Working Working Gunin Handgun in Firearm Handgun Household Household

Percent

Allgun 74.1 35.0 acquisitions (N=786) Gun acquired 82.4 48.8 2 years ago (N=251)

Handgun acquired 76.9 65.0 2 years ago (N=128)

Note: Doesnot include casesin whichthe respondentwas issuedmost recently acquiredgun for work.

that 340,700 thefts occurred annually "Was this theft incident reported to the ... we estimate from 1987 to 1992 in the U.S. in which police or did they find out in some one or more firearms were stolen. other way?" Since these items refer to that there Separate data from North Carolina the household gun population, we use were 21 1,000 suggest that on average 1.5 firearms the NSPOF household weights in cal- handguns and are stolen per theft (Cook, Molliconi, culating our estimates. However, we 382,000 long and Cole 1995, 82) which, if general- remind the reader that in Chapter 2 we izable, implies that about half a mil- produced some evidence to suggest guns stolen lion guns are stolen each year that a large number of firearms may be during non- nationwide. stored in American households with- commercial out the full knowledge of all of the adult As part of the NSPOF survey, all re- theJts [in spondents were asked, "In the past 12 members living in the home. Presum- months, have you, or has anyone in ably, adults who are not aware of the 19941. your household, had any firearms sto- presence of a firearm in the home will len from them? Please do not include also be unaware that the mystery fire- theft that might have occurred at work." arm has been stolen. As such, these Answers in the affirmative were fol- estimates are almost surely likely to be lowed by the questions "What types of a lower bound for the number of thefts guns were stolen, and how many?" and in the U.S. each year.

PoliceFoundation Based on the NSPOF, we estimate that Based on NSPOF responses, we estimate 0.9 percent of all gun-owning house- that 3.4 percent of gun-owning households (269,000) experienced during the holds (about 1 million) disposed of a previous year a theft in which a fire- gun in the previous year. Most reported arm was stolen. Using the information selling the gun or giving it as a gift. Since on the number of firearms that were respondents were not asked to provide taken from theft victims, we estimate the number of guns that were disposed that there were 21 1,000 handguns and of during the past 12 months, the most 382,000 long guns stolen during non- we can do is produce a lower bound commercial thefts during the past year, for the number of used guns that circu- for a total of 593,000 stolen guns. These lated. Our lower-bound figures suggest estimates are subject to considerable that half a million guns were sold, an- .. .600,000 sampling error, and are not inconsis- other quarter-million were bartered for used tent with the 511,000 per year figure something else, and 160,000 were given estimated by Cook, Molliconi, and Cole away. Of the firearms that were kept in handguns (1995,82) from NCVS data for the years circulation (transferred to another owner and 3 70,000 1987 to 1992. in some way), over 60 percent were used handguns. We estimate that 600,000 used handguns and 370,000 used long guns long guns Disposing of Guns Earlier we examined transactions in sec- changed owners during the past year changed ond-hand guns from the purchasers' re- through these private transactions. owners ports. Data from the NSPOF also allow during the us to estimate used-gun sales from the NSPOF figures indicate that 6,000 fire- sellers' reports. Each respondent was arms were confiscated during the past past year asked "In the past 12 months, other than year by police or other authorities, dra- through by theft, have you or anyone in your matically lower than previous estimates ...private household gotten rid of or otherwise lost of up to 200,000 confiscated guns per a gun you or they owned?"Respondents year (Cook 1993). The underestimate transactions. who indicated that they or someone else may be due either to under- had transferred a handgun were then representation of criminals in the asked, "Please think of the gun you most sample or reluctance on the part of the recently got rid of. How did you get rid respondents to admit to criminal in- of this gun?" [emphasis in original] and volvement (see Cook 1985). We also "What type of gun did you discard of or estimate that 36,000 guns were thrown lose recently?' away by their owners.

Endnotes 1. For a full discussion of this formula, please see the technical report. 2. For a fuller description of this analysis, please see the technical report.

Guns in America Introduction youth, either in military service or Almost every- (more commonly) from growing up one who A number of surveys have documented with guns in the home. the patterns and motivations for gun currently ownership in the United States. The In addition to information on current gun ownership, the NSPOF includes owns a gun results from the NSPOF tend to reinforce previous findings. Most long gun items on previous ownership. From bad some these we find that while 42 percent of owners enjoy the sporting uses of guns, experience men own guns now, another 24 per- such as hunting and target shooting. with guns as For those who own handguns, self-pro- cent owned them previously and then gave them up. Thus two-thirds of men tection is the primary motivation. The a youth, people who keep guns for self-protec- have been gun owners. The cohorts either in tion are for the most part those who who are currently middle aged (40 through 64) have had the greatest life- military also make use of them in recreation. As a result, the patterns of handgun time involvement with guns. service or ownership closely parallel patterns of About 5 percent of the nonowners were (more gun ownership overall: gun ownership planning to acquire a gun during the commonly) rates are highest among men, people year following the survey for self-de- living in rural areas, and people of fense purposes. Blacks and youths from growing middle age and higher income. Almost were especially likely to indicate ac- up with guns everyone who currently owns a gun quisition plans, as were those who had in the home. had some experience with guns as a been robbed or attacked recently. Patterns of Gun Ownership GunOwnership by Sex The NSPOF asked all respondents the question "Do you or any members of Percent your household 18 years of age or older currently have any firearms in your 50 42 home, car, or elsewhere around your 1 home? Do not include airguns, toys, models, or starter pistols" [emphasis in original].Follow-up questions included, "Are you planning to get a firearm for protection against crime any time in the next 12 months?" [emphasis in origi- ...g un nal], "How many of the firearms in your ownership household are currently in working order-that is, they can be fired?" and is much more "Does the gun/do any of these guns Males Females prevalent in belong to you personally?" rural areas Thus, we are only able to identify indi- and small vidual ownership in cases where the men, but just 9 percent of women, own respondent answered the preliminary a gun. There is a similar disparity for towns than in question concerning the number of lifetime ownership and for current cities .... Those working guns. In what follows we treat handgun ownership. aged 40 ownership of a gun that is not in working order as equivalent to not having a With respect to race, whites are sub- through 64 gun. stantially more likely to own guns than have a blacks, and blacks more likely than From NSPOF data, we estimate that Hispanics. Most of the white-black dif- substantially one-quarter of the adult public currently ference evaporates in the case of hand- higher rate own a working firearm, and 40 per- gun ownership. cent have owned one at some point in than the other their lives. The majority of all gun own- It comes as no surprise that gun own- age groups. ers own a handgun, usually in addi- ership is much more prevalent in rural tion to one or more long guns. The areas and srnall towns than in cities. distribution of gun ownership is far Perhaps more surprising is the age pat- from uniform. We explore the patterns tern, where we see that those aged 40 in Table 4.1, demographic characteris- through 64 have a substantially higher tics; Table 4.2, socioeconomic charac- rate than the other age groups. teristics; and Table 4.3, other variables. Table 4.2 displays patterns of gun own- The most notable demographic divide ership along socioeconomic dimen- in gun owning is by sex: 42 percent of sions. The results may be surprising to

Guns in America GunOwnership Patterns: Demographic Characteristics

Characteristic Currently Currently EverOwned Own Own a Gun Gun Handgun (Lifetime)

Percent

Total(N=2,447)

Sex: Male(N=l, 125) Female(N=1,320)

Age: 18-24 (N=235) 25-39 (N=909) 40-64 (N=976) 65+ (N=351)

Race: White(N=1,521) Black(N=424) Hispanic(N=404)

Maritalstatus: Nevermarried (N=497) Divorced(N=265) Separated(N=88) Widowed(N=169) Married(N=1,398)

Community: Rural(N=4 1 0) Small city(N=729) Medium city(N=409) Suburbs (N=307) Largecity (N=554)

Note:Sample N's reported representnumber of respondentsin groupwho provided valid response to questionof whether theythemselves owneda working gun.

Police Foundation GunOwnership Patterns: Socioeconomic Characteristics

Characteristic Currently Currently EverOwned Own Own a Gun Gun Handgun (Lifetime)

Percent

Education: < Highschool (N=349) 17.9 HS diploma(N=817) 23.6 Somecollege (N=691) 29.6 Collegedegree (N=277) 26.5 Graduatedegree (N=2 14) 31 .O

Income(thousands): $0-1 0 (N=281) 10.8 $1 0-20 (N=414) 17.6 $20-30 (N=369) 33.5 $30-50 (N=553) 30.5 $50-75 (N=296) 26.5 $75 and over (N=197) 32.5

Professional,technical (N=44 1) 27.4 Manager,administrator, sales workers(N=416) 26.3 Clerical(N=281) 18.6 Craftsmen,operatives (N=495) 36.8 Farmer,farm laborers(N=235) 61.3 Serviceworkers (N=235) 10.9

Militaryservice: Veteran(N=345) 50.5 Armedforces personnel on active duty(N=132) 21.1 None(N= 1,934) 20.6

*Definedto correspond to Kleck991, (1 p. 56-7). Definitions osfollows (NSPOF categories in parentheses): Professional,technical (professional specialty, or technicians andrelated support); Manager,administrator, sales workers(Executive, administrative, and managerial,or sales);Clerical (administrative support, includingclerical); Craftsmen,Operatives (mechanics and repairers, construction trades, extractive and precisionproduction, machine operators, assemblers andinspectors, transportation and materialmoving, handlers,equipment cleaners, helpers,and laborers); Farmers, farm laborers (farmoperations, other agriculture orrelated, forestry, logging,fishers, hunters and trappers); Service workers (service, except protective).

Guns in America GunOwnership Patterns: OtherVariables

Currently Currently EverOwned Own Gun OwnHandgun Gun a (Lifetime)

Percent

Parentshad gun inhome: Yes (N=1,376) No (N= 1,048)

Political views: Liberal(N=608) Moderate(N=66 1 ) Conservative(N=962)

Arrestedfor non-trafficoffense: Yes (N=160) No (N=2,2 19) anyone whose concept has been rience in the decision to own a gun. ... those who formed by images of gun violence in Those whose parents kept guns are the inner city. In fact, gun ownership three times as likely as others to own admit to ever is more prevalent among the middle one themselves. In fact, 80 percent of having been class than among the poor. In particu- all current gun owners report that their arrested for lar, the prevalence of ownership is parents kept a gun in the home. greater for those with some college a nontraffic education than those without, and Two other associations are of general offense have greater for those whose household in- interest. With respect to political views, a relatively come exceeds $20,000 than those with conservatives are more likely to own a lower incomes. With respect to occu- gun than moderates or liberals. And it high pation, most farmers and farm work- is somewhat worrisome that those who ownership admit to ever having been arrested for ers own a gun, and skilled blue-collar rate (3 7 workers (craftsmen, operatives) are also a nontraffic offense have a relatively relatively likely to own a firearnl. Half high ownership rate (37 percent) com- percent) of all veterans own a gun. pared to the norm. compared to

Table 4.3 offers results for three unre- Finally, Table 4.4 presents gun- and the norm. lated variables. First, we see the ap- handgun-ownership rates stratified by parent importance of childhood expe- responses to various attitudinal items

Police Foundation in the NSPOF. Respondents who feel this age. Since younger cohorts have safe alone at home in the evening, or not yet reached all of the ages listed alone out in the neighborhood at night, along the top of the table, the number are slightly more likely to own a gun of cohorts included in the comparisons than respondents who do not feel safe. decreases as the age cutoff increases. Respondents who have been the vic- Gun ownership is becoming less com- tim of a burglary or robbery over the mon among men. Gun acquisition is past 12 months are more likely to own highest among those men who came a gun than other adults. As is usually of age just after World War I1 (born the case, handgun ownership patterns 1930 through 1943), followed by baby match patterns of ownership for all boomers born between 1944 and 1953. guns. Relative to these men, more recent Gun cohorts were less likely to have ac- ownership It comes as no surprise that respon- quired a gun by the age of 21 or 31. dents who oppose laws to ban private is becoming gun ownership are more likely to own For women the largest change appears less common guns than proponents of such mea- to have occurred between the oldest sures. What does come as a surprise is cohort and those born after 1930. Be- among men. that 8 percent of respondents who sup- yond those cohorts there is no obvi- Gun port such measures personally own a ous pattern. These results are consis- acquisition gun. Finally, respondents who believe tent with Smith and Smith's (1995, 147) that crime rates in their neighborhoods findings that the proportion of women is bigbest are increasing over the past 12 months who own a gun has been fairly con- among those are more likely to own a gun than those stant from 1980 through 1994. men who who report a steady or decreasing came of age crime rate. Reasons for Ownership Previous surveys suggest that more just aJter Gun Ownership Across Cohorts than half of all gun owners have their World War I%;... Respondents who reported ever own- firearms pm'mam'ly for hunting or other ing a gun were also asked, "How old sporting purposes and that a vast ma- were you when you got your first gun?" jority of gun owners indicate that sport- As shown in Table 4.5, data on the age ing use is one reason they own a gun. of first gun acquisition provide us with At the same time, a significant propor- some clues about whether the preva- tion of respondents report protection lence of gun ownership differs across as a reason for owning a gun. Another cohorts of American adults. Several age theme that has emerged from the lit- cutoffs are presented along the top of erature is that handguns and long guns the table, while each cell in the table are purchased for different reasons. The presents the proportion of the respec- majority of handguns are kept prim- tive cohort that acquired a firearm by arily for protection against crime.

Guns in America GunOwnership Patterns and Experiences with Crime

Experiencewith Crime OwnGun OwnHandgun

Percent

Feelingsof safetywhile aloneat homeat night: Very or somewhatsafe (N=2,196) Very or somewhatunsafe (N=234)

Feelingsof safetywhile aloneout in the neighborhood: Very or somewhatsafe (N=1,789) Very or somewhatunsafe (N=570) Three-

Homehas been burgledin past 12 months: quarters Yes (N=9 1) of all long No (N=2,353) gun owners Robbedlattackedin past 12 months: report having Yes (N= 142) No (N=2,302) a long gun

Perceivedtrend in crimerates in primarily for neighborhoodover past 12 months: sporting Goingdown (N=159) Stay same (N=1,7 13) purposes ... Goingup (N=484)

Note: Forthe N's here, we usethe numberof respondentsin a group that gave aus valid response to the questionof whether theythemselves owneda working gun.

Reasons for Owning- a Gun only long guns, only handguns, or one Handgun-owning respondents were or more of each. The latter group is asked, "What is the most important rea- included twice, since they provided son why you own a handgun?" Simi- reasons for both types of guns. larly, respondents who own at least one Three-quarters of a11 Iong gun owners long gun were asked, "What is the most report having a long gun primarily for important reason you own a long gun?" sporting purposes, including hunting In Table 4.6, we tabulate their re- and target shooting. The large majority sponses according to the composition of handgun owners have a handgun of their gun collection: those who own primarily for self-defense. Note, how-

Police Foundation GunOwnership by Cohort

Sexand Age OwnedGun OwnedGun OwnedGun OwnedGun (in 1994) BeforeAge 21 BeforeAge 31 BeforeAge 41 BeforeAge 51

Percent Distribution

Males, 2 1-30 48.3 Males, 31-40 44.1 54.5 Males, 4 1-50 51.9 62.7 69.1 Males, 51-64 59.3 72.8 76.4 78.1 Males, 65 + 47.9 60.9 66.0 69.3

Females, 2 1-30 5.2 Females, 3 1-40 5.2 16.4 Females, 4 1-50 4.2 10.9 17.1 Females, 5 1-64 9.4 13.2 17.6 22.4 Females, 65 + 2.8 5.4 7.1 8.9

Note:All respondents who haveever owneda firearmwere asked,"How old were youwhen you got your first gun?"

Focusing ever, that for a few respondents hunt- for female handgun owners than for ing or target shooting is the most im- males (84 versus 57 percent). on handgun portant reason. Note also that in the owners ..., "handgun only" group, for whom hand- Reasons for Not Owning a Gun 63 percent gun ownership is mostly utilitarian, 40 Respondents who did not personally own gun percent are women, compared to their own a gun at the time of the NSPOF a lower than 15 percent representation interview were asked, "Why don't you primarily for in the other two groups. personally own a gun now?" Chilton interviewers were instructed to probe protection Overall, 46 percent of gun owners own for additional reasons, using the fol- against crime. some kind of gun primarily for protec- low up question "Why else don't you tion against crime. The figure is 41 own a gun?" Up to five reasons were percent for males, and 67 percent for recorded for each respondent. females. (Respondents were not asked about secondary reasons for owning a We have grouped their reasons into gun.) Focusing on handgun owners seven categories, as presented in Table (irrespective of whether or not a long 4.7. The most prevalent reason for not gun is also owned), 63 percent own a owning a firearm was affordability; 27 gun primarily for protection against percent of respondents listed this as crime. The proportion is much higher the most important reason for not own-

Guns in America Primary Reasonsfor Owning a Gun

Own Own OwnBoth Own Both Hand- Long (GaveReason (Gave Reason gun Gun for Owning for Owning Only Only Handgun) Long Gun) (N=206) (N=266) (N=31 4) (N=314)

PercentDistribution

Self-defense 74.4 14.9 55.9 12.4 Hunting 0.5 69.9 12.2 65.8 Targetlsport shooting 10.8 6.1 13.3 9.3 Gun collection 0.9 0.4 2.4 2.3 The most Job-related 4.7 0.3 2.9 1.2 prevalent Other reason 7.9 8.4 13.0 8.9 reason for Note: Table excludes the response "don't know." not owning a firearm was affordability ... ing a gun, and 33 percent of respon- When we compare male and female roughly one- dents who do not currently own a fire- nonowners, we find proportionately third of arm give cost as one reason for not fewer women than men are dissuaded having a gun. Only a small (12 per- from owning a gun by price, though gunless adults cent) proportion of this cost-conscious women are more likely to report the are at least group had any concern that might stop presence of children, tlie dangers of open to the them from obtaining a gun if they could gun ownership, or moral objections as find one cheap enough or if their in- a reason for not having a firearm. possibility of come increased. These results are calculated using obtaining one There is a large group of people, how- adults without guns who live in homes and might do ever, who are actively opposed to hav- both with and without a gun. When so if their ing a gun in their homes, because they the sample is restricted to only those view guns as too dangerous or for some gunless adults living in gunless house- financial other reason. Twenty-two percent of holds, the proportions of respondents problems nonowners indicated that guns are dan- who cite expense, the danger of guns, eased or their gerous, and 17 percent mentioned their and the presence of children as a rea- children as a reason for not having a son for not having a gun all increase. motivation gun. Others (21 percent) were opposed With respect to the potential for growth became to guns for various principled reasons. in gun ownership in the United States stronger.

Police Foundation Reasonsfor NotOwning a Gun

Percent

Can't affordone, too expensive Gunsare dangerousto have Opposedto guns* Havechildren Haveno needfor one 9.4 Someone elsein householdowns gun Can't shoot,not trained

*Includesthe responses: Moral objections,guns are immoral; againstguns; don't believein guns; don't like guns; don't wanta gun, not interested;don't want a gunin the house;had a badexperience with a gun; gunsdon't solve problems;guns onlyfor the police,

Overal& we conclude that many of those who the adult public said they were. Since 4.7percent currently lack a gun (roughly one-third the handgun is the weapon of choice of gunless adults) are at least open to for self-defense, we believe that most of the adult the possibility of obtaining one and might of those who said yes intended to ob- public said do so if their financial problems eased tain a handgun. For that reason, Table they [did not or their motivation became stronger. For 4.8 compares the prevalence of plans many the motivation may come from to obtain a self-protection gun with the own a gun an increased concern about crime. In- actual prevalence of handguns. deed, nearly 5 percent of respondents but]. .. were If these plans had been realized, the reported that they were planning to planning to overall handgun ownership rate would obtain a gun for protection against have increased by about 30 percent obtain a gun crime soon, within the subsequent 12 (from 16 percent to over 20 percent) months. within the next during the year following the survey. 12 months for For blacks, handgun ownership would protection Plans for Purchase have nearly doubled. Of course, noth- Nonowners were asked whether they ing like this actually occurred, and re- against were planning to obtain a gun within spondents' statements on this item crime. .. the next 12 months for protection should not be taken literally. But they against crime. Overall, 4.7 percent of do offer some insight into the link be-

Guns in America Current Ownershipand Plans for Acquiringa Gunfor Protection: Individual Characteristics

Characteristic Currently NonownersWho Own Plan toGet Gun Handgun in NextYear

Percentof AllAdults

Total(N=2,447)

Sex: Male(N=l, 125) Female(N=1,320)

Race: White(N=1,521) Black(N=424) Hispanic(N=404)

Arrestedfor nontrafficoffense: Yes (N= 160) No (N=2,2 19)

Notes:N's representnumber of respondentsin groupwho providedvalid response to questionof whether theyown a working gun.Only thosewho said they did notown a gunwere asked aboutplans for acquiringone. tween crime and the demand for guns, Also of interest is the relationship be- and reinforce our previous findings, tween current handgun ownership and which suggest the potential for fairly concern about crime. Respondents who substantial increases in protective gun felt unsafe at home or in their neigh- ownership. borhoods had about the same rate of current ownership as those who felt Table 4.9 develops this linkage. Con- safe, but they were much more likely cerns about crime and personal safety to report that they planned to obtain a were closely associated with plans to gun. Those who reported being vic- obtain a gun for protection. For ex- tims of burglary or robbery during ample, 7 percent of those who believed the previous year were somewhat crime was increasing in their neighbor- more likely than others to own a hood planned to obtain a gun, com- handgun. Whether they obtained that pared with just 4 percent of those who gun before or after the crime cannot believed crime was not increasing. be determined.

PoliceFoundation Current Ownershipand Plansfor Acquiringa Gun for Protection: Experience withCrime

Experience Own Handgun NonownersWho Planto Get Gun in NextYear

Percentof All Adults

Feelingsof safetywhile alone at homeat night: Very or somewhatsafe (N=2,196) Very or somewhatunsafe (N=234)

Feelingsof safetywhile alone out in the neighborhoodat night: Very or somewhatsafe (N=1,789) Very or somewhat unsafe(N=570)

Home hasbeen burgled in past 12 months: Yes (N=91) No (N=2,353)

Robbedlattackedin the past 12 months: Yes (N= 142) NO(N=2,302)

Favorslaw makingillegal for private citizensto own gun: Yes (N=725) NO(N=1,634)

Perceivedtrend in crime rates in neighborhoodover past 12 months: Goingdown (N=159) Stayingthe same (N=1,7 13) Goingup (N=484)

Note:N's representnumber of respondentsin groupwho providedvalid responseto questionof whether they own a workinggun,

Guns in America Perceptionsof Safety AmongUnarmed Residents of ArmedHouseholds

Perceptionof Safety Males Females Total KnowingThat Someone (N=43) (N=3 12) (N=355) Elsein the Household Hasa Gun

Percent Distribution

Notat all safelnotvery safe 2.3 Somewhatsafe 23.5 Very safe, extremely safe 74.3

Feelings of Safety with a has a gun? Do you feel ... ?" The re- Gun in the Home sponses that were read to participants While most gun owners reported in- were "extremely safe, very safe, some- creased feelings of safety from their what safe, not very safe, and not at all Most adults own firearms, in a recent telephone safe." who live in a survey Hemenway, Solnick, and Azrael As seen in Table 4.10, only a small gun-owning (1995, 124) found that 85 percent of minority of gunless adults (5 percent adults who did not own guns reported for women, 2 percent for men) report household, that they would feel less safe if more feeling "not at all safe" or "not very who do not members of their community obtained safe" knowing that someone else in the themselves a firearm. Thus, it appears that the ac- household owns a gun. Most adults quisition of guns adversely affects other who live in a gun-owning household, own a gun, members of the community. But what who do not themselves own a gun, believe they effect is there on other members of the believe they are very safe. The ques- are very safe. same household? Do gunless adults in tion is a bit ambiguous. We do not gun-owning households feel safer or know whether they are reporting that less safe with a gun in the home? they are not worried about the possi- Respondents who reported a firearm bility of an accident with the gun or in the household but indicated that they whether they feel the gun provides did not themselves own a gun were protection against outsiders. In either asked, "How safe do you feel know- case, conflict within households over ing that someone in your household guns is apparently rather rare.

PoliceFoundation Guns in America Introduction who said yes were then asked about In this chapter, we examine a collec- the sources from which training was tion of topics related to the use and received. Chapter 3 presented some misuse of firearms: results from these items on training in ... tbree- connection with our analysis of gun- the prevalence of firearms training .I storage practices. We now develop a quarters of among the adult population in the more comprehensive picture of how United States, and in particular for American men Americans learn the skills necessary to gun owners and those living with and over handle guns safely. guns in the home; one-third of the recreational uses of guns, in As indicated in Tables 5.1 and 5.2, women have particular hunting and sport or tar- about three-quarters of American men get shooting; and over one-third of women have re- received some ceived some instruction or training in firearm accidents; and instruction or II how to use a gun. Over half of all men .I gun carrying. have received "formal" training-that training in is, training from a professional instruc- bow to use Firearms Training tor provided by law enforcement or a gun. The following question was asked of other security agencies, the National all respondents in the NSPOF: "Have Rifle Association, gun clubs, the Scouts, you ever had any instruction or train- the National Safety Council, or (most ing on how to use guns?" Respondents commonly) the armed forces. Most FirearmsTraining

Males Females Total (N=1,186) (N=1,382) (N=2,568)

Percent

Personallyowns firearm 41.8 9.0 24.6 Reportsfirearm in household 48.5 Receivedinstructionltraining on howto use guns 75.7 Receivedinstruction in "formal" trainingprogram 55.6 Receivedinstruction informally Source:* Parent 12.7 11.8 12.2 Other relative 5.3 13.5 9.6 Friendlacquaintance/neighbor 4.3 4.9 4.6 Other 0.7 0.0 0.4

* Includes asmany as threesources per respondent.

Almost all women, on the other hand, receive have participated in a formal training gun owners... their training informally from a parent program (Table 5.3). (Only about 23 or other relative. percent of other adults have received have received The military is an important source of formal training.) Among gun owners, some sort of training for gun owners. All told, 29 men are much more likely than women training and million people received training from the to have had formal training. Regard- less of gender, gun owners with for- 58percent of military, 9 million from a law enforcement or private security agency, 7 mil- mal training tend to be better educated them have lion from the National Rifle Association, and have solnewhat higher incomes than those who lack formal training. participated 5 million in school, and 4 million in in a formal Scouts. Another 16 million adults participated in other formal training pro- training Recreational Uses grams, such as those sponsored by the of Firearms National Safety Council (Table 5.2). program. While hunting with a gun has been de- Who gets trained? Almost all gun own- clining somewhat in popularity over the ers (87 percent) have received some last several decades, rnillions of Ameri- sort of training and 58 percent of them cans still hunt for sport or to supple-

Guns in America PopulationEstimates for FirearmsTraining

Source Males Females Total

Millions

Military 27.1 1.9 29.0 Law enforcement1 private security NationalRifle Association 5.6 Gun clublshooting range 0.8 Otherformal training (such as NationalSafety Council) 13.1 Scouts 3.9 Schools 3.3

Totalwho havereceived "formal" firearms training 50.2 10.7 60.8

Note: "Formal"training program includestraining from military,law enforcement/securitypersonnel, NRA instructor,gun club,other formaltraining, Scouts, or school program.

ment the household larder. The NSPOF As seen in Table 5.4, our estimates in- ... 23 million included several items on this and other dicate that about 35 percent of gun about sporting uses, including "Thinking about owners hunted in 1994. Thirty-five per- just the past 12 months, have you used cent of owners engaged in sport shoot- half-f gun any guns in this household to go hunting. These estimates imply that about owners ing?" Respondents were also asked, 8 percent of all adults (16 million) went participated "How about target or sport shooting? In hunting at least once in 1994, and 16 the past 12 months have you used any million Americans participated in sport in a gun guns in this household to do any kind shooting. Taking account of the sub- sport during of target or sport shooting, besides hunt- stantial overlap between the two ing?" Respondents who answered yes to groups, 23 million-about half-of gun 1994. either question were then asked to pro- owners participated in a gun sport vide the number of days over the past during 1994. 12 months that they had used a gun to hunt or engage in sport shooting, and In Table 5.5, we see that rifles and shot- also to describe the type of gun the re- guns are equally popular with hunt- spondent used "most frequently" for this ers, while hunting with a handgun is activity over the last year. quite rare. Only 1.5 percent of hunters

Police Foundation GunOwnership andTraining

Nonowners Nonowners Gun in Gun-Free in Households Owners Households withGuns (N=1,262) (N=396) (N=789)

Percent

Has receivedtraining 41.3 56.4 87.3 Has received 'formal"training 23.3 22.0 58.1

Prevalenceof RecreationalGun Use

All Gun Estimated Adults Owners Adults

Percent Millions

Huntedduring past year Wentsport shooting duringpast year Eitherhunted or went sport shooting

Gunsin America Typeof GunUsed by RecreationalGun Users

Males Females All Adults

Percent Distribution

Gunused most frequently for huntingin past12 months: (N=226) (N=30) (N=256) Rifle 1.7 0.0 1.5 Handgun Shotgun Other gun

Gunused most frequently for sportshooting in pasf 12 months: Handgun Rifle Shotgun Other gun

used a handgun most frequently. On Gun Accidents ... rijles and the other hand, half of sport shooters There is considerable uncertainty re- typically use a handgun of some sort. shotguns are garding the number of accidental gun- equally The descriptive statistics presented in shot woundings that occur each year. Table 5.6 suggest that sport shooters Part of the problem is definition. Of popular with and hunters are in general fairly repre- the 40,230 people killed by gunshot in hunters, sentative of all gun owners with respect 1993, 1,740 were classified as accidents, while bunting to education and income. Recreational as opposed to suicides or homicides. shooters (sport or hunting) are slightly At the margins, that distinction is hazy with a less likely than other gun owners to be even in principle. Is a gunshot wound handgun is black. Hunters are more likely to have suffered while playing Russian roulette quite rare. been raised in rural areas or to live in a suicide or an accident? What about a rural areas as adults, and they are less shooting that is unintentional but likely to live in the Northeast. grossly negligent, as when one person

Police Foundation Characteristicsof RecreationalGun UsersVersus Others

Characteristic Sporting Gun Non-Gun Non-Sporting Owners Owners GunOwners (N=41 5) (N=1,658) (N=492) - PercentDistribution

Sex: Male Female 17.9 Race: White 87.9 Black 4.4 Hispanic 2.9 Other 4.9 Education: Less than high school 7.5 High school 28.8 Some college 34.6 College degree (bachelor's) 12.4 Advanced degree 16.6 Armed forcespersonnel: Veteran 27.9 Income(thousands): $0-20 20.3 20-50 47.7 50 and over 31.9 Don't know, missing 6.7 Communitywhere residing: Rural 32.2 Small town 34.0 Medium city 9.3 Suburbs 9.6 Large city 15.0 Communitywhere raised: Rural 42.2 Small town 29.2 Medium city 10.2 Suburbs 10.6 Large city 7.7 Censusregion: Northeast 12.2 Midwest 23.5 South 41.6 West 22.7 Receivedfirearms training 93.4 Average age (years) 42.0 45.7 50.1 Note:Non-owning group also excludes nonownerswho have recreational useover past 12 months

Gunsin America Numberof Householdswith Gun Accidents

RatePer 1,000 Thousands

Gun accident during past12 months

Gun accident tochild age <18 years during past12 months

Note: Thirteenrespondents reportedan accident (rowI), andtwo respondents reportedan accident toa child(row 2).

shoots another while fooling with a gun Gun Accidents in the NSPOF he thought was not loaded? In practice, The NSPOF asked each participant, "In the difficulty of classification is greatly the past 12 months, have you or any- compounded by the problem of obtain- one in your household been a victim ing an accurate report of what happened. of an accidental gunshot wound?" Since ... a member the question refers to all household The number of accidental deaths is members, this is one of the rare in- of one-third known with greater precision than the stances in this report in which we em- number of nonfatal woundings simply of one percent ploy the household weights to produce because deaths are almost always of us. our estimates. known and investigated by the medi- households cal examiner and police. On the other Based on reported accidents, we esti- hand, woundings may go unnoticed by mate that a member of one-third of one was officials, or if known may nonetheless percent of U.S. households was acci- accidentally go uninvestigated and unrecorded. For dentally wounded in 1994. That esti- wounded example, Annest et al. (1995) used data mate is based on just 13 respondents 1974 ... from emergency-room records to esti- in the NSPOF sample of 2,568. Using in mate that there were 99,000 nonfatal the NSPOF sampling weights, these though [this] woundings each year, but this estimate very small proportions imply a large estimate is misses cases that were not treated in (in an absolute sense) number of acci- based on emergency rooms, and does not dis- dents: 315,000 per year (Table 5.7). The tinguish between accidental and inten- standard error for this estimate is just 13 tional woundings. 160,000, which implies a rather wide respondents ... range of statistically plausible values. Gaining a reliable estimate of the number of accidental gunshot wounds is The uncertainty about this estimate hence of considerable interest. Unfor- stems not only from sampling error, but tunately, however, the NSPOF does not also from the accuracy of reports by support a reliable estimate, as explained respondents (Table 5.8). Possible below. sources of error include "telescoping"

Police Foundation GunAccidents Reportedin NSPOF

Victim Age Sex Education Income Part Where Circum- TYpe (Thousands) of Accident stances of Body Took of Gun Wounded Place Accident Involved

1. Respondent

R = Respondent.

Gunsin America (reporting an event that occurred more by state laws except in the case of law than a year ago), and failure to report enforcement officers and (in many an accident that did occur in the year- states) those who have obtained a spe- long window (perhaps due to the cial license. The effects (positive and respondent's unwillingness to divulge negative) of legislation that increases an event in which he was implicated). the ability of citizens to carry concealed Respondents may also misclassify as firearms are the topic of ongoing de- unintentional incidents that would be bate (McDowall, Loftin, and Wiersema considered intentional by law enforce- 1995; Polsby 1995), but a number of ment authorities (for example, 2 of the states have enacted such legislation 13 events were drive-by shootings). In within the last few years. general a survey of this sort is a problematic device for measuring a rare Gun Carrying in the NSPOF Carrying a event. We return to this matter in greater The NSPOF data provide two separate detail in the next chapter's discussion gun away opportunities to examine gun-carrying of defensive gun uses. behavior among the adult population. from home, For what it is worth, we note that the First, as discussed earlier in the report, especially in 13 respondents who reported accidents the NSPOF survey explores the char- an urban do not represent a cross section of gun- acteristics, acquisition, and uses of one owning households. Of these 13, 11 gun that is randomly selected from area, is are either black or Hispanic, and most among each gun-owning respondent's problematic are of less-than-average income and stock. One of the questions asked of because it education. With only one clear excep- each respondent is "Where do you usu- tion, the accidents did not result from ally keep this gun?" with two of the places the hunting or target shooting. Seven of the possible response categories given by public at risk 10 guns that respondents could iden- "with you, on you" and "in car, truck." ifthe carrier tify by type were handguns. Most of Thus, analysis of NPSOF data allows the accidents did not occur in the us to estimate the proportion of the is reckless or respondent's home. gun stock (and separately of the hand- criminal. gun stock) that is "usually" carried, and Gun Carrying to distinguish between guns that are Carrying a gun away from home, es- carried directly on the owners from pecially in an urban area, is problem- those that are stored in motor vehicles. atic because it places the public at risk Second, the NSPOF also contains a se- if the carrier is reckless or criminal. For ries of questions directly concerning that reason, carrying a gun in a vehicle gun carrying for protection. Respon- or especially on one's person is sub- dents are asked whether the gun car- jected to a variety of state and local rying was in connection with their regulations. In particular, carrying a work, how many days during the past concealed gun is generally prohibited year they carried a gun, whether they

Police Foundation Numberof GunsCarried Regularly

How Males Females Total Carried (N=562) (N=168) (N=730)

Percent

Guns carriedregularly: On person orin vehicle 15.9 18.4 16.4 On person 9.8 4.3 8.7 In vehicle 6.1 14.1 7.7

Millions

Handguns carriedregularly: On person orin vehicle 7.8 On person 4.8 In vehicle 3.0

carried the gun on their person or in a ally) for the purpose of self-protection. motor vehicle, and exactly what type Tables 5.10 and 5.1 1 report the results. of gun they carried for protection. About 14.2 million adults carried a gun for protection at least once during the The NSPOF Table 5.9 contains our estimates for the 12 months preceding the survey-9.8 proportion of tlie handgun stock that data provide million men and 4.4 million women. is regularly kept on the respondent or Four million of these indicated that they two separate in the respondent's car-that is, "car- carried a gun for protection "in con- opportunities ried." We estimate that 10 million hand- nection with work." Two-thirds of those guns, one-sixth of the total stock, are to examine who carried a gun kept it in their ve- regularly carried, lialf on the person and hicles, while the others sometimes car- gun-carrying half in a vehicle. An additional half ried it on their persons. behavior million long guns are carried regularly. among the The second approach to measuring the The occupations for respondents who adult prevalence of carrying is probably of reported carrying a gun in connection population. greater interest, since it provides a bet- with work were quite diverse. Some- ter sense of how many people carry what surprisingly, only a quarter of this guns, and allows us to focus on those group was employed in the protective who carry firearms (at least occasion- service field. The executive/manageriaI,

Guns in America Prevalenceof GunCarrying

Whyand How Carried Males Females All Adults (N=1 ,I82) (N=1,380) (N=2,563)

Percent

Guncarried: 10.9 Relatedto work 3.3 Not relatedto work 7.6

On person 1.8 In vehicle 4.7 On person andin vehicle 1.1

Millions

Guncarried regularly: 9.8 Relatedto work 3.0 Not relatedto work 6.9 On person 1.6 In vehicle 4.2 On personand in vehicle 1.O About 14.2 million professional specialty, and transposta- 30 days per year, but a substantial tiodmaterial-moving occupations each minority rarely leaves home without adults accounted for 10 percent. Whether re- a gun. The guns in question are al- carried a gun spondents were formally required by most always handguns (93 percent). for protection their employers to carry firearms as past Some correlates of gun carrying (not of their occupational duties, or if inat least once shown in the tables) are worth not- stead they carry guns on their own ini- ing. Males who have carried guns during the tiative because of perceived danger, is over the past year are about two and 12 months not clear from the survey data. In any one-half times as likely to have been event, we estimate that 3 million adults preceding arrested for a nontraffic offense as who were not in law enforcement or other men (15 percent versus 6 per- the survey- security carried firearms for protection cent). And a disproportionate share 9.8 million on the job. of gun carriers reside in the South, men and Table 5.11 analyzes gun carrying that where the prevalence of carrying is is not related to work. The majority almost double that of the rest of the 4.4 million of those who carry do so less than nation. women.

Police Foundation Frequencyof Gun CarryingNot Relatedto Work

PercentDistribution

Numberof days in past 12 months gun carried onperson: (N=61) 0-7 8-30 31-179 180-365 Every day

Averagenumber of days in past 12 months gun carriedon person 124days

Numberof days in past 12 months gun carriedin motor vehicle:(N=133) 0-7 8-30 31-1 79 180-365 Everyday

Averagenumber of days in past12 months gun carriedin motor vehicle 99.9 days

Guns in America Introduction conducted explicitly for this purpose (Kleck and Gertz 1995). The 2.5 mil- The annual number of defensive gun lion figure has been picked up by the uses (DGUs) is frequently invoked as press and now appears regularly in The NSPOF a measure of the benefits of private gun newspaper articles, letters to the edi- ownership. It is typically compared to data indicate tor, and editorials, and even in Con- the costs as measured by the number gressional Research Service briefs for that in 1994 of violent crimes committed with a fire- public policymakers. at least 1.5 arm each year. The National Crime Vic- timization Survey (NCVS) provides a The NSPOF survey is quite similar to million adults relatively uncontroversial estimate of that conducted by Kleck and Gertz used a gun the number of gun crimes-1.3 million (1995), and provides a basis for repli- defensiuely in 1994 (BJS 1996). The NCVS also pro- cating their estimate. The NSPOF data vides estimates of the number of DGUs, indicate that in 1994 at least 1.5 mil- against recently averaging about 65,000 per lion adults used a gun defensively another year (McDowall and Wiersema 1994). against another person, a figure that is ... But other surveys provide a basis for much closer to Kleck and Gertz's 2.5 person far-higher estimates (Kleck 1988). The million figure than to the NCVS-based most recent and noteworthy estimate estimates. Further, many of the NSPOF of the number of DGUs is 2.5 million respondents who indicated a DGU in per year, based on data from a nation- the preceding year said that they had ally representative telephone survey also used their gun defensively on one or more other occasions. Taking ac- dent was the victim or the perpetrator, count of these multiple reporters, the and whether the respondent's actions NSPOF data suggest that from 4 to 23 were otherwise legal, reasonable, and million DGUs occurred in 1994 (de- in the public interest. pending on which definition of a DGU is used). We begin by reviewing the previous literature on defensive gun uses in the Our discussion of these results focuses U.S. The third section presents esti- on two issues. The first is whether they mates of DGU incidents based on data are credible. Respondents who re- from the NSPOF, following the meth- ported a DGU were asked a number ods used in earlier analyses based on of questions about the circumstances surveys of this sort. The fourth section Most and results of their action. That infor- provides a discussion of the results. mation provides the basis for estirnat- commentators ing a number of statistics for which the have assumed true values are known with some de- How Many Defensive Gun that. gree of accuracy. For example, the Uses? Previous Findings .. NSPOF data imply that there were over Previous estimates of the number of [defensive 100,000 criminals shot by their victims DGUs come from surveys of nation- gun uses] are in 1994. That figure can be compared ally representative samples. Here we actions that with estimates from other sources about review the results from the National the number of people treated annually Crime Victimization Survey (NCVS), would be for gunshot wo~tndsin the United and then from several telephone sur- endorsed by States. This and other such compari- veys conducted by private polling an impartial sons, as well as puzzling inconsisten- firms. cies in several DGU reports, suggest observer who that the NSPOF data on DGUs is grossly NCVS-Based Estirnates knew all the in error. The NCVS is conducted by the Census Bureau for the U.S. Bureau of Justice facts. The second issue we explore is the Statistics, and involves in-person inter- value of these estimates for the ongo- views with every person age 12 and ing debate over the public value of above in a nationally representative private gun ownership. Most comtnen- sample of 56,000 households. Each tators have assumed that the DGUs household is interviewed once every reported by survey respondents are six months, and households are re- actions that would be endorsed by an tained in the sample for seven inter- impartial observer who knew all the views over the course of three years. facts. Yet the sketchy and unverified The NCVS asks respondents who have accounts available from surveys leave been the victim of a crime in the pre- considerable uncertainty about what ceding six months whether they "did actually happened, whether the respon- or tried to do [anything] about the in-

Guns in America cident while it was going on?" If so, Previous Telephone Survey-Based respondents are asked to describe their Estimates actions; among the possible response From 1976 to 1994, various one-shot codes are "attacked offender with gun; commercial surveys have included fired gun" and "threatened offender questions about DGUs, though none with gun." of the surveys were designed exclu- In research based on the NCVS, DGUs sively to examine this issue (Kleck and have been defined as those instances Gertz 1995, 157). The surveys differ for which the respondent reported re- along various dimensions: sample popu- sisting by either attacking the offender lation (non-institutionalized adults ver- with the gun or threatening to do so. sus registered voters, national samples The following estimates of the annual versus citizens from a particular state); DGU count have been published: whether the DGU questions were asked ... the of all survey participants, or only those NCVS is NCVS who met specific criteria such as gun Estimate CircumstancesSource ownership; whether a distinction was the 'kold 68,000 Assaultand Kleck (1988) made between uses against animals and standard" robbery, uses against people; and the time pe- of crinzinal 1979-85 riod over which respondents were asked 80,000 All violent crimes Cook (1991) victimization and burglary, to recall defensive gun uses (lifetime, past 1979-87 five years, or past one year). surveys in 65,000 All violent crimes McDowall and burglary, & Wiersema Kleck and Gertz (1995, 182-183) com- terms of such 1987-90 (1 994) pute the number of defensive gun uses criteria as against people suggested by each of sample size, The reliability of NCVS-based estimates these studies. The estimates range from has been questioned by Kleck and 770,000 to 3.6 million defensive gun response Gertz (1995, 154-5), who developed uses per year. They focus attention on rate, and several arguments for why the NCVS two of the surveys which, they sug- may understate the true count. Below, gest, are of particularly high quality. methodologi- we evaluate some of the issues they The 1981 survey by Hart Research As- cal sophisti- raise. For now it is sufficient to note sociates "implied a minimum of about .. that the NCVS is the "gold standard" of 640,000 annual DGUs involving hand- cation. criminal victimization surveys in terms guns. .. (p. 158)." Extrapolating the Hart of such criteria as sample size, response handgun estimates to all gun uses, they rate, and methodological sophistication. estimate 1.8 million defensive gun uses Estimates of DGUs based on the NCVS per year. cannot be lightly dismissed. On the The most recent telephone survey was other hand, the arguments by Kleck and conducted by Kleck and Gertz (1995, Gertz deserve close attention. 161) explicitly for the purpose of

Police Foundation estimating the annual incidence of de- The DGU estimates were standardized fensive gun uses. Kleck and Gertz to show the estimated number of DGUs oversampled males in the South and against humans that did not involve West regions in order to produce an actions by police, military, or protec- acceptably large number of DGUs, and tive service personnel acting in the line weighted the data to generate nation- of duty. Kleck and Gertz (1995,162-63) ally representative population estimates applied additional, more stringent, cri- ... a series of (161). A total of 4,977 households were teria in producing their own estimates contacted in the Kleck and Gertz sur- for the annual number of "genuine" one-shot vey, of which 1,832 cases completed DGUs, as follows: commercial the full questionnaire-all respondents (1) The defensive use "involved actual reporting a DGU plus a randomly se- telephone contact with a person, rather than lected third of respondents who did not merely investigating suspicious cir- surveys o$.. report a DGU. An additional 3,145 were cumstances"; [defensive gun terminated by Kleck and Gertz's inter- uses] have viewers once it was determined that (2) "The defender could state a spe- they had not participated in a DGU. cific crime which he thought was produced being committed at the time of the estimates that Kleck and Gertz's DGU question reads incident"; as follows (p. 161): "Within the past are one or fiveyears, have you yourself or another (3) The respondent "used" the gun ("at two orders of member of your household used a gun, a minimum it had to be used as magnitude even if it was not fired, for self-protec- part of a threat against a person, tion or for the protection of property either by verbally referring to the larger than at home, work, or elsewhere? Please gun ... or by pointing it at an ad- those do not include military service, police versary"). produced work, or work as a security guard" [em- Incidents that met all these criteria were using ... phasis in original]. Respondents an- used to calculate what they termed A- swering in the affirmative were then type estimates for one- and five-year National asked whether the DGU was used to prevalence. Kleck and Gertz also pro- Crime protect against an animal or person, duced more conservative B estimates Victimization and also to provide the number of that apply the additional DGUs in which the respondent was restriction: Survey data. involved over the past five years. (4) The respondent was not employed From the sample of almost 5,000 adults, by the police, military, or protec- 244 respondents indicated some kind tive service industry (regardless of of defensive gun use over the past five whether the most recent DGU oc- years. Of this group, 22 (9 percent) curred at work), and the record indicated the most recent defensive gun from the interview was complete use was against an animal. in all relevant respects.

Guns in America Kleck and Gertz's A-type estimate for involved, including the respondent's ac- the number of adults involved in a tions with the gun, the location and DGU during the previous year is the other circumstances of the incident, and well-known 2.5 million. Their B-type the respondent's relationship to the estimate is 2.2 million. perpetrator. In sum, a series of one-shot commercial The Chilton interviewers were also telephone surveys of DGUs have pro- asked to provide their own assessment duced estimates that are one or two or- of whether the respondent was invent- ders of magnitude larger than those pro- ing the most recent DGU incident. duced using the NCVS data. In the next NSPOF Estimates section, we use the NSPOF to develop Given that the NSPOF is quite similar estimates for the number of defensive to the survey reported in Kleck and The NSPOP. .. gun uses and users as a first step in at- Gertz with respect to the instrument, tempting to resolve this discrepancy. sampling procedure, and interviewing includes ... method, we would expect similar re- the number of How Many Defensive sults on the number of defensive gun ...[d efensiue Gun Uses? NSPOF-Based users each year. As it turns out, we find Estimates that NSPOF-based estimates of the gun uses] number of defensive gun users DGU Questions in the NSPOF during the (DGUers) are lower but compatible (in Each of the 2,568 respondents in the preceding a statistical sense) to those produced NSPOF were asked the question: by Kleck and Gertz. The NSPOF has year by those "Within the past 12 months, have you an advantage over the Kleck and Gertz yourself used a gun, even if it was not respondents survey in that it includes an item on fired, to protect yourself or someone the number of DGUs during the pre- who had at else, or for the protection of property ceding year by those respondents who least one... at home, work, or elsewhere?" Answers had at least one; thus we are able to in the affirmative were followed with estimate the number of defensive gun "How many different times did you use uses, which is several times as large as a gun, even if it was not fired, to pro- the number of users. tect yourself or property in the past 12 months?" [emphasis in original]. Nega- Table 6.1 contains results from the tive answers to the first DGU question NSPOF on the number of defensive gun were followed by "Have you ever used users each year. We exclude from our a gun to defend yourself or someone calculations those respondents whom else?" Respondents who answered yes the Chilton interviewers suspected of to either of these DGU questions were fabricating the most recent DGU inci- then asked a sequence of 30 additional dent. As shown in the table, 54 respon- questions concerning the most recent dents reported a defensive gun use DGU in which the respondent was during the past 12 months, which

Police Foundation DefensiveGun Uses Reportedfor Preceding12 Months

Definition (CumulativeExclusions) Percent Millions

Total (N=54) 1.93

Exclude ifagainst animals (N=45) 1.64 Exclude ifmilitary use (N=38) 1.44 Exclude ifno reportof specific crime (N=37) 1.29 Excludeif no reportof specific use of gun (N=26) Excludeif did not see perpetrator(N=19) 0.77 Exclude ifwork-related DGU (even if not militarylprotective service job) (N=l8) 0.75

... almost projects to 3.7 million adults. A major- number of DGUs in the preceding year ity of these may be excluded for the provides the basis for estimating the halfof reasons indicated, as in Kleck and population total, which turns out to be respondents Gertz (see above). In what follows, 23 million. [who reported we use several operational definitions of DGU, with different sets of the There were 112 respondents who re- a defensive above exclusions. ported at least one DGU against a per- gun use] son during the previous fiveyears. Table 6.2 provides a variety of estimates reported They represent 7.8 million adults, or of the number of DGUers and DGUs. 4.1 percent of the population (plus or multiple The first row includes all NSPOF re- minus 0.6 percent). DGUs over spondents who reported a DGU against a person. There were 45 such respon- In the third column of Table 6.2, we the past year; dents for the preceding year, represent- apply the Kleck and Gertz (1995) cri- one woman ing 3.12 million adults, or 1.64 percent teria for "genuine" DGUs (type A), leav- reported 52 of the adult population. As it turns out, ing us with just 19 respondents. They almost half of these respondents re- represent 1.5 million defensive users. DGUs. ported multiple DGUs over the past This estimate is directly comparable to year; one woman reported 52 DGUs. the well-known Kleck and Gertz esti- Incorporating the information on the mate of 2.5 million, shown in the last

Gunsin America DefensiveGun Use, One-and Five-Year-RecallPeriod

All Selected Kleckand NSPOF NSPOFCases Gertz A-Type A-Type

1 year (N=45) (N=19) (N=66)

Defensivegun users (millions) 3.12 1.46 2.55 Defensivegun users (as percentof population) 1.64 0.77 1.33 Defensivegun uses (millions) 23.0 4.7 2.6

5 years (N=113) (N=5 1) (N= 165)

Defensivegun users (millions) 7.87 3.23 6.37 Defensivegun users 4.14 1.70 3.32 (as percentof population)

Standarderror (0.56) (0.36)

Note:In Kleckand Gertz's 1995 DGUstudy, A-type estimates meet certain criteria.See text for explanation column. While ours is smaller, it is sta- the DGUs involved more than one per- ,,.in over 40 tistically plausible that the difference petrator; in most cases (69 percent), the is due to sampling error. Note that perpetrator(s) were strangers to the percent of when we include the multiple DGUs victim. defensive reported by half our 19 respondents, Handguns were used by defenders in uses the gun our estimate increases to 4.7 million about three-quarters of these incidents, DGUs. had been kept and in over 40 percent of defensive uses the gun had been kept either di- either directly Circumstances and Outcomes rectly on the respondent or in the on the While the NSPOF includes a number respondent's vehicle, as shown in Table respondent of items on the circumstances and out- 6.5. Usually the victim confronted the comes of each DGU, our exploration perpetrator with a loaded gun. In the or in the is limited by the small sample size. We cases in which the gun was not already respondent's focus on the 85 respondents who re- loaded (27 percent), most respondents .. port a civilian DGU against a person proceeded to load the gun before vehicle. during the past 5 years. facing the perpetrator. In 15 percent of As shown in Table 6.3, 60 percent of the cases, the gun that was used in the DGUs occurred in or near the victim's defense did not belong to anyone in home. As shown in Table 6.4, half of the respondent's household.

PoliceFoundation Circumstancesof Defensive Gun Use Against Humans, Previous Five Years*

NSPOFQuestion Percent Distribution

Where didthe defensive gun use take place? (N=82) Insiderespondent's home Near respondent'shome Inlnearfriendlrelative's home 13.7 At or near work 3.8 Commercial place(bar, gas station, shoppingcenter) 7.2 Parkinglotlgarage 8.2 Streetlpublictransportation 6.3 Other 1.1

*ExcludingLaw Enforcement

Characteristicsof Perpetrator in DefensiveGun Use Against Humans, Previous Five Years*

NSPOFQuestion Percent Distribution

Didrespondent see person defending against? (N=83) Yes 77.6

How manypeople wasrespondent defending against?(N=67) 1 49.5 2 16.3 3 or more 34.2

Respondent'srelationship to perpetrator:(N=84) Stranger Friendlrelative Boylgirlfriend (current orex)

*ExcludingLaw Enforcement.

Gunsin America Characteristicsof Gun in DefensiveGun Use Against Humans, Previous Five Years*

Characteristic Percent Distribution

Typeof gunin defensivegun use: (N=81) Handgun Longgun Other

Ownerof gunin defensivegun use: (N=84) Respondent Someonein respondenrshousehold Someoneout of respondent'shousehold

Whenrespondent first wanted to usegun for protection, wherewas gun stored? (N=79) Bedroom Gun cabinet Other closet Other householdlocation Withlonrespondent In carltruck

Wasgun already loaded?(N=82) Yes No, but respondentloaded gun duringconfrontation No

*ExcludingLaw Enforcement. As shown in Table 6.6, the defender of DGUs the victim captured and held The fired his or her gun in 27 percent of the perpetrator at gunpoint until the these incidents (combined "fire warn- police could arrive. perpetrator ing shots" and "fire at perpetrator" per- was wounded centages, though some respondents re- The perpetrator was armed in 40 per- by the crime ported firing both warning shots and airn- cent of these cases. Half of armed per- ing at the perpetrator). Forty percent of petrators had a gun, and in 30 percent victim in eight these were "warning shots," and about of the cases in which the perpetrator percent of all a third were aimed at the perpetrator had a gun (6 percent of the total) the ...[d efensiue but missed. The perpetrator was victim reported having been fired upon. wounded by the crime victim in eight In 45 percent of DGUs the respondents gun uses]. percent of all DGUs. In nine percent believed that they or someone else would

PoliceFoundation Circumstancesand Outcomes of DefensiveGun Use Against Humans, Previous5 Years*

PercentDistribution

Did perpetratorknow respondent had gun?(N=79) Yes

Whatdid respondentdo in the defensivegun use?(N=85) Tell perpetratorrespondent had gun Showgun toperpetrator Point gunat perpetrator Usegun asclub to strike Firewarning shots Fireat perpetrator Captureperpetrator, hold until policearrive Woundlkillperpetrator

Whatwould have happenedif respondenthad no1 usedgun? (N=79) Improved situation Madeno difference Made situation worse

How likely thatsomeone would have been killed if respondenthad notused gun? (N=76) Very unlikely Somewhat unlikely Likely Somewhat likely Very likely

*ExcludingLaw Enforcement, Continued

Guns in America Continued

PercentDistribution

Did perpetratorthreaten, attack,injure respondent? (N=85) Noneof these Threatenedonly Attackedlnotinjured Attackedand injured

Whowas first to attack withphysical force? (N=17) Respondent Perpetrator Someoneelse

Did perpetrator have weapon?(N=84) Yes No Don't know

Whatkind of weapondid perpetrator have?(N=38) Gun Knifelsharpobject Bluntlotherobject Don't know

Did perpetrator shoot you/someoneat else? (N=21) Yes

Did perpetratorget awaywith moneylproperty?(N=40) Yes

Do thepolice knowabout this incident? (N=84) Yes

Police Foundation have been killed by the perpetrator had each year by armed citizens. Further, they not used a gun in self-defense. guns are used far more often to defend against crime than to perpetrate The police were informed about the incrime. On the other hand, if we reject cident in slightly over half of these cases. these estimates in favor of those based on NCVS data, the reverse is true. It is Defensive Gun Users thus of considerable interest and impor- Table 6.7 presents descriptive statistics tance to check the reasonableness of the for three distinct groups: the 85 respon- NSPOF estimates before embracing dents who reported a defensive gun them. use against a human during the past five years; gun owners who have never As a guide to how to proceed, we note reported a DGU; and those respondents Max Singer's discussion of "mythical [Defensive who did not own a gun. In compari- numbers": "The main point of this ar- gun users are] son with gun owners, proportionately ticle may well be to illustrate how far ... two and one- more DGUers are female, minority, one can go in bounding a problem by unmarried, and living in an urban area. taking numbers seriously, seeing what halftimes as These findings are all consistent with they imply, checking various implica- likely as other earlier studies of DGU reporters (Kleck tions against each other and against gun owners to and Gertz 1995, 178-9). general knowledge (such as the number of persons or households in the Defensive gun users tend to be bave been city). Small efforts in this direction can young-half under 35, which is nine go a long way to help ordinary people arrested for a years below the average age for other and responsible officials to cope with nontra ff ic gun owners. DGUers are also two and experts of various kinds" (Singer 1971, one-half times as likely as other gun offense, and 9). In this section we follow Singer's owners to have been arrested for a four times as advice, comparing some of the esti- nontraffic offense, and four times as mates from the NSPOF against other likely to bave likely to have been arrested as adults statistics. The results suggest that the who do not own a gun. been arrested DGU estimates are far too high. as adults A more detailed description of DGUers is not possible due to the small sample size. We begin by noting that if only a small who do not fraction of violent crimes result in self- own a gun. defense with a gun (an uncontroversial Resolving the NCVS-NSPOF Discrepancy assumption) then the number of DGUs will necessarily be much less than the Can It Be? Some Troubling number of violent crimes. Table 6.8 Implications presents some implications of the If these numbers are credible, we are NSPOF estimates for annual DGUs. The led to conclude that millions of at- NSPOF estimates suggest that 130,000 tempted assaults and thefts are foiled criminals are wounded or killed by

Guns in America Demographic Characteristicsof Gun Owners, DefensiveGun Users, and Non-Owners

Characteristic Gun Owners DGUs, Non-Gun Who Have Last5 Years, OwnersWho NeverHad ExcludingLaw HaveNever DGU Enforcement HadDGU (N=644) (N=85) (N=1,576)

PercentDistribution Sex: Male Female Race: White Black Other Age: 18-34 29.0 54.4 37.0 35 and over 70.3 43.8 60.7 Averageage 45.8 36.3 44.0

MaritalStatus: Married Widowed Divorced/Separated Nevermarried

Community: Rural Smalltown/city Medium city Suburbsof large city Largecity Income (thousands): $0-30 $30 and over Don't know/missing

Children under18 in household 36.8 44.2 46.9 Respondenthas ever beenarrested for nontrafficoffense 7.8 20.5 4.9

Source:U.S. Bureauof the Censusestimates fromStatistical Abstract of the United States1995. Note: "DGUs" are those who reportat least one defensivegun use against aperson during the preceding five years, not includingon-the-job DGUs by law enforcement officersor protective service workers. Column categoriesare mutually exclusive;DGUs that do not own firearmsare excluded fromthird column.

PoliceFoundation DefensiveGun Use Comparedto TotalCrime Counts

Crime NSPOFDGUs NCVSEstimate Implied by Crime of CrimeCount, "Defense TY Pe 1994 Rate" (Thousands) (Thousands) (Percent)

Rape +Attempted Rape 322 316 102 Assault 834 9,128 9 AggravatedAssault 462 2,478 19 Robbery 466 1,299 36

... the total civilian gun defenders. In contrast, es- law enforcement. We find that possi- number of timates based on data from public- bility rather unlikely. health surveillance systems suggest that As seen in Table 6.9, the NSPOF data people the total number of people nonfatally also imply that as many as 630,000 lives nonfatally shot by a firearm nationwide and are saved each year by defensive gun shot by a treated in an emergency room or hos- uses. By comparison, there were 22,076 pital is about 100,000 (Annest et al. people murdered in 1994 (FBI 1995, firearm 1995). That figure includes assaults, 18). Since the number of homicides is nationwide accidents, and suicide attempts. Add- generally regarded as accurate, we can and treated in ing an additional 16,000 who are shot only think of two explanations to rec- and killed in assaults still leaves us short oncile these two statistics, the first of an emergency of the estimate from the NSPOF for the which is absurd: (1) victims of serious room or number of people shot just in self-de- (potentially lethal) criminal attacks have hospital is fense! Thus if the NSPOF results are firearms available and successfully correct, it must be true that most per- ward off their attackers in about 97 about petrators who are shot during a crimi- percent of all cases or (2) the NSPOF 100, 000. .. nal encounter never receive emergency estimates of the number of lives saved, room treatment for their wounds, and, if not the DGU estimates themselves, incidentally, never become known to are greatly exaggerated.

Gunsin America DefensiveGun Use Reports: Lives Saved

Thousands

How likelythat someonewould have been killed had gun notbeen useddefensively? "Very likely" 'Somewhat likely" "Likely" Total numberof lives saved Total numberof homicidessaved (1994)*

Impliedsuccessful defense rate** 97%

*Homicide estimatesfor 1994taken from FBI's UniformCrime Reports(1 995, 18). **Impliedsuccessful defense rate representspercent of all potentiallylethal attacks that are successfully defended byarmed victim. Calculated as ( A / A + B ), with A = numberof successfulgun defenses by victim in potentially lethalattack (NSPOF estimate), and B = numberof homicides(FBI count).

The evidence of positive bias in the Some Explanations DGU estimates is still stronger when With a sample size of 2,568, each we recall that the DGU estimates are NSPOF respondent represents from calculated using only the most recently 70,000 to 80,000 citizens on average reported DGU incidents of NSPOF re- using the projection weights discussed spondents. As we have seen, about half earlier. As we have seen, the most re- of the respondents who reported a cent NCVS-based estimates suggest DGU indicated that there had been two 65,000 defensive gun uses by citizens or more in the preceding year. While against crime each year (McDowall and we have no details on the circum- Wiersema 1995); on the basis of the stances of those additional DGUs, pre- NCVS figures, we would have expected sumably if the respondents had been one respondent from the entire NSPOF asked they would have reported addi- sample to have reported a defensive tional violent crimes, wounded perpe- gun use during the past year. Instead, trators, and lethal attacks foiled. The 19 reported at least one DGU that meets already improbable figures for the num- our stringent criteria. In this section, ber of crimes defended against with a we explore possible explanations for gun could be magnified still farther. these differences.

Police Foundation Sequence of Questions. The NCVS threaten, attack, or injure you?" A total asks respondents whether they have of six respondents who indicated that been victims of a crime during a speci- the circumstance of the DGU was rape, fied time, and, if so, whether and what robbery, or attack (question 72) defensive actions were taken during the responded "no" to question 75. crime. As a result, the opportunity to The NCVS has a more systematic ap- discuss a defensive gun use is only proach to inquiring about victimization, made available to NCVS participants if and some of these DGUers would not a crime has first been reported. In the have been classified as victims in the NSPOF (as with most telephone surveys), NCVS interview. While the NCVS pro- all respondents are asked whether a gun cedures will eliminate some faulty re- has been used defensively during the ports of victimization, it may also be indicated period of time. In the true that some respondents will choose NSPOF.. .all By construction of our selection crite- not to report crimes that actually respondents ria, each of the 19 NSPOF respondents occurred. are asked indicate that some form of crime was Survey Environment. The NCVS in- involved in their most recent DGU. A whether a terviewing environment is different in small portion of the discrepancy be- potentially important ways from that gun bas tween the NSPOF and NCVS estimates of the one-shot telephone-survey in- been used may be accounted for by the fact that terviews like those of the NSPOF. The in three cases the most serious crime defensively NCVS is conducted face-to-face in the reported is "trespass," a crime which is respondents' homes. The interviewers during the not included in the NCVS. identify themselves as federal employ- ees (working for the U.S. Bureau of indicated Some of the DGU reporters provide a the Census), and promise that all an- period of contradictory account of what happened. swers will be kept confidential. In con- In question 72 they are asked "Which of time. trast, the NSPOF interviewers identified the following best describes what was themselves as from "Chilton Research happening when you used the gun de- Services" and conducted all interviews fensively?" They are given nine options, over the telephone without any prom- and are permitted to answer "yes" to any ise of confidentiality. Which type of in- number of them. Three responded "yes" terview would respondents trust more? to several categories of serious crime (rape, robbery) but also said "yes" to Our presumption is that some respon- the category "no crime was involved." dents would feel more comfortable Another apparent inconsistency ap- speaking with someone in person, es- pears when we compare the responses pecially with the guarantee, but that oth- to this question with the responses to ers would feel more comfortable on the question 75, "Did the perpetrator telephone. Kleck and Gertz's (pp. 154-

Guns in America 6) intuition is quite different than ours. using the five-year-recall period pro- They assert that respondents will be duces annual estimates that are dramati- far less likely to disclose sensitive in- cally smaller than annual DGU esti- formation to the NCVS interviewer than mates derived from the one-year-recall to a telephone interviewer who says period (1.46 million versus 648,000, she is working for a private firm. Kleck using the conservative count for our and Gertz assert that the commercial most stringent criteria). As seen in the telephone surveys produce a much table, this phenomenon is common to higher prevalence of DGUs than the other telephone gun surveys (Kleck and NCVS precisely because there are many Gertz 1995, 165). We would not, of respondents who are unwilling to dis- course, expect the five-year-recall pe- cuss legally dubious actions with gov- riod to produce estimates of defensive ernment workers, but are willing to gun users that are five times as large as Our discuss them with a stranger on the the one-year-recall period, given that telephone. We know of no evidence higher-risk individuals may be victim- presumption that would test this conjecture. ized several times. At the same time, is that some the observation that the one-year-re- In some respects the NCVS is inarguably respondents call period estimates are twice as large superior. While the NSPOF has a sample as the annual estimates produced by would feel of 2,568,with a response rate of 44 or 59 the five-year recall suggests a problem percent, the NCVS has a sample size of more with respondents' memories. It may be 120,000 in 56,000 housing units, and re- comfortable that respondents in the NSPOF include ceived responses from residents in 96 DGUs that occurred more than a year percent of targeted households (BJS speaking with ago in the 12-month-recall question (a 1996). Thus, not only is the NCVS ex- someone in phenomenon known as "telescoping"). pected to provide more reliable estimates To the extent to which this occurs, the person, due to the sheer size of the sample (that one-year-recall period will produce is, NCVS-based estimates are less sensi- especially overestimates of the number of DGUs tive to a few aberrant responses), but with the each year. the differences in response rates also guarantee [of suggest that the NCVS is closer to a truly The NCVS guards against this phenom- representative sample of U.S. adults than enon by re-interviewing respondents c0q-J are telephone surveys. every six months, and using the previ- but that ous NCVS interview as a benchmark for others would Telescoping. Following the conven- respondents for the six-month-recall pe- tion in the literature, our estimates fo- riod. The first interview with each NCVS feel more cus on the number of defensive gun participant is "unbounded," and has been comfortable users and uses during the past year. found to produce far larger estimates for However, as seen in Table 6.2, divid- the six-month recall than subsequent on the ing the estimates for DGUs and DGUers (bounded) interviews (Cantor 1989). telep hone.

Police Foundation False Positives. Prevalence estimates a survey respondent who had recently based on interview data are subject to heard a bump in the night and checked both false negatives (a respondent fails it out, gun in hand, may report having to report a relevant instance) and false scared off a trespasser even though in positives (where a respondent reports fact he or she did not see anyone at a relevant instance that did not actu- the time. ally occur, or did not occur in the relevant time frame). If the true preva- The possibility that some respondents lence is low, as in the case of DGUs, may be confused by the question or then in a sense there is much greater about their own experiences is sug- scope for false positives than false nega- gested by the rather high incidence of tives--only a relatively few respondents mental illness and substance abuse in me are logically capable of giving a false the United States. Recent estimates from negative, whereas anyone who did not the National Institute for Mental Health 51.3 million American possibility use a gun defensively can give a false suggest that adults aged 18 and over have "one or that some positive. If the true prevalence is 1/1,000, more mental or addictive disorders" and the false-positive rate is 2/1,000, then respondents (Bounrdon et al. 1994, 23). Thus, at the estimated DGU rate will be at least any point in time a large proportion of may be twice the true level even if none of the American adults are either under the confused by true DGUers choose to rqort their eqe- influence of some intoxicant or suffer- rience (Hemenway 1996). the question ing from a mental disorder, and in ei- or about Is there any reason to believe that some ther case may be unreliable reporters their own respondents will report DGUs that did in a survey. A representative sample not occur? In addition to the telescop- of American adults will include these experiences is ing problem discussed above, respon- individuals. suggested dents may falsely report because they are confused, have a distorted memory, An additional source of false positives is by the strategic behavior by gun advocates. or are simply having fun with the in- Those who are well informed about the rather high terviewer. incidence of gun-control debate will know that the Research on survey methodology sug- number of DGUs is relevant and may mental illness gests that respondents have a desire to be tempted to enhance that estimate and substance make themselves "look good" in the through their own response to the abuse in the eyes of the interviewer (Sudman and survey. Bradburn 1974,40). Fighting off a crimi- The purpose of this discussion is not United States. nal attack is (in most circumstances) a to claim that every citizen reporting a heroic act. There may be a temptation DGU is mentally impaired or invent- for some respondents either to make ing the incident for whatever reason. something up or else to change the Rather, our intention is to note that a details of an actual event. For example,

Guns in America representative survey of 2,568 Arneri- is, they have public merit in ways that cans that asks questions about any topic other private uses of guns (target shoot- will include at least a handful of people ing, hunting) do not (Cook and Moore who are drunk, have an erratic memory 1995). The cost of any regulation that or an axe to grind, or who are just hav- will deprive some law-abiding citizens ing fun. Given that our estimate of over of guns must be reckoned accordingly. 4 million defensive gun uses rests on If, as suggested by the NSPOF data, it just 19 responses, a handful of false is quite likely that a law-abiding gun positives would make a big difference. owner will have occasion to use the gun in self-defense against a robber or Of course it is possible that there are burglar, then the social cost of restrict- also one or more false negatives in this ing ownership and use may be sub- survey. We focus on the problem of stantial. On the other hand, if the likeli- false positives because of the logic of The hood of virtuous self-defense is minute, estimating rare events, and because we controversy as suggested by the NCVS data, then we have been persuaded by the evidence reach quite a different conclusion. offered earlier that the NSPOF estimates over the overstate the true incidence by a very The discussion above has focused on frequency wide margin. demonstrating that despite a number with which of surveys that seem relevant, includ- Finally, while the NSPOF estimate ap- ing the NSPOF, we remain highly un- guns are used pears too high, that does not imply that certain about the actual number of in self-defense the far lower NCVS estimate is correct. genuine DGUs that occur each year. The fact that the NCVS only asks DGU is animated The number that is in wide circulation, questions for those who report a crime 2.5 million, is lower than our best esti- by the notion eliminates the false-positive problem, mate based on a literal interpretation that such but may cause the NCVS to miss of NSPOF data. But there are numer- DGUers who do not remember or uses are vital ous reasons, both empirical and con- choose not to report the crime. The ceptual, to believe that this NSPOF es- and virtuous. rather frustrating conclusion is that the timate is far higher than the underly- available survey data leave considering reality. The tmth eludes this method able uncertainty about the "true" num- of measurement, because even a hand- ber of DGUs. ful of false reports are sufficient to greatly distort the conclusion. Interpretation of Defensive But there is a more fundamental prob- Gun Use Estimates lem here. Even if we could design a The controversy over the frequency questionnaire so cleverly as to weed with which guns are used in self- out misinformation, there would remain defense is animated by the notion that a problem in interpreting the results. such uses are vital and virtuous: that Does the number of DGUs serve as a

Police Foundation measure of the public benefit of pri- not only to the tragic cases in which vate gun possession, even in principle? someone shoots a member of their fam- When it comes to DGUs, is more bet- ily after hearing a noise in the night, but ter? We note several problems: also those cases in which the intruder is perhaps trespassing but poses no physi- 1. Gun use may take the place of other cal threat to the household. means of avoiding trouble. Someone who has a gun handy will be inclined 3. 7%e number of DGUs tells us little to use it when there is a perceived about the most important effects on threat to person or property. But other crime of widespread gun ownership. means of defense, such as calling for When a large percentage of households help or leaving the scene, may be just and even people on the street are as effective. Access to a firearm may armed, that circumstance presumably ... surveys are encourage some people to be less pru- has an important effect on the behav- a decidedly dent about avoiding confrontations and ior of predatory criminals. Some may flawed may enable or embolden others to es- be deterred or diverted to other types calate confrontations. The logic here of crime. Others may change their tac- method for extends to preventive activities as well. tics, acquiring a gun themselves or in learning Gun possession may encourage some some other way seeking to preempt about the people to invest less in other means of gun use by the intended victim (Cook self-protection or to be less vigilant in 1991). Such consequences presumably frequency avoiding unsafe situations. The NSPOF, have an important effect on criminal with which as with other surveys, does not pro- victimization rates, but are in no way innocent vide sufficient information about what reflected in the DGU count. alternative courses of action were avail- To sum up, surveys are a decidedly victims of able to its respondents. Without this flawed method for learning about the information, it is not possible to deter- crirne use a frequency with which innocent victims mine whether many of these gun uses gun to defend of crime use a gun to defend them- were beneficial to society. themselves. selves. On the other hand, even if we 2. Readiness to use guns in self-defense could develop a reliable estimate of may lead tofatal m&takes. Someone who this frequency, it would only be of mar- keeps a gun handy for dealing with in- ginal relevance to the ongoing debate truders and other predators may end up over the appropriate regulation of fire- shooting the wrong person. We refer here arms commerce, possession, and use.

Gunsin America VII

"One of the few constants in American Support for Handgun public opinion over the last two de- Control Measures: How Registration cades has been that three-fourths of the Much and from Whom? population supports gun control," and police Each respondent in the NSPOF was wrote Tom W. Smith in 1980. Now, asked the following questions: "Wbuld background years later, this consistently high level you favor or oppose a law which re- of support remains. In the 1994 Gen- checks have quired private citizens to register hand- eral Social Survey, 78 percent favored long been in guns they own?" "Would you favor or a law that would require a person to oppose a law which made it illegal for favor, but an obtain a police permit before buying a any private citizen, other than law en- gun (BJS 1995, 193). But there are lim- outright ban forcement officers and licensed secu- its to how far this majority has been on handgun rity guards, to own a handgun?" Simi- willing to go. Registration and police lar items have been included in a vari- ownership background checks have long been in ety of national surveys over the years favor, but an outright ban on handgun has been (Kleck 1991). ownership has been opposed by a clear opposed by majority (BJS 1995, 191). The NSPOF The Gallup Monthly Poll from time to a clear includes two items on attitudes toward time has asked, "V170uld you favor or gun-control measures, and the results oppose the registration of all hand- majority .... conform with this general pattern. guns?" During the 1980s support in- Supportfor Handgun Registration Supportfor HandgunBan

Percent Percent

loo 1 loo 1

In favor Not in favor in favor Notin favor

... every creased from 66 percent (1982) to 81 Table 7.1 details the patterns of support percent (1990) and has remained at that for these two measures across socio- subgroup is high level since then. The NSPOF-based demographic groups, using NSPOF data. solidly in estimate is lower, with 71 percent indi- Overall, we see every subgroup is sol- support of cating support for handgun registration. idly in support of registration, whereas registration, The Gallup Poll has also included an- no subgroup shows majority support for other NSPOF item, the ban on hand- a handgun ban (though nearly half of whereas no gun ownership, for a number of years. Hispanics favor such a ban). The pat- subgroup While a majority were in favor when terns of support for the two policies are shows the question was first asked in 1959, closely linked for several dimensions, the proportion dropped to as low as beginning with the great sexual divide: majority 31 percent in the mid-1970s, and has women are more supportive of both support for a plateaued at about 40 percent since policies then men by about 14 percent- handgun 1981 (Kleck 1991, 366, 378; BJS 1995, age points. Whites are less likely than 191). Again the NSPOF estimate is African-Americans and Hispanics to fa- ban .... lower than the most recent Gallup re- vor either handgun registration or a ban. sults, with just 32 percent supporting a No clear pattern emerges across age ban. groups.

Gunsin America Viewson GunControl by Sociodemographic Characteristics

Characteristic FavorHandgun Registration FavorHandgun Ban

Percent

Total (N=2,568) 70.5

Sex: Male (N=l, 186) Female(N=1,382)

Race: White (N=1,626) Black (N=430) Hispanic(N=407)

Age: 20-24 (N=195) 25-34 (N=581) 35-49 (N=873) 50-64 (N=465) 65 plus (N=360)

Education: Less than highschool (N=358) 74.1 High school (N=858) 65.7 Somecollege (N=73 1 ) 71.2 Collegedegree (bachelor's) (N=291) 78.5 Some post-college(N=70) 84.3 Advanceddegree (N=224) 67.4

Continued

Police Foundation Viewson GunControl by Sociodemographic Characteristics (Continued)

Characteristic FavorHandgun Registration FavorHandgun Ban

Percent

Community: Rural (N=446) Small town (N=777) Medium city(N=42 1) Suburbs(N=3 19) Large city (N=566)

Census Region: Northeast(N=476) Midwest(N=620) South (N=972) West (N=500)

Total familyincome (in thousands): $0-1 0 (N=293) 10-20 (N=422) 20-30 (N=382) 30-50 (N=579) 50-75 (N=3 12) 75 plus (N=200)

Education and Education and income both evince a either measure than residents of towns income both U-shaped pattern of support for a hand- and cities. And the South and the gun ban. High-school dropouts and Northeast divide over these issues in evince a those with postgraduate education are the expected way. U-s haped most supportive, while those with pattern of "some college" are least supportive. Gun Involvement and Similarly, those in the midrange of in- Support for Handgun Control support for a come ($20-50,000) are least support- In Table 7.2, we disaggregate respon- ive, with the poor and the rich some- handgun ban. dents on the basis of their involvement what more so. with guns, beginning with their cur- Rural communities, where the gun cul- rent ownership. As expected, adults ture is strongest, offer less support for who either own guns or live in a house-

Gunsin America Viewson Gun Control by Extentof Involvementwith Guns

Favor FavorMaking Handgun HandgunsIllegal RegistrationLaw for PrivateCitizens

Percent

R ownsgun (N=789) Male (N=602) Female(N=187)

R does not own gun(N=1,658) Male(N=525) Female(N=1 ,133)

R ownsgun, employed in protectiveservice job (N=38)

R ownsgun, not employed in protectiveservice job (N=744)

R does not owngun, lives in householdwith gun(N=396) Male(N=60) Female(N=336)

Parents keptgun in home: Yes (N=1,454) NO(N=1,087)

R huntedin last year (N=258)

Rwent sportshooting last year (N=296)

R carried gunfor protection last year (N=268)

Militaryservice: R Veteran(N=380) R Armedforces personnel on active duty(N=135) Never in armed forces(N=2,013)

R = Respondent

Police Foundation hold with a gun are less likely to sup- or carry a gun for protection. The low port either handgun control measure level of support among veterans may than other adults. Yet note that over reflect their active involvement with half of all gun-owning men and nearly guns, as documented earlier. three-quarters of gun-owning women support handgun registration. And it is Support for Handgun Control surely intriguing that 11 percent of gun and Concern about Crime owners and 26 percent of nonowners Do concerns about crime, or actual vic- who live with an owner favor a ban timization experiences, increase or de- on handguns. In fact, 10 percent of crease a respondent's support for hand- handgun owners favor a ban. Support gun control? Handguns are the pre- for handgun registration and a hand- ferred gun both for perpetrating crime gun ban are highest among those gun ... over and for defending against it, and in that owners employed in protective service sense restrictions on handguns cut both half of all occupations, though the sample size ways. What we actually find in tlie gun-owning for this group is somewhat small. NSPOF results is an interesting differ- men and How can we explain why a subset of ence between popular views of the two handgun owners would actually en- control measures. neap-2~+ three- dorse an outright ban on handguns? quarters of In Table 7.3, concern about crime is They may have been confused about related to support for the handgun con- gun-owning the question. Or they may have been trol measures. In comparison to those indicating a genuine willingness to ex- women with a more sanguine view, those who change their own handgun for the support believe that crime is going up in their chance to live in a handgun-free neighborhood are more likely to sup- handgun community. port handgun registration but less likely registration. Tliose respondents who are most ac- to support a handgun ban. This pat- tively involved with guns are less likely tern may reflect a belief that registra- than other gun owners to support ei- tion would give an advantage to law- ther control measure. In Table 7.2 we abiding citizens over the criminal ele- see the relatively low levels of support ment but that an outright ban would by those who hunt, go sport shooting, have the opposite effect.

Guns in America Viewson GunControl in Relationto Concernsabout Crime

Favor FavorMaking Handgun HandgunsIllegal Registration Law for Private Citizens

Percent

Perceivedtrend incrime rates inneighborhood over past12 months: Going down (N=165) 63.8 Staying the same (N=1,801) 71 .O ... 11 percent Going up (N=505) 69.1 of sun Robbedlattackedin past12 months (N=145) owners and

Feelingsof safetywhile out alone %percent of in the neighborhood at night: nonowners Very or somewhat safe (N=1,888) 69.5 Very or somewhat unsafe(N=588) 73.8 who live with an owner favor a ban on handguns.

PoliceFoundation Gunsin America VIII

The NSPOF provides uniquely detailed The most important methodological The most data on the stock of firearms in private problem explored in this report is the hands, the uses and misuses of these likelihood of a large positive bias in important firearms, and the recent transactions in estimates of the incidence of rare ...problem which firearms have changed hands. events. We have offered evidence that explored in These survey data are not without the NSPOF-based estimates of the num- flaws. We have taken care to document ber of defensive gun uses and of the this report is the methods used in gathering these number of firearm accidents are too the likelihood data and point out potential problems. high by an order of magnitude or more, of a large The response rate is uncomfortably and that many of the DGU reports con- low, although in line with other tele- tain puzzling internal contradictions or positive bias phone surveys of this sort. More no- otherwise don't make sense. The prob- in estimates table for students of survey design, we lem comes down to a matter of arith- of the find that female respondents are sub- metic. If the true incidence is 1 in 1,000, stantially less likely to report their say, then even a small false-positive rate incidence husband's guns than are the male own- among the other 999 will result in an of rare ers themselves. As a result, we limit estimate that is far too high. False posi- most of our analysis to respondents' tives may originate from telescoping, events ...[s uch reports concerning their own guns, confusion, a desire to impress the in- as] defensive which appear accurate. terviewer, and other causes. This gun uses. problem is generic to any method of much more likely than long guns measurement but has not been ad- to be kept unlocked and ready for equately acknowledged or dealt with use in the home as well as to be in the field of survey research. Thus, carried in public, but they are much one of our most important conclusions less likely to be used in sporting is negative: do not place faith in esti- activities. Despite this difference in mates of the number of defensive gun purpose, the demographic and so- uses-or indeed, in any survey-based cioeconomic patterns of ownership estimate of the incidence of a rare are very similar for handguns and event-unless there is a credible long guns. In fact, most handgun mechanism for screening out false re- owners also own one or more long ports. Note that the NCVS deals with guns. It seems fair to conclude that ... do not the false-positive problem by preempt- the more fundamental divide among ing it: only respondents who say they people is not with respect to their place faith in were the victim of a crime are asked felt needs (sports or self-protection) estimates of about self-defense measures, so the vast but rather with respect to their level ... defensive majority of respondents never have a of comfort and experience with chance to offer a report, false or other- guns. Those who like guns, have gun uses-or wise, on defensive gun uses. experience with them, and have the indeed, in any means to buy them tend to keep a For measuring other characteristics of number of them for various pur- survey-based gun ownership and use, we do find poses. Most of the adult public turns the NSPOF credible. Among the more estimate of elsewhere for recreation and self- notable findings: the incidence protection. II While there are enough guns in of a rare Over time, the self-protection and private hands to provide every sporting uses have been changing event-unless adult with one, in fact the pattern in relative importance in motivating there is of ownership is quite concentrated. gun acquisition and use. Perhaps as Only a quarter of adults own a gun, a credible a result of the increasing urbaniza- and on the average owners have tion of the United States, the overall mechanism about four guns each. Just 10 mil- prevalence of ownership of guns for screening lion adults own over half of the 200 appears to be declining, as is par- million guns. out false ticipation in hunting. Proportionately reports. II When asked, handgun owners usu- fewer households own guns now ally give self-protection as their than in the 1960s and 1970s, and the primary motive, while long gun younger cohorts are entering into owners mention hunting or target gun ownership at slower rates than shooting. Other findings support previous cohorts. When people do this basic division: handguns are acquire guns now, it is more likely

Guns in America to be motivated by self-defense than erally-licensed firearm dealer, a fig- in the past. The mix of new guns ure that is higher than previously sold is now equally divided between thought. The remaining gun trans- handguns and long guns, whereas actions are of special concern be- Our 25 years ago there were twice as cause they are less subject to extraordinary many long guns sold (Cook 1993). regulatory scrutiny. collective The NSPOF findings suggest that about 5 percent of adults are plan- These and a number of other findings involvement ning to acquire a gun for self-pro- serve as a useful reference for those With - tection even though they do not own who wish to understand the current reflects and a gun now. roles played by guns in the day-to-day life of America's households, and to has a large The concern for self-protection ascertain some of the trends. The data helps account for the unsafe stor- influence on may also be useful in judging the im- age methods and the widespread pact of proposed changes in how gov- life-and-death practice of carrying guns in public. ernment regulates gun transactions and matters, Of particular interest is the fact that use. Our extraordinary collective in- those owners who have received including volvement with guns reflects and has formal firearms training are no less a large influence on life-and-death crime, likely than others to store their guns matters, including crime, suicide, and unsafely, unlocked and loaded. accidents. Data of the sort collected by suicide, and We find that 60 to 70 percent of all the NSPOF provide a basis for informed accidents. firearm acquisitions involve a fed- action in this area of public concern.

Police Foundation Guns in America Adams, Kenneth. 1996. Guns and gun control. In Americans' view of crime and justice, ed. Timothy J. Flanagan and Dennis R. Longmire. Newberry Park, CA: Sage. Annest, Joseph L., James A. Mercy, Delinda R. Gibson, and George W. Ryan. 1995. National estimates of nonfatal firearm-related injuries: Beyond the tip of the iceberg. Journal of the American Medical Association 273(22). Bea, Keith. 1994. Gun control. Congressional Research Service Issue Brief (Sep- tember 19). Bourndon, Karen, Donald Rae, William Narrow, Ronald Manderscheid, and Darrel Regier. 1994. National prevalence and treatment of mental and addictive disorders. In Mental Health, United States, 1994, 22-35. Wash- ington, DC: U.S. Department of Health and Human Services, Center for Mental Health Services. Brick, J. Michael, Joseph Waksberg, Dale Kulp, and Amy Starer. 1995. Bias in list-assisted telephone samples. Public Opinion Quarterly 59:21&35. Cantor, David. 1989. Substantive implications of longitudinal design features: The National Crime Survey as a case study. In Panel Sum/eys, ed. D. Kasprzyk et al. New York: John Wiley. Chilton Research Services. 1995. National Study of Private Ownership of Fire- arms in the United States: Methodology report. Radnor, PA: Chilton Research Services. Clotfelter, Charles T., and Philip J. Cook. 1989. Selling hope: State lotteries in America. Harvard University Press: Cambridge, MA. Colombotos, J. 1969. Personal versus telephone interviews: Effect on responses. Public Health Reports 84:773-82. Cook, Philip J. 1981. The "Saturday Night Special": An assessment of alternative definitions from a policy perspective. Journal of Criminal Law and Crimi- nology 72(4): 1735-45. -. 1985.The case of the missing victims: Gunshot woundings in the National Crime Survey. Journal of Quantitative Criminologyl(l):91-102. -. 1991. The technology of personal violence. In Crime and Justice: An Annual Review of Research 14:l-71. Chicago: University of Chicago Press. -. 1993. Notes on the availability and prevalence of firearms. American Journal of PreventiveMedicine 9(3).

Cook, Philip J., Stephanie Molliconi, and Thomas B. Cole. 1995. Regulating gun markets. neJournal of Criminal Law and Criminology 86(1):59-92.

Cooper, Sanford. 1964. Random sampling by telephone-an improved method. Journal of Marketing Research 1:45-8. Deming, W. Edwards. El9441 1978. On errors in surveys. In Sociological methods: A sourcebook,ed. Norman K. Denzin, 233-47. New York: McGraw-Hill. Freeman, Howard, Jill Kiecolt, William Nicholls, and Merrill Shanks. 1982. Telephone sampling bias in surveying disability. Public Opinion Quarterly 46:392-407. Frey, James H. 1989. Survey Research by Telephone.2nd ed. Newbury Park, CA: Sage Publications. Groves, R.M. and R.L. Kahn. 1979. Surveys by telephone:A national comparison with personal interviews. New York: Academic Press. Hemenway, David. 1995. Guns, public health, and public safety. In Guns and the Constitution: The myth of Second Amendmentprotection for firearms in America, ed. Dennis Henigan, E. Bruce Nicholson, and David Hemenway, 49-82. Northampton, MA: Alethia Press.

-. 1996. Survey research and self-defense gun use: An explanation of ex- treme overestimates. Harvard University working paper. Hemenway, David, and Deborah Azrael. 1995. Use of guns in self-defense: Results of a national telephone survey. Harvard University working paper. Hemenway, David, and Elizabeth Richardson. 1995. Who owns semi-automatic firearms? Harvard University working paper. Hemenway, David, Sara J. Solnick, and Deborah Azrael. 1995. Firearm training and storage. Journal of the American Medical Association 273(1):46-50.

Guns in America Hochstim, J.R. 1962. Comparison of three information-gathering strategies in a population study of sociomedical variables. Proceedings of the American Statistical Association, Social Statistics Section 154-9.

-. 1967. A critical comparison of three strategies of collecting data from households. Journal of the American Statistical Association 62:976-89. Jones, Edward. 1981. The District of Columbia's Firearms Control Regulations Act of 1975: The toughest handgun control law in the United States-or is it? Annals of the AAPSS 455:138-49. Jordan, L.A., A.C. Marcus, and L.G. Reeder. 1980. Response styles in telephone and household interviewing: A field experiment. Public Opinion Quarterly 44:210-22.

Kellermann, A.L., F. P. Rivara, J. Banton, D. Reay, and C.L. Flinger. 1991. Validat- ing survey responses to questions about gun ownership among owners of registered handguns. American Journal of Epidemiology 131 : 1080-4. Kleck, Gary. 1988. Crime control through the private use of armed force. Social Problems 35(1):1-21.

-. 1991. Point blank: Guns and violence in America. New York: Aldine De Gsuyter.

-. 1993. Bad data and the "Evil Empire": Interpreting poll data on gun control. Vzolence and Victims 8(4): 367-376. Kleck, Gary, and Mark Gertz. 1995. Armed resistance to crime: The prevalence and nature of self-defense with a gun. The Journal of Criminal Law and Criminology 86(1): 150-87. Kopel, David. 1993. Hold your fire: Gun control won't stop rising violence. Policy Review 58-65. Kviz, F. J. 1977. Toward a standard definition of response rate. Public Opinion Quarterly 41 :265-7. Leuthold, David, and Raymond Scheele. 1971. Patterns of bias in samples based on telephone directories. Public Opinion Quarterly 3 5:249-57. Lizotte, Alan, David Bordua, and Carolyn White. 1981. Firearms ownership for sport and protection: Two not so divergent models. American Sociological Review 46:499-503.

Police Foundation Mayer, Jane, and Doyle McManus. 1988. Landslide: The unmaking of the Presi- dent, 29841988. Boston: Houghton-Mifflin.

McDowall, David, and Brian Wiersema. 1994. The incidence of defensive firearm use by U.S. crime victims, 1987 through 1990. American Journal of Public Health 1982-4.

Nelson, David, Joyce Grant-Worley, Kenneth Powell, James Mercy, and Deborah Holtzman. 1996. Population estimates of firearm storage practices and firearm carrying in Oregon. Journal of the American Medical Association 275(22): 1744-8.

O'Higgins, Niall. 1994. YTS, employment and sample selection bias. Oxford Economic Papers 46(4):605-28.

Orwin, Robert, and Robert Boruch. 1982. RRT meets RRD: Statistical strategies for assuring response privacy in telephone surveys. Public Opinion Quar- terly 46560-71.

Palmer, Gladys. 1943. Factors in the variability of response in enumerative studies. Journal of the American Statistical Association 38: 143-52. Potthoff, Richard, Kenneth Manton, and Max Woodbury. 1993. Correcting for nonavailability bias in surveys by weighting based on number of callbacks. Journal of the American Statistical Association 88 (424):1197-207. Salmon, Charles, and John Spicer Nichols. 1983. The next-birthday method of respondent selection. Public Opinion Quarterly 47:270-6. Sims, Barbara. Forthcoming. The National Opinion Survey of Crime and Justice, 1995: Development and methods. In Americans view crime andjustice, ed. Timothy J. Flanagan and Dennis R. Longmire. Newberry Park, CA: Sage. Singer, Max. 1971. The vitality of mythical numbers. The Public Interest 23:3-9. Slansky, Paul. 1989. i%e clothes have no emperor: A chronicle of the American 80s. New York: Simon and Schuster. Smith, Tom W. 1980. The 75% solution: An analysis of the structure of attitudes on gun control, 1959-1977. The Journal of Criminal Law and Criminology 71(3):3OO-16.

-. 1985. An analysis of the accuracy of spousal reports. National Opinion Research Center: General Social Science Survey Methodological Report Num- ber 35. Chicago: University of Chicago.

Guns in America Smith, Tom W., and Robert J. Smith. 1995. Changes in firearms ownership among women, 1980-1994. TheJournal of Criminal Law and Criminology 86(1): 133-49.

Sudman, Seymour. 1966. Probability sampling with quotas. Journal of the Ameri- can Statistical Association 749-71.

-. 1976. Applied sampling. New York: Academic Press.

Sudman, Seymour, and Norman M. Bradburn. 1974. Response effects in surveys: A review and synthesis. Chicago: Aldine.

Troldahl, V., and R. Carter. 1964. Random selection of respondents within households in phone surveys. Journal of Marketing Research 1 :71-6.

U.S. Bureau of the Census. 1995. Statistical abstract of the United States: 1995. 115th ed. Washington, DC.

-. 1994. Monthly Vital Statistics Report (October) 42(13):4.

U.S. Department of Education. 1994. Working with NCES's NELS:88 Data Sets- Potential Problems and Answers. Paper prepared for the 1994 AERA Annual Meeting, New Orleans, LA.

U.S. Department of Justice. 1994. Criminal victimization in the United States, 1992: A national crime victimization survey report. Government Printing Office: Washington, DC. - 1995. Criminal victimization 1993. Bureau of Justice Statistics Bulletin. Government Printing Office: Washington, DC.

U.S. Department of Justice. Federal Bureau of Investigation. 1994. Uniform crime reports for the United States, 1994. Washington, DC: Government Printing Office.

Waksberg, Joseph. 1978. Sampling methods for random digit dialing. Journal of the American Statistical Association 73:40-6.

Weil, Douglas S., and David Hemenway. 1992. Loaded guns in the home: Analysis of a national random survey of gun owners. Journal of the American Medi- cal Association 267(22):3033-7.

Wolfgang, Marvin E. 1958. Patterns in criminal homicide. New York, NY: Wiley. Wright, James. 1979. Public opinion and gun control: A comparison of results from two national surveys. Arnherst: University of Massachusetts.

Police Foundation -. 1984. The ownership of firearms for reasons of self defense. In Firearms and violence: Issues ofpublicpolicy, ed. Donald Kates, 301-27. Cambridge, MA: Ballinger. -. 1988. Second thoughts about gun control. Be Public Interest 23-39.

-. 1995. Ten essential observations on guns in America. Society 63-8. Wright, James, and Peter Rossi. 1986. Armed and considered dangerous: A Sur- vey of felons and theirfirearms. New York: Aldine De Gruyter. Wright, James, Peter Rossi and Kathleen Daly. 1983. Under the gun: Weapons, crime, and violence in America. New York: Aldine Publishing Company. Wright, James, Joseph Sheley, and M. Dwayne Smith. 1992. Kids, guns, and killing fields. Society 85-9. Zawitz, Marianne W. 1995. Firearms, crime and criminal justice: Guns used in crime. Bureau of Jz~sticeStatistics Selected Findings (July).

Zimring, Franklin E. 1972. The medium is the message: Firearm calibre as a determinant of death from assault. Journal of Legal Studies 1:97-124.

-. 1991. Firearms, violence and public policy. Scientific American (Novem- ber) 48-54.

Guns in America About the Authors

Philip J. Cook is ITTlSanford Professor of Public Policy Studies at Duke Univer- sity. The "technology" of personal violence has been a major concern of his research for the past 20 years. He has studied and written extensively on two basic scientific issues that arise in any debate over gun control: What difference does the type of weapon make in (potentially) violent encounters, and what difference does gun availability make in the choices of violent criminals.

Jens Ludwig is Assistant Professor of Public Policy at Georgetown University. His primary interest is in evaluating the effects of government policy on crime, education, and poverty. His recent projects include re-examining the relationship between permissive gun carrying laws and homicide.